{
  "best_global_step": null,
  "best_metric": null,
  "best_model_checkpoint": null,
  "epoch": 3.0,
  "eval_steps": 100.0,
  "global_step": 21405,
  "is_hyper_param_search": false,
  "is_local_process_zero": true,
  "is_world_process_zero": true,
  "log_history": [
    {
      "epoch": 0.0001401541695865452,
      "grad_norm": Infinity,
      "learning_rate": 0.0,
      "loss": 8.4692,
      "step": 1
    },
    {
      "epoch": 0.0002803083391730904,
      "grad_norm": Infinity,
      "learning_rate": 0.0,
      "loss": 11.0279,
      "step": 2
    },
    {
      "epoch": 0.0004204625087596356,
      "grad_norm": 7.121354103088379,
      "learning_rate": 0.0,
      "loss": 8.7646,
      "step": 3
    },
    {
      "epoch": 0.0005606166783461808,
      "grad_norm": 7.236179828643799,
      "learning_rate": 6e-07,
      "loss": 8.7277,
      "step": 4
    },
    {
      "epoch": 0.000700770847932726,
      "grad_norm": 6.580241680145264,
      "learning_rate": 1.2e-06,
      "loss": 7.9166,
      "step": 5
    },
    {
      "epoch": 0.0008409250175192712,
      "grad_norm": 6.403110027313232,
      "learning_rate": 1.8e-06,
      "loss": 7.7798,
      "step": 6
    },
    {
      "epoch": 0.0009810791871058165,
      "grad_norm": 7.469630241394043,
      "learning_rate": 2.4e-06,
      "loss": 8.9914,
      "step": 7
    },
    {
      "epoch": 0.0011212333566923615,
      "grad_norm": 7.176066875457764,
      "learning_rate": 2.9999999999999997e-06,
      "loss": 8.8701,
      "step": 8
    },
    {
      "epoch": 0.0012613875262789068,
      "grad_norm": 7.3517656326293945,
      "learning_rate": 3.6e-06,
      "loss": 8.9468,
      "step": 9
    },
    {
      "epoch": 0.001401541695865452,
      "grad_norm": 7.987253189086914,
      "learning_rate": 4.2e-06,
      "loss": 9.5461,
      "step": 10
    },
    {
      "epoch": 0.0015416958654519972,
      "grad_norm": 7.5793657302856445,
      "learning_rate": 4.8e-06,
      "loss": 9.1411,
      "step": 11
    },
    {
      "epoch": 0.0016818500350385423,
      "grad_norm": 7.367099761962891,
      "learning_rate": 5.399999999999999e-06,
      "loss": 8.8992,
      "step": 12
    },
    {
      "epoch": 0.0018220042046250876,
      "grad_norm": 6.14939546585083,
      "learning_rate": 5.999999999999999e-06,
      "loss": 7.4343,
      "step": 13
    },
    {
      "epoch": 0.001962158374211633,
      "grad_norm": 7.571694374084473,
      "learning_rate": 6.599999999999999e-06,
      "loss": 9.3066,
      "step": 14
    },
    {
      "epoch": 0.002102312543798178,
      "grad_norm": 6.577233791351318,
      "learning_rate": 7.2e-06,
      "loss": 7.9043,
      "step": 15
    },
    {
      "epoch": 0.002242466713384723,
      "grad_norm": 8.27171516418457,
      "learning_rate": 7.799999999999998e-06,
      "loss": 10.096,
      "step": 16
    },
    {
      "epoch": 0.0023826208829712684,
      "grad_norm": 6.9634599685668945,
      "learning_rate": 8.4e-06,
      "loss": 8.2555,
      "step": 17
    },
    {
      "epoch": 0.0025227750525578137,
      "grad_norm": 6.922667026519775,
      "learning_rate": 8.999999999999999e-06,
      "loss": 8.4747,
      "step": 18
    },
    {
      "epoch": 0.002662929222144359,
      "grad_norm": 7.340466022491455,
      "learning_rate": 9.6e-06,
      "loss": 8.8149,
      "step": 19
    },
    {
      "epoch": 0.002803083391730904,
      "grad_norm": 6.512653350830078,
      "learning_rate": 1.02e-05,
      "loss": 7.8311,
      "step": 20
    },
    {
      "epoch": 0.002943237561317449,
      "grad_norm": 8.32393741607666,
      "learning_rate": 1.0799999999999998e-05,
      "loss": 9.9733,
      "step": 21
    },
    {
      "epoch": 0.0030833917309039944,
      "grad_norm": 7.370021820068359,
      "learning_rate": 1.14e-05,
      "loss": 9.0241,
      "step": 22
    },
    {
      "epoch": 0.0032235459004905397,
      "grad_norm": 7.775641441345215,
      "learning_rate": 1.1999999999999999e-05,
      "loss": 9.5082,
      "step": 23
    },
    {
      "epoch": 0.0033637000700770846,
      "grad_norm": 7.187759876251221,
      "learning_rate": 1.26e-05,
      "loss": 8.6962,
      "step": 24
    },
    {
      "epoch": 0.00350385423966363,
      "grad_norm": 9.324918746948242,
      "learning_rate": 1.3199999999999997e-05,
      "loss": 11.2191,
      "step": 25
    },
    {
      "epoch": 0.003644008409250175,
      "grad_norm": 7.992179870605469,
      "learning_rate": 1.3799999999999998e-05,
      "loss": 9.4027,
      "step": 26
    },
    {
      "epoch": 0.0037841625788367205,
      "grad_norm": 8.400440216064453,
      "learning_rate": 1.44e-05,
      "loss": 10.1472,
      "step": 27
    },
    {
      "epoch": 0.003924316748423266,
      "grad_norm": 7.050783634185791,
      "learning_rate": 1.4999999999999999e-05,
      "loss": 8.353,
      "step": 28
    },
    {
      "epoch": 0.004064470918009811,
      "grad_norm": 7.598066806793213,
      "learning_rate": 1.5599999999999996e-05,
      "loss": 9.0813,
      "step": 29
    },
    {
      "epoch": 0.004204625087596356,
      "grad_norm": 7.8444600105285645,
      "learning_rate": 1.6199999999999997e-05,
      "loss": 9.2958,
      "step": 30
    },
    {
      "epoch": 0.004344779257182901,
      "grad_norm": 7.099885940551758,
      "learning_rate": 1.68e-05,
      "loss": 8.4595,
      "step": 31
    },
    {
      "epoch": 0.004484933426769446,
      "grad_norm": 7.714644908905029,
      "learning_rate": 1.74e-05,
      "loss": 9.2685,
      "step": 32
    },
    {
      "epoch": 0.0046250875963559914,
      "grad_norm": 7.360316276550293,
      "learning_rate": 1.7999999999999997e-05,
      "loss": 8.7488,
      "step": 33
    },
    {
      "epoch": 0.004765241765942537,
      "grad_norm": 7.624013900756836,
      "learning_rate": 1.8599999999999998e-05,
      "loss": 9.0688,
      "step": 34
    },
    {
      "epoch": 0.004905395935529082,
      "grad_norm": 8.676628112792969,
      "learning_rate": 1.92e-05,
      "loss": 10.4329,
      "step": 35
    },
    {
      "epoch": 0.005045550105115627,
      "grad_norm": 7.6324005126953125,
      "learning_rate": 1.98e-05,
      "loss": 8.8521,
      "step": 36
    },
    {
      "epoch": 0.005185704274702173,
      "grad_norm": 7.269668102264404,
      "learning_rate": 2.04e-05,
      "loss": 8.5564,
      "step": 37
    },
    {
      "epoch": 0.005325858444288718,
      "grad_norm": 7.439398288726807,
      "learning_rate": 2.1e-05,
      "loss": 8.7445,
      "step": 38
    },
    {
      "epoch": 0.005466012613875262,
      "grad_norm": 7.443110942840576,
      "learning_rate": 2.1599999999999996e-05,
      "loss": 8.5582,
      "step": 39
    },
    {
      "epoch": 0.005606166783461808,
      "grad_norm": 8.377077102661133,
      "learning_rate": 2.2199999999999998e-05,
      "loss": 9.6897,
      "step": 40
    },
    {
      "epoch": 0.005746320953048353,
      "grad_norm": 7.658986568450928,
      "learning_rate": 2.28e-05,
      "loss": 8.7662,
      "step": 41
    },
    {
      "epoch": 0.005886475122634898,
      "grad_norm": 11.107864379882812,
      "learning_rate": 2.34e-05,
      "loss": 11.7328,
      "step": 42
    },
    {
      "epoch": 0.006026629292221444,
      "grad_norm": 9.172978401184082,
      "learning_rate": 2.3999999999999997e-05,
      "loss": 10.4654,
      "step": 43
    },
    {
      "epoch": 0.006166783461807989,
      "grad_norm": 7.016298294067383,
      "learning_rate": 2.4599999999999998e-05,
      "loss": 7.8736,
      "step": 44
    },
    {
      "epoch": 0.006306937631394534,
      "grad_norm": 9.631633758544922,
      "learning_rate": 2.52e-05,
      "loss": 10.5877,
      "step": 45
    },
    {
      "epoch": 0.0064470918009810795,
      "grad_norm": 7.38787317276001,
      "learning_rate": 2.5799999999999997e-05,
      "loss": 8.5266,
      "step": 46
    },
    {
      "epoch": 0.006587245970567625,
      "grad_norm": Infinity,
      "learning_rate": 2.6399999999999995e-05,
      "loss": 16.7878,
      "step": 47
    },
    {
      "epoch": 0.006727400140154169,
      "grad_norm": 7.5063323974609375,
      "learning_rate": 2.6399999999999995e-05,
      "loss": 8.6264,
      "step": 48
    },
    {
      "epoch": 0.0068675543097407145,
      "grad_norm": 6.67501974105835,
      "learning_rate": 2.6999999999999996e-05,
      "loss": 7.7156,
      "step": 49
    },
    {
      "epoch": 0.00700770847932726,
      "grad_norm": 8.071187019348145,
      "learning_rate": 2.7599999999999997e-05,
      "loss": 10.1454,
      "step": 50
    },
    {
      "epoch": 0.007147862648913805,
      "grad_norm": 7.581518173217773,
      "learning_rate": 2.8199999999999998e-05,
      "loss": 8.4984,
      "step": 51
    },
    {
      "epoch": 0.00728801681850035,
      "grad_norm": 6.454829692840576,
      "learning_rate": 2.88e-05,
      "loss": 7.3631,
      "step": 52
    },
    {
      "epoch": 0.007428170988086896,
      "grad_norm": 8.451108932495117,
      "learning_rate": 2.94e-05,
      "loss": 9.2767,
      "step": 53
    },
    {
      "epoch": 0.007568325157673441,
      "grad_norm": 7.960541248321533,
      "learning_rate": 2.9999999999999997e-05,
      "loss": 8.7883,
      "step": 54
    },
    {
      "epoch": 0.007708479327259986,
      "grad_norm": 8.233039855957031,
      "learning_rate": 3.06e-05,
      "loss": 9.1653,
      "step": 55
    },
    {
      "epoch": 0.007848633496846532,
      "grad_norm": 10.144512176513672,
      "learning_rate": 3.119999999999999e-05,
      "loss": 10.5917,
      "step": 56
    },
    {
      "epoch": 0.007988787666433076,
      "grad_norm": 8.107776641845703,
      "learning_rate": 3.1799999999999994e-05,
      "loss": 8.7712,
      "step": 57
    },
    {
      "epoch": 0.008128941836019622,
      "grad_norm": 11.103604316711426,
      "learning_rate": 3.2399999999999995e-05,
      "loss": 11.2103,
      "step": 58
    },
    {
      "epoch": 0.008269096005606167,
      "grad_norm": 8.37151050567627,
      "learning_rate": 3.2999999999999996e-05,
      "loss": 9.1048,
      "step": 59
    },
    {
      "epoch": 0.008409250175192713,
      "grad_norm": 8.462974548339844,
      "learning_rate": 3.36e-05,
      "loss": 9.0559,
      "step": 60
    },
    {
      "epoch": 0.008549404344779257,
      "grad_norm": 7.523645877838135,
      "learning_rate": 3.42e-05,
      "loss": 8.2492,
      "step": 61
    },
    {
      "epoch": 0.008689558514365802,
      "grad_norm": 7.640683650970459,
      "learning_rate": 3.48e-05,
      "loss": 8.2912,
      "step": 62
    },
    {
      "epoch": 0.008829712683952348,
      "grad_norm": 7.68369197845459,
      "learning_rate": 3.539999999999999e-05,
      "loss": 8.1744,
      "step": 63
    },
    {
      "epoch": 0.008969866853538892,
      "grad_norm": 9.849550247192383,
      "learning_rate": 3.5999999999999994e-05,
      "loss": 10.102,
      "step": 64
    },
    {
      "epoch": 0.009110021023125438,
      "grad_norm": 7.607925891876221,
      "learning_rate": 3.6599999999999995e-05,
      "loss": 8.1789,
      "step": 65
    },
    {
      "epoch": 0.009250175192711983,
      "grad_norm": 7.810117244720459,
      "learning_rate": 3.7199999999999996e-05,
      "loss": 8.4259,
      "step": 66
    },
    {
      "epoch": 0.009390329362298529,
      "grad_norm": 7.368885517120361,
      "learning_rate": 3.78e-05,
      "loss": 7.7666,
      "step": 67
    },
    {
      "epoch": 0.009530483531885073,
      "grad_norm": 8.448766708374023,
      "learning_rate": 3.84e-05,
      "loss": 8.9383,
      "step": 68
    },
    {
      "epoch": 0.009670637701471618,
      "grad_norm": 9.512053489685059,
      "learning_rate": 3.9e-05,
      "loss": 9.8389,
      "step": 69
    },
    {
      "epoch": 0.009810791871058164,
      "grad_norm": 9.501230239868164,
      "learning_rate": 3.96e-05,
      "loss": 9.6005,
      "step": 70
    },
    {
      "epoch": 0.009950946040644709,
      "grad_norm": 10.725296020507812,
      "learning_rate": 4.02e-05,
      "loss": 9.7433,
      "step": 71
    },
    {
      "epoch": 0.010091100210231255,
      "grad_norm": 7.951533317565918,
      "learning_rate": 4.08e-05,
      "loss": 8.1849,
      "step": 72
    },
    {
      "epoch": 0.010231254379817799,
      "grad_norm": 6.88252592086792,
      "learning_rate": 4.14e-05,
      "loss": 7.277,
      "step": 73
    },
    {
      "epoch": 0.010371408549404345,
      "grad_norm": 8.40587329864502,
      "learning_rate": 4.2e-05,
      "loss": 8.5527,
      "step": 74
    },
    {
      "epoch": 0.01051156271899089,
      "grad_norm": 9.101225852966309,
      "learning_rate": 4.259999999999999e-05,
      "loss": 9.049,
      "step": 75
    },
    {
      "epoch": 0.010651716888577436,
      "grad_norm": 9.011225700378418,
      "learning_rate": 4.319999999999999e-05,
      "loss": 8.7113,
      "step": 76
    },
    {
      "epoch": 0.01079187105816398,
      "grad_norm": 8.225573539733887,
      "learning_rate": 4.3799999999999994e-05,
      "loss": 8.441,
      "step": 77
    },
    {
      "epoch": 0.010932025227750525,
      "grad_norm": 8.690326690673828,
      "learning_rate": 4.4399999999999995e-05,
      "loss": 8.4745,
      "step": 78
    },
    {
      "epoch": 0.011072179397337071,
      "grad_norm": 9.590045928955078,
      "learning_rate": 4.4999999999999996e-05,
      "loss": 9.2402,
      "step": 79
    },
    {
      "epoch": 0.011212333566923615,
      "grad_norm": 10.65174388885498,
      "learning_rate": 4.56e-05,
      "loss": 9.9752,
      "step": 80
    },
    {
      "epoch": 0.011352487736510162,
      "grad_norm": 9.207296371459961,
      "learning_rate": 4.62e-05,
      "loss": 8.864,
      "step": 81
    },
    {
      "epoch": 0.011492641906096706,
      "grad_norm": 7.806180477142334,
      "learning_rate": 4.68e-05,
      "loss": 7.7845,
      "step": 82
    },
    {
      "epoch": 0.011632796075683252,
      "grad_norm": 8.521522521972656,
      "learning_rate": 4.7399999999999993e-05,
      "loss": 8.3196,
      "step": 83
    },
    {
      "epoch": 0.011772950245269797,
      "grad_norm": 8.61214542388916,
      "learning_rate": 4.7999999999999994e-05,
      "loss": 8.3665,
      "step": 84
    },
    {
      "epoch": 0.011913104414856343,
      "grad_norm": 10.06464958190918,
      "learning_rate": 4.8599999999999995e-05,
      "loss": 8.9888,
      "step": 85
    },
    {
      "epoch": 0.012053258584442887,
      "grad_norm": 8.386164665222168,
      "learning_rate": 4.9199999999999997e-05,
      "loss": 8.0003,
      "step": 86
    },
    {
      "epoch": 0.012193412754029432,
      "grad_norm": 8.232111930847168,
      "learning_rate": 4.98e-05,
      "loss": 7.7958,
      "step": 87
    },
    {
      "epoch": 0.012333566923615978,
      "grad_norm": 8.352997779846191,
      "learning_rate": 5.04e-05,
      "loss": 8.1255,
      "step": 88
    },
    {
      "epoch": 0.012473721093202522,
      "grad_norm": 10.184182167053223,
      "learning_rate": 5.1e-05,
      "loss": 9.0337,
      "step": 89
    },
    {
      "epoch": 0.012613875262789068,
      "grad_norm": 8.01484203338623,
      "learning_rate": 5.1599999999999994e-05,
      "loss": 7.4799,
      "step": 90
    },
    {
      "epoch": 0.012754029432375613,
      "grad_norm": 8.27849292755127,
      "learning_rate": 5.2199999999999995e-05,
      "loss": 7.7335,
      "step": 91
    },
    {
      "epoch": 0.012894183601962159,
      "grad_norm": 9.302130699157715,
      "learning_rate": 5.279999999999999e-05,
      "loss": 8.2463,
      "step": 92
    },
    {
      "epoch": 0.013034337771548703,
      "grad_norm": 7.507938385009766,
      "learning_rate": 5.339999999999999e-05,
      "loss": 6.9302,
      "step": 93
    },
    {
      "epoch": 0.01317449194113525,
      "grad_norm": 8.125649452209473,
      "learning_rate": 5.399999999999999e-05,
      "loss": 7.3503,
      "step": 94
    },
    {
      "epoch": 0.013314646110721794,
      "grad_norm": 8.156755447387695,
      "learning_rate": 5.459999999999999e-05,
      "loss": 7.396,
      "step": 95
    },
    {
      "epoch": 0.013454800280308338,
      "grad_norm": 8.820056915283203,
      "learning_rate": 5.519999999999999e-05,
      "loss": 7.7505,
      "step": 96
    },
    {
      "epoch": 0.013594954449894885,
      "grad_norm": 7.855623722076416,
      "learning_rate": 5.5799999999999994e-05,
      "loss": 7.1811,
      "step": 97
    },
    {
      "epoch": 0.013735108619481429,
      "grad_norm": 9.308487892150879,
      "learning_rate": 5.6399999999999995e-05,
      "loss": 7.7698,
      "step": 98
    },
    {
      "epoch": 0.013875262789067975,
      "grad_norm": 8.365824699401855,
      "learning_rate": 5.6999999999999996e-05,
      "loss": 7.3467,
      "step": 99
    },
    {
      "epoch": 0.01401541695865452,
      "grad_norm": 9.018302917480469,
      "learning_rate": 5.76e-05,
      "loss": 7.5294,
      "step": 100
    },
    {
      "epoch": 0.014155571128241066,
      "grad_norm": 8.283246994018555,
      "learning_rate": 5.82e-05,
      "loss": 7.1433,
      "step": 101
    },
    {
      "epoch": 0.01429572529782761,
      "grad_norm": 10.114821434020996,
      "learning_rate": 5.88e-05,
      "loss": 8.0797,
      "step": 102
    },
    {
      "epoch": 0.014435879467414156,
      "grad_norm": 10.132929801940918,
      "learning_rate": 5.94e-05,
      "loss": 8.2213,
      "step": 103
    },
    {
      "epoch": 0.0145760336370007,
      "grad_norm": 9.059672355651855,
      "learning_rate": 5.9999999999999995e-05,
      "loss": 7.5188,
      "step": 104
    },
    {
      "epoch": 0.014716187806587245,
      "grad_norm": 11.894686698913574,
      "learning_rate": 6.0599999999999996e-05,
      "loss": 8.9093,
      "step": 105
    },
    {
      "epoch": 0.014856341976173791,
      "grad_norm": 8.074777603149414,
      "learning_rate": 6.12e-05,
      "loss": 6.8733,
      "step": 106
    },
    {
      "epoch": 0.014996496145760336,
      "grad_norm": 10.521833419799805,
      "learning_rate": 6.18e-05,
      "loss": 7.8667,
      "step": 107
    },
    {
      "epoch": 0.015136650315346882,
      "grad_norm": 10.695642471313477,
      "learning_rate": 6.239999999999999e-05,
      "loss": 8.2731,
      "step": 108
    },
    {
      "epoch": 0.015276804484933426,
      "grad_norm": 9.828344345092773,
      "learning_rate": 6.299999999999999e-05,
      "loss": 7.5958,
      "step": 109
    },
    {
      "epoch": 0.015416958654519973,
      "grad_norm": 9.938995361328125,
      "learning_rate": 6.359999999999999e-05,
      "loss": 7.6425,
      "step": 110
    },
    {
      "epoch": 0.015557112824106517,
      "grad_norm": 8.546896934509277,
      "learning_rate": 6.419999999999999e-05,
      "loss": 6.9451,
      "step": 111
    },
    {
      "epoch": 0.015697266993693063,
      "grad_norm": 9.221542358398438,
      "learning_rate": 6.479999999999999e-05,
      "loss": 7.2392,
      "step": 112
    },
    {
      "epoch": 0.015837421163279606,
      "grad_norm": 9.153562545776367,
      "learning_rate": 6.539999999999999e-05,
      "loss": 7.0752,
      "step": 113
    },
    {
      "epoch": 0.015977575332866152,
      "grad_norm": 12.386898040771484,
      "learning_rate": 6.599999999999999e-05,
      "loss": 6.8654,
      "step": 114
    },
    {
      "epoch": 0.0161177295024527,
      "grad_norm": 8.489028930664062,
      "learning_rate": 6.659999999999999e-05,
      "loss": 6.547,
      "step": 115
    },
    {
      "epoch": 0.016257883672039244,
      "grad_norm": 8.682656288146973,
      "learning_rate": 6.72e-05,
      "loss": 6.7826,
      "step": 116
    },
    {
      "epoch": 0.016398037841625787,
      "grad_norm": 7.733676433563232,
      "learning_rate": 6.78e-05,
      "loss": 6.0851,
      "step": 117
    },
    {
      "epoch": 0.016538192011212333,
      "grad_norm": 9.6995849609375,
      "learning_rate": 6.84e-05,
      "loss": 7.3635,
      "step": 118
    },
    {
      "epoch": 0.01667834618079888,
      "grad_norm": 12.979042053222656,
      "learning_rate": 6.9e-05,
      "loss": 6.9633,
      "step": 119
    },
    {
      "epoch": 0.016818500350385426,
      "grad_norm": NaN,
      "learning_rate": 6.96e-05,
      "loss": 7.0443,
      "step": 120
    },
    {
      "epoch": 0.01695865451997197,
      "grad_norm": 13.420258522033691,
      "learning_rate": 6.96e-05,
      "loss": 5.9981,
      "step": 121
    },
    {
      "epoch": 0.017098808689558515,
      "grad_norm": 12.565690994262695,
      "learning_rate": 7.02e-05,
      "loss": 4.2515,
      "step": 122
    },
    {
      "epoch": 0.01723896285914506,
      "grad_norm": 19.81829261779785,
      "learning_rate": 7.079999999999999e-05,
      "loss": 4.9932,
      "step": 123
    },
    {
      "epoch": 0.017379117028731603,
      "grad_norm": 12.81844711303711,
      "learning_rate": 7.139999999999999e-05,
      "loss": 4.2595,
      "step": 124
    },
    {
      "epoch": 0.01751927119831815,
      "grad_norm": 11.58743667602539,
      "learning_rate": 7.199999999999999e-05,
      "loss": 3.9826,
      "step": 125
    },
    {
      "epoch": 0.017659425367904696,
      "grad_norm": 13.66759204864502,
      "learning_rate": 7.259999999999999e-05,
      "loss": 4.0842,
      "step": 126
    },
    {
      "epoch": 0.017799579537491242,
      "grad_norm": 10.878744125366211,
      "learning_rate": 7.319999999999999e-05,
      "loss": 3.7609,
      "step": 127
    },
    {
      "epoch": 0.017939733707077785,
      "grad_norm": 10.014312744140625,
      "learning_rate": 7.379999999999999e-05,
      "loss": 3.6731,
      "step": 128
    },
    {
      "epoch": 0.01807988787666433,
      "grad_norm": 8.86495590209961,
      "learning_rate": 7.439999999999999e-05,
      "loss": 3.5619,
      "step": 129
    },
    {
      "epoch": 0.018220042046250877,
      "grad_norm": 10.835160255432129,
      "learning_rate": 7.5e-05,
      "loss": 3.6498,
      "step": 130
    },
    {
      "epoch": 0.01836019621583742,
      "grad_norm": 8.720636367797852,
      "learning_rate": 7.56e-05,
      "loss": 3.5168,
      "step": 131
    },
    {
      "epoch": 0.018500350385423966,
      "grad_norm": 7.215508937835693,
      "learning_rate": 7.62e-05,
      "loss": 3.394,
      "step": 132
    },
    {
      "epoch": 0.018640504555010512,
      "grad_norm": 7.7803168296813965,
      "learning_rate": 7.68e-05,
      "loss": 3.4178,
      "step": 133
    },
    {
      "epoch": 0.018780658724597058,
      "grad_norm": 6.6492085456848145,
      "learning_rate": 7.74e-05,
      "loss": 3.3398,
      "step": 134
    },
    {
      "epoch": 0.0189208128941836,
      "grad_norm": 9.923211097717285,
      "learning_rate": 7.8e-05,
      "loss": 3.5073,
      "step": 135
    },
    {
      "epoch": 0.019060967063770147,
      "grad_norm": 8.024308204650879,
      "learning_rate": 7.86e-05,
      "loss": 3.3961,
      "step": 136
    },
    {
      "epoch": 0.019201121233356693,
      "grad_norm": 7.969802379608154,
      "learning_rate": 7.92e-05,
      "loss": 3.368,
      "step": 137
    },
    {
      "epoch": 0.019341275402943236,
      "grad_norm": 5.528815746307373,
      "learning_rate": 7.98e-05,
      "loss": 3.2395,
      "step": 138
    },
    {
      "epoch": 0.019481429572529782,
      "grad_norm": 6.242894172668457,
      "learning_rate": 8.04e-05,
      "loss": 3.2522,
      "step": 139
    },
    {
      "epoch": 0.019621583742116328,
      "grad_norm": 6.63530158996582,
      "learning_rate": 8.1e-05,
      "loss": 3.2979,
      "step": 140
    },
    {
      "epoch": 0.019761737911702874,
      "grad_norm": 4.684387683868408,
      "learning_rate": 8.16e-05,
      "loss": 3.1842,
      "step": 141
    },
    {
      "epoch": 0.019901892081289417,
      "grad_norm": 4.867077350616455,
      "learning_rate": 8.22e-05,
      "loss": 3.1842,
      "step": 142
    },
    {
      "epoch": 0.020042046250875963,
      "grad_norm": 4.582318305969238,
      "learning_rate": 8.28e-05,
      "loss": 3.1722,
      "step": 143
    },
    {
      "epoch": 0.02018220042046251,
      "grad_norm": 5.831948757171631,
      "learning_rate": 8.34e-05,
      "loss": 3.2393,
      "step": 144
    },
    {
      "epoch": 0.020322354590049056,
      "grad_norm": 2.2314395904541016,
      "learning_rate": 8.4e-05,
      "loss": 3.0924,
      "step": 145
    },
    {
      "epoch": 0.020462508759635598,
      "grad_norm": 4.964643955230713,
      "learning_rate": 8.459999999999998e-05,
      "loss": 3.1685,
      "step": 146
    },
    {
      "epoch": 0.020602662929222144,
      "grad_norm": 4.167355060577393,
      "learning_rate": 8.519999999999998e-05,
      "loss": 3.1267,
      "step": 147
    },
    {
      "epoch": 0.02074281709880869,
      "grad_norm": 3.614588975906372,
      "learning_rate": 8.579999999999998e-05,
      "loss": 3.1363,
      "step": 148
    },
    {
      "epoch": 0.020882971268395233,
      "grad_norm": 2.30169677734375,
      "learning_rate": 8.639999999999999e-05,
      "loss": 3.0837,
      "step": 149
    },
    {
      "epoch": 0.02102312543798178,
      "grad_norm": 4.596937656402588,
      "learning_rate": 8.699999999999999e-05,
      "loss": 3.1314,
      "step": 150
    },
    {
      "epoch": 0.021163279607568326,
      "grad_norm": 3.200518846511841,
      "learning_rate": 8.759999999999999e-05,
      "loss": 3.0718,
      "step": 151
    },
    {
      "epoch": 0.021303433777154872,
      "grad_norm": 2.7224369049072266,
      "learning_rate": 8.819999999999999e-05,
      "loss": 3.0051,
      "step": 152
    },
    {
      "epoch": 0.021443587946741414,
      "grad_norm": 2.585502862930298,
      "learning_rate": 8.879999999999999e-05,
      "loss": 3.0545,
      "step": 153
    },
    {
      "epoch": 0.02158374211632796,
      "grad_norm": 4.326929092407227,
      "learning_rate": 8.939999999999999e-05,
      "loss": 3.0337,
      "step": 154
    },
    {
      "epoch": 0.021723896285914507,
      "grad_norm": 2.0838465690612793,
      "learning_rate": 8.999999999999999e-05,
      "loss": 2.9955,
      "step": 155
    },
    {
      "epoch": 0.02186405045550105,
      "grad_norm": 1.980881929397583,
      "learning_rate": 9.059999999999999e-05,
      "loss": 2.9981,
      "step": 156
    },
    {
      "epoch": 0.022004204625087596,
      "grad_norm": 2.443854808807373,
      "learning_rate": 9.12e-05,
      "loss": 2.9851,
      "step": 157
    },
    {
      "epoch": 0.022144358794674142,
      "grad_norm": 6.009809970855713,
      "learning_rate": 9.18e-05,
      "loss": 3.0649,
      "step": 158
    },
    {
      "epoch": 0.022284512964260688,
      "grad_norm": 1.1571614742279053,
      "learning_rate": 9.24e-05,
      "loss": 2.969,
      "step": 159
    },
    {
      "epoch": 0.02242466713384723,
      "grad_norm": 1.4274930953979492,
      "learning_rate": 9.3e-05,
      "loss": 2.9455,
      "step": 160
    },
    {
      "epoch": 0.022564821303433777,
      "grad_norm": 2.8754162788391113,
      "learning_rate": 9.36e-05,
      "loss": 2.9669,
      "step": 161
    },
    {
      "epoch": 0.022704975473020323,
      "grad_norm": 1.7089864015579224,
      "learning_rate": 9.419999999999999e-05,
      "loss": 2.9549,
      "step": 162
    },
    {
      "epoch": 0.02284512964260687,
      "grad_norm": 2.965763807296753,
      "learning_rate": 9.479999999999999e-05,
      "loss": 2.9707,
      "step": 163
    },
    {
      "epoch": 0.022985283812193412,
      "grad_norm": 1.996239185333252,
      "learning_rate": 9.539999999999999e-05,
      "loss": 2.9744,
      "step": 164
    },
    {
      "epoch": 0.023125437981779958,
      "grad_norm": 0.8861143589019775,
      "learning_rate": 9.599999999999999e-05,
      "loss": 2.921,
      "step": 165
    },
    {
      "epoch": 0.023265592151366504,
      "grad_norm": 1.0977675914764404,
      "learning_rate": 9.659999999999999e-05,
      "loss": 2.9073,
      "step": 166
    },
    {
      "epoch": 0.023405746320953047,
      "grad_norm": 0.6487427949905396,
      "learning_rate": 9.719999999999999e-05,
      "loss": 2.9416,
      "step": 167
    },
    {
      "epoch": 0.023545900490539593,
      "grad_norm": 1.6304970979690552,
      "learning_rate": 9.779999999999999e-05,
      "loss": 2.9264,
      "step": 168
    },
    {
      "epoch": 0.02368605466012614,
      "grad_norm": 1.029008150100708,
      "learning_rate": 9.839999999999999e-05,
      "loss": 2.9081,
      "step": 169
    },
    {
      "epoch": 0.023826208829712685,
      "grad_norm": 1.8654495477676392,
      "learning_rate": 9.9e-05,
      "loss": 2.9003,
      "step": 170
    },
    {
      "epoch": 0.023966362999299228,
      "grad_norm": 1.41410493850708,
      "learning_rate": 9.96e-05,
      "loss": 2.91,
      "step": 171
    },
    {
      "epoch": 0.024106517168885774,
      "grad_norm": 1.3176707029342651,
      "learning_rate": 0.0001002,
      "loss": 2.9348,
      "step": 172
    },
    {
      "epoch": 0.02424667133847232,
      "grad_norm": 1.9054174423217773,
      "learning_rate": 0.0001008,
      "loss": 2.9161,
      "step": 173
    },
    {
      "epoch": 0.024386825508058863,
      "grad_norm": 1.1568753719329834,
      "learning_rate": 0.0001014,
      "loss": 2.9115,
      "step": 174
    },
    {
      "epoch": 0.02452697967764541,
      "grad_norm": 3.0676801204681396,
      "learning_rate": 0.000102,
      "loss": 2.9599,
      "step": 175
    },
    {
      "epoch": 0.024667133847231956,
      "grad_norm": 0.8326209783554077,
      "learning_rate": 0.0001026,
      "loss": 2.8861,
      "step": 176
    },
    {
      "epoch": 0.0248072880168185,
      "grad_norm": 0.7381007075309753,
      "learning_rate": 0.00010319999999999999,
      "loss": 2.8958,
      "step": 177
    },
    {
      "epoch": 0.024947442186405044,
      "grad_norm": 0.5153982639312744,
      "learning_rate": 0.00010379999999999999,
      "loss": 2.9183,
      "step": 178
    },
    {
      "epoch": 0.02508759635599159,
      "grad_norm": 1.976210117340088,
      "learning_rate": 0.00010439999999999999,
      "loss": 2.8974,
      "step": 179
    },
    {
      "epoch": 0.025227750525578137,
      "grad_norm": 1.870713233947754,
      "learning_rate": 0.00010499999999999999,
      "loss": 2.9263,
      "step": 180
    },
    {
      "epoch": 0.02536790469516468,
      "grad_norm": 2.3896396160125732,
      "learning_rate": 0.00010559999999999998,
      "loss": 2.8912,
      "step": 181
    },
    {
      "epoch": 0.025508058864751226,
      "grad_norm": 1.2706975936889648,
      "learning_rate": 0.00010619999999999998,
      "loss": 2.9116,
      "step": 182
    },
    {
      "epoch": 0.025648213034337772,
      "grad_norm": 0.9402898550033569,
      "learning_rate": 0.00010679999999999998,
      "loss": 2.9286,
      "step": 183
    },
    {
      "epoch": 0.025788367203924318,
      "grad_norm": 0.7575398087501526,
      "learning_rate": 0.00010739999999999998,
      "loss": 2.9138,
      "step": 184
    },
    {
      "epoch": 0.02592852137351086,
      "grad_norm": 3.0395188331604004,
      "learning_rate": 0.00010799999999999998,
      "loss": 3.023,
      "step": 185
    },
    {
      "epoch": 0.026068675543097407,
      "grad_norm": 0.7392745018005371,
      "learning_rate": 0.00010859999999999998,
      "loss": 2.956,
      "step": 186
    },
    {
      "epoch": 0.026208829712683953,
      "grad_norm": 1.332564353942871,
      "learning_rate": 0.00010919999999999998,
      "loss": 2.9403,
      "step": 187
    },
    {
      "epoch": 0.0263489838822705,
      "grad_norm": 0.6688811182975769,
      "learning_rate": 0.00010979999999999999,
      "loss": 2.8767,
      "step": 188
    },
    {
      "epoch": 0.026489138051857042,
      "grad_norm": 0.882061243057251,
      "learning_rate": 0.00011039999999999999,
      "loss": 2.9085,
      "step": 189
    },
    {
      "epoch": 0.026629292221443588,
      "grad_norm": 0.7065404653549194,
      "learning_rate": 0.00011099999999999999,
      "loss": 2.9152,
      "step": 190
    },
    {
      "epoch": 0.026769446391030134,
      "grad_norm": 1.5446754693984985,
      "learning_rate": 0.00011159999999999999,
      "loss": 2.8788,
      "step": 191
    },
    {
      "epoch": 0.026909600560616677,
      "grad_norm": 0.5920804738998413,
      "learning_rate": 0.00011219999999999999,
      "loss": 2.9091,
      "step": 192
    },
    {
      "epoch": 0.027049754730203223,
      "grad_norm": 0.573976993560791,
      "learning_rate": 0.00011279999999999999,
      "loss": 2.9037,
      "step": 193
    },
    {
      "epoch": 0.02718990889978977,
      "grad_norm": 0.7928887009620667,
      "learning_rate": 0.00011339999999999999,
      "loss": 2.9208,
      "step": 194
    },
    {
      "epoch": 0.027330063069376315,
      "grad_norm": 1.1057881116867065,
      "learning_rate": 0.00011399999999999999,
      "loss": 2.855,
      "step": 195
    },
    {
      "epoch": 0.027470217238962858,
      "grad_norm": 1.963627815246582,
      "learning_rate": 0.0001146,
      "loss": 2.8812,
      "step": 196
    },
    {
      "epoch": 0.027610371408549404,
      "grad_norm": 1.3813062906265259,
      "learning_rate": 0.0001152,
      "loss": 2.8663,
      "step": 197
    },
    {
      "epoch": 0.02775052557813595,
      "grad_norm": 3.114259719848633,
      "learning_rate": 0.0001158,
      "loss": 2.9357,
      "step": 198
    },
    {
      "epoch": 0.027890679747722493,
      "grad_norm": 1.8745334148406982,
      "learning_rate": 0.0001164,
      "loss": 2.8679,
      "step": 199
    },
    {
      "epoch": 0.02803083391730904,
      "grad_norm": 4.405505180358887,
      "learning_rate": 0.000117,
      "loss": 3.0878,
      "step": 200
    },
    {
      "epoch": 0.028170988086895585,
      "grad_norm": 1.3865611553192139,
      "learning_rate": 0.0001176,
      "loss": 2.8596,
      "step": 201
    },
    {
      "epoch": 0.02831114225648213,
      "grad_norm": 0.9126668572425842,
      "learning_rate": 0.0001182,
      "loss": 2.9487,
      "step": 202
    },
    {
      "epoch": 0.028451296426068674,
      "grad_norm": 0.6223608255386353,
      "learning_rate": 0.0001188,
      "loss": 2.8767,
      "step": 203
    },
    {
      "epoch": 0.02859145059565522,
      "grad_norm": 2.065746545791626,
      "learning_rate": 0.0001194,
      "loss": 2.8937,
      "step": 204
    },
    {
      "epoch": 0.028731604765241767,
      "grad_norm": 1.4100656509399414,
      "learning_rate": 0.00011999999999999999,
      "loss": 2.9177,
      "step": 205
    },
    {
      "epoch": 0.028871758934828313,
      "grad_norm": 0.936177134513855,
      "learning_rate": 0.00012059999999999999,
      "loss": 2.8652,
      "step": 206
    },
    {
      "epoch": 0.029011913104414856,
      "grad_norm": 0.4871051609516144,
      "learning_rate": 0.00012119999999999999,
      "loss": 2.8553,
      "step": 207
    },
    {
      "epoch": 0.0291520672740014,
      "grad_norm": 3.69014573097229,
      "learning_rate": 0.00012179999999999999,
      "loss": 2.9089,
      "step": 208
    },
    {
      "epoch": 0.029292221443587948,
      "grad_norm": 0.4774470925331116,
      "learning_rate": 0.0001224,
      "loss": 2.8655,
      "step": 209
    },
    {
      "epoch": 0.02943237561317449,
      "grad_norm": 0.7280520796775818,
      "learning_rate": 0.00012299999999999998,
      "loss": 2.8594,
      "step": 210
    },
    {
      "epoch": 0.029572529782761037,
      "grad_norm": 1.935536503791809,
      "learning_rate": 0.0001236,
      "loss": 2.8728,
      "step": 211
    },
    {
      "epoch": 0.029712683952347583,
      "grad_norm": 0.8123719692230225,
      "learning_rate": 0.00012419999999999998,
      "loss": 2.9089,
      "step": 212
    },
    {
      "epoch": 0.02985283812193413,
      "grad_norm": 0.9120416641235352,
      "learning_rate": 0.00012479999999999997,
      "loss": 2.8723,
      "step": 213
    },
    {
      "epoch": 0.029992992291520672,
      "grad_norm": 0.31208303570747375,
      "learning_rate": 0.00012539999999999999,
      "loss": 2.8848,
      "step": 214
    },
    {
      "epoch": 0.030133146461107218,
      "grad_norm": 1.156059741973877,
      "learning_rate": 0.00012599999999999997,
      "loss": 2.862,
      "step": 215
    },
    {
      "epoch": 0.030273300630693764,
      "grad_norm": 0.5333488583564758,
      "learning_rate": 0.0001266,
      "loss": 2.8534,
      "step": 216
    },
    {
      "epoch": 0.030413454800280307,
      "grad_norm": 1.356495976448059,
      "learning_rate": 0.00012719999999999997,
      "loss": 2.8656,
      "step": 217
    },
    {
      "epoch": 0.030553608969866853,
      "grad_norm": 0.7143720388412476,
      "learning_rate": 0.0001278,
      "loss": 2.8471,
      "step": 218
    },
    {
      "epoch": 0.0306937631394534,
      "grad_norm": 1.2398115396499634,
      "learning_rate": 0.00012839999999999998,
      "loss": 2.8298,
      "step": 219
    },
    {
      "epoch": 0.030833917309039945,
      "grad_norm": 1.4916234016418457,
      "learning_rate": 0.000129,
      "loss": 2.8283,
      "step": 220
    },
    {
      "epoch": 0.030974071478626488,
      "grad_norm": 1.075265884399414,
      "learning_rate": 0.00012959999999999998,
      "loss": 2.8709,
      "step": 221
    },
    {
      "epoch": 0.031114225648213034,
      "grad_norm": 0.7169482707977295,
      "learning_rate": 0.0001302,
      "loss": 2.8371,
      "step": 222
    },
    {
      "epoch": 0.03125437981779958,
      "grad_norm": 1.6476532220840454,
      "learning_rate": 0.00013079999999999998,
      "loss": 2.8555,
      "step": 223
    },
    {
      "epoch": 0.031394533987386126,
      "grad_norm": 1.5766204595565796,
      "learning_rate": 0.0001314,
      "loss": 2.8857,
      "step": 224
    },
    {
      "epoch": 0.03153468815697267,
      "grad_norm": 1.301588535308838,
      "learning_rate": 0.00013199999999999998,
      "loss": 2.8482,
      "step": 225
    },
    {
      "epoch": 0.03167484232655921,
      "grad_norm": 0.4020247459411621,
      "learning_rate": 0.0001326,
      "loss": 2.8864,
      "step": 226
    },
    {
      "epoch": 0.03181499649614576,
      "grad_norm": 1.7609339952468872,
      "learning_rate": 0.00013319999999999999,
      "loss": 2.8646,
      "step": 227
    },
    {
      "epoch": 0.031955150665732304,
      "grad_norm": 0.6501938104629517,
      "learning_rate": 0.0001338,
      "loss": 2.8688,
      "step": 228
    },
    {
      "epoch": 0.032095304835318854,
      "grad_norm": 0.5484358072280884,
      "learning_rate": 0.0001344,
      "loss": 2.8752,
      "step": 229
    },
    {
      "epoch": 0.0322354590049054,
      "grad_norm": 0.7517778873443604,
      "learning_rate": 0.000135,
      "loss": 2.8287,
      "step": 230
    },
    {
      "epoch": 0.03237561317449194,
      "grad_norm": 0.590181827545166,
      "learning_rate": 0.0001356,
      "loss": 2.8778,
      "step": 231
    },
    {
      "epoch": 0.03251576734407849,
      "grad_norm": 1.5680372714996338,
      "learning_rate": 0.0001362,
      "loss": 2.8207,
      "step": 232
    },
    {
      "epoch": 0.03265592151366503,
      "grad_norm": 1.1434855461120605,
      "learning_rate": 0.0001368,
      "loss": 2.8577,
      "step": 233
    },
    {
      "epoch": 0.032796075683251574,
      "grad_norm": 1.7322049140930176,
      "learning_rate": 0.0001374,
      "loss": 2.8589,
      "step": 234
    },
    {
      "epoch": 0.032936229852838124,
      "grad_norm": 0.3845532238483429,
      "learning_rate": 0.000138,
      "loss": 2.8265,
      "step": 235
    },
    {
      "epoch": 0.03307638402242467,
      "grad_norm": 1.3557482957839966,
      "learning_rate": 0.0001386,
      "loss": 2.8262,
      "step": 236
    },
    {
      "epoch": 0.03321653819201121,
      "grad_norm": 1.2092598676681519,
      "learning_rate": 0.0001392,
      "loss": 2.8347,
      "step": 237
    },
    {
      "epoch": 0.03335669236159776,
      "grad_norm": 3.0183780193328857,
      "learning_rate": 0.00013979999999999998,
      "loss": 2.8297,
      "step": 238
    },
    {
      "epoch": 0.0334968465311843,
      "grad_norm": 1.6880500316619873,
      "learning_rate": 0.0001404,
      "loss": 2.9211,
      "step": 239
    },
    {
      "epoch": 0.03363700070077085,
      "grad_norm": 0.8898597359657288,
      "learning_rate": 0.00014099999999999998,
      "loss": 2.8295,
      "step": 240
    },
    {
      "epoch": 0.033777154870357394,
      "grad_norm": 0.5321073532104492,
      "learning_rate": 0.00014159999999999997,
      "loss": 2.8406,
      "step": 241
    },
    {
      "epoch": 0.03391730903994394,
      "grad_norm": 0.7906889915466309,
      "learning_rate": 0.0001422,
      "loss": 2.8915,
      "step": 242
    },
    {
      "epoch": 0.034057463209530486,
      "grad_norm": 0.5971015095710754,
      "learning_rate": 0.00014279999999999997,
      "loss": 2.8852,
      "step": 243
    },
    {
      "epoch": 0.03419761737911703,
      "grad_norm": 2.295685291290283,
      "learning_rate": 0.0001434,
      "loss": 2.8779,
      "step": 244
    },
    {
      "epoch": 0.03433777154870357,
      "grad_norm": 0.9585694670677185,
      "learning_rate": 0.00014399999999999998,
      "loss": 2.9366,
      "step": 245
    },
    {
      "epoch": 0.03447792571829012,
      "grad_norm": 1.4037121534347534,
      "learning_rate": 0.0001446,
      "loss": 2.8852,
      "step": 246
    },
    {
      "epoch": 0.034618079887876664,
      "grad_norm": 0.8975903987884521,
      "learning_rate": 0.00014519999999999998,
      "loss": 2.8585,
      "step": 247
    },
    {
      "epoch": 0.03475823405746321,
      "grad_norm": 2.904129981994629,
      "learning_rate": 0.0001458,
      "loss": 2.9175,
      "step": 248
    },
    {
      "epoch": 0.034898388227049756,
      "grad_norm": 0.9668857455253601,
      "learning_rate": 0.00014639999999999998,
      "loss": 2.9313,
      "step": 249
    },
    {
      "epoch": 0.0350385423966363,
      "grad_norm": 0.9312328100204468,
      "learning_rate": 0.000147,
      "loss": 3.0427,
      "step": 250
    },
    {
      "epoch": 0.03517869656622284,
      "grad_norm": 0.9613952040672302,
      "learning_rate": 0.00014759999999999998,
      "loss": 2.888,
      "step": 251
    },
    {
      "epoch": 0.03531885073580939,
      "grad_norm": 0.5064305067062378,
      "learning_rate": 0.0001482,
      "loss": 2.882,
      "step": 252
    },
    {
      "epoch": 0.035459004905395934,
      "grad_norm": 0.9140638113021851,
      "learning_rate": 0.00014879999999999998,
      "loss": 2.8515,
      "step": 253
    },
    {
      "epoch": 0.035599159074982484,
      "grad_norm": 0.742790162563324,
      "learning_rate": 0.0001494,
      "loss": 2.8343,
      "step": 254
    },
    {
      "epoch": 0.035739313244569026,
      "grad_norm": 0.44202330708503723,
      "learning_rate": 0.00015,
      "loss": 2.8345,
      "step": 255
    },
    {
      "epoch": 0.03587946741415557,
      "grad_norm": 0.3355579674243927,
      "learning_rate": 0.00015059999999999997,
      "loss": 2.8469,
      "step": 256
    },
    {
      "epoch": 0.03601962158374212,
      "grad_norm": 1.9614343643188477,
      "learning_rate": 0.0001512,
      "loss": 2.841,
      "step": 257
    },
    {
      "epoch": 0.03615977575332866,
      "grad_norm": 2.2115304470062256,
      "learning_rate": 0.00015179999999999998,
      "loss": 2.852,
      "step": 258
    },
    {
      "epoch": 0.036299929922915204,
      "grad_norm": 1.1922167539596558,
      "learning_rate": 0.0001524,
      "loss": 2.864,
      "step": 259
    },
    {
      "epoch": 0.036440084092501754,
      "grad_norm": 0.9516031742095947,
      "learning_rate": 0.00015299999999999998,
      "loss": 2.8438,
      "step": 260
    },
    {
      "epoch": 0.0365802382620883,
      "grad_norm": 0.5913712978363037,
      "learning_rate": 0.0001536,
      "loss": 2.8534,
      "step": 261
    },
    {
      "epoch": 0.03672039243167484,
      "grad_norm": 1.1199309825897217,
      "learning_rate": 0.00015419999999999998,
      "loss": 2.8423,
      "step": 262
    },
    {
      "epoch": 0.03686054660126139,
      "grad_norm": 0.5776010751724243,
      "learning_rate": 0.0001548,
      "loss": 2.8878,
      "step": 263
    },
    {
      "epoch": 0.03700070077084793,
      "grad_norm": 0.5159766674041748,
      "learning_rate": 0.00015539999999999998,
      "loss": 2.8519,
      "step": 264
    },
    {
      "epoch": 0.03714085494043448,
      "grad_norm": 0.5820059180259705,
      "learning_rate": 0.000156,
      "loss": 2.8866,
      "step": 265
    },
    {
      "epoch": 0.037281009110021024,
      "grad_norm": 1.8129342794418335,
      "learning_rate": 0.00015659999999999998,
      "loss": 2.8437,
      "step": 266
    },
    {
      "epoch": 0.03742116327960757,
      "grad_norm": 0.5767428278923035,
      "learning_rate": 0.0001572,
      "loss": 2.8326,
      "step": 267
    },
    {
      "epoch": 0.037561317449194116,
      "grad_norm": 0.392638623714447,
      "learning_rate": 0.0001578,
      "loss": 2.8714,
      "step": 268
    },
    {
      "epoch": 0.03770147161878066,
      "grad_norm": 0.7401202917098999,
      "learning_rate": 0.0001584,
      "loss": 2.8517,
      "step": 269
    },
    {
      "epoch": 0.0378416257883672,
      "grad_norm": 0.4700906574726105,
      "learning_rate": 0.000159,
      "loss": 2.8904,
      "step": 270
    },
    {
      "epoch": 0.03798177995795375,
      "grad_norm": 2.120917320251465,
      "learning_rate": 0.0001596,
      "loss": 2.8843,
      "step": 271
    },
    {
      "epoch": 0.038121934127540294,
      "grad_norm": 0.46234700083732605,
      "learning_rate": 0.0001602,
      "loss": 2.815,
      "step": 272
    },
    {
      "epoch": 0.03826208829712684,
      "grad_norm": 1.6301603317260742,
      "learning_rate": 0.0001608,
      "loss": 2.8317,
      "step": 273
    },
    {
      "epoch": 0.038402242466713386,
      "grad_norm": 0.7664803862571716,
      "learning_rate": 0.0001614,
      "loss": 2.8285,
      "step": 274
    },
    {
      "epoch": 0.03854239663629993,
      "grad_norm": 0.5692226886749268,
      "learning_rate": 0.000162,
      "loss": 2.8247,
      "step": 275
    },
    {
      "epoch": 0.03868255080588647,
      "grad_norm": 1.199317216873169,
      "learning_rate": 0.0001626,
      "loss": 2.8341,
      "step": 276
    },
    {
      "epoch": 0.03882270497547302,
      "grad_norm": 1.7699816226959229,
      "learning_rate": 0.0001632,
      "loss": 2.888,
      "step": 277
    },
    {
      "epoch": 0.038962859145059564,
      "grad_norm": 1.5419954061508179,
      "learning_rate": 0.0001638,
      "loss": 2.8286,
      "step": 278
    },
    {
      "epoch": 0.039103013314646114,
      "grad_norm": 2.050661087036133,
      "learning_rate": 0.0001644,
      "loss": 2.8947,
      "step": 279
    },
    {
      "epoch": 0.039243167484232656,
      "grad_norm": 0.802770733833313,
      "learning_rate": 0.000165,
      "loss": 2.8268,
      "step": 280
    },
    {
      "epoch": 0.0393833216538192,
      "grad_norm": 1.0551692247390747,
      "learning_rate": 0.0001656,
      "loss": 2.8512,
      "step": 281
    },
    {
      "epoch": 0.03952347582340575,
      "grad_norm": 0.5748604536056519,
      "learning_rate": 0.0001662,
      "loss": 2.8443,
      "step": 282
    },
    {
      "epoch": 0.03966362999299229,
      "grad_norm": 1.8012722730636597,
      "learning_rate": 0.0001668,
      "loss": 2.8339,
      "step": 283
    },
    {
      "epoch": 0.039803784162578834,
      "grad_norm": 2.066579818725586,
      "learning_rate": 0.0001674,
      "loss": 2.8402,
      "step": 284
    },
    {
      "epoch": 0.039943938332165384,
      "grad_norm": 0.491862416267395,
      "learning_rate": 0.000168,
      "loss": 2.8373,
      "step": 285
    },
    {
      "epoch": 0.040084092501751926,
      "grad_norm": 2.1423556804656982,
      "learning_rate": 0.0001686,
      "loss": 2.8744,
      "step": 286
    },
    {
      "epoch": 0.04022424667133847,
      "grad_norm": 18.57097053527832,
      "learning_rate": 0.00016919999999999997,
      "loss": 3.6457,
      "step": 287
    },
    {
      "epoch": 0.04036440084092502,
      "grad_norm": 1.6656326055526733,
      "learning_rate": 0.00016979999999999998,
      "loss": 2.8279,
      "step": 288
    },
    {
      "epoch": 0.04050455501051156,
      "grad_norm": 2.3413543701171875,
      "learning_rate": 0.00017039999999999997,
      "loss": 2.8795,
      "step": 289
    },
    {
      "epoch": 0.04064470918009811,
      "grad_norm": 2.042393207550049,
      "learning_rate": 0.00017099999999999998,
      "loss": 2.8585,
      "step": 290
    },
    {
      "epoch": 0.040784863349684654,
      "grad_norm": 0.9782095551490784,
      "learning_rate": 0.00017159999999999997,
      "loss": 2.8575,
      "step": 291
    },
    {
      "epoch": 0.040925017519271197,
      "grad_norm": 0.4077852666378021,
      "learning_rate": 0.00017219999999999998,
      "loss": 2.8529,
      "step": 292
    },
    {
      "epoch": 0.041065171688857746,
      "grad_norm": 1.9978406429290771,
      "learning_rate": 0.00017279999999999997,
      "loss": 2.8674,
      "step": 293
    },
    {
      "epoch": 0.04120532585844429,
      "grad_norm": 1.3952635526657104,
      "learning_rate": 0.00017339999999999996,
      "loss": 2.8605,
      "step": 294
    },
    {
      "epoch": 0.04134548002803083,
      "grad_norm": 0.8728106021881104,
      "learning_rate": 0.00017399999999999997,
      "loss": 2.8661,
      "step": 295
    },
    {
      "epoch": 0.04148563419761738,
      "grad_norm": 0.7876748442649841,
      "learning_rate": 0.00017459999999999996,
      "loss": 2.818,
      "step": 296
    },
    {
      "epoch": 0.041625788367203924,
      "grad_norm": 1.246867299079895,
      "learning_rate": 0.00017519999999999998,
      "loss": 2.8833,
      "step": 297
    },
    {
      "epoch": 0.04176594253679047,
      "grad_norm": 0.8078440427780151,
      "learning_rate": 0.00017579999999999996,
      "loss": 2.9058,
      "step": 298
    },
    {
      "epoch": 0.041906096706377016,
      "grad_norm": 1.0067757368087769,
      "learning_rate": 0.00017639999999999998,
      "loss": 2.7852,
      "step": 299
    },
    {
      "epoch": 0.04204625087596356,
      "grad_norm": 1.1474734544754028,
      "learning_rate": 0.00017699999999999997,
      "loss": 2.9193,
      "step": 300
    },
    {
      "epoch": 0.04218640504555011,
      "grad_norm": 2.7537102699279785,
      "learning_rate": 0.00017759999999999998,
      "loss": 2.8676,
      "step": 301
    },
    {
      "epoch": 0.04232655921513665,
      "grad_norm": 0.8940155506134033,
      "learning_rate": 0.00017819999999999997,
      "loss": 2.872,
      "step": 302
    },
    {
      "epoch": 0.042466713384723194,
      "grad_norm": 0.7787176966667175,
      "learning_rate": 0.00017879999999999998,
      "loss": 2.8232,
      "step": 303
    },
    {
      "epoch": 0.042606867554309744,
      "grad_norm": 1.5186501741409302,
      "learning_rate": 0.00017939999999999997,
      "loss": 2.8408,
      "step": 304
    },
    {
      "epoch": 0.042747021723896286,
      "grad_norm": 1.1387662887573242,
      "learning_rate": 0.00017999999999999998,
      "loss": 2.8433,
      "step": 305
    },
    {
      "epoch": 0.04288717589348283,
      "grad_norm": 2.0307185649871826,
      "learning_rate": 0.00018059999999999997,
      "loss": 2.9168,
      "step": 306
    },
    {
      "epoch": 0.04302733006306938,
      "grad_norm": 0.6528588533401489,
      "learning_rate": 0.00018119999999999999,
      "loss": 2.8807,
      "step": 307
    },
    {
      "epoch": 0.04316748423265592,
      "grad_norm": 0.45445582270622253,
      "learning_rate": 0.00018179999999999997,
      "loss": 2.8789,
      "step": 308
    },
    {
      "epoch": 0.043307638402242464,
      "grad_norm": 0.5610876083374023,
      "learning_rate": 0.0001824,
      "loss": 2.828,
      "step": 309
    },
    {
      "epoch": 0.043447792571829014,
      "grad_norm": 2.383838653564453,
      "learning_rate": 0.00018299999999999998,
      "loss": 2.8649,
      "step": 310
    },
    {
      "epoch": 0.043587946741415556,
      "grad_norm": 1.3355058431625366,
      "learning_rate": 0.0001836,
      "loss": 2.8304,
      "step": 311
    },
    {
      "epoch": 0.0437281009110021,
      "grad_norm": 0.5329657196998596,
      "learning_rate": 0.00018419999999999998,
      "loss": 2.8369,
      "step": 312
    },
    {
      "epoch": 0.04386825508058865,
      "grad_norm": 0.422691285610199,
      "learning_rate": 0.0001848,
      "loss": 2.8273,
      "step": 313
    },
    {
      "epoch": 0.04400840925017519,
      "grad_norm": 0.4409281313419342,
      "learning_rate": 0.00018539999999999998,
      "loss": 2.8523,
      "step": 314
    },
    {
      "epoch": 0.04414856341976174,
      "grad_norm": 2.9379351139068604,
      "learning_rate": 0.000186,
      "loss": 2.8821,
      "step": 315
    },
    {
      "epoch": 0.044288717589348284,
      "grad_norm": 0.3748665153980255,
      "learning_rate": 0.00018659999999999998,
      "loss": 2.8506,
      "step": 316
    },
    {
      "epoch": 0.044428871758934826,
      "grad_norm": 0.529017448425293,
      "learning_rate": 0.0001872,
      "loss": 2.852,
      "step": 317
    },
    {
      "epoch": 0.044569025928521376,
      "grad_norm": 1.6215025186538696,
      "learning_rate": 0.00018779999999999998,
      "loss": 2.8704,
      "step": 318
    },
    {
      "epoch": 0.04470918009810792,
      "grad_norm": 1.9423863887786865,
      "learning_rate": 0.00018839999999999997,
      "loss": 2.8542,
      "step": 319
    },
    {
      "epoch": 0.04484933426769446,
      "grad_norm": 0.6593060493469238,
      "learning_rate": 0.00018899999999999999,
      "loss": 2.8426,
      "step": 320
    },
    {
      "epoch": 0.04498948843728101,
      "grad_norm": 0.34480753540992737,
      "learning_rate": 0.00018959999999999997,
      "loss": 2.8502,
      "step": 321
    },
    {
      "epoch": 0.045129642606867554,
      "grad_norm": 1.324047327041626,
      "learning_rate": 0.0001902,
      "loss": 2.8123,
      "step": 322
    },
    {
      "epoch": 0.045269796776454096,
      "grad_norm": 0.7530344724655151,
      "learning_rate": 0.00019079999999999998,
      "loss": 2.8417,
      "step": 323
    },
    {
      "epoch": 0.045409950946040646,
      "grad_norm": 0.4720246195793152,
      "learning_rate": 0.0001914,
      "loss": 2.841,
      "step": 324
    },
    {
      "epoch": 0.04555010511562719,
      "grad_norm": 1.2542595863342285,
      "learning_rate": 0.00019199999999999998,
      "loss": 2.8635,
      "step": 325
    },
    {
      "epoch": 0.04569025928521374,
      "grad_norm": 1.5343085527420044,
      "learning_rate": 0.0001926,
      "loss": 2.8487,
      "step": 326
    },
    {
      "epoch": 0.04583041345480028,
      "grad_norm": 0.613078236579895,
      "learning_rate": 0.00019319999999999998,
      "loss": 2.8425,
      "step": 327
    },
    {
      "epoch": 0.045970567624386824,
      "grad_norm": 0.5128901600837708,
      "learning_rate": 0.0001938,
      "loss": 2.8412,
      "step": 328
    },
    {
      "epoch": 0.046110721793973374,
      "grad_norm": 1.5029518604278564,
      "learning_rate": 0.00019439999999999998,
      "loss": 2.8883,
      "step": 329
    },
    {
      "epoch": 0.046250875963559916,
      "grad_norm": 1.5685569047927856,
      "learning_rate": 0.000195,
      "loss": 2.8264,
      "step": 330
    },
    {
      "epoch": 0.04639103013314646,
      "grad_norm": 0.4594239890575409,
      "learning_rate": 0.00019559999999999998,
      "loss": 2.8357,
      "step": 331
    },
    {
      "epoch": 0.04653118430273301,
      "grad_norm": 1.7477678060531616,
      "learning_rate": 0.0001962,
      "loss": 2.8293,
      "step": 332
    },
    {
      "epoch": 0.04667133847231955,
      "grad_norm": 1.4861485958099365,
      "learning_rate": 0.00019679999999999999,
      "loss": 2.8524,
      "step": 333
    },
    {
      "epoch": 0.046811492641906094,
      "grad_norm": 0.48972010612487793,
      "learning_rate": 0.0001974,
      "loss": 2.8838,
      "step": 334
    },
    {
      "epoch": 0.046951646811492644,
      "grad_norm": 0.6660546660423279,
      "learning_rate": 0.000198,
      "loss": 2.8777,
      "step": 335
    },
    {
      "epoch": 0.047091800981079186,
      "grad_norm": 2.614840030670166,
      "learning_rate": 0.0001986,
      "loss": 2.8734,
      "step": 336
    },
    {
      "epoch": 0.04723195515066573,
      "grad_norm": 2.2148666381835938,
      "learning_rate": 0.0001992,
      "loss": 2.8422,
      "step": 337
    },
    {
      "epoch": 0.04737210932025228,
      "grad_norm": 1.6337977647781372,
      "learning_rate": 0.0001998,
      "loss": 2.8671,
      "step": 338
    },
    {
      "epoch": 0.04751226348983882,
      "grad_norm": 2.3236374855041504,
      "learning_rate": 0.0002004,
      "loss": 2.8773,
      "step": 339
    },
    {
      "epoch": 0.04765241765942537,
      "grad_norm": 0.4809541702270508,
      "learning_rate": 0.000201,
      "loss": 2.8244,
      "step": 340
    },
    {
      "epoch": 0.047792571829011914,
      "grad_norm": 1.523079752922058,
      "learning_rate": 0.0002016,
      "loss": 2.8679,
      "step": 341
    },
    {
      "epoch": 0.047932725998598456,
      "grad_norm": 1.1739789247512817,
      "learning_rate": 0.0002022,
      "loss": 2.8072,
      "step": 342
    },
    {
      "epoch": 0.048072880168185006,
      "grad_norm": 0.6344507932662964,
      "learning_rate": 0.0002028,
      "loss": 2.8647,
      "step": 343
    },
    {
      "epoch": 0.04821303433777155,
      "grad_norm": 1.3769617080688477,
      "learning_rate": 0.00020339999999999998,
      "loss": 2.8539,
      "step": 344
    },
    {
      "epoch": 0.04835318850735809,
      "grad_norm": 0.4988667964935303,
      "learning_rate": 0.000204,
      "loss": 2.8529,
      "step": 345
    },
    {
      "epoch": 0.04849334267694464,
      "grad_norm": 0.8496665358543396,
      "learning_rate": 0.00020459999999999999,
      "loss": 2.8972,
      "step": 346
    },
    {
      "epoch": 0.048633496846531184,
      "grad_norm": 1.0026979446411133,
      "learning_rate": 0.0002052,
      "loss": 2.8639,
      "step": 347
    },
    {
      "epoch": 0.048773651016117726,
      "grad_norm": 1.7278631925582886,
      "learning_rate": 0.0002058,
      "loss": 2.8485,
      "step": 348
    },
    {
      "epoch": 0.048913805185704276,
      "grad_norm": 0.6872090101242065,
      "learning_rate": 0.00020639999999999998,
      "loss": 2.899,
      "step": 349
    },
    {
      "epoch": 0.04905395935529082,
      "grad_norm": 1.0069712400436401,
      "learning_rate": 0.00020699999999999996,
      "loss": 2.8375,
      "step": 350
    },
    {
      "epoch": 0.04919411352487737,
      "grad_norm": 0.4872021973133087,
      "learning_rate": 0.00020759999999999998,
      "loss": 2.8516,
      "step": 351
    },
    {
      "epoch": 0.04933426769446391,
      "grad_norm": 1.8567527532577515,
      "learning_rate": 0.00020819999999999996,
      "loss": 2.8876,
      "step": 352
    },
    {
      "epoch": 0.049474421864050454,
      "grad_norm": 1.2505429983139038,
      "learning_rate": 0.00020879999999999998,
      "loss": 2.8486,
      "step": 353
    },
    {
      "epoch": 0.049614576033637,
      "grad_norm": 0.7958128452301025,
      "learning_rate": 0.00020939999999999997,
      "loss": 2.8298,
      "step": 354
    },
    {
      "epoch": 0.049754730203223546,
      "grad_norm": 0.5082005858421326,
      "learning_rate": 0.00020999999999999998,
      "loss": 2.8309,
      "step": 355
    },
    {
      "epoch": 0.04989488437281009,
      "grad_norm": 1.5224041938781738,
      "learning_rate": 0.00021059999999999997,
      "loss": 2.8561,
      "step": 356
    },
    {
      "epoch": 0.05003503854239664,
      "grad_norm": 0.5568313598632812,
      "learning_rate": 0.00021119999999999996,
      "loss": 2.8423,
      "step": 357
    },
    {
      "epoch": 0.05017519271198318,
      "grad_norm": 1.4406063556671143,
      "learning_rate": 0.00021179999999999997,
      "loss": 2.7996,
      "step": 358
    },
    {
      "epoch": 0.050315346881569724,
      "grad_norm": 2.8648271560668945,
      "learning_rate": 0.00021239999999999996,
      "loss": 2.8842,
      "step": 359
    },
    {
      "epoch": 0.050455501051156273,
      "grad_norm": 1.8214137554168701,
      "learning_rate": 0.00021299999999999997,
      "loss": 2.8375,
      "step": 360
    },
    {
      "epoch": 0.050595655220742816,
      "grad_norm": 1.0080620050430298,
      "learning_rate": 0.00021359999999999996,
      "loss": 2.8281,
      "step": 361
    },
    {
      "epoch": 0.05073580939032936,
      "grad_norm": 0.5720990896224976,
      "learning_rate": 0.00021419999999999998,
      "loss": 2.827,
      "step": 362
    },
    {
      "epoch": 0.05087596355991591,
      "grad_norm": 1.4787616729736328,
      "learning_rate": 0.00021479999999999996,
      "loss": 2.897,
      "step": 363
    },
    {
      "epoch": 0.05101611772950245,
      "grad_norm": 0.944041907787323,
      "learning_rate": 0.00021539999999999998,
      "loss": 2.8256,
      "step": 364
    },
    {
      "epoch": 0.051156271899089,
      "grad_norm": 0.9127120971679688,
      "learning_rate": 0.00021599999999999996,
      "loss": 2.8517,
      "step": 365
    },
    {
      "epoch": 0.051296426068675544,
      "grad_norm": 0.46851101517677307,
      "learning_rate": 0.00021659999999999998,
      "loss": 2.8267,
      "step": 366
    },
    {
      "epoch": 0.051436580238262086,
      "grad_norm": 0.5274301171302795,
      "learning_rate": 0.00021719999999999997,
      "loss": 2.8684,
      "step": 367
    },
    {
      "epoch": 0.051576734407848636,
      "grad_norm": 0.6189049482345581,
      "learning_rate": 0.00021779999999999998,
      "loss": 2.8831,
      "step": 368
    },
    {
      "epoch": 0.05171688857743518,
      "grad_norm": 1.1459282636642456,
      "learning_rate": 0.00021839999999999997,
      "loss": 2.8211,
      "step": 369
    },
    {
      "epoch": 0.05185704274702172,
      "grad_norm": 1.3347804546356201,
      "learning_rate": 0.00021899999999999998,
      "loss": 2.8871,
      "step": 370
    },
    {
      "epoch": 0.05199719691660827,
      "grad_norm": 0.47682449221611023,
      "learning_rate": 0.00021959999999999997,
      "loss": 2.8804,
      "step": 371
    },
    {
      "epoch": 0.052137351086194814,
      "grad_norm": 2.0062739849090576,
      "learning_rate": 0.00022019999999999999,
      "loss": 2.8677,
      "step": 372
    },
    {
      "epoch": 0.052277505255781356,
      "grad_norm": 0.9368986487388611,
      "learning_rate": 0.00022079999999999997,
      "loss": 2.8389,
      "step": 373
    },
    {
      "epoch": 0.052417659425367906,
      "grad_norm": 0.8859605193138123,
      "learning_rate": 0.0002214,
      "loss": 2.8442,
      "step": 374
    },
    {
      "epoch": 0.05255781359495445,
      "grad_norm": 0.8973392844200134,
      "learning_rate": 0.00022199999999999998,
      "loss": 2.8508,
      "step": 375
    },
    {
      "epoch": 0.052697967764541,
      "grad_norm": 0.552753210067749,
      "learning_rate": 0.0002226,
      "loss": 2.8554,
      "step": 376
    },
    {
      "epoch": 0.05283812193412754,
      "grad_norm": 1.781485915184021,
      "learning_rate": 0.00022319999999999998,
      "loss": 2.8305,
      "step": 377
    },
    {
      "epoch": 0.052978276103714084,
      "grad_norm": 1.4769580364227295,
      "learning_rate": 0.0002238,
      "loss": 2.8233,
      "step": 378
    },
    {
      "epoch": 0.05311843027330063,
      "grad_norm": 0.480735182762146,
      "learning_rate": 0.00022439999999999998,
      "loss": 2.7994,
      "step": 379
    },
    {
      "epoch": 0.053258584442887176,
      "grad_norm": 0.44265174865722656,
      "learning_rate": 0.000225,
      "loss": 2.8467,
      "step": 380
    },
    {
      "epoch": 0.05339873861247372,
      "grad_norm": 2.6101057529449463,
      "learning_rate": 0.00022559999999999998,
      "loss": 2.8585,
      "step": 381
    },
    {
      "epoch": 0.05353889278206027,
      "grad_norm": 0.7626937031745911,
      "learning_rate": 0.00022619999999999997,
      "loss": 2.8747,
      "step": 382
    },
    {
      "epoch": 0.05367904695164681,
      "grad_norm": 2.9585444927215576,
      "learning_rate": 0.00022679999999999998,
      "loss": 2.8791,
      "step": 383
    },
    {
      "epoch": 0.053819201121233354,
      "grad_norm": 0.573103666305542,
      "learning_rate": 0.00022739999999999997,
      "loss": 2.8262,
      "step": 384
    },
    {
      "epoch": 0.0539593552908199,
      "grad_norm": 1.5471686124801636,
      "learning_rate": 0.00022799999999999999,
      "loss": 2.8254,
      "step": 385
    },
    {
      "epoch": 0.054099509460406446,
      "grad_norm": 0.5622177124023438,
      "learning_rate": 0.00022859999999999997,
      "loss": 2.8457,
      "step": 386
    },
    {
      "epoch": 0.054239663629992996,
      "grad_norm": 0.8726809024810791,
      "learning_rate": 0.0002292,
      "loss": 2.823,
      "step": 387
    },
    {
      "epoch": 0.05437981779957954,
      "grad_norm": 0.6712299585342407,
      "learning_rate": 0.00022979999999999997,
      "loss": 2.8658,
      "step": 388
    },
    {
      "epoch": 0.05451997196916608,
      "grad_norm": 1.3056963682174683,
      "learning_rate": 0.0002304,
      "loss": 2.8463,
      "step": 389
    },
    {
      "epoch": 0.05466012613875263,
      "grad_norm": 2.0756328105926514,
      "learning_rate": 0.00023099999999999998,
      "loss": 2.8616,
      "step": 390
    },
    {
      "epoch": 0.05480028030833917,
      "grad_norm": 0.638837993144989,
      "learning_rate": 0.0002316,
      "loss": 2.8496,
      "step": 391
    },
    {
      "epoch": 0.054940434477925716,
      "grad_norm": 0.5293677449226379,
      "learning_rate": 0.00023219999999999998,
      "loss": 2.8431,
      "step": 392
    },
    {
      "epoch": 0.055080588647512266,
      "grad_norm": 1.3139581680297852,
      "learning_rate": 0.0002328,
      "loss": 2.9194,
      "step": 393
    },
    {
      "epoch": 0.05522074281709881,
      "grad_norm": 1.280122995376587,
      "learning_rate": 0.00023339999999999998,
      "loss": 2.8804,
      "step": 394
    },
    {
      "epoch": 0.05536089698668535,
      "grad_norm": 1.2172834873199463,
      "learning_rate": 0.000234,
      "loss": 2.8266,
      "step": 395
    },
    {
      "epoch": 0.0555010511562719,
      "grad_norm": 1.5580065250396729,
      "learning_rate": 0.00023459999999999998,
      "loss": 2.9107,
      "step": 396
    },
    {
      "epoch": 0.055641205325858444,
      "grad_norm": 0.5468613505363464,
      "learning_rate": 0.0002352,
      "loss": 2.8887,
      "step": 397
    },
    {
      "epoch": 0.055781359495444986,
      "grad_norm": 1.5642807483673096,
      "learning_rate": 0.00023579999999999999,
      "loss": 2.8515,
      "step": 398
    },
    {
      "epoch": 0.055921513665031536,
      "grad_norm": 1.492555022239685,
      "learning_rate": 0.0002364,
      "loss": 2.8246,
      "step": 399
    },
    {
      "epoch": 0.05606166783461808,
      "grad_norm": 2.943617343902588,
      "learning_rate": 0.000237,
      "loss": 2.8775,
      "step": 400
    },
    {
      "epoch": 0.05620182200420463,
      "grad_norm": 0.8271451592445374,
      "learning_rate": 0.0002376,
      "loss": 2.8883,
      "step": 401
    },
    {
      "epoch": 0.05634197617379117,
      "grad_norm": 1.8347681760787964,
      "learning_rate": 0.0002382,
      "loss": 2.8397,
      "step": 402
    },
    {
      "epoch": 0.056482130343377714,
      "grad_norm": 1.0476518869400024,
      "learning_rate": 0.0002388,
      "loss": 2.854,
      "step": 403
    },
    {
      "epoch": 0.05662228451296426,
      "grad_norm": 0.5847808122634888,
      "learning_rate": 0.0002394,
      "loss": 2.8279,
      "step": 404
    },
    {
      "epoch": 0.056762438682550806,
      "grad_norm": 1.2081626653671265,
      "learning_rate": 0.00023999999999999998,
      "loss": 2.8142,
      "step": 405
    },
    {
      "epoch": 0.05690259285213735,
      "grad_norm": 1.6852529048919678,
      "learning_rate": 0.0002406,
      "loss": 2.8055,
      "step": 406
    },
    {
      "epoch": 0.0570427470217239,
      "grad_norm": 1.3509167432785034,
      "learning_rate": 0.00024119999999999998,
      "loss": 2.8701,
      "step": 407
    },
    {
      "epoch": 0.05718290119131044,
      "grad_norm": 2.65307354927063,
      "learning_rate": 0.0002418,
      "loss": 2.8621,
      "step": 408
    },
    {
      "epoch": 0.057323055360896984,
      "grad_norm": 0.5879318118095398,
      "learning_rate": 0.00024239999999999998,
      "loss": 2.8647,
      "step": 409
    },
    {
      "epoch": 0.05746320953048353,
      "grad_norm": 2.9156978130340576,
      "learning_rate": 0.000243,
      "loss": 2.8549,
      "step": 410
    },
    {
      "epoch": 0.057603363700070076,
      "grad_norm": 1.1443407535552979,
      "learning_rate": 0.00024359999999999999,
      "loss": 2.8621,
      "step": 411
    },
    {
      "epoch": 0.057743517869656626,
      "grad_norm": 1.2485133409500122,
      "learning_rate": 0.00024419999999999997,
      "loss": 2.8343,
      "step": 412
    },
    {
      "epoch": 0.05788367203924317,
      "grad_norm": 1.3243160247802734,
      "learning_rate": 0.0002448,
      "loss": 2.8165,
      "step": 413
    },
    {
      "epoch": 0.05802382620882971,
      "grad_norm": 0.671999990940094,
      "learning_rate": 0.00024539999999999995,
      "loss": 2.8125,
      "step": 414
    },
    {
      "epoch": 0.05816398037841626,
      "grad_norm": 2.0717403888702393,
      "learning_rate": 0.00024599999999999996,
      "loss": 2.87,
      "step": 415
    },
    {
      "epoch": 0.0583041345480028,
      "grad_norm": 1.918505072593689,
      "learning_rate": 0.0002466,
      "loss": 2.8406,
      "step": 416
    },
    {
      "epoch": 0.058444288717589346,
      "grad_norm": 0.8821729421615601,
      "learning_rate": 0.0002472,
      "loss": 2.8408,
      "step": 417
    },
    {
      "epoch": 0.058584442887175896,
      "grad_norm": 1.0364701747894287,
      "learning_rate": 0.00024779999999999995,
      "loss": 2.8325,
      "step": 418
    },
    {
      "epoch": 0.05872459705676244,
      "grad_norm": 1.1036708354949951,
      "learning_rate": 0.00024839999999999997,
      "loss": 2.8547,
      "step": 419
    },
    {
      "epoch": 0.05886475122634898,
      "grad_norm": 0.47669097781181335,
      "learning_rate": 0.000249,
      "loss": 2.7921,
      "step": 420
    },
    {
      "epoch": 0.05900490539593553,
      "grad_norm": 0.5236237645149231,
      "learning_rate": 0.00024959999999999994,
      "loss": 2.8615,
      "step": 421
    },
    {
      "epoch": 0.05914505956552207,
      "grad_norm": 0.7237966060638428,
      "learning_rate": 0.00025019999999999996,
      "loss": 2.8656,
      "step": 422
    },
    {
      "epoch": 0.059285213735108616,
      "grad_norm": 0.8609464168548584,
      "learning_rate": 0.00025079999999999997,
      "loss": 2.8718,
      "step": 423
    },
    {
      "epoch": 0.059425367904695166,
      "grad_norm": 0.683222234249115,
      "learning_rate": 0.0002514,
      "loss": 2.8529,
      "step": 424
    },
    {
      "epoch": 0.05956552207428171,
      "grad_norm": 1.165490746498108,
      "learning_rate": 0.00025199999999999995,
      "loss": 2.8316,
      "step": 425
    },
    {
      "epoch": 0.05970567624386826,
      "grad_norm": 0.49919211864471436,
      "learning_rate": 0.00025259999999999996,
      "loss": 2.8195,
      "step": 426
    },
    {
      "epoch": 0.0598458304134548,
      "grad_norm": 0.2870229184627533,
      "learning_rate": 0.0002532,
      "loss": 2.8689,
      "step": 427
    },
    {
      "epoch": 0.059985984583041344,
      "grad_norm": 0.8433550596237183,
      "learning_rate": 0.0002538,
      "loss": 2.8503,
      "step": 428
    },
    {
      "epoch": 0.06012613875262789,
      "grad_norm": 2.0954675674438477,
      "learning_rate": 0.00025439999999999995,
      "loss": 2.8753,
      "step": 429
    },
    {
      "epoch": 0.060266292922214436,
      "grad_norm": 1.3331342935562134,
      "learning_rate": 0.00025499999999999996,
      "loss": 2.8756,
      "step": 430
    },
    {
      "epoch": 0.06040644709180098,
      "grad_norm": 0.755756139755249,
      "learning_rate": 0.0002556,
      "loss": 2.8152,
      "step": 431
    },
    {
      "epoch": 0.06054660126138753,
      "grad_norm": 0.48160722851753235,
      "learning_rate": 0.0002562,
      "loss": 2.8827,
      "step": 432
    },
    {
      "epoch": 0.06068675543097407,
      "grad_norm": 1.1558563709259033,
      "learning_rate": 0.00025679999999999995,
      "loss": 2.842,
      "step": 433
    },
    {
      "epoch": 0.060826909600560614,
      "grad_norm": 0.4860924482345581,
      "learning_rate": 0.00025739999999999997,
      "loss": 2.8391,
      "step": 434
    },
    {
      "epoch": 0.06096706377014716,
      "grad_norm": 2.3018696308135986,
      "learning_rate": 0.000258,
      "loss": 2.8771,
      "step": 435
    },
    {
      "epoch": 0.061107217939733706,
      "grad_norm": 0.5955384373664856,
      "learning_rate": 0.0002586,
      "loss": 2.8905,
      "step": 436
    },
    {
      "epoch": 0.061247372109320256,
      "grad_norm": 1.4681602716445923,
      "learning_rate": 0.00025919999999999996,
      "loss": 2.8492,
      "step": 437
    },
    {
      "epoch": 0.0613875262789068,
      "grad_norm": 1.8079183101654053,
      "learning_rate": 0.00025979999999999997,
      "loss": 2.8385,
      "step": 438
    },
    {
      "epoch": 0.06152768044849334,
      "grad_norm": 0.6035788655281067,
      "learning_rate": 0.0002604,
      "loss": 2.8479,
      "step": 439
    },
    {
      "epoch": 0.06166783461807989,
      "grad_norm": 0.7603915333747864,
      "learning_rate": 0.000261,
      "loss": 2.8522,
      "step": 440
    },
    {
      "epoch": 0.06180798878766643,
      "grad_norm": 1.7382203340530396,
      "learning_rate": 0.00026159999999999996,
      "loss": 2.8522,
      "step": 441
    },
    {
      "epoch": 0.061948142957252976,
      "grad_norm": 0.7512733340263367,
      "learning_rate": 0.0002622,
      "loss": 2.8483,
      "step": 442
    },
    {
      "epoch": 0.062088297126839526,
      "grad_norm": 2.5912866592407227,
      "learning_rate": 0.0002628,
      "loss": 2.8408,
      "step": 443
    },
    {
      "epoch": 0.06222845129642607,
      "grad_norm": 0.9884958267211914,
      "learning_rate": 0.00026339999999999995,
      "loss": 2.8384,
      "step": 444
    },
    {
      "epoch": 0.06236860546601261,
      "grad_norm": 2.7016091346740723,
      "learning_rate": 0.00026399999999999997,
      "loss": 2.9177,
      "step": 445
    },
    {
      "epoch": 0.06250875963559915,
      "grad_norm": 1.9022507667541504,
      "learning_rate": 0.0002646,
      "loss": 2.9078,
      "step": 446
    },
    {
      "epoch": 0.06264891380518571,
      "grad_norm": 0.9832592010498047,
      "learning_rate": 0.0002652,
      "loss": 2.8912,
      "step": 447
    },
    {
      "epoch": 0.06278906797477225,
      "grad_norm": 0.9287002682685852,
      "learning_rate": 0.00026579999999999996,
      "loss": 2.8694,
      "step": 448
    },
    {
      "epoch": 0.0629292221443588,
      "grad_norm": 1.1300129890441895,
      "learning_rate": 0.00026639999999999997,
      "loss": 2.9087,
      "step": 449
    },
    {
      "epoch": 0.06306937631394534,
      "grad_norm": 2.3987514972686768,
      "learning_rate": 0.000267,
      "loss": 2.9048,
      "step": 450
    },
    {
      "epoch": 0.06320953048353188,
      "grad_norm": 1.8747663497924805,
      "learning_rate": 0.0002676,
      "loss": 2.8403,
      "step": 451
    },
    {
      "epoch": 0.06334968465311842,
      "grad_norm": 1.1864196062088013,
      "learning_rate": 0.00026819999999999996,
      "loss": 2.8734,
      "step": 452
    },
    {
      "epoch": 0.06348983882270498,
      "grad_norm": 0.40319573879241943,
      "learning_rate": 0.0002688,
      "loss": 2.8677,
      "step": 453
    },
    {
      "epoch": 0.06362999299229152,
      "grad_norm": 3.450862169265747,
      "learning_rate": 0.0002694,
      "loss": 2.8852,
      "step": 454
    },
    {
      "epoch": 0.06377014716187807,
      "grad_norm": 0.3499014675617218,
      "learning_rate": 0.00027,
      "loss": 2.8179,
      "step": 455
    },
    {
      "epoch": 0.06391030133146461,
      "grad_norm": 2.264272689819336,
      "learning_rate": 0.00027059999999999996,
      "loss": 2.9051,
      "step": 456
    },
    {
      "epoch": 0.06405045550105115,
      "grad_norm": 2.137887954711914,
      "learning_rate": 0.0002712,
      "loss": 2.8614,
      "step": 457
    },
    {
      "epoch": 0.06419060967063771,
      "grad_norm": 0.527977705001831,
      "learning_rate": 0.0002718,
      "loss": 2.8431,
      "step": 458
    },
    {
      "epoch": 0.06433076384022425,
      "grad_norm": 1.5349394083023071,
      "learning_rate": 0.0002724,
      "loss": 2.8297,
      "step": 459
    },
    {
      "epoch": 0.0644709180098108,
      "grad_norm": 0.48932817578315735,
      "learning_rate": 0.00027299999999999997,
      "loss": 2.8337,
      "step": 460
    },
    {
      "epoch": 0.06461107217939734,
      "grad_norm": 0.44389232993125916,
      "learning_rate": 0.0002736,
      "loss": 2.8659,
      "step": 461
    },
    {
      "epoch": 0.06475122634898388,
      "grad_norm": 0.4781602621078491,
      "learning_rate": 0.0002742,
      "loss": 2.8562,
      "step": 462
    },
    {
      "epoch": 0.06489138051857042,
      "grad_norm": 1.7393834590911865,
      "learning_rate": 0.0002748,
      "loss": 2.822,
      "step": 463
    },
    {
      "epoch": 0.06503153468815698,
      "grad_norm": 1.2854914665222168,
      "learning_rate": 0.00027539999999999997,
      "loss": 2.8046,
      "step": 464
    },
    {
      "epoch": 0.06517168885774352,
      "grad_norm": 0.983287513256073,
      "learning_rate": 0.000276,
      "loss": 2.8248,
      "step": 465
    },
    {
      "epoch": 0.06531184302733006,
      "grad_norm": 1.7185975313186646,
      "learning_rate": 0.0002766,
      "loss": 2.8546,
      "step": 466
    },
    {
      "epoch": 0.0654519971969166,
      "grad_norm": 2.0991129875183105,
      "learning_rate": 0.0002772,
      "loss": 2.843,
      "step": 467
    },
    {
      "epoch": 0.06559215136650315,
      "grad_norm": 1.4711074829101562,
      "learning_rate": 0.0002778,
      "loss": 2.8146,
      "step": 468
    },
    {
      "epoch": 0.0657323055360897,
      "grad_norm": 0.789509117603302,
      "learning_rate": 0.0002784,
      "loss": 2.8189,
      "step": 469
    },
    {
      "epoch": 0.06587245970567625,
      "grad_norm": 0.5332419872283936,
      "learning_rate": 0.000279,
      "loss": 2.8548,
      "step": 470
    },
    {
      "epoch": 0.06601261387526279,
      "grad_norm": 1.6290559768676758,
      "learning_rate": 0.00027959999999999997,
      "loss": 2.8416,
      "step": 471
    },
    {
      "epoch": 0.06615276804484933,
      "grad_norm": 1.8316640853881836,
      "learning_rate": 0.0002802,
      "loss": 2.8407,
      "step": 472
    },
    {
      "epoch": 0.06629292221443588,
      "grad_norm": 1.1368061304092407,
      "learning_rate": 0.0002808,
      "loss": 2.8067,
      "step": 473
    },
    {
      "epoch": 0.06643307638402242,
      "grad_norm": 1.3229907751083374,
      "learning_rate": 0.00028139999999999996,
      "loss": 2.8425,
      "step": 474
    },
    {
      "epoch": 0.06657323055360898,
      "grad_norm": 1.197740077972412,
      "learning_rate": 0.00028199999999999997,
      "loss": 2.8299,
      "step": 475
    },
    {
      "epoch": 0.06671338472319552,
      "grad_norm": 0.820569634437561,
      "learning_rate": 0.0002826,
      "loss": 2.8054,
      "step": 476
    },
    {
      "epoch": 0.06685353889278206,
      "grad_norm": 1.2852872610092163,
      "learning_rate": 0.00028319999999999994,
      "loss": 2.8526,
      "step": 477
    },
    {
      "epoch": 0.0669936930623686,
      "grad_norm": 0.5447834730148315,
      "learning_rate": 0.00028379999999999996,
      "loss": 2.8504,
      "step": 478
    },
    {
      "epoch": 0.06713384723195515,
      "grad_norm": 0.9654664397239685,
      "learning_rate": 0.0002844,
      "loss": 2.8271,
      "step": 479
    },
    {
      "epoch": 0.0672740014015417,
      "grad_norm": 1.7507197856903076,
      "learning_rate": 0.000285,
      "loss": 2.8422,
      "step": 480
    },
    {
      "epoch": 0.06741415557112825,
      "grad_norm": 1.2217130661010742,
      "learning_rate": 0.00028559999999999995,
      "loss": 2.8106,
      "step": 481
    },
    {
      "epoch": 0.06755430974071479,
      "grad_norm": 0.8571678400039673,
      "learning_rate": 0.00028619999999999996,
      "loss": 2.8634,
      "step": 482
    },
    {
      "epoch": 0.06769446391030133,
      "grad_norm": 1.1033737659454346,
      "learning_rate": 0.0002868,
      "loss": 2.8464,
      "step": 483
    },
    {
      "epoch": 0.06783461807988787,
      "grad_norm": 1.261634349822998,
      "learning_rate": 0.00028739999999999994,
      "loss": 2.8423,
      "step": 484
    },
    {
      "epoch": 0.06797477224947442,
      "grad_norm": 0.5701166391372681,
      "learning_rate": 0.00028799999999999995,
      "loss": 2.8253,
      "step": 485
    },
    {
      "epoch": 0.06811492641906097,
      "grad_norm": 0.7327373623847961,
      "learning_rate": 0.00028859999999999997,
      "loss": 2.8154,
      "step": 486
    },
    {
      "epoch": 0.06825508058864752,
      "grad_norm": 1.3246729373931885,
      "learning_rate": 0.0002892,
      "loss": 2.8537,
      "step": 487
    },
    {
      "epoch": 0.06839523475823406,
      "grad_norm": 0.9053966403007507,
      "learning_rate": 0.00028979999999999994,
      "loss": 2.8544,
      "step": 488
    },
    {
      "epoch": 0.0685353889278206,
      "grad_norm": 0.7421877980232239,
      "learning_rate": 0.00029039999999999996,
      "loss": 2.8277,
      "step": 489
    },
    {
      "epoch": 0.06867554309740714,
      "grad_norm": 0.4885466992855072,
      "learning_rate": 0.00029099999999999997,
      "loss": 2.8848,
      "step": 490
    },
    {
      "epoch": 0.06881569726699369,
      "grad_norm": 0.458811491727829,
      "learning_rate": 0.0002916,
      "loss": 2.8791,
      "step": 491
    },
    {
      "epoch": 0.06895585143658024,
      "grad_norm": 1.3304370641708374,
      "learning_rate": 0.00029219999999999995,
      "loss": 2.887,
      "step": 492
    },
    {
      "epoch": 0.06909600560616679,
      "grad_norm": 1.2078003883361816,
      "learning_rate": 0.00029279999999999996,
      "loss": 2.8148,
      "step": 493
    },
    {
      "epoch": 0.06923615977575333,
      "grad_norm": 1.063692569732666,
      "learning_rate": 0.0002934,
      "loss": 2.8957,
      "step": 494
    },
    {
      "epoch": 0.06937631394533987,
      "grad_norm": 0.6019052267074585,
      "learning_rate": 0.000294,
      "loss": 2.8422,
      "step": 495
    },
    {
      "epoch": 0.06951646811492641,
      "grad_norm": 1.3142274618148804,
      "learning_rate": 0.00029459999999999995,
      "loss": 2.8856,
      "step": 496
    },
    {
      "epoch": 0.06965662228451297,
      "grad_norm": 2.929884195327759,
      "learning_rate": 0.00029519999999999997,
      "loss": 2.8955,
      "step": 497
    },
    {
      "epoch": 0.06979677645409951,
      "grad_norm": 0.9539110660552979,
      "learning_rate": 0.0002958,
      "loss": 2.8155,
      "step": 498
    },
    {
      "epoch": 0.06993693062368606,
      "grad_norm": 3.578932762145996,
      "learning_rate": 0.0002964,
      "loss": 2.9668,
      "step": 499
    },
    {
      "epoch": 0.0700770847932726,
      "grad_norm": 2.2652204036712646,
      "learning_rate": 0.00029699999999999996,
      "loss": 2.847,
      "step": 500
    },
    {
      "epoch": 0.07021723896285914,
      "grad_norm": 0.4257313907146454,
      "learning_rate": 0.00029759999999999997,
      "loss": 2.854,
      "step": 501
    },
    {
      "epoch": 0.07035739313244568,
      "grad_norm": 0.9282848834991455,
      "learning_rate": 0.0002982,
      "loss": 2.8182,
      "step": 502
    },
    {
      "epoch": 0.07049754730203224,
      "grad_norm": 0.6917478442192078,
      "learning_rate": 0.0002988,
      "loss": 2.82,
      "step": 503
    },
    {
      "epoch": 0.07063770147161878,
      "grad_norm": 1.0468342304229736,
      "learning_rate": 0.00029939999999999996,
      "loss": 2.859,
      "step": 504
    },
    {
      "epoch": 0.07077785564120533,
      "grad_norm": 0.547046959400177,
      "learning_rate": 0.0003,
      "loss": 2.7904,
      "step": 505
    },
    {
      "epoch": 0.07091800981079187,
      "grad_norm": 2.35677433013916,
      "learning_rate": 0.0002999856493661803,
      "loss": 2.8388,
      "step": 506
    },
    {
      "epoch": 0.07105816398037841,
      "grad_norm": 0.5450826287269592,
      "learning_rate": 0.0002999712987323607,
      "loss": 2.8602,
      "step": 507
    },
    {
      "epoch": 0.07119831814996497,
      "grad_norm": 1.2112269401550293,
      "learning_rate": 0.000299956948098541,
      "loss": 2.8173,
      "step": 508
    },
    {
      "epoch": 0.07133847231955151,
      "grad_norm": 0.8833276033401489,
      "learning_rate": 0.00029994259746472134,
      "loss": 2.8481,
      "step": 509
    },
    {
      "epoch": 0.07147862648913805,
      "grad_norm": 0.4758220314979553,
      "learning_rate": 0.00029992824683090167,
      "loss": 2.8409,
      "step": 510
    },
    {
      "epoch": 0.0716187806587246,
      "grad_norm": 1.2521374225616455,
      "learning_rate": 0.000299913896197082,
      "loss": 2.8397,
      "step": 511
    },
    {
      "epoch": 0.07175893482831114,
      "grad_norm": 1.1220290660858154,
      "learning_rate": 0.00029989954556326233,
      "loss": 2.813,
      "step": 512
    },
    {
      "epoch": 0.07189908899789768,
      "grad_norm": 2.468276262283325,
      "learning_rate": 0.0002998851949294427,
      "loss": 2.8517,
      "step": 513
    },
    {
      "epoch": 0.07203924316748424,
      "grad_norm": 2.1448941230773926,
      "learning_rate": 0.00029987084429562304,
      "loss": 2.8776,
      "step": 514
    },
    {
      "epoch": 0.07217939733707078,
      "grad_norm": 0.7374310493469238,
      "learning_rate": 0.00029985649366180337,
      "loss": 2.8523,
      "step": 515
    },
    {
      "epoch": 0.07231955150665732,
      "grad_norm": 1.072592854499817,
      "learning_rate": 0.0002998421430279837,
      "loss": 2.874,
      "step": 516
    },
    {
      "epoch": 0.07245970567624387,
      "grad_norm": 1.4461050033569336,
      "learning_rate": 0.0002998277923941641,
      "loss": 2.851,
      "step": 517
    },
    {
      "epoch": 0.07259985984583041,
      "grad_norm": 1.34160578250885,
      "learning_rate": 0.0002998134417603444,
      "loss": 2.8605,
      "step": 518
    },
    {
      "epoch": 0.07274001401541696,
      "grad_norm": 2.1639254093170166,
      "learning_rate": 0.00029979909112652474,
      "loss": 2.8328,
      "step": 519
    },
    {
      "epoch": 0.07288016818500351,
      "grad_norm": 0.6973652243614197,
      "learning_rate": 0.00029978474049270506,
      "loss": 2.8316,
      "step": 520
    },
    {
      "epoch": 0.07302032235459005,
      "grad_norm": 0.37315288186073303,
      "learning_rate": 0.0002997703898588854,
      "loss": 2.8319,
      "step": 521
    },
    {
      "epoch": 0.0731604765241766,
      "grad_norm": 0.37722694873809814,
      "learning_rate": 0.0002997560392250657,
      "loss": 2.8682,
      "step": 522
    },
    {
      "epoch": 0.07330063069376314,
      "grad_norm": 1.1548172235488892,
      "learning_rate": 0.0002997416885912461,
      "loss": 2.8356,
      "step": 523
    },
    {
      "epoch": 0.07344078486334968,
      "grad_norm": 2.373615264892578,
      "learning_rate": 0.00029972733795742643,
      "loss": 2.8617,
      "step": 524
    },
    {
      "epoch": 0.07358093903293624,
      "grad_norm": 1.537207007408142,
      "learning_rate": 0.00029971298732360676,
      "loss": 2.8203,
      "step": 525
    },
    {
      "epoch": 0.07372109320252278,
      "grad_norm": 0.3691214621067047,
      "learning_rate": 0.00029969863668978714,
      "loss": 2.8562,
      "step": 526
    },
    {
      "epoch": 0.07386124737210932,
      "grad_norm": 0.6084275841712952,
      "learning_rate": 0.0002996842860559675,
      "loss": 2.7979,
      "step": 527
    },
    {
      "epoch": 0.07400140154169586,
      "grad_norm": 2.023315668106079,
      "learning_rate": 0.0002996699354221478,
      "loss": 2.8636,
      "step": 528
    },
    {
      "epoch": 0.0741415557112824,
      "grad_norm": 1.2384412288665771,
      "learning_rate": 0.00029965558478832813,
      "loss": 2.8563,
      "step": 529
    },
    {
      "epoch": 0.07428170988086896,
      "grad_norm": 0.9338071942329407,
      "learning_rate": 0.00029964123415450846,
      "loss": 2.8837,
      "step": 530
    },
    {
      "epoch": 0.0744218640504555,
      "grad_norm": 1.2200719118118286,
      "learning_rate": 0.0002996268835206888,
      "loss": 2.8741,
      "step": 531
    },
    {
      "epoch": 0.07456201822004205,
      "grad_norm": 1.0358259677886963,
      "learning_rate": 0.0002996125328868691,
      "loss": 2.8885,
      "step": 532
    },
    {
      "epoch": 0.07470217238962859,
      "grad_norm": 0.7400373816490173,
      "learning_rate": 0.0002995981822530495,
      "loss": 2.8382,
      "step": 533
    },
    {
      "epoch": 0.07484232655921513,
      "grad_norm": 2.8748083114624023,
      "learning_rate": 0.00029958383161922983,
      "loss": 2.8345,
      "step": 534
    },
    {
      "epoch": 0.07498248072880168,
      "grad_norm": 2.020559787750244,
      "learning_rate": 0.00029956948098541016,
      "loss": 2.885,
      "step": 535
    },
    {
      "epoch": 0.07512263489838823,
      "grad_norm": 3.0204548835754395,
      "learning_rate": 0.0002995551303515905,
      "loss": 2.8575,
      "step": 536
    },
    {
      "epoch": 0.07526278906797478,
      "grad_norm": 2.7605319023132324,
      "learning_rate": 0.00029954077971777087,
      "loss": 2.885,
      "step": 537
    },
    {
      "epoch": 0.07540294323756132,
      "grad_norm": 1.4202461242675781,
      "learning_rate": 0.0002995264290839512,
      "loss": 2.8707,
      "step": 538
    },
    {
      "epoch": 0.07554309740714786,
      "grad_norm": 0.6885740756988525,
      "learning_rate": 0.0002995120784501315,
      "loss": 2.8068,
      "step": 539
    },
    {
      "epoch": 0.0756832515767344,
      "grad_norm": 2.3399600982666016,
      "learning_rate": 0.00029949772781631185,
      "loss": 2.8882,
      "step": 540
    },
    {
      "epoch": 0.07582340574632096,
      "grad_norm": 1.3935046195983887,
      "learning_rate": 0.0002994833771824922,
      "loss": 2.8035,
      "step": 541
    },
    {
      "epoch": 0.0759635599159075,
      "grad_norm": 1.8794046640396118,
      "learning_rate": 0.00029946902654867256,
      "loss": 2.8572,
      "step": 542
    },
    {
      "epoch": 0.07610371408549405,
      "grad_norm": 2.469773769378662,
      "learning_rate": 0.0002994546759148529,
      "loss": 2.8994,
      "step": 543
    },
    {
      "epoch": 0.07624386825508059,
      "grad_norm": 1.048001766204834,
      "learning_rate": 0.0002994403252810332,
      "loss": 2.8137,
      "step": 544
    },
    {
      "epoch": 0.07638402242466713,
      "grad_norm": 0.8849368691444397,
      "learning_rate": 0.00029942597464721355,
      "loss": 2.8401,
      "step": 545
    },
    {
      "epoch": 0.07652417659425367,
      "grad_norm": 1.7072055339813232,
      "learning_rate": 0.0002994116240133939,
      "loss": 2.8557,
      "step": 546
    },
    {
      "epoch": 0.07666433076384023,
      "grad_norm": 1.740373134613037,
      "learning_rate": 0.0002993972733795742,
      "loss": 2.8559,
      "step": 547
    },
    {
      "epoch": 0.07680448493342677,
      "grad_norm": 1.8785969018936157,
      "learning_rate": 0.0002993829227457546,
      "loss": 2.8995,
      "step": 548
    },
    {
      "epoch": 0.07694463910301332,
      "grad_norm": 1.5170527696609497,
      "learning_rate": 0.0002993685721119349,
      "loss": 2.8803,
      "step": 549
    },
    {
      "epoch": 0.07708479327259986,
      "grad_norm": 1.2071936130523682,
      "learning_rate": 0.00029935422147811525,
      "loss": 2.8162,
      "step": 550
    },
    {
      "epoch": 0.0772249474421864,
      "grad_norm": 2.1247119903564453,
      "learning_rate": 0.0002993398708442956,
      "loss": 2.8225,
      "step": 551
    },
    {
      "epoch": 0.07736510161177294,
      "grad_norm": 0.5597912073135376,
      "learning_rate": 0.00029932552021047596,
      "loss": 2.8832,
      "step": 552
    },
    {
      "epoch": 0.0775052557813595,
      "grad_norm": 0.9041293859481812,
      "learning_rate": 0.0002993111695766563,
      "loss": 2.789,
      "step": 553
    },
    {
      "epoch": 0.07764540995094604,
      "grad_norm": 0.4143102467060089,
      "learning_rate": 0.0002992968189428366,
      "loss": 2.8113,
      "step": 554
    },
    {
      "epoch": 0.07778556412053259,
      "grad_norm": 0.9289295673370361,
      "learning_rate": 0.00029928246830901694,
      "loss": 2.8364,
      "step": 555
    },
    {
      "epoch": 0.07792571829011913,
      "grad_norm": 0.8136385083198547,
      "learning_rate": 0.00029926811767519727,
      "loss": 2.791,
      "step": 556
    },
    {
      "epoch": 0.07806587245970567,
      "grad_norm": 3.271901845932007,
      "learning_rate": 0.0002992537670413776,
      "loss": 2.9064,
      "step": 557
    },
    {
      "epoch": 0.07820602662929223,
      "grad_norm": 2.2082877159118652,
      "learning_rate": 0.000299239416407558,
      "loss": 2.8753,
      "step": 558
    },
    {
      "epoch": 0.07834618079887877,
      "grad_norm": 0.38986533880233765,
      "learning_rate": 0.0002992250657737383,
      "loss": 2.8168,
      "step": 559
    },
    {
      "epoch": 0.07848633496846531,
      "grad_norm": 1.8162808418273926,
      "learning_rate": 0.00029921071513991864,
      "loss": 2.8152,
      "step": 560
    },
    {
      "epoch": 0.07862648913805186,
      "grad_norm": 0.8369203209877014,
      "learning_rate": 0.000299196364506099,
      "loss": 2.8359,
      "step": 561
    },
    {
      "epoch": 0.0787666433076384,
      "grad_norm": 0.4745757281780243,
      "learning_rate": 0.00029918201387227935,
      "loss": 2.8316,
      "step": 562
    },
    {
      "epoch": 0.07890679747722494,
      "grad_norm": 1.5335533618927002,
      "learning_rate": 0.0002991676632384597,
      "loss": 2.8297,
      "step": 563
    },
    {
      "epoch": 0.0790469516468115,
      "grad_norm": 1.0716702938079834,
      "learning_rate": 0.00029915331260464,
      "loss": 2.8689,
      "step": 564
    },
    {
      "epoch": 0.07918710581639804,
      "grad_norm": 1.863500952720642,
      "learning_rate": 0.00029913896197082034,
      "loss": 2.8462,
      "step": 565
    },
    {
      "epoch": 0.07932725998598458,
      "grad_norm": 1.3357001543045044,
      "learning_rate": 0.00029912461133700067,
      "loss": 2.7989,
      "step": 566
    },
    {
      "epoch": 0.07946741415557113,
      "grad_norm": 0.7774486541748047,
      "learning_rate": 0.000299110260703181,
      "loss": 2.8174,
      "step": 567
    },
    {
      "epoch": 0.07960756832515767,
      "grad_norm": 0.6659305691719055,
      "learning_rate": 0.0002990959100693614,
      "loss": 2.8042,
      "step": 568
    },
    {
      "epoch": 0.07974772249474422,
      "grad_norm": 0.5252679586410522,
      "learning_rate": 0.0002990815594355417,
      "loss": 2.8254,
      "step": 569
    },
    {
      "epoch": 0.07988787666433077,
      "grad_norm": 1.5993984937667847,
      "learning_rate": 0.00029906720880172203,
      "loss": 2.8631,
      "step": 570
    },
    {
      "epoch": 0.08002803083391731,
      "grad_norm": 1.2399847507476807,
      "learning_rate": 0.0002990528581679024,
      "loss": 2.8652,
      "step": 571
    },
    {
      "epoch": 0.08016818500350385,
      "grad_norm": 1.3382014036178589,
      "learning_rate": 0.00029903850753408275,
      "loss": 2.8044,
      "step": 572
    },
    {
      "epoch": 0.0803083391730904,
      "grad_norm": 0.36626508831977844,
      "learning_rate": 0.0002990241569002631,
      "loss": 2.8503,
      "step": 573
    },
    {
      "epoch": 0.08044849334267694,
      "grad_norm": 0.5636441707611084,
      "learning_rate": 0.0002990098062664434,
      "loss": 2.8446,
      "step": 574
    },
    {
      "epoch": 0.0805886475122635,
      "grad_norm": 0.975190281867981,
      "learning_rate": 0.00029899545563262373,
      "loss": 2.8376,
      "step": 575
    },
    {
      "epoch": 0.08072880168185004,
      "grad_norm": 1.7194899320602417,
      "learning_rate": 0.00029898110499880406,
      "loss": 2.876,
      "step": 576
    },
    {
      "epoch": 0.08086895585143658,
      "grad_norm": 1.046378254890442,
      "learning_rate": 0.00029896675436498444,
      "loss": 2.8959,
      "step": 577
    },
    {
      "epoch": 0.08100911002102312,
      "grad_norm": 0.40222662687301636,
      "learning_rate": 0.00029895240373116477,
      "loss": 2.8193,
      "step": 578
    },
    {
      "epoch": 0.08114926419060967,
      "grad_norm": 1.0115432739257812,
      "learning_rate": 0.0002989380530973451,
      "loss": 2.8054,
      "step": 579
    },
    {
      "epoch": 0.08128941836019622,
      "grad_norm": 0.8710485696792603,
      "learning_rate": 0.0002989237024635255,
      "loss": 2.8417,
      "step": 580
    },
    {
      "epoch": 0.08142957252978276,
      "grad_norm": 0.5470033288002014,
      "learning_rate": 0.0002989093518297058,
      "loss": 2.7937,
      "step": 581
    },
    {
      "epoch": 0.08156972669936931,
      "grad_norm": 2.266864776611328,
      "learning_rate": 0.00029889500119588614,
      "loss": 2.8503,
      "step": 582
    },
    {
      "epoch": 0.08170988086895585,
      "grad_norm": 0.5184596180915833,
      "learning_rate": 0.00029888065056206647,
      "loss": 2.8387,
      "step": 583
    },
    {
      "epoch": 0.08185003503854239,
      "grad_norm": 1.1281994581222534,
      "learning_rate": 0.0002988662999282468,
      "loss": 2.8761,
      "step": 584
    },
    {
      "epoch": 0.08199018920812894,
      "grad_norm": 0.36221277713775635,
      "learning_rate": 0.0002988519492944271,
      "loss": 2.8019,
      "step": 585
    },
    {
      "epoch": 0.08213034337771549,
      "grad_norm": 0.40339556336402893,
      "learning_rate": 0.00029883759866060745,
      "loss": 2.8583,
      "step": 586
    },
    {
      "epoch": 0.08227049754730204,
      "grad_norm": 0.8000438809394836,
      "learning_rate": 0.00029882324802678784,
      "loss": 2.8069,
      "step": 587
    },
    {
      "epoch": 0.08241065171688858,
      "grad_norm": 0.7097018361091614,
      "learning_rate": 0.00029880889739296817,
      "loss": 2.8529,
      "step": 588
    },
    {
      "epoch": 0.08255080588647512,
      "grad_norm": 0.8318480849266052,
      "learning_rate": 0.0002987945467591485,
      "loss": 2.8447,
      "step": 589
    },
    {
      "epoch": 0.08269096005606166,
      "grad_norm": 0.8203009366989136,
      "learning_rate": 0.0002987801961253289,
      "loss": 2.8843,
      "step": 590
    },
    {
      "epoch": 0.08283111422564822,
      "grad_norm": 0.8691034317016602,
      "learning_rate": 0.0002987658454915092,
      "loss": 2.8418,
      "step": 591
    },
    {
      "epoch": 0.08297126839523476,
      "grad_norm": 0.8345413208007812,
      "learning_rate": 0.00029875149485768953,
      "loss": 2.8885,
      "step": 592
    },
    {
      "epoch": 0.0831114225648213,
      "grad_norm": 0.5269362330436707,
      "learning_rate": 0.00029873714422386986,
      "loss": 2.8721,
      "step": 593
    },
    {
      "epoch": 0.08325157673440785,
      "grad_norm": 0.8924299478530884,
      "learning_rate": 0.0002987227935900502,
      "loss": 2.8992,
      "step": 594
    },
    {
      "epoch": 0.08339173090399439,
      "grad_norm": 0.5429219007492065,
      "learning_rate": 0.0002987084429562305,
      "loss": 2.8454,
      "step": 595
    },
    {
      "epoch": 0.08353188507358093,
      "grad_norm": 0.7319108843803406,
      "learning_rate": 0.0002986940923224109,
      "loss": 2.8692,
      "step": 596
    },
    {
      "epoch": 0.08367203924316749,
      "grad_norm": 1.0079227685928345,
      "learning_rate": 0.00029867974168859123,
      "loss": 2.9254,
      "step": 597
    },
    {
      "epoch": 0.08381219341275403,
      "grad_norm": 1.6758877038955688,
      "learning_rate": 0.00029866539105477156,
      "loss": 2.8259,
      "step": 598
    },
    {
      "epoch": 0.08395234758234058,
      "grad_norm": 1.7681196928024292,
      "learning_rate": 0.0002986510404209519,
      "loss": 2.865,
      "step": 599
    },
    {
      "epoch": 0.08409250175192712,
      "grad_norm": 1.3744769096374512,
      "learning_rate": 0.0002986366897871322,
      "loss": 2.8459,
      "step": 600
    },
    {
      "epoch": 0.08423265592151366,
      "grad_norm": 0.495762437582016,
      "learning_rate": 0.0002986223391533126,
      "loss": 2.8304,
      "step": 601
    },
    {
      "epoch": 0.08437281009110022,
      "grad_norm": 1.6722474098205566,
      "learning_rate": 0.00029860798851949293,
      "loss": 2.8219,
      "step": 602
    },
    {
      "epoch": 0.08451296426068676,
      "grad_norm": 0.4749285876750946,
      "learning_rate": 0.00029859363788567326,
      "loss": 2.8286,
      "step": 603
    },
    {
      "epoch": 0.0846531184302733,
      "grad_norm": 0.6313146948814392,
      "learning_rate": 0.0002985792872518536,
      "loss": 2.8612,
      "step": 604
    },
    {
      "epoch": 0.08479327259985985,
      "grad_norm": 0.769989013671875,
      "learning_rate": 0.0002985649366180339,
      "loss": 2.8284,
      "step": 605
    },
    {
      "epoch": 0.08493342676944639,
      "grad_norm": 0.5903698205947876,
      "learning_rate": 0.0002985505859842143,
      "loss": 2.8803,
      "step": 606
    },
    {
      "epoch": 0.08507358093903293,
      "grad_norm": 0.622111976146698,
      "learning_rate": 0.0002985362353503946,
      "loss": 2.8348,
      "step": 607
    },
    {
      "epoch": 0.08521373510861949,
      "grad_norm": 0.9654092192649841,
      "learning_rate": 0.00029852188471657495,
      "loss": 2.8219,
      "step": 608
    },
    {
      "epoch": 0.08535388927820603,
      "grad_norm": 0.739698052406311,
      "learning_rate": 0.0002985075340827553,
      "loss": 2.8007,
      "step": 609
    },
    {
      "epoch": 0.08549404344779257,
      "grad_norm": 0.7439742088317871,
      "learning_rate": 0.0002984931834489356,
      "loss": 2.8283,
      "step": 610
    },
    {
      "epoch": 0.08563419761737912,
      "grad_norm": 0.5097655653953552,
      "learning_rate": 0.00029847883281511594,
      "loss": 2.8618,
      "step": 611
    },
    {
      "epoch": 0.08577435178696566,
      "grad_norm": 0.557064950466156,
      "learning_rate": 0.0002984644821812963,
      "loss": 2.8195,
      "step": 612
    },
    {
      "epoch": 0.0859145059565522,
      "grad_norm": 0.40500202775001526,
      "learning_rate": 0.00029845013154747665,
      "loss": 2.818,
      "step": 613
    },
    {
      "epoch": 0.08605466012613876,
      "grad_norm": 1.3171998262405396,
      "learning_rate": 0.000298435780913657,
      "loss": 2.835,
      "step": 614
    },
    {
      "epoch": 0.0861948142957253,
      "grad_norm": 0.8443118929862976,
      "learning_rate": 0.00029842143027983736,
      "loss": 2.838,
      "step": 615
    },
    {
      "epoch": 0.08633496846531184,
      "grad_norm": 0.5561873912811279,
      "learning_rate": 0.0002984070796460177,
      "loss": 2.8168,
      "step": 616
    },
    {
      "epoch": 0.08647512263489839,
      "grad_norm": 1.3321802616119385,
      "learning_rate": 0.000298392729012198,
      "loss": 2.7918,
      "step": 617
    },
    {
      "epoch": 0.08661527680448493,
      "grad_norm": 0.6840429306030273,
      "learning_rate": 0.00029837837837837835,
      "loss": 2.8632,
      "step": 618
    },
    {
      "epoch": 0.08675543097407148,
      "grad_norm": 0.5101912021636963,
      "learning_rate": 0.0002983640277445587,
      "loss": 2.8235,
      "step": 619
    },
    {
      "epoch": 0.08689558514365803,
      "grad_norm": 0.330929160118103,
      "learning_rate": 0.000298349677110739,
      "loss": 2.8051,
      "step": 620
    },
    {
      "epoch": 0.08703573931324457,
      "grad_norm": 1.2912763357162476,
      "learning_rate": 0.00029833532647691933,
      "loss": 2.8092,
      "step": 621
    },
    {
      "epoch": 0.08717589348283111,
      "grad_norm": 0.45430678129196167,
      "learning_rate": 0.0002983209758430997,
      "loss": 2.8301,
      "step": 622
    },
    {
      "epoch": 0.08731604765241766,
      "grad_norm": 1.664919137954712,
      "learning_rate": 0.00029830662520928004,
      "loss": 2.8641,
      "step": 623
    },
    {
      "epoch": 0.0874562018220042,
      "grad_norm": 0.5690026879310608,
      "learning_rate": 0.0002982922745754604,
      "loss": 2.8253,
      "step": 624
    },
    {
      "epoch": 0.08759635599159075,
      "grad_norm": 0.49163636565208435,
      "learning_rate": 0.00029827792394164076,
      "loss": 2.7965,
      "step": 625
    },
    {
      "epoch": 0.0877365101611773,
      "grad_norm": 0.5771698951721191,
      "learning_rate": 0.0002982635733078211,
      "loss": 2.811,
      "step": 626
    },
    {
      "epoch": 0.08787666433076384,
      "grad_norm": 1.500550389289856,
      "learning_rate": 0.0002982492226740014,
      "loss": 2.8566,
      "step": 627
    },
    {
      "epoch": 0.08801681850035038,
      "grad_norm": 1.1520962715148926,
      "learning_rate": 0.00029823487204018174,
      "loss": 2.8206,
      "step": 628
    },
    {
      "epoch": 0.08815697266993693,
      "grad_norm": 0.9202871322631836,
      "learning_rate": 0.00029822052140636207,
      "loss": 2.8741,
      "step": 629
    },
    {
      "epoch": 0.08829712683952348,
      "grad_norm": 0.7967159152030945,
      "learning_rate": 0.0002982061707725424,
      "loss": 2.8376,
      "step": 630
    },
    {
      "epoch": 0.08843728100911002,
      "grad_norm": 1.4931164979934692,
      "learning_rate": 0.0002981918201387228,
      "loss": 2.8284,
      "step": 631
    },
    {
      "epoch": 0.08857743517869657,
      "grad_norm": 0.48530471324920654,
      "learning_rate": 0.0002981774695049031,
      "loss": 2.8405,
      "step": 632
    },
    {
      "epoch": 0.08871758934828311,
      "grad_norm": 0.8628910183906555,
      "learning_rate": 0.00029816311887108344,
      "loss": 2.8141,
      "step": 633
    },
    {
      "epoch": 0.08885774351786965,
      "grad_norm": 0.8304014205932617,
      "learning_rate": 0.0002981487682372638,
      "loss": 2.8201,
      "step": 634
    },
    {
      "epoch": 0.0889978976874562,
      "grad_norm": 0.3863701820373535,
      "learning_rate": 0.00029813441760344415,
      "loss": 2.835,
      "step": 635
    },
    {
      "epoch": 0.08913805185704275,
      "grad_norm": 0.5625427961349487,
      "learning_rate": 0.0002981200669696245,
      "loss": 2.8139,
      "step": 636
    },
    {
      "epoch": 0.0892782060266293,
      "grad_norm": 0.5027657151222229,
      "learning_rate": 0.0002981057163358048,
      "loss": 2.772,
      "step": 637
    },
    {
      "epoch": 0.08941836019621584,
      "grad_norm": 1.7406222820281982,
      "learning_rate": 0.00029809136570198514,
      "loss": 2.8658,
      "step": 638
    },
    {
      "epoch": 0.08955851436580238,
      "grad_norm": 1.1116559505462646,
      "learning_rate": 0.00029807701506816546,
      "loss": 2.7864,
      "step": 639
    },
    {
      "epoch": 0.08969866853538892,
      "grad_norm": 0.6801770329475403,
      "learning_rate": 0.0002980626644343458,
      "loss": 2.8095,
      "step": 640
    },
    {
      "epoch": 0.08983882270497548,
      "grad_norm": 1.4479053020477295,
      "learning_rate": 0.0002980483138005262,
      "loss": 2.8471,
      "step": 641
    },
    {
      "epoch": 0.08997897687456202,
      "grad_norm": 0.5120633244514465,
      "learning_rate": 0.0002980339631667065,
      "loss": 2.8464,
      "step": 642
    },
    {
      "epoch": 0.09011913104414856,
      "grad_norm": 0.45446351170539856,
      "learning_rate": 0.00029801961253288683,
      "loss": 2.7928,
      "step": 643
    },
    {
      "epoch": 0.09025928521373511,
      "grad_norm": 0.9317551851272583,
      "learning_rate": 0.0002980052618990672,
      "loss": 2.8274,
      "step": 644
    },
    {
      "epoch": 0.09039943938332165,
      "grad_norm": 0.6883323192596436,
      "learning_rate": 0.00029799091126524754,
      "loss": 2.8461,
      "step": 645
    },
    {
      "epoch": 0.09053959355290819,
      "grad_norm": 0.7593468427658081,
      "learning_rate": 0.00029797656063142787,
      "loss": 2.7726,
      "step": 646
    },
    {
      "epoch": 0.09067974772249475,
      "grad_norm": 0.7465790510177612,
      "learning_rate": 0.0002979622099976082,
      "loss": 2.7725,
      "step": 647
    },
    {
      "epoch": 0.09081990189208129,
      "grad_norm": 1.7886598110198975,
      "learning_rate": 0.00029794785936378853,
      "loss": 2.8947,
      "step": 648
    },
    {
      "epoch": 0.09096005606166783,
      "grad_norm": 0.6924257874488831,
      "learning_rate": 0.00029793350872996886,
      "loss": 2.8607,
      "step": 649
    },
    {
      "epoch": 0.09110021023125438,
      "grad_norm": 1.339035987854004,
      "learning_rate": 0.00029791915809614924,
      "loss": 2.7822,
      "step": 650
    },
    {
      "epoch": 0.09124036440084092,
      "grad_norm": 0.43366163969039917,
      "learning_rate": 0.00029790480746232957,
      "loss": 2.7786,
      "step": 651
    },
    {
      "epoch": 0.09138051857042748,
      "grad_norm": 1.2496447563171387,
      "learning_rate": 0.0002978904568285099,
      "loss": 2.8206,
      "step": 652
    },
    {
      "epoch": 0.09152067274001402,
      "grad_norm": 1.8453083038330078,
      "learning_rate": 0.0002978761061946902,
      "loss": 2.8424,
      "step": 653
    },
    {
      "epoch": 0.09166082690960056,
      "grad_norm": 1.6689263582229614,
      "learning_rate": 0.0002978617555608706,
      "loss": 2.7919,
      "step": 654
    },
    {
      "epoch": 0.0918009810791871,
      "grad_norm": 2.8990964889526367,
      "learning_rate": 0.00029784740492705094,
      "loss": 2.8228,
      "step": 655
    },
    {
      "epoch": 0.09194113524877365,
      "grad_norm": 2.011852741241455,
      "learning_rate": 0.00029783305429323127,
      "loss": 2.8267,
      "step": 656
    },
    {
      "epoch": 0.09208128941836019,
      "grad_norm": 0.7403373718261719,
      "learning_rate": 0.0002978187036594116,
      "loss": 2.7918,
      "step": 657
    },
    {
      "epoch": 0.09222144358794675,
      "grad_norm": 0.7828468680381775,
      "learning_rate": 0.0002978043530255919,
      "loss": 2.8375,
      "step": 658
    },
    {
      "epoch": 0.09236159775753329,
      "grad_norm": 0.8046102523803711,
      "learning_rate": 0.00029779000239177225,
      "loss": 2.8533,
      "step": 659
    },
    {
      "epoch": 0.09250175192711983,
      "grad_norm": 1.337938904762268,
      "learning_rate": 0.00029777565175795264,
      "loss": 2.7909,
      "step": 660
    },
    {
      "epoch": 0.09264190609670638,
      "grad_norm": 0.45919352769851685,
      "learning_rate": 0.00029776130112413296,
      "loss": 2.7858,
      "step": 661
    },
    {
      "epoch": 0.09278206026629292,
      "grad_norm": 0.9781817197799683,
      "learning_rate": 0.0002977469504903133,
      "loss": 2.7464,
      "step": 662
    },
    {
      "epoch": 0.09292221443587946,
      "grad_norm": 0.5097095370292664,
      "learning_rate": 0.0002977325998564936,
      "loss": 2.7552,
      "step": 663
    },
    {
      "epoch": 0.09306236860546602,
      "grad_norm": 0.9436579346656799,
      "learning_rate": 0.000297718249222674,
      "loss": 2.7777,
      "step": 664
    },
    {
      "epoch": 0.09320252277505256,
      "grad_norm": 1.041588306427002,
      "learning_rate": 0.00029770389858885433,
      "loss": 2.8027,
      "step": 665
    },
    {
      "epoch": 0.0933426769446391,
      "grad_norm": 2.2087807655334473,
      "learning_rate": 0.00029768954795503466,
      "loss": 2.7533,
      "step": 666
    },
    {
      "epoch": 0.09348283111422565,
      "grad_norm": 0.5946235656738281,
      "learning_rate": 0.000297675197321215,
      "loss": 2.7246,
      "step": 667
    },
    {
      "epoch": 0.09362298528381219,
      "grad_norm": 1.53168785572052,
      "learning_rate": 0.0002976608466873953,
      "loss": 2.7586,
      "step": 668
    },
    {
      "epoch": 0.09376313945339874,
      "grad_norm": 0.6736552119255066,
      "learning_rate": 0.0002976464960535757,
      "loss": 2.758,
      "step": 669
    },
    {
      "epoch": 0.09390329362298529,
      "grad_norm": 2.449899196624756,
      "learning_rate": 0.00029763214541975603,
      "loss": 2.7873,
      "step": 670
    },
    {
      "epoch": 0.09404344779257183,
      "grad_norm": 1.226711630821228,
      "learning_rate": 0.00029761779478593636,
      "loss": 2.7737,
      "step": 671
    },
    {
      "epoch": 0.09418360196215837,
      "grad_norm": 1.5849742889404297,
      "learning_rate": 0.0002976034441521167,
      "loss": 2.7496,
      "step": 672
    },
    {
      "epoch": 0.09432375613174492,
      "grad_norm": 1.3360331058502197,
      "learning_rate": 0.000297589093518297,
      "loss": 2.7938,
      "step": 673
    },
    {
      "epoch": 0.09446391030133146,
      "grad_norm": 0.40812867879867554,
      "learning_rate": 0.00029757474288447734,
      "loss": 2.735,
      "step": 674
    },
    {
      "epoch": 0.09460406447091801,
      "grad_norm": 2.3891429901123047,
      "learning_rate": 0.00029756039225065767,
      "loss": 2.7499,
      "step": 675
    },
    {
      "epoch": 0.09474421864050456,
      "grad_norm": 1.6537995338439941,
      "learning_rate": 0.00029754604161683805,
      "loss": 2.732,
      "step": 676
    },
    {
      "epoch": 0.0948843728100911,
      "grad_norm": 0.6037951111793518,
      "learning_rate": 0.0002975316909830184,
      "loss": 2.7448,
      "step": 677
    },
    {
      "epoch": 0.09502452697967764,
      "grad_norm": 2.345980644226074,
      "learning_rate": 0.0002975173403491987,
      "loss": 2.7403,
      "step": 678
    },
    {
      "epoch": 0.09516468114926419,
      "grad_norm": 1.643855094909668,
      "learning_rate": 0.0002975029897153791,
      "loss": 2.7589,
      "step": 679
    },
    {
      "epoch": 0.09530483531885074,
      "grad_norm": 1.035865306854248,
      "learning_rate": 0.0002974886390815594,
      "loss": 2.7081,
      "step": 680
    },
    {
      "epoch": 0.09544498948843728,
      "grad_norm": 0.4680897891521454,
      "learning_rate": 0.00029747428844773975,
      "loss": 2.6718,
      "step": 681
    },
    {
      "epoch": 0.09558514365802383,
      "grad_norm": 1.733031153678894,
      "learning_rate": 0.0002974599378139201,
      "loss": 2.703,
      "step": 682
    },
    {
      "epoch": 0.09572529782761037,
      "grad_norm": 0.830502986907959,
      "learning_rate": 0.0002974455871801004,
      "loss": 2.645,
      "step": 683
    },
    {
      "epoch": 0.09586545199719691,
      "grad_norm": 0.5354193449020386,
      "learning_rate": 0.00029743123654628074,
      "loss": 2.6269,
      "step": 684
    },
    {
      "epoch": 0.09600560616678346,
      "grad_norm": 0.7488672733306885,
      "learning_rate": 0.0002974168859124611,
      "loss": 2.6441,
      "step": 685
    },
    {
      "epoch": 0.09614576033637001,
      "grad_norm": 0.8250200152397156,
      "learning_rate": 0.00029740253527864145,
      "loss": 2.6969,
      "step": 686
    },
    {
      "epoch": 0.09628591450595655,
      "grad_norm": 3.3549208641052246,
      "learning_rate": 0.0002973881846448218,
      "loss": 2.767,
      "step": 687
    },
    {
      "epoch": 0.0964260686755431,
      "grad_norm": 1.539199709892273,
      "learning_rate": 0.0002973738340110021,
      "loss": 2.6454,
      "step": 688
    },
    {
      "epoch": 0.09656622284512964,
      "grad_norm": 1.9080121517181396,
      "learning_rate": 0.0002973594833771825,
      "loss": 2.6029,
      "step": 689
    },
    {
      "epoch": 0.09670637701471618,
      "grad_norm": 1.0665572881698608,
      "learning_rate": 0.0002973451327433628,
      "loss": 2.6352,
      "step": 690
    },
    {
      "epoch": 0.09684653118430274,
      "grad_norm": 1.162095308303833,
      "learning_rate": 0.00029733078210954315,
      "loss": 2.6054,
      "step": 691
    },
    {
      "epoch": 0.09698668535388928,
      "grad_norm": 0.8056411743164062,
      "learning_rate": 0.0002973164314757235,
      "loss": 2.5109,
      "step": 692
    },
    {
      "epoch": 0.09712683952347582,
      "grad_norm": 2.13920521736145,
      "learning_rate": 0.0002973020808419038,
      "loss": 2.5706,
      "step": 693
    },
    {
      "epoch": 0.09726699369306237,
      "grad_norm": 0.9956919550895691,
      "learning_rate": 0.00029728773020808413,
      "loss": 2.4241,
      "step": 694
    },
    {
      "epoch": 0.09740714786264891,
      "grad_norm": 1.7647011280059814,
      "learning_rate": 0.0002972733795742645,
      "loss": 2.553,
      "step": 695
    },
    {
      "epoch": 0.09754730203223545,
      "grad_norm": 1.1218814849853516,
      "learning_rate": 0.00029725902894044484,
      "loss": 2.4433,
      "step": 696
    },
    {
      "epoch": 0.09768745620182201,
      "grad_norm": 1.4899669885635376,
      "learning_rate": 0.00029724467830662517,
      "loss": 2.5247,
      "step": 697
    },
    {
      "epoch": 0.09782761037140855,
      "grad_norm": 0.9114047884941101,
      "learning_rate": 0.00029723032767280555,
      "loss": 2.4372,
      "step": 698
    },
    {
      "epoch": 0.0979677645409951,
      "grad_norm": 1.6940371990203857,
      "learning_rate": 0.0002972159770389859,
      "loss": 2.5265,
      "step": 699
    },
    {
      "epoch": 0.09810791871058164,
      "grad_norm": 1.4025936126708984,
      "learning_rate": 0.0002972016264051662,
      "loss": 2.4353,
      "step": 700
    },
    {
      "epoch": 0.09824807288016818,
      "grad_norm": 1.901639461517334,
      "learning_rate": 0.00029718727577134654,
      "loss": 2.3507,
      "step": 701
    },
    {
      "epoch": 0.09838822704975474,
      "grad_norm": 0.7116155028343201,
      "learning_rate": 0.00029717292513752687,
      "loss": 2.41,
      "step": 702
    },
    {
      "epoch": 0.09852838121934128,
      "grad_norm": 0.8871269226074219,
      "learning_rate": 0.0002971585745037072,
      "loss": 2.2023,
      "step": 703
    },
    {
      "epoch": 0.09866853538892782,
      "grad_norm": 0.8017802834510803,
      "learning_rate": 0.0002971442238698876,
      "loss": 2.2224,
      "step": 704
    },
    {
      "epoch": 0.09880868955851436,
      "grad_norm": 1.6170402765274048,
      "learning_rate": 0.0002971298732360679,
      "loss": 2.1752,
      "step": 705
    },
    {
      "epoch": 0.09894884372810091,
      "grad_norm": 1.2236593961715698,
      "learning_rate": 0.00029711552260224824,
      "loss": 2.2152,
      "step": 706
    },
    {
      "epoch": 0.09908899789768745,
      "grad_norm": 0.8129385113716125,
      "learning_rate": 0.00029710117196842857,
      "loss": 2.1591,
      "step": 707
    },
    {
      "epoch": 0.099229152067274,
      "grad_norm": 0.7264969944953918,
      "learning_rate": 0.00029708682133460895,
      "loss": 2.1175,
      "step": 708
    },
    {
      "epoch": 0.09936930623686055,
      "grad_norm": 1.3700792789459229,
      "learning_rate": 0.0002970724707007893,
      "loss": 2.0763,
      "step": 709
    },
    {
      "epoch": 0.09950946040644709,
      "grad_norm": 0.8801718950271606,
      "learning_rate": 0.0002970581200669696,
      "loss": 1.9882,
      "step": 710
    },
    {
      "epoch": 0.09964961457603363,
      "grad_norm": 1.5149402618408203,
      "learning_rate": 0.00029704376943314993,
      "loss": 1.9982,
      "step": 711
    },
    {
      "epoch": 0.09978976874562018,
      "grad_norm": 0.797324001789093,
      "learning_rate": 0.00029702941879933026,
      "loss": 1.8953,
      "step": 712
    },
    {
      "epoch": 0.09992992291520673,
      "grad_norm": 0.704853355884552,
      "learning_rate": 0.0002970150681655106,
      "loss": 1.8684,
      "step": 713
    },
    {
      "epoch": 0.10007007708479328,
      "grad_norm": 1.9356935024261475,
      "learning_rate": 0.000297000717531691,
      "loss": 1.8943,
      "step": 714
    },
    {
      "epoch": 0.10021023125437982,
      "grad_norm": 1.0025293827056885,
      "learning_rate": 0.0002969863668978713,
      "loss": 1.8833,
      "step": 715
    },
    {
      "epoch": 0.10035038542396636,
      "grad_norm": 0.7290270924568176,
      "learning_rate": 0.00029697201626405163,
      "loss": 1.7453,
      "step": 716
    },
    {
      "epoch": 0.1004905395935529,
      "grad_norm": 0.9768915176391602,
      "learning_rate": 0.000296957665630232,
      "loss": 1.7353,
      "step": 717
    },
    {
      "epoch": 0.10063069376313945,
      "grad_norm": 0.9743488430976868,
      "learning_rate": 0.00029694331499641234,
      "loss": 1.6711,
      "step": 718
    },
    {
      "epoch": 0.100770847932726,
      "grad_norm": 0.8709413409233093,
      "learning_rate": 0.00029692896436259267,
      "loss": 1.6368,
      "step": 719
    },
    {
      "epoch": 0.10091100210231255,
      "grad_norm": 0.7288438677787781,
      "learning_rate": 0.000296914613728773,
      "loss": 1.5678,
      "step": 720
    },
    {
      "epoch": 0.10105115627189909,
      "grad_norm": 1.2401670217514038,
      "learning_rate": 0.00029690026309495333,
      "loss": 1.5383,
      "step": 721
    },
    {
      "epoch": 0.10119131044148563,
      "grad_norm": 0.7433953881263733,
      "learning_rate": 0.00029688591246113366,
      "loss": 1.3487,
      "step": 722
    },
    {
      "epoch": 0.10133146461107218,
      "grad_norm": 1.1703156232833862,
      "learning_rate": 0.000296871561827314,
      "loss": 1.4389,
      "step": 723
    },
    {
      "epoch": 0.10147161878065872,
      "grad_norm": 1.0181443691253662,
      "learning_rate": 0.00029685721119349437,
      "loss": 1.2741,
      "step": 724
    },
    {
      "epoch": 0.10161177295024527,
      "grad_norm": 0.8912602663040161,
      "learning_rate": 0.0002968428605596747,
      "loss": 1.3337,
      "step": 725
    },
    {
      "epoch": 0.10175192711983182,
      "grad_norm": 1.1339879035949707,
      "learning_rate": 0.000296828509925855,
      "loss": 1.2564,
      "step": 726
    },
    {
      "epoch": 0.10189208128941836,
      "grad_norm": 1.0206938982009888,
      "learning_rate": 0.00029681415929203535,
      "loss": 1.2673,
      "step": 727
    },
    {
      "epoch": 0.1020322354590049,
      "grad_norm": 0.7952700853347778,
      "learning_rate": 0.00029679980865821574,
      "loss": 1.1733,
      "step": 728
    },
    {
      "epoch": 0.10217238962859145,
      "grad_norm": 1.1448856592178345,
      "learning_rate": 0.00029678545802439606,
      "loss": 1.1501,
      "step": 729
    },
    {
      "epoch": 0.102312543798178,
      "grad_norm": 0.8677394390106201,
      "learning_rate": 0.0002967711073905764,
      "loss": 1.1525,
      "step": 730
    },
    {
      "epoch": 0.10245269796776454,
      "grad_norm": 0.9588388204574585,
      "learning_rate": 0.0002967567567567567,
      "loss": 1.0873,
      "step": 731
    },
    {
      "epoch": 0.10259285213735109,
      "grad_norm": 0.9860378503799438,
      "learning_rate": 0.00029674240612293705,
      "loss": 1.1744,
      "step": 732
    },
    {
      "epoch": 0.10273300630693763,
      "grad_norm": 0.8007164001464844,
      "learning_rate": 0.00029672805548911743,
      "loss": 1.0184,
      "step": 733
    },
    {
      "epoch": 0.10287316047652417,
      "grad_norm": 0.9642511606216431,
      "learning_rate": 0.00029671370485529776,
      "loss": 0.9591,
      "step": 734
    },
    {
      "epoch": 0.10301331464611072,
      "grad_norm": 0.9371851086616516,
      "learning_rate": 0.0002966993542214781,
      "loss": 1.0169,
      "step": 735
    },
    {
      "epoch": 0.10315346881569727,
      "grad_norm": 1.197338581085205,
      "learning_rate": 0.0002966850035876584,
      "loss": 1.0416,
      "step": 736
    },
    {
      "epoch": 0.10329362298528381,
      "grad_norm": 0.8738648891448975,
      "learning_rate": 0.00029667065295383875,
      "loss": 1.0081,
      "step": 737
    },
    {
      "epoch": 0.10343377715487036,
      "grad_norm": 0.8442432284355164,
      "learning_rate": 0.0002966563023200191,
      "loss": 0.7859,
      "step": 738
    },
    {
      "epoch": 0.1035739313244569,
      "grad_norm": 0.9249336123466492,
      "learning_rate": 0.00029664195168619946,
      "loss": 0.9638,
      "step": 739
    },
    {
      "epoch": 0.10371408549404344,
      "grad_norm": 0.9109815359115601,
      "learning_rate": 0.0002966276010523798,
      "loss": 0.9012,
      "step": 740
    },
    {
      "epoch": 0.10385423966363,
      "grad_norm": 0.6473962664604187,
      "learning_rate": 0.0002966132504185601,
      "loss": 0.799,
      "step": 741
    },
    {
      "epoch": 0.10399439383321654,
      "grad_norm": 0.7730115056037903,
      "learning_rate": 0.00029659889978474044,
      "loss": 0.7371,
      "step": 742
    },
    {
      "epoch": 0.10413454800280308,
      "grad_norm": 1.0130146741867065,
      "learning_rate": 0.00029658454915092083,
      "loss": 0.8051,
      "step": 743
    },
    {
      "epoch": 0.10427470217238963,
      "grad_norm": 1.0508068799972534,
      "learning_rate": 0.00029657019851710116,
      "loss": 0.8287,
      "step": 744
    },
    {
      "epoch": 0.10441485634197617,
      "grad_norm": 0.9912812113761902,
      "learning_rate": 0.0002965558478832815,
      "loss": 0.7549,
      "step": 745
    },
    {
      "epoch": 0.10455501051156271,
      "grad_norm": 1.7838891744613647,
      "learning_rate": 0.0002965414972494618,
      "loss": 0.7836,
      "step": 746
    },
    {
      "epoch": 0.10469516468114927,
      "grad_norm": 1.3858511447906494,
      "learning_rate": 0.00029652714661564214,
      "loss": 0.7246,
      "step": 747
    },
    {
      "epoch": 0.10483531885073581,
      "grad_norm": 1.1503989696502686,
      "learning_rate": 0.00029651279598182247,
      "loss": 0.7644,
      "step": 748
    },
    {
      "epoch": 0.10497547302032235,
      "grad_norm": 2.166940689086914,
      "learning_rate": 0.00029649844534800285,
      "loss": 1.0979,
      "step": 749
    },
    {
      "epoch": 0.1051156271899089,
      "grad_norm": 4.8708672523498535,
      "learning_rate": 0.0002964840947141832,
      "loss": 1.3126,
      "step": 750
    },
    {
      "epoch": 0.10525578135949544,
      "grad_norm": 1.5065680742263794,
      "learning_rate": 0.0002964697440803635,
      "loss": 1.0039,
      "step": 751
    },
    {
      "epoch": 0.105395935529082,
      "grad_norm": 1.3567593097686768,
      "learning_rate": 0.0002964553934465439,
      "loss": 0.7777,
      "step": 752
    },
    {
      "epoch": 0.10553608969866854,
      "grad_norm": 1.324194073677063,
      "learning_rate": 0.0002964410428127242,
      "loss": 0.8186,
      "step": 753
    },
    {
      "epoch": 0.10567624386825508,
      "grad_norm": 0.7472732067108154,
      "learning_rate": 0.00029642669217890455,
      "loss": 0.7239,
      "step": 754
    },
    {
      "epoch": 0.10581639803784162,
      "grad_norm": 0.6719563603401184,
      "learning_rate": 0.0002964123415450849,
      "loss": 0.6175,
      "step": 755
    },
    {
      "epoch": 0.10595655220742817,
      "grad_norm": 0.5964769721031189,
      "learning_rate": 0.0002963979909112652,
      "loss": 0.639,
      "step": 756
    },
    {
      "epoch": 0.10609670637701471,
      "grad_norm": 0.9903078675270081,
      "learning_rate": 0.00029638364027744554,
      "loss": 0.7978,
      "step": 757
    },
    {
      "epoch": 0.10623686054660127,
      "grad_norm": 1.1828066110610962,
      "learning_rate": 0.00029636928964362586,
      "loss": 0.791,
      "step": 758
    },
    {
      "epoch": 0.10637701471618781,
      "grad_norm": 0.5313505530357361,
      "learning_rate": 0.00029635493900980625,
      "loss": 0.6287,
      "step": 759
    },
    {
      "epoch": 0.10651716888577435,
      "grad_norm": 0.8737642168998718,
      "learning_rate": 0.0002963405883759866,
      "loss": 0.5805,
      "step": 760
    },
    {
      "epoch": 0.1066573230553609,
      "grad_norm": 0.9216781258583069,
      "learning_rate": 0.0002963262377421669,
      "loss": 0.664,
      "step": 761
    },
    {
      "epoch": 0.10679747722494744,
      "grad_norm": 0.8127090930938721,
      "learning_rate": 0.0002963118871083473,
      "loss": 0.6276,
      "step": 762
    },
    {
      "epoch": 0.106937631394534,
      "grad_norm": 1.3506488800048828,
      "learning_rate": 0.0002962975364745276,
      "loss": 0.5912,
      "step": 763
    },
    {
      "epoch": 0.10707778556412054,
      "grad_norm": 1.2340604066848755,
      "learning_rate": 0.00029628318584070794,
      "loss": 0.7388,
      "step": 764
    },
    {
      "epoch": 0.10721793973370708,
      "grad_norm": 0.669952929019928,
      "learning_rate": 0.00029626883520688827,
      "loss": 0.5855,
      "step": 765
    },
    {
      "epoch": 0.10735809390329362,
      "grad_norm": 0.6108424067497253,
      "learning_rate": 0.0002962544845730686,
      "loss": 0.6295,
      "step": 766
    },
    {
      "epoch": 0.10749824807288016,
      "grad_norm": 1.2293285131454468,
      "learning_rate": 0.00029624013393924893,
      "loss": 0.6832,
      "step": 767
    },
    {
      "epoch": 0.10763840224246671,
      "grad_norm": 0.7247191667556763,
      "learning_rate": 0.0002962257833054293,
      "loss": 0.532,
      "step": 768
    },
    {
      "epoch": 0.10777855641205326,
      "grad_norm": 0.8430277109146118,
      "learning_rate": 0.00029621143267160964,
      "loss": 0.5374,
      "step": 769
    },
    {
      "epoch": 0.1079187105816398,
      "grad_norm": 0.5940694212913513,
      "learning_rate": 0.00029619708203778997,
      "loss": 0.5505,
      "step": 770
    },
    {
      "epoch": 0.10805886475122635,
      "grad_norm": 1.552704095840454,
      "learning_rate": 0.00029618273140397035,
      "loss": 0.5974,
      "step": 771
    },
    {
      "epoch": 0.10819901892081289,
      "grad_norm": 0.9161866903305054,
      "learning_rate": 0.0002961683807701507,
      "loss": 0.5543,
      "step": 772
    },
    {
      "epoch": 0.10833917309039943,
      "grad_norm": 0.663697361946106,
      "learning_rate": 0.000296154030136331,
      "loss": 0.4751,
      "step": 773
    },
    {
      "epoch": 0.10847932725998599,
      "grad_norm": 0.7411726117134094,
      "learning_rate": 0.00029613967950251134,
      "loss": 0.5135,
      "step": 774
    },
    {
      "epoch": 0.10861948142957253,
      "grad_norm": 0.6696982979774475,
      "learning_rate": 0.00029612532886869167,
      "loss": 0.5468,
      "step": 775
    },
    {
      "epoch": 0.10875963559915908,
      "grad_norm": 0.7200729250907898,
      "learning_rate": 0.000296110978234872,
      "loss": 0.5524,
      "step": 776
    },
    {
      "epoch": 0.10889978976874562,
      "grad_norm": 0.7118507027626038,
      "learning_rate": 0.0002960966276010523,
      "loss": 0.5293,
      "step": 777
    },
    {
      "epoch": 0.10903994393833216,
      "grad_norm": 0.9761529564857483,
      "learning_rate": 0.0002960822769672327,
      "loss": 0.5712,
      "step": 778
    },
    {
      "epoch": 0.1091800981079187,
      "grad_norm": 0.6491013169288635,
      "learning_rate": 0.00029606792633341303,
      "loss": 0.5138,
      "step": 779
    },
    {
      "epoch": 0.10932025227750526,
      "grad_norm": 1.068501591682434,
      "learning_rate": 0.00029605357569959336,
      "loss": 0.4659,
      "step": 780
    },
    {
      "epoch": 0.1094604064470918,
      "grad_norm": 1.220663070678711,
      "learning_rate": 0.00029603922506577375,
      "loss": 0.5934,
      "step": 781
    },
    {
      "epoch": 0.10960056061667835,
      "grad_norm": 1.1378167867660522,
      "learning_rate": 0.0002960248744319541,
      "loss": 0.4591,
      "step": 782
    },
    {
      "epoch": 0.10974071478626489,
      "grad_norm": 0.7223802208900452,
      "learning_rate": 0.0002960105237981344,
      "loss": 0.4731,
      "step": 783
    },
    {
      "epoch": 0.10988086895585143,
      "grad_norm": 0.5333732962608337,
      "learning_rate": 0.00029599617316431473,
      "loss": 0.4557,
      "step": 784
    },
    {
      "epoch": 0.11002102312543798,
      "grad_norm": 0.7551818490028381,
      "learning_rate": 0.00029598182253049506,
      "loss": 0.5145,
      "step": 785
    },
    {
      "epoch": 0.11016117729502453,
      "grad_norm": 1.171321988105774,
      "learning_rate": 0.0002959674718966754,
      "loss": 0.5494,
      "step": 786
    },
    {
      "epoch": 0.11030133146461107,
      "grad_norm": 0.6282427906990051,
      "learning_rate": 0.00029595312126285577,
      "loss": 0.4243,
      "step": 787
    },
    {
      "epoch": 0.11044148563419762,
      "grad_norm": 3.9415132999420166,
      "learning_rate": 0.0002959387706290361,
      "loss": 0.7097,
      "step": 788
    },
    {
      "epoch": 0.11058163980378416,
      "grad_norm": 1.1901054382324219,
      "learning_rate": 0.00029592441999521643,
      "loss": 0.4911,
      "step": 789
    },
    {
      "epoch": 0.1107217939733707,
      "grad_norm": 1.140281081199646,
      "learning_rate": 0.00029591006936139676,
      "loss": 0.5038,
      "step": 790
    },
    {
      "epoch": 0.11086194814295726,
      "grad_norm": 0.7657198905944824,
      "learning_rate": 0.00029589571872757714,
      "loss": 0.5436,
      "step": 791
    },
    {
      "epoch": 0.1110021023125438,
      "grad_norm": 1.3570201396942139,
      "learning_rate": 0.00029588136809375747,
      "loss": 0.5629,
      "step": 792
    },
    {
      "epoch": 0.11114225648213034,
      "grad_norm": 1.1226831674575806,
      "learning_rate": 0.0002958670174599378,
      "loss": 0.567,
      "step": 793
    },
    {
      "epoch": 0.11128241065171689,
      "grad_norm": 2.085481882095337,
      "learning_rate": 0.0002958526668261181,
      "loss": 0.6337,
      "step": 794
    },
    {
      "epoch": 0.11142256482130343,
      "grad_norm": 1.2003912925720215,
      "learning_rate": 0.00029583831619229845,
      "loss": 0.5762,
      "step": 795
    },
    {
      "epoch": 0.11156271899088997,
      "grad_norm": 0.7766419053077698,
      "learning_rate": 0.0002958239655584788,
      "loss": 0.4641,
      "step": 796
    },
    {
      "epoch": 0.11170287316047653,
      "grad_norm": 1.336724877357483,
      "learning_rate": 0.00029580961492465917,
      "loss": 0.5853,
      "step": 797
    },
    {
      "epoch": 0.11184302733006307,
      "grad_norm": 2.394106388092041,
      "learning_rate": 0.0002957952642908395,
      "loss": 0.7521,
      "step": 798
    },
    {
      "epoch": 0.11198318149964961,
      "grad_norm": 2.7776308059692383,
      "learning_rate": 0.0002957809136570198,
      "loss": 0.8316,
      "step": 799
    },
    {
      "epoch": 0.11212333566923616,
      "grad_norm": 3.571431875228882,
      "learning_rate": 0.00029576656302320015,
      "loss": 0.962,
      "step": 800
    },
    {
      "epoch": 0.1122634898388227,
      "grad_norm": 1.0274524688720703,
      "learning_rate": 0.0002957522123893805,
      "loss": 0.6233,
      "step": 801
    },
    {
      "epoch": 0.11240364400840926,
      "grad_norm": 0.9401232600212097,
      "learning_rate": 0.0002957378617555608,
      "loss": 0.5457,
      "step": 802
    },
    {
      "epoch": 0.1125437981779958,
      "grad_norm": 0.7741809487342834,
      "learning_rate": 0.0002957235111217412,
      "loss": 0.5304,
      "step": 803
    },
    {
      "epoch": 0.11268395234758234,
      "grad_norm": 0.7393455505371094,
      "learning_rate": 0.0002957091604879215,
      "loss": 0.5699,
      "step": 804
    },
    {
      "epoch": 0.11282410651716888,
      "grad_norm": 0.6533899903297424,
      "learning_rate": 0.00029569480985410185,
      "loss": 0.4363,
      "step": 805
    },
    {
      "epoch": 0.11296426068675543,
      "grad_norm": 0.8278070092201233,
      "learning_rate": 0.00029568045922028223,
      "loss": 0.5238,
      "step": 806
    },
    {
      "epoch": 0.11310441485634197,
      "grad_norm": 0.6343386173248291,
      "learning_rate": 0.00029566610858646256,
      "loss": 0.4623,
      "step": 807
    },
    {
      "epoch": 0.11324456902592853,
      "grad_norm": 0.7172487378120422,
      "learning_rate": 0.0002956517579526429,
      "loss": 0.4413,
      "step": 808
    },
    {
      "epoch": 0.11338472319551507,
      "grad_norm": 1.3451247215270996,
      "learning_rate": 0.0002956374073188232,
      "loss": 0.5434,
      "step": 809
    },
    {
      "epoch": 0.11352487736510161,
      "grad_norm": 0.6080853343009949,
      "learning_rate": 0.00029562305668500355,
      "loss": 0.4458,
      "step": 810
    },
    {
      "epoch": 0.11366503153468815,
      "grad_norm": 0.9870772361755371,
      "learning_rate": 0.0002956087060511839,
      "loss": 0.5647,
      "step": 811
    },
    {
      "epoch": 0.1138051857042747,
      "grad_norm": 0.8646849393844604,
      "learning_rate": 0.0002955943554173642,
      "loss": 0.4547,
      "step": 812
    },
    {
      "epoch": 0.11394533987386125,
      "grad_norm": 0.9252731800079346,
      "learning_rate": 0.0002955800047835446,
      "loss": 0.395,
      "step": 813
    },
    {
      "epoch": 0.1140854940434478,
      "grad_norm": 0.5344290733337402,
      "learning_rate": 0.0002955656541497249,
      "loss": 0.4093,
      "step": 814
    },
    {
      "epoch": 0.11422564821303434,
      "grad_norm": 0.7339096069335938,
      "learning_rate": 0.00029555130351590524,
      "loss": 0.506,
      "step": 815
    },
    {
      "epoch": 0.11436580238262088,
      "grad_norm": 0.45406267046928406,
      "learning_rate": 0.0002955369528820856,
      "loss": 0.3366,
      "step": 816
    },
    {
      "epoch": 0.11450595655220742,
      "grad_norm": 0.5326426029205322,
      "learning_rate": 0.00029552260224826595,
      "loss": 0.4403,
      "step": 817
    },
    {
      "epoch": 0.11464611072179397,
      "grad_norm": 0.48331156373023987,
      "learning_rate": 0.0002955082516144463,
      "loss": 0.3566,
      "step": 818
    },
    {
      "epoch": 0.11478626489138052,
      "grad_norm": 0.7945199608802795,
      "learning_rate": 0.0002954939009806266,
      "loss": 0.4666,
      "step": 819
    },
    {
      "epoch": 0.11492641906096707,
      "grad_norm": 0.7521847486495972,
      "learning_rate": 0.00029547955034680694,
      "loss": 0.4222,
      "step": 820
    },
    {
      "epoch": 0.11506657323055361,
      "grad_norm": 0.7186934947967529,
      "learning_rate": 0.00029546519971298727,
      "loss": 0.3886,
      "step": 821
    },
    {
      "epoch": 0.11520672740014015,
      "grad_norm": 0.8452358841896057,
      "learning_rate": 0.00029545084907916765,
      "loss": 0.4749,
      "step": 822
    },
    {
      "epoch": 0.1153468815697267,
      "grad_norm": 0.5826500058174133,
      "learning_rate": 0.000295436498445348,
      "loss": 0.3369,
      "step": 823
    },
    {
      "epoch": 0.11548703573931325,
      "grad_norm": 0.5616267323493958,
      "learning_rate": 0.0002954221478115283,
      "loss": 0.4216,
      "step": 824
    },
    {
      "epoch": 0.1156271899088998,
      "grad_norm": 1.2843900918960571,
      "learning_rate": 0.0002954077971777087,
      "loss": 0.4776,
      "step": 825
    },
    {
      "epoch": 0.11576734407848634,
      "grad_norm": 0.7610683441162109,
      "learning_rate": 0.000295393446543889,
      "loss": 0.3296,
      "step": 826
    },
    {
      "epoch": 0.11590749824807288,
      "grad_norm": 1.0371572971343994,
      "learning_rate": 0.00029537909591006935,
      "loss": 0.4196,
      "step": 827
    },
    {
      "epoch": 0.11604765241765942,
      "grad_norm": 0.5327167510986328,
      "learning_rate": 0.0002953647452762497,
      "loss": 0.38,
      "step": 828
    },
    {
      "epoch": 0.11618780658724596,
      "grad_norm": 0.7355878353118896,
      "learning_rate": 0.00029535039464243,
      "loss": 0.4441,
      "step": 829
    },
    {
      "epoch": 0.11632796075683252,
      "grad_norm": 0.5058201551437378,
      "learning_rate": 0.00029533604400861033,
      "loss": 0.4299,
      "step": 830
    },
    {
      "epoch": 0.11646811492641906,
      "grad_norm": 0.7439339756965637,
      "learning_rate": 0.00029532169337479066,
      "loss": 0.369,
      "step": 831
    },
    {
      "epoch": 0.1166082690960056,
      "grad_norm": 0.6938100457191467,
      "learning_rate": 0.00029530734274097104,
      "loss": 0.37,
      "step": 832
    },
    {
      "epoch": 0.11674842326559215,
      "grad_norm": 0.5563863515853882,
      "learning_rate": 0.0002952929921071514,
      "loss": 0.2806,
      "step": 833
    },
    {
      "epoch": 0.11688857743517869,
      "grad_norm": 0.8294783234596252,
      "learning_rate": 0.0002952786414733317,
      "loss": 0.5081,
      "step": 834
    },
    {
      "epoch": 0.11702873160476523,
      "grad_norm": 0.5650732517242432,
      "learning_rate": 0.0002952642908395121,
      "loss": 0.3773,
      "step": 835
    },
    {
      "epoch": 0.11716888577435179,
      "grad_norm": 0.713193416595459,
      "learning_rate": 0.0002952499402056924,
      "loss": 0.3845,
      "step": 836
    },
    {
      "epoch": 0.11730903994393833,
      "grad_norm": 0.7967789173126221,
      "learning_rate": 0.00029523558957187274,
      "loss": 0.4202,
      "step": 837
    },
    {
      "epoch": 0.11744919411352488,
      "grad_norm": 0.6098101139068604,
      "learning_rate": 0.00029522123893805307,
      "loss": 0.4096,
      "step": 838
    },
    {
      "epoch": 0.11758934828311142,
      "grad_norm": 0.8075692057609558,
      "learning_rate": 0.0002952068883042334,
      "loss": 0.318,
      "step": 839
    },
    {
      "epoch": 0.11772950245269796,
      "grad_norm": 0.9130297899246216,
      "learning_rate": 0.00029519253767041373,
      "loss": 0.2853,
      "step": 840
    },
    {
      "epoch": 0.11786965662228452,
      "grad_norm": 1.4829981327056885,
      "learning_rate": 0.0002951781870365941,
      "loss": 0.494,
      "step": 841
    },
    {
      "epoch": 0.11800981079187106,
      "grad_norm": 0.7205755710601807,
      "learning_rate": 0.00029516383640277444,
      "loss": 0.3676,
      "step": 842
    },
    {
      "epoch": 0.1181499649614576,
      "grad_norm": 1.1660726070404053,
      "learning_rate": 0.00029514948576895477,
      "loss": 0.3481,
      "step": 843
    },
    {
      "epoch": 0.11829011913104415,
      "grad_norm": 2.141003131866455,
      "learning_rate": 0.00029513513513513515,
      "loss": 0.4338,
      "step": 844
    },
    {
      "epoch": 0.11843027330063069,
      "grad_norm": 1.3865693807601929,
      "learning_rate": 0.0002951207845013155,
      "loss": 0.673,
      "step": 845
    },
    {
      "epoch": 0.11857042747021723,
      "grad_norm": 1.0828391313552856,
      "learning_rate": 0.0002951064338674958,
      "loss": 0.5126,
      "step": 846
    },
    {
      "epoch": 0.11871058163980379,
      "grad_norm": 1.4041882753372192,
      "learning_rate": 0.00029509208323367614,
      "loss": 0.4598,
      "step": 847
    },
    {
      "epoch": 0.11885073580939033,
      "grad_norm": 1.1987242698669434,
      "learning_rate": 0.00029507773259985646,
      "loss": 0.4092,
      "step": 848
    },
    {
      "epoch": 0.11899088997897687,
      "grad_norm": 1.8898736238479614,
      "learning_rate": 0.0002950633819660368,
      "loss": 0.5179,
      "step": 849
    },
    {
      "epoch": 0.11913104414856342,
      "grad_norm": 2.226663827896118,
      "learning_rate": 0.0002950490313322171,
      "loss": 0.7412,
      "step": 850
    },
    {
      "epoch": 0.11927119831814996,
      "grad_norm": 0.7179242372512817,
      "learning_rate": 0.0002950346806983975,
      "loss": 0.451,
      "step": 851
    },
    {
      "epoch": 0.11941135248773652,
      "grad_norm": 0.9541609883308411,
      "learning_rate": 0.00029502033006457783,
      "loss": 0.426,
      "step": 852
    },
    {
      "epoch": 0.11955150665732306,
      "grad_norm": 0.8750336170196533,
      "learning_rate": 0.00029500597943075816,
      "loss": 0.4546,
      "step": 853
    },
    {
      "epoch": 0.1196916608269096,
      "grad_norm": 0.44224628806114197,
      "learning_rate": 0.00029499162879693854,
      "loss": 0.4102,
      "step": 854
    },
    {
      "epoch": 0.11983181499649614,
      "grad_norm": 0.7057518362998962,
      "learning_rate": 0.00029497727816311887,
      "loss": 0.4254,
      "step": 855
    },
    {
      "epoch": 0.11997196916608269,
      "grad_norm": 0.6637173295021057,
      "learning_rate": 0.0002949629275292992,
      "loss": 0.3963,
      "step": 856
    },
    {
      "epoch": 0.12011212333566923,
      "grad_norm": 0.9395194053649902,
      "learning_rate": 0.00029494857689547953,
      "loss": 0.3936,
      "step": 857
    },
    {
      "epoch": 0.12025227750525579,
      "grad_norm": 0.42872217297554016,
      "learning_rate": 0.00029493422626165986,
      "loss": 0.2949,
      "step": 858
    },
    {
      "epoch": 0.12039243167484233,
      "grad_norm": 0.49341869354248047,
      "learning_rate": 0.0002949198756278402,
      "loss": 0.309,
      "step": 859
    },
    {
      "epoch": 0.12053258584442887,
      "grad_norm": 0.6677924394607544,
      "learning_rate": 0.00029490552499402057,
      "loss": 0.4954,
      "step": 860
    },
    {
      "epoch": 0.12067274001401541,
      "grad_norm": 0.6404086351394653,
      "learning_rate": 0.0002948911743602009,
      "loss": 0.3655,
      "step": 861
    },
    {
      "epoch": 0.12081289418360196,
      "grad_norm": 0.45750537514686584,
      "learning_rate": 0.0002948768237263812,
      "loss": 0.2629,
      "step": 862
    },
    {
      "epoch": 0.12095304835318851,
      "grad_norm": 0.7091655731201172,
      "learning_rate": 0.00029486247309256156,
      "loss": 0.3562,
      "step": 863
    },
    {
      "epoch": 0.12109320252277506,
      "grad_norm": 0.7352771759033203,
      "learning_rate": 0.0002948481224587419,
      "loss": 0.4015,
      "step": 864
    },
    {
      "epoch": 0.1212333566923616,
      "grad_norm": 0.4700618386268616,
      "learning_rate": 0.0002948337718249222,
      "loss": 0.3375,
      "step": 865
    },
    {
      "epoch": 0.12137351086194814,
      "grad_norm": 0.6827343106269836,
      "learning_rate": 0.00029481942119110254,
      "loss": 0.4001,
      "step": 866
    },
    {
      "epoch": 0.12151366503153468,
      "grad_norm": 0.6253527402877808,
      "learning_rate": 0.0002948050705572829,
      "loss": 0.3327,
      "step": 867
    },
    {
      "epoch": 0.12165381920112123,
      "grad_norm": 0.5779173970222473,
      "learning_rate": 0.00029479071992346325,
      "loss": 0.3193,
      "step": 868
    },
    {
      "epoch": 0.12179397337070778,
      "grad_norm": 0.8246491551399231,
      "learning_rate": 0.0002947763692896436,
      "loss": 0.3463,
      "step": 869
    },
    {
      "epoch": 0.12193412754029433,
      "grad_norm": 0.8151153326034546,
      "learning_rate": 0.00029476201865582396,
      "loss": 0.3097,
      "step": 870
    },
    {
      "epoch": 0.12207428170988087,
      "grad_norm": 0.5913914442062378,
      "learning_rate": 0.0002947476680220043,
      "loss": 0.3373,
      "step": 871
    },
    {
      "epoch": 0.12221443587946741,
      "grad_norm": 0.76949143409729,
      "learning_rate": 0.0002947333173881846,
      "loss": 0.2267,
      "step": 872
    },
    {
      "epoch": 0.12235459004905395,
      "grad_norm": 0.4098033010959625,
      "learning_rate": 0.00029471896675436495,
      "loss": 0.3199,
      "step": 873
    },
    {
      "epoch": 0.12249474421864051,
      "grad_norm": 0.7013890147209167,
      "learning_rate": 0.0002947046161205453,
      "loss": 0.3377,
      "step": 874
    },
    {
      "epoch": 0.12263489838822705,
      "grad_norm": 0.41884222626686096,
      "learning_rate": 0.0002946902654867256,
      "loss": 0.2448,
      "step": 875
    },
    {
      "epoch": 0.1227750525578136,
      "grad_norm": 0.4423123598098755,
      "learning_rate": 0.000294675914852906,
      "loss": 0.2241,
      "step": 876
    },
    {
      "epoch": 0.12291520672740014,
      "grad_norm": 0.49844491481781006,
      "learning_rate": 0.0002946615642190863,
      "loss": 0.2462,
      "step": 877
    },
    {
      "epoch": 0.12305536089698668,
      "grad_norm": 0.7215579152107239,
      "learning_rate": 0.00029464721358526665,
      "loss": 0.2465,
      "step": 878
    },
    {
      "epoch": 0.12319551506657322,
      "grad_norm": 0.7288787364959717,
      "learning_rate": 0.00029463286295144703,
      "loss": 0.2142,
      "step": 879
    },
    {
      "epoch": 0.12333566923615978,
      "grad_norm": 0.6424992084503174,
      "learning_rate": 0.00029461851231762736,
      "loss": 0.2674,
      "step": 880
    },
    {
      "epoch": 0.12347582340574632,
      "grad_norm": 0.5809525847434998,
      "learning_rate": 0.0002946041616838077,
      "loss": 0.2308,
      "step": 881
    },
    {
      "epoch": 0.12361597757533287,
      "grad_norm": 0.5676800608634949,
      "learning_rate": 0.000294589811049988,
      "loss": 0.2667,
      "step": 882
    },
    {
      "epoch": 0.12375613174491941,
      "grad_norm": 0.8107067346572876,
      "learning_rate": 0.00029457546041616834,
      "loss": 0.3554,
      "step": 883
    },
    {
      "epoch": 0.12389628591450595,
      "grad_norm": 0.7875389456748962,
      "learning_rate": 0.00029456110978234867,
      "loss": 0.3677,
      "step": 884
    },
    {
      "epoch": 0.12403644008409251,
      "grad_norm": 1.7270439863204956,
      "learning_rate": 0.000294546759148529,
      "loss": 0.3136,
      "step": 885
    },
    {
      "epoch": 0.12417659425367905,
      "grad_norm": 0.5410801768302917,
      "learning_rate": 0.0002945324085147094,
      "loss": 0.2402,
      "step": 886
    },
    {
      "epoch": 0.1243167484232656,
      "grad_norm": 0.6522603631019592,
      "learning_rate": 0.0002945180578808897,
      "loss": 0.3489,
      "step": 887
    },
    {
      "epoch": 0.12445690259285214,
      "grad_norm": 0.9362044334411621,
      "learning_rate": 0.00029450370724707004,
      "loss": 0.3319,
      "step": 888
    },
    {
      "epoch": 0.12459705676243868,
      "grad_norm": 1.1232060194015503,
      "learning_rate": 0.0002944893566132504,
      "loss": 0.2982,
      "step": 889
    },
    {
      "epoch": 0.12473721093202522,
      "grad_norm": 0.7525444030761719,
      "learning_rate": 0.00029447500597943075,
      "loss": 0.3529,
      "step": 890
    },
    {
      "epoch": 0.12487736510161178,
      "grad_norm": 0.5798377990722656,
      "learning_rate": 0.0002944606553456111,
      "loss": 0.3072,
      "step": 891
    },
    {
      "epoch": 0.1250175192711983,
      "grad_norm": 1.2238162755966187,
      "learning_rate": 0.0002944463047117914,
      "loss": 0.4976,
      "step": 892
    },
    {
      "epoch": 0.12515767344078485,
      "grad_norm": 0.6953086256980896,
      "learning_rate": 0.00029443195407797174,
      "loss": 0.27,
      "step": 893
    },
    {
      "epoch": 0.12529782761037142,
      "grad_norm": 0.41317448019981384,
      "learning_rate": 0.00029441760344415207,
      "loss": 0.2135,
      "step": 894
    },
    {
      "epoch": 0.12543798177995796,
      "grad_norm": 1.0593888759613037,
      "learning_rate": 0.00029440325281033245,
      "loss": 0.2639,
      "step": 895
    },
    {
      "epoch": 0.1255781359495445,
      "grad_norm": 1.0215004682540894,
      "learning_rate": 0.0002943889021765128,
      "loss": 0.2244,
      "step": 896
    },
    {
      "epoch": 0.12571829011913105,
      "grad_norm": 1.18927800655365,
      "learning_rate": 0.0002943745515426931,
      "loss": 0.3516,
      "step": 897
    },
    {
      "epoch": 0.1258584442887176,
      "grad_norm": 1.8573057651519775,
      "learning_rate": 0.00029436020090887343,
      "loss": 0.4673,
      "step": 898
    },
    {
      "epoch": 0.12599859845830413,
      "grad_norm": 1.3714181184768677,
      "learning_rate": 0.0002943458502750538,
      "loss": 0.2739,
      "step": 899
    },
    {
      "epoch": 0.12613875262789068,
      "grad_norm": 2.108304500579834,
      "learning_rate": 0.00029433149964123415,
      "loss": 0.4144,
      "step": 900
    },
    {
      "epoch": 0.12627890679747722,
      "grad_norm": 0.6717967987060547,
      "learning_rate": 0.0002943171490074145,
      "loss": 0.315,
      "step": 901
    },
    {
      "epoch": 0.12641906096706376,
      "grad_norm": 0.7445950508117676,
      "learning_rate": 0.0002943027983735948,
      "loss": 0.3095,
      "step": 902
    },
    {
      "epoch": 0.1265592151366503,
      "grad_norm": 0.7132701873779297,
      "learning_rate": 0.00029428844773977513,
      "loss": 0.2809,
      "step": 903
    },
    {
      "epoch": 0.12669936930623685,
      "grad_norm": 0.8447664976119995,
      "learning_rate": 0.00029427409710595546,
      "loss": 0.3618,
      "step": 904
    },
    {
      "epoch": 0.12683952347582342,
      "grad_norm": 0.8230907917022705,
      "learning_rate": 0.00029425974647213584,
      "loss": 0.2536,
      "step": 905
    },
    {
      "epoch": 0.12697967764540996,
      "grad_norm": 1.2871246337890625,
      "learning_rate": 0.00029424539583831617,
      "loss": 0.4985,
      "step": 906
    },
    {
      "epoch": 0.1271198318149965,
      "grad_norm": 0.39408767223358154,
      "learning_rate": 0.0002942310452044965,
      "loss": 0.2835,
      "step": 907
    },
    {
      "epoch": 0.12725998598458305,
      "grad_norm": 0.4728454351425171,
      "learning_rate": 0.0002942166945706769,
      "loss": 0.3499,
      "step": 908
    },
    {
      "epoch": 0.1274001401541696,
      "grad_norm": 1.3943095207214355,
      "learning_rate": 0.0002942023439368572,
      "loss": 0.3032,
      "step": 909
    },
    {
      "epoch": 0.12754029432375613,
      "grad_norm": 0.6031758785247803,
      "learning_rate": 0.00029418799330303754,
      "loss": 0.2535,
      "step": 910
    },
    {
      "epoch": 0.12768044849334267,
      "grad_norm": 0.8355891108512878,
      "learning_rate": 0.00029417364266921787,
      "loss": 0.3006,
      "step": 911
    },
    {
      "epoch": 0.12782060266292922,
      "grad_norm": 0.7548608779907227,
      "learning_rate": 0.0002941592920353982,
      "loss": 0.3633,
      "step": 912
    },
    {
      "epoch": 0.12796075683251576,
      "grad_norm": 0.839979350566864,
      "learning_rate": 0.0002941449414015785,
      "loss": 0.3087,
      "step": 913
    },
    {
      "epoch": 0.1281009110021023,
      "grad_norm": 0.6054525971412659,
      "learning_rate": 0.0002941305907677589,
      "loss": 0.2655,
      "step": 914
    },
    {
      "epoch": 0.12824106517168885,
      "grad_norm": 0.6736500859260559,
      "learning_rate": 0.00029411624013393924,
      "loss": 0.3425,
      "step": 915
    },
    {
      "epoch": 0.12838121934127542,
      "grad_norm": 0.6754444241523743,
      "learning_rate": 0.00029410188950011957,
      "loss": 0.3403,
      "step": 916
    },
    {
      "epoch": 0.12852137351086196,
      "grad_norm": 0.580228328704834,
      "learning_rate": 0.0002940875388662999,
      "loss": 0.2767,
      "step": 917
    },
    {
      "epoch": 0.1286615276804485,
      "grad_norm": 1.5820302963256836,
      "learning_rate": 0.0002940731882324803,
      "loss": 0.3742,
      "step": 918
    },
    {
      "epoch": 0.12880168185003504,
      "grad_norm": 0.7250005602836609,
      "learning_rate": 0.0002940588375986606,
      "loss": 0.2758,
      "step": 919
    },
    {
      "epoch": 0.1289418360196216,
      "grad_norm": 0.9791303873062134,
      "learning_rate": 0.00029404448696484093,
      "loss": 0.3885,
      "step": 920
    },
    {
      "epoch": 0.12908199018920813,
      "grad_norm": 0.3761806786060333,
      "learning_rate": 0.00029403013633102126,
      "loss": 0.2247,
      "step": 921
    },
    {
      "epoch": 0.12922214435879467,
      "grad_norm": 0.4687352478504181,
      "learning_rate": 0.0002940157856972016,
      "loss": 0.2165,
      "step": 922
    },
    {
      "epoch": 0.12936229852838121,
      "grad_norm": 0.7041298747062683,
      "learning_rate": 0.0002940014350633819,
      "loss": 0.2938,
      "step": 923
    },
    {
      "epoch": 0.12950245269796776,
      "grad_norm": 0.5893972516059875,
      "learning_rate": 0.0002939870844295623,
      "loss": 0.2835,
      "step": 924
    },
    {
      "epoch": 0.1296426068675543,
      "grad_norm": 0.9080823063850403,
      "learning_rate": 0.00029397273379574263,
      "loss": 0.2739,
      "step": 925
    },
    {
      "epoch": 0.12978276103714084,
      "grad_norm": 1.0706671476364136,
      "learning_rate": 0.00029395838316192296,
      "loss": 0.2591,
      "step": 926
    },
    {
      "epoch": 0.1299229152067274,
      "grad_norm": 0.9298347234725952,
      "learning_rate": 0.0002939440325281033,
      "loss": 0.347,
      "step": 927
    },
    {
      "epoch": 0.13006306937631396,
      "grad_norm": 0.8759827017784119,
      "learning_rate": 0.0002939296818942836,
      "loss": 0.3267,
      "step": 928
    },
    {
      "epoch": 0.1302032235459005,
      "grad_norm": 0.4727971851825714,
      "learning_rate": 0.00029391533126046395,
      "loss": 0.2267,
      "step": 929
    },
    {
      "epoch": 0.13034337771548704,
      "grad_norm": 0.754137396812439,
      "learning_rate": 0.00029390098062664433,
      "loss": 0.2867,
      "step": 930
    },
    {
      "epoch": 0.13048353188507358,
      "grad_norm": 0.5513168573379517,
      "learning_rate": 0.00029388662999282466,
      "loss": 0.2234,
      "step": 931
    },
    {
      "epoch": 0.13062368605466013,
      "grad_norm": 0.6047320365905762,
      "learning_rate": 0.000293872279359005,
      "loss": 0.2005,
      "step": 932
    },
    {
      "epoch": 0.13076384022424667,
      "grad_norm": 1.0456677675247192,
      "learning_rate": 0.0002938579287251853,
      "loss": 0.3261,
      "step": 933
    },
    {
      "epoch": 0.1309039943938332,
      "grad_norm": 0.5368068218231201,
      "learning_rate": 0.0002938435780913657,
      "loss": 0.3023,
      "step": 934
    },
    {
      "epoch": 0.13104414856341975,
      "grad_norm": 0.32677504420280457,
      "learning_rate": 0.000293829227457546,
      "loss": 0.2282,
      "step": 935
    },
    {
      "epoch": 0.1311843027330063,
      "grad_norm": 0.8589951395988464,
      "learning_rate": 0.00029381487682372635,
      "loss": 0.2576,
      "step": 936
    },
    {
      "epoch": 0.13132445690259284,
      "grad_norm": 0.8622442483901978,
      "learning_rate": 0.0002938005261899067,
      "loss": 0.2456,
      "step": 937
    },
    {
      "epoch": 0.1314646110721794,
      "grad_norm": 0.7158130407333374,
      "learning_rate": 0.000293786175556087,
      "loss": 0.3435,
      "step": 938
    },
    {
      "epoch": 0.13160476524176595,
      "grad_norm": 0.965425968170166,
      "learning_rate": 0.00029377182492226734,
      "loss": 0.327,
      "step": 939
    },
    {
      "epoch": 0.1317449194113525,
      "grad_norm": 0.8095971941947937,
      "learning_rate": 0.0002937574742884477,
      "loss": 0.2978,
      "step": 940
    },
    {
      "epoch": 0.13188507358093904,
      "grad_norm": 0.8291435241699219,
      "learning_rate": 0.00029374312365462805,
      "loss": 0.3361,
      "step": 941
    },
    {
      "epoch": 0.13202522775052558,
      "grad_norm": 0.5811228156089783,
      "learning_rate": 0.0002937287730208084,
      "loss": 0.2784,
      "step": 942
    },
    {
      "epoch": 0.13216538192011212,
      "grad_norm": 0.7495807409286499,
      "learning_rate": 0.00029371442238698876,
      "loss": 0.2987,
      "step": 943
    },
    {
      "epoch": 0.13230553608969867,
      "grad_norm": 0.980492115020752,
      "learning_rate": 0.0002937000717531691,
      "loss": 0.4014,
      "step": 944
    },
    {
      "epoch": 0.1324456902592852,
      "grad_norm": 0.9034101366996765,
      "learning_rate": 0.0002936857211193494,
      "loss": 0.1718,
      "step": 945
    },
    {
      "epoch": 0.13258584442887175,
      "grad_norm": 1.4471925497055054,
      "learning_rate": 0.00029367137048552975,
      "loss": 0.3174,
      "step": 946
    },
    {
      "epoch": 0.1327259985984583,
      "grad_norm": 3.6291677951812744,
      "learning_rate": 0.0002936570198517101,
      "loss": 0.4778,
      "step": 947
    },
    {
      "epoch": 0.13286615276804484,
      "grad_norm": 0.8271286487579346,
      "learning_rate": 0.0002936426692178904,
      "loss": 0.2167,
      "step": 948
    },
    {
      "epoch": 0.1330063069376314,
      "grad_norm": 4.98697566986084,
      "learning_rate": 0.00029362831858407073,
      "loss": 0.879,
      "step": 949
    },
    {
      "epoch": 0.13314646110721795,
      "grad_norm": 3.502244710922241,
      "learning_rate": 0.0002936139679502511,
      "loss": 1.2223,
      "step": 950
    },
    {
      "epoch": 0.1332866152768045,
      "grad_norm": 0.7069107890129089,
      "learning_rate": 0.00029359961731643144,
      "loss": 0.3105,
      "step": 951
    },
    {
      "epoch": 0.13342676944639104,
      "grad_norm": 0.5436837673187256,
      "learning_rate": 0.0002935852666826118,
      "loss": 0.2988,
      "step": 952
    },
    {
      "epoch": 0.13356692361597758,
      "grad_norm": 0.6500002145767212,
      "learning_rate": 0.00029357091604879216,
      "loss": 0.405,
      "step": 953
    },
    {
      "epoch": 0.13370707778556412,
      "grad_norm": 0.39849650859832764,
      "learning_rate": 0.0002935565654149725,
      "loss": 0.2854,
      "step": 954
    },
    {
      "epoch": 0.13384723195515066,
      "grad_norm": 0.5219706296920776,
      "learning_rate": 0.0002935422147811528,
      "loss": 0.2648,
      "step": 955
    },
    {
      "epoch": 0.1339873861247372,
      "grad_norm": 0.6016669869422913,
      "learning_rate": 0.00029352786414733314,
      "loss": 0.3059,
      "step": 956
    },
    {
      "epoch": 0.13412754029432375,
      "grad_norm": 0.6434925198554993,
      "learning_rate": 0.00029351351351351347,
      "loss": 0.3159,
      "step": 957
    },
    {
      "epoch": 0.1342676944639103,
      "grad_norm": 0.9862679839134216,
      "learning_rate": 0.0002934991628796938,
      "loss": 0.3284,
      "step": 958
    },
    {
      "epoch": 0.13440784863349683,
      "grad_norm": 0.5369930267333984,
      "learning_rate": 0.0002934848122458742,
      "loss": 0.3041,
      "step": 959
    },
    {
      "epoch": 0.1345480028030834,
      "grad_norm": 0.680402934551239,
      "learning_rate": 0.0002934704616120545,
      "loss": 0.3951,
      "step": 960
    },
    {
      "epoch": 0.13468815697266995,
      "grad_norm": 0.49302855134010315,
      "learning_rate": 0.00029345611097823484,
      "loss": 0.3065,
      "step": 961
    },
    {
      "epoch": 0.1348283111422565,
      "grad_norm": 0.3662201762199402,
      "learning_rate": 0.0002934417603444152,
      "loss": 0.2148,
      "step": 962
    },
    {
      "epoch": 0.13496846531184303,
      "grad_norm": 1.3564612865447998,
      "learning_rate": 0.00029342740971059555,
      "loss": 0.2587,
      "step": 963
    },
    {
      "epoch": 0.13510861948142958,
      "grad_norm": 0.9542006254196167,
      "learning_rate": 0.0002934130590767759,
      "loss": 0.3687,
      "step": 964
    },
    {
      "epoch": 0.13524877365101612,
      "grad_norm": 0.8196474313735962,
      "learning_rate": 0.0002933987084429562,
      "loss": 0.273,
      "step": 965
    },
    {
      "epoch": 0.13538892782060266,
      "grad_norm": 0.4538367986679077,
      "learning_rate": 0.00029338435780913654,
      "loss": 0.1551,
      "step": 966
    },
    {
      "epoch": 0.1355290819901892,
      "grad_norm": 0.537915050983429,
      "learning_rate": 0.00029337000717531686,
      "loss": 0.3536,
      "step": 967
    },
    {
      "epoch": 0.13566923615977575,
      "grad_norm": 0.5604998469352722,
      "learning_rate": 0.0002933556565414972,
      "loss": 0.2973,
      "step": 968
    },
    {
      "epoch": 0.1358093903293623,
      "grad_norm": 0.6562977433204651,
      "learning_rate": 0.0002933413059076776,
      "loss": 0.2933,
      "step": 969
    },
    {
      "epoch": 0.13594954449894883,
      "grad_norm": 0.5113474130630493,
      "learning_rate": 0.0002933269552738579,
      "loss": 0.3259,
      "step": 970
    },
    {
      "epoch": 0.1360896986685354,
      "grad_norm": 0.576876699924469,
      "learning_rate": 0.00029331260464003823,
      "loss": 0.2725,
      "step": 971
    },
    {
      "epoch": 0.13622985283812195,
      "grad_norm": 0.7563326954841614,
      "learning_rate": 0.0002932982540062186,
      "loss": 0.2997,
      "step": 972
    },
    {
      "epoch": 0.1363700070077085,
      "grad_norm": 0.30106568336486816,
      "learning_rate": 0.00029328390337239894,
      "loss": 0.1572,
      "step": 973
    },
    {
      "epoch": 0.13651016117729503,
      "grad_norm": 1.384362816810608,
      "learning_rate": 0.00029326955273857927,
      "loss": 0.2946,
      "step": 974
    },
    {
      "epoch": 0.13665031534688157,
      "grad_norm": 0.5033469200134277,
      "learning_rate": 0.0002932552021047596,
      "loss": 0.2046,
      "step": 975
    },
    {
      "epoch": 0.13679046951646812,
      "grad_norm": 0.6536031365394592,
      "learning_rate": 0.00029324085147093993,
      "loss": 0.3499,
      "step": 976
    },
    {
      "epoch": 0.13693062368605466,
      "grad_norm": 0.5294567942619324,
      "learning_rate": 0.00029322650083712026,
      "loss": 0.2348,
      "step": 977
    },
    {
      "epoch": 0.1370707778556412,
      "grad_norm": 0.9334960579872131,
      "learning_rate": 0.00029321215020330064,
      "loss": 0.2194,
      "step": 978
    },
    {
      "epoch": 0.13721093202522774,
      "grad_norm": 0.5313480496406555,
      "learning_rate": 0.00029319779956948097,
      "loss": 0.1931,
      "step": 979
    },
    {
      "epoch": 0.1373510861948143,
      "grad_norm": 0.4705626368522644,
      "learning_rate": 0.0002931834489356613,
      "loss": 0.1537,
      "step": 980
    },
    {
      "epoch": 0.13749124036440083,
      "grad_norm": 0.7432977557182312,
      "learning_rate": 0.0002931690983018417,
      "loss": 0.3833,
      "step": 981
    },
    {
      "epoch": 0.13763139453398737,
      "grad_norm": 0.6052164435386658,
      "learning_rate": 0.000293154747668022,
      "loss": 0.2779,
      "step": 982
    },
    {
      "epoch": 0.13777154870357394,
      "grad_norm": 0.671550452709198,
      "learning_rate": 0.00029314039703420234,
      "loss": 0.3163,
      "step": 983
    },
    {
      "epoch": 0.13791170287316049,
      "grad_norm": 0.8387758731842041,
      "learning_rate": 0.00029312604640038267,
      "loss": 0.2762,
      "step": 984
    },
    {
      "epoch": 0.13805185704274703,
      "grad_norm": 0.6830567121505737,
      "learning_rate": 0.000293111695766563,
      "loss": 0.3395,
      "step": 985
    },
    {
      "epoch": 0.13819201121233357,
      "grad_norm": 0.9115906357765198,
      "learning_rate": 0.0002930973451327433,
      "loss": 0.2845,
      "step": 986
    },
    {
      "epoch": 0.1383321653819201,
      "grad_norm": 1.1519874334335327,
      "learning_rate": 0.00029308299449892365,
      "loss": 0.2206,
      "step": 987
    },
    {
      "epoch": 0.13847231955150666,
      "grad_norm": 0.5526348352432251,
      "learning_rate": 0.00029306864386510404,
      "loss": 0.2042,
      "step": 988
    },
    {
      "epoch": 0.1386124737210932,
      "grad_norm": 0.9735465049743652,
      "learning_rate": 0.00029305429323128436,
      "loss": 0.3277,
      "step": 989
    },
    {
      "epoch": 0.13875262789067974,
      "grad_norm": 0.7076131701469421,
      "learning_rate": 0.0002930399425974647,
      "loss": 0.2557,
      "step": 990
    },
    {
      "epoch": 0.13889278206026628,
      "grad_norm": 0.8041265606880188,
      "learning_rate": 0.000293025591963645,
      "loss": 0.2214,
      "step": 991
    },
    {
      "epoch": 0.13903293622985283,
      "grad_norm": 0.5854584574699402,
      "learning_rate": 0.00029301124132982535,
      "loss": 0.2721,
      "step": 992
    },
    {
      "epoch": 0.13917309039943937,
      "grad_norm": 0.9238445162773132,
      "learning_rate": 0.0002929968906960057,
      "loss": 0.2782,
      "step": 993
    },
    {
      "epoch": 0.13931324456902594,
      "grad_norm": 2.3666505813598633,
      "learning_rate": 0.00029298254006218606,
      "loss": 0.411,
      "step": 994
    },
    {
      "epoch": 0.13945339873861248,
      "grad_norm": 1.6303638219833374,
      "learning_rate": 0.0002929681894283664,
      "loss": 0.3095,
      "step": 995
    },
    {
      "epoch": 0.13959355290819903,
      "grad_norm": 0.9080213904380798,
      "learning_rate": 0.0002929538387945467,
      "loss": 0.2901,
      "step": 996
    },
    {
      "epoch": 0.13973370707778557,
      "grad_norm": 1.881140112876892,
      "learning_rate": 0.0002929394881607271,
      "loss": 0.3931,
      "step": 997
    },
    {
      "epoch": 0.1398738612473721,
      "grad_norm": 1.4179917573928833,
      "learning_rate": 0.00029292513752690743,
      "loss": 0.3517,
      "step": 998
    },
    {
      "epoch": 0.14001401541695865,
      "grad_norm": 1.5313589572906494,
      "learning_rate": 0.00029291078689308776,
      "loss": 0.2621,
      "step": 999
    },
    {
      "epoch": 0.1401541695865452,
      "grad_norm": 2.040766716003418,
      "learning_rate": 0.0002928964362592681,
      "loss": 0.2916,
      "step": 1000
    },
    {
      "epoch": 0.14029432375613174,
      "grad_norm": 0.6564692854881287,
      "learning_rate": 0.0002928820856254484,
      "loss": 0.2523,
      "step": 1001
    },
    {
      "epoch": 0.14043447792571828,
      "grad_norm": 0.7949646711349487,
      "learning_rate": 0.00029286773499162874,
      "loss": 0.2727,
      "step": 1002
    },
    {
      "epoch": 0.14057463209530482,
      "grad_norm": 0.9629088640213013,
      "learning_rate": 0.00029285338435780907,
      "loss": 0.2348,
      "step": 1003
    },
    {
      "epoch": 0.14071478626489137,
      "grad_norm": 0.9612164497375488,
      "learning_rate": 0.00029283903372398945,
      "loss": 0.2592,
      "step": 1004
    },
    {
      "epoch": 0.14085494043447794,
      "grad_norm": 0.8088884353637695,
      "learning_rate": 0.0002928246830901698,
      "loss": 0.2401,
      "step": 1005
    },
    {
      "epoch": 0.14099509460406448,
      "grad_norm": 0.6948018074035645,
      "learning_rate": 0.0002928103324563501,
      "loss": 0.2583,
      "step": 1006
    },
    {
      "epoch": 0.14113524877365102,
      "grad_norm": 0.5473814606666565,
      "learning_rate": 0.0002927959818225305,
      "loss": 0.2481,
      "step": 1007
    },
    {
      "epoch": 0.14127540294323757,
      "grad_norm": 0.4820880889892578,
      "learning_rate": 0.0002927816311887108,
      "loss": 0.2994,
      "step": 1008
    },
    {
      "epoch": 0.1414155571128241,
      "grad_norm": 0.46331194043159485,
      "learning_rate": 0.00029276728055489115,
      "loss": 0.1575,
      "step": 1009
    },
    {
      "epoch": 0.14155571128241065,
      "grad_norm": 0.6499099731445312,
      "learning_rate": 0.0002927529299210715,
      "loss": 0.2881,
      "step": 1010
    },
    {
      "epoch": 0.1416958654519972,
      "grad_norm": 1.3492460250854492,
      "learning_rate": 0.0002927385792872518,
      "loss": 0.2339,
      "step": 1011
    },
    {
      "epoch": 0.14183601962158374,
      "grad_norm": 0.37316715717315674,
      "learning_rate": 0.00029272422865343214,
      "loss": 0.1807,
      "step": 1012
    },
    {
      "epoch": 0.14197617379117028,
      "grad_norm": 0.9624108076095581,
      "learning_rate": 0.0002927098780196125,
      "loss": 0.3353,
      "step": 1013
    },
    {
      "epoch": 0.14211632796075682,
      "grad_norm": 0.8035147786140442,
      "learning_rate": 0.00029269552738579285,
      "loss": 0.2025,
      "step": 1014
    },
    {
      "epoch": 0.14225648213034336,
      "grad_norm": 0.4931930899620056,
      "learning_rate": 0.0002926811767519732,
      "loss": 0.2892,
      "step": 1015
    },
    {
      "epoch": 0.14239663629992994,
      "grad_norm": 0.5515649318695068,
      "learning_rate": 0.00029266682611815356,
      "loss": 0.219,
      "step": 1016
    },
    {
      "epoch": 0.14253679046951648,
      "grad_norm": 0.5605678558349609,
      "learning_rate": 0.0002926524754843339,
      "loss": 0.2067,
      "step": 1017
    },
    {
      "epoch": 0.14267694463910302,
      "grad_norm": 0.772860050201416,
      "learning_rate": 0.0002926381248505142,
      "loss": 0.2783,
      "step": 1018
    },
    {
      "epoch": 0.14281709880868956,
      "grad_norm": 0.5818120837211609,
      "learning_rate": 0.00029262377421669455,
      "loss": 0.2423,
      "step": 1019
    },
    {
      "epoch": 0.1429572529782761,
      "grad_norm": 0.8352418541908264,
      "learning_rate": 0.0002926094235828749,
      "loss": 0.1817,
      "step": 1020
    },
    {
      "epoch": 0.14309740714786265,
      "grad_norm": 0.4489438235759735,
      "learning_rate": 0.0002925950729490552,
      "loss": 0.2192,
      "step": 1021
    },
    {
      "epoch": 0.1432375613174492,
      "grad_norm": 0.6198673248291016,
      "learning_rate": 0.00029258072231523553,
      "loss": 0.185,
      "step": 1022
    },
    {
      "epoch": 0.14337771548703573,
      "grad_norm": 0.5971577167510986,
      "learning_rate": 0.0002925663716814159,
      "loss": 0.3252,
      "step": 1023
    },
    {
      "epoch": 0.14351786965662228,
      "grad_norm": 0.8665616512298584,
      "learning_rate": 0.00029255202104759624,
      "loss": 0.2205,
      "step": 1024
    },
    {
      "epoch": 0.14365802382620882,
      "grad_norm": 1.0874003171920776,
      "learning_rate": 0.00029253767041377657,
      "loss": 0.2067,
      "step": 1025
    },
    {
      "epoch": 0.14379817799579536,
      "grad_norm": 0.5821864604949951,
      "learning_rate": 0.00029252331977995695,
      "loss": 0.287,
      "step": 1026
    },
    {
      "epoch": 0.14393833216538193,
      "grad_norm": 0.3562723994255066,
      "learning_rate": 0.0002925089691461373,
      "loss": 0.216,
      "step": 1027
    },
    {
      "epoch": 0.14407848633496848,
      "grad_norm": 0.5792339444160461,
      "learning_rate": 0.0002924946185123176,
      "loss": 0.2107,
      "step": 1028
    },
    {
      "epoch": 0.14421864050455502,
      "grad_norm": 0.6203877925872803,
      "learning_rate": 0.00029248026787849794,
      "loss": 0.1572,
      "step": 1029
    },
    {
      "epoch": 0.14435879467414156,
      "grad_norm": 0.5179003477096558,
      "learning_rate": 0.00029246591724467827,
      "loss": 0.2192,
      "step": 1030
    },
    {
      "epoch": 0.1444989488437281,
      "grad_norm": 0.4894259572029114,
      "learning_rate": 0.0002924515666108586,
      "loss": 0.2466,
      "step": 1031
    },
    {
      "epoch": 0.14463910301331465,
      "grad_norm": 0.6231358647346497,
      "learning_rate": 0.000292437215977039,
      "loss": 0.1881,
      "step": 1032
    },
    {
      "epoch": 0.1447792571829012,
      "grad_norm": 0.5783756971359253,
      "learning_rate": 0.0002924228653432193,
      "loss": 0.2478,
      "step": 1033
    },
    {
      "epoch": 0.14491941135248773,
      "grad_norm": 1.7217965126037598,
      "learning_rate": 0.00029240851470939964,
      "loss": 0.1941,
      "step": 1034
    },
    {
      "epoch": 0.14505956552207427,
      "grad_norm": 0.38905397057533264,
      "learning_rate": 0.00029239416407558,
      "loss": 0.1675,
      "step": 1035
    },
    {
      "epoch": 0.14519971969166082,
      "grad_norm": 0.638530969619751,
      "learning_rate": 0.00029237981344176035,
      "loss": 0.2163,
      "step": 1036
    },
    {
      "epoch": 0.14533987386124736,
      "grad_norm": 0.4328227937221527,
      "learning_rate": 0.0002923654628079407,
      "loss": 0.147,
      "step": 1037
    },
    {
      "epoch": 0.14548002803083393,
      "grad_norm": 1.1534838676452637,
      "learning_rate": 0.000292351112174121,
      "loss": 0.184,
      "step": 1038
    },
    {
      "epoch": 0.14562018220042047,
      "grad_norm": 0.5368075370788574,
      "learning_rate": 0.00029233676154030133,
      "loss": 0.1647,
      "step": 1039
    },
    {
      "epoch": 0.14576033637000702,
      "grad_norm": 0.3025093078613281,
      "learning_rate": 0.00029232241090648166,
      "loss": 0.0858,
      "step": 1040
    },
    {
      "epoch": 0.14590049053959356,
      "grad_norm": 0.7990778684616089,
      "learning_rate": 0.000292308060272662,
      "loss": 0.2087,
      "step": 1041
    },
    {
      "epoch": 0.1460406447091801,
      "grad_norm": 1.5308758020401,
      "learning_rate": 0.0002922937096388424,
      "loss": 0.3255,
      "step": 1042
    },
    {
      "epoch": 0.14618079887876664,
      "grad_norm": 2.1593265533447266,
      "learning_rate": 0.0002922793590050227,
      "loss": 0.2336,
      "step": 1043
    },
    {
      "epoch": 0.1463209530483532,
      "grad_norm": 1.7676151990890503,
      "learning_rate": 0.00029226500837120303,
      "loss": 0.3244,
      "step": 1044
    },
    {
      "epoch": 0.14646110721793973,
      "grad_norm": 0.5936906337738037,
      "learning_rate": 0.0002922506577373834,
      "loss": 0.1466,
      "step": 1045
    },
    {
      "epoch": 0.14660126138752627,
      "grad_norm": 0.7183467745780945,
      "learning_rate": 0.00029223630710356374,
      "loss": 0.1865,
      "step": 1046
    },
    {
      "epoch": 0.14674141555711281,
      "grad_norm": 1.5002034902572632,
      "learning_rate": 0.00029222195646974407,
      "loss": 0.2901,
      "step": 1047
    },
    {
      "epoch": 0.14688156972669936,
      "grad_norm": 0.51473468542099,
      "learning_rate": 0.0002922076058359244,
      "loss": 0.1346,
      "step": 1048
    },
    {
      "epoch": 0.14702172389628593,
      "grad_norm": 1.5852375030517578,
      "learning_rate": 0.00029219325520210473,
      "loss": 0.2854,
      "step": 1049
    },
    {
      "epoch": 0.14716187806587247,
      "grad_norm": 1.9233812093734741,
      "learning_rate": 0.00029217890456828506,
      "loss": 0.2318,
      "step": 1050
    },
    {
      "epoch": 0.147302032235459,
      "grad_norm": 0.4522407352924347,
      "learning_rate": 0.00029216455393446544,
      "loss": 0.1817,
      "step": 1051
    },
    {
      "epoch": 0.14744218640504556,
      "grad_norm": 0.5314527153968811,
      "learning_rate": 0.00029215020330064577,
      "loss": 0.2001,
      "step": 1052
    },
    {
      "epoch": 0.1475823405746321,
      "grad_norm": 0.45209234952926636,
      "learning_rate": 0.0002921358526668261,
      "loss": 0.2472,
      "step": 1053
    },
    {
      "epoch": 0.14772249474421864,
      "grad_norm": 0.3947276473045349,
      "learning_rate": 0.0002921215020330064,
      "loss": 0.2458,
      "step": 1054
    },
    {
      "epoch": 0.14786264891380518,
      "grad_norm": 0.7615723013877869,
      "learning_rate": 0.00029210715139918675,
      "loss": 0.2498,
      "step": 1055
    },
    {
      "epoch": 0.14800280308339173,
      "grad_norm": 0.4284062683582306,
      "learning_rate": 0.0002920928007653671,
      "loss": 0.2253,
      "step": 1056
    },
    {
      "epoch": 0.14814295725297827,
      "grad_norm": 0.690561830997467,
      "learning_rate": 0.0002920784501315474,
      "loss": 0.2275,
      "step": 1057
    },
    {
      "epoch": 0.1482831114225648,
      "grad_norm": 0.3718281090259552,
      "learning_rate": 0.0002920640994977278,
      "loss": 0.1717,
      "step": 1058
    },
    {
      "epoch": 0.14842326559215135,
      "grad_norm": 0.882584810256958,
      "learning_rate": 0.0002920497488639081,
      "loss": 0.2968,
      "step": 1059
    },
    {
      "epoch": 0.14856341976173792,
      "grad_norm": 0.8743811249732971,
      "learning_rate": 0.00029203539823008845,
      "loss": 0.2523,
      "step": 1060
    },
    {
      "epoch": 0.14870357393132447,
      "grad_norm": 0.6886244416236877,
      "learning_rate": 0.00029202104759626883,
      "loss": 0.2255,
      "step": 1061
    },
    {
      "epoch": 0.148843728100911,
      "grad_norm": 0.2720070481300354,
      "learning_rate": 0.00029200669696244916,
      "loss": 0.1723,
      "step": 1062
    },
    {
      "epoch": 0.14898388227049755,
      "grad_norm": 0.4363411068916321,
      "learning_rate": 0.0002919923463286295,
      "loss": 0.2229,
      "step": 1063
    },
    {
      "epoch": 0.1491240364400841,
      "grad_norm": 0.4638037085533142,
      "learning_rate": 0.0002919779956948098,
      "loss": 0.1472,
      "step": 1064
    },
    {
      "epoch": 0.14926419060967064,
      "grad_norm": 0.9076966643333435,
      "learning_rate": 0.00029196364506099015,
      "loss": 0.295,
      "step": 1065
    },
    {
      "epoch": 0.14940434477925718,
      "grad_norm": 0.4324796199798584,
      "learning_rate": 0.0002919492944271705,
      "loss": 0.1955,
      "step": 1066
    },
    {
      "epoch": 0.14954449894884372,
      "grad_norm": 2.00901460647583,
      "learning_rate": 0.00029193494379335086,
      "loss": 0.223,
      "step": 1067
    },
    {
      "epoch": 0.14968465311843027,
      "grad_norm": 0.5622586011886597,
      "learning_rate": 0.0002919205931595312,
      "loss": 0.1851,
      "step": 1068
    },
    {
      "epoch": 0.1498248072880168,
      "grad_norm": 0.6686916947364807,
      "learning_rate": 0.0002919062425257115,
      "loss": 0.2852,
      "step": 1069
    },
    {
      "epoch": 0.14996496145760335,
      "grad_norm": 1.70041823387146,
      "learning_rate": 0.0002918918918918919,
      "loss": 0.1535,
      "step": 1070
    },
    {
      "epoch": 0.15010511562718992,
      "grad_norm": 0.522042453289032,
      "learning_rate": 0.00029187754125807223,
      "loss": 0.2031,
      "step": 1071
    },
    {
      "epoch": 0.15024526979677646,
      "grad_norm": 1.0612695217132568,
      "learning_rate": 0.00029186319062425256,
      "loss": 0.3035,
      "step": 1072
    },
    {
      "epoch": 0.150385423966363,
      "grad_norm": 0.5846889019012451,
      "learning_rate": 0.0002918488399904329,
      "loss": 0.2559,
      "step": 1073
    },
    {
      "epoch": 0.15052557813594955,
      "grad_norm": 0.45221787691116333,
      "learning_rate": 0.0002918344893566132,
      "loss": 0.2564,
      "step": 1074
    },
    {
      "epoch": 0.1506657323055361,
      "grad_norm": 0.4605032801628113,
      "learning_rate": 0.00029182013872279354,
      "loss": 0.2377,
      "step": 1075
    },
    {
      "epoch": 0.15080588647512264,
      "grad_norm": 0.6658544540405273,
      "learning_rate": 0.00029180578808897387,
      "loss": 0.3347,
      "step": 1076
    },
    {
      "epoch": 0.15094604064470918,
      "grad_norm": 0.8150544166564941,
      "learning_rate": 0.00029179143745515425,
      "loss": 0.2513,
      "step": 1077
    },
    {
      "epoch": 0.15108619481429572,
      "grad_norm": 0.32213959097862244,
      "learning_rate": 0.0002917770868213346,
      "loss": 0.1681,
      "step": 1078
    },
    {
      "epoch": 0.15122634898388226,
      "grad_norm": 1.1388335227966309,
      "learning_rate": 0.0002917627361875149,
      "loss": 0.3986,
      "step": 1079
    },
    {
      "epoch": 0.1513665031534688,
      "grad_norm": 0.9581372737884521,
      "learning_rate": 0.0002917483855536953,
      "loss": 0.2778,
      "step": 1080
    },
    {
      "epoch": 0.15150665732305535,
      "grad_norm": 0.6253133416175842,
      "learning_rate": 0.0002917340349198756,
      "loss": 0.2072,
      "step": 1081
    },
    {
      "epoch": 0.15164681149264192,
      "grad_norm": 0.34929677844047546,
      "learning_rate": 0.00029171968428605595,
      "loss": 0.1321,
      "step": 1082
    },
    {
      "epoch": 0.15178696566222846,
      "grad_norm": 0.3293485641479492,
      "learning_rate": 0.0002917053336522363,
      "loss": 0.1635,
      "step": 1083
    },
    {
      "epoch": 0.151927119831815,
      "grad_norm": 1.2029085159301758,
      "learning_rate": 0.0002916909830184166,
      "loss": 0.3263,
      "step": 1084
    },
    {
      "epoch": 0.15206727400140155,
      "grad_norm": 0.3753032982349396,
      "learning_rate": 0.00029167663238459694,
      "loss": 0.1284,
      "step": 1085
    },
    {
      "epoch": 0.1522074281709881,
      "grad_norm": 0.6432974338531494,
      "learning_rate": 0.0002916622817507773,
      "loss": 0.2181,
      "step": 1086
    },
    {
      "epoch": 0.15234758234057463,
      "grad_norm": 0.7160272598266602,
      "learning_rate": 0.00029164793111695765,
      "loss": 0.2578,
      "step": 1087
    },
    {
      "epoch": 0.15248773651016118,
      "grad_norm": 0.7639457583427429,
      "learning_rate": 0.000291633580483138,
      "loss": 0.2939,
      "step": 1088
    },
    {
      "epoch": 0.15262789067974772,
      "grad_norm": 0.699263870716095,
      "learning_rate": 0.0002916192298493183,
      "loss": 0.2517,
      "step": 1089
    },
    {
      "epoch": 0.15276804484933426,
      "grad_norm": 0.5161824822425842,
      "learning_rate": 0.0002916048792154987,
      "loss": 0.1953,
      "step": 1090
    },
    {
      "epoch": 0.1529081990189208,
      "grad_norm": 0.4656703770160675,
      "learning_rate": 0.000291590528581679,
      "loss": 0.2245,
      "step": 1091
    },
    {
      "epoch": 0.15304835318850735,
      "grad_norm": 0.5346180200576782,
      "learning_rate": 0.00029157617794785934,
      "loss": 0.2257,
      "step": 1092
    },
    {
      "epoch": 0.15318850735809392,
      "grad_norm": 0.6907939314842224,
      "learning_rate": 0.00029156182731403967,
      "loss": 0.2394,
      "step": 1093
    },
    {
      "epoch": 0.15332866152768046,
      "grad_norm": 0.690552294254303,
      "learning_rate": 0.00029154747668022,
      "loss": 0.2181,
      "step": 1094
    },
    {
      "epoch": 0.153468815697267,
      "grad_norm": 0.43683817982673645,
      "learning_rate": 0.00029153312604640033,
      "loss": 0.1848,
      "step": 1095
    },
    {
      "epoch": 0.15360896986685355,
      "grad_norm": 0.48682838678359985,
      "learning_rate": 0.0002915187754125807,
      "loss": 0.1116,
      "step": 1096
    },
    {
      "epoch": 0.1537491240364401,
      "grad_norm": 1.0500648021697998,
      "learning_rate": 0.00029150442477876104,
      "loss": 0.3946,
      "step": 1097
    },
    {
      "epoch": 0.15388927820602663,
      "grad_norm": 2.116490364074707,
      "learning_rate": 0.00029149007414494137,
      "loss": 0.5003,
      "step": 1098
    },
    {
      "epoch": 0.15402943237561317,
      "grad_norm": 2.300340175628662,
      "learning_rate": 0.00029147572351112175,
      "loss": 0.638,
      "step": 1099
    },
    {
      "epoch": 0.15416958654519972,
      "grad_norm": 1.7466180324554443,
      "learning_rate": 0.0002914613728773021,
      "loss": 0.2625,
      "step": 1100
    },
    {
      "epoch": 0.15430974071478626,
      "grad_norm": 0.42691710591316223,
      "learning_rate": 0.0002914470222434824,
      "loss": 0.2048,
      "step": 1101
    },
    {
      "epoch": 0.1544498948843728,
      "grad_norm": 0.4420432448387146,
      "learning_rate": 0.00029143267160966274,
      "loss": 0.2027,
      "step": 1102
    },
    {
      "epoch": 0.15459004905395934,
      "grad_norm": 0.5113061666488647,
      "learning_rate": 0.00029141832097584307,
      "loss": 0.1723,
      "step": 1103
    },
    {
      "epoch": 0.1547302032235459,
      "grad_norm": 0.6693875789642334,
      "learning_rate": 0.0002914039703420234,
      "loss": 0.176,
      "step": 1104
    },
    {
      "epoch": 0.15487035739313246,
      "grad_norm": 0.41856876015663147,
      "learning_rate": 0.0002913896197082038,
      "loss": 0.1345,
      "step": 1105
    },
    {
      "epoch": 0.155010511562719,
      "grad_norm": 0.8738510012626648,
      "learning_rate": 0.0002913752690743841,
      "loss": 0.2735,
      "step": 1106
    },
    {
      "epoch": 0.15515066573230554,
      "grad_norm": 0.8897049427032471,
      "learning_rate": 0.00029136091844056443,
      "loss": 0.2832,
      "step": 1107
    },
    {
      "epoch": 0.15529081990189209,
      "grad_norm": 0.48510468006134033,
      "learning_rate": 0.00029134656780674476,
      "loss": 0.1509,
      "step": 1108
    },
    {
      "epoch": 0.15543097407147863,
      "grad_norm": 0.94893878698349,
      "learning_rate": 0.00029133221717292515,
      "loss": 0.2569,
      "step": 1109
    },
    {
      "epoch": 0.15557112824106517,
      "grad_norm": 0.5670791268348694,
      "learning_rate": 0.0002913178665391055,
      "loss": 0.1519,
      "step": 1110
    },
    {
      "epoch": 0.1557112824106517,
      "grad_norm": 0.6965529918670654,
      "learning_rate": 0.0002913035159052858,
      "loss": 0.1842,
      "step": 1111
    },
    {
      "epoch": 0.15585143658023826,
      "grad_norm": 0.3937253952026367,
      "learning_rate": 0.00029128916527146613,
      "loss": 0.1999,
      "step": 1112
    },
    {
      "epoch": 0.1559915907498248,
      "grad_norm": 0.33624860644340515,
      "learning_rate": 0.00029127481463764646,
      "loss": 0.191,
      "step": 1113
    },
    {
      "epoch": 0.15613174491941134,
      "grad_norm": 0.9810729026794434,
      "learning_rate": 0.0002912604640038268,
      "loss": 0.1928,
      "step": 1114
    },
    {
      "epoch": 0.15627189908899788,
      "grad_norm": 0.638692319393158,
      "learning_rate": 0.00029124611337000717,
      "loss": 0.1824,
      "step": 1115
    },
    {
      "epoch": 0.15641205325858445,
      "grad_norm": 1.0449035167694092,
      "learning_rate": 0.0002912317627361875,
      "loss": 0.2524,
      "step": 1116
    },
    {
      "epoch": 0.156552207428171,
      "grad_norm": 0.7117040753364563,
      "learning_rate": 0.00029121741210236783,
      "loss": 0.1824,
      "step": 1117
    },
    {
      "epoch": 0.15669236159775754,
      "grad_norm": 0.479743629693985,
      "learning_rate": 0.00029120306146854816,
      "loss": 0.2058,
      "step": 1118
    },
    {
      "epoch": 0.15683251576734408,
      "grad_norm": 0.689148485660553,
      "learning_rate": 0.0002911887108347285,
      "loss": 0.2216,
      "step": 1119
    },
    {
      "epoch": 0.15697266993693063,
      "grad_norm": 0.6677058339118958,
      "learning_rate": 0.0002911743602009088,
      "loss": 0.1864,
      "step": 1120
    },
    {
      "epoch": 0.15711282410651717,
      "grad_norm": 0.5702966451644897,
      "learning_rate": 0.0002911600095670892,
      "loss": 0.178,
      "step": 1121
    },
    {
      "epoch": 0.1572529782761037,
      "grad_norm": 0.7367022037506104,
      "learning_rate": 0.0002911456589332695,
      "loss": 0.1856,
      "step": 1122
    },
    {
      "epoch": 0.15739313244569025,
      "grad_norm": 0.5543919205665588,
      "learning_rate": 0.00029113130829944985,
      "loss": 0.1388,
      "step": 1123
    },
    {
      "epoch": 0.1575332866152768,
      "grad_norm": 0.4027252793312073,
      "learning_rate": 0.0002911169576656302,
      "loss": 0.1598,
      "step": 1124
    },
    {
      "epoch": 0.15767344078486334,
      "grad_norm": 0.5057381391525269,
      "learning_rate": 0.00029110260703181057,
      "loss": 0.1759,
      "step": 1125
    },
    {
      "epoch": 0.15781359495444988,
      "grad_norm": 0.36679747700691223,
      "learning_rate": 0.0002910882563979909,
      "loss": 0.1271,
      "step": 1126
    },
    {
      "epoch": 0.15795374912403645,
      "grad_norm": 0.36856910586357117,
      "learning_rate": 0.0002910739057641712,
      "loss": 0.0974,
      "step": 1127
    },
    {
      "epoch": 0.158093903293623,
      "grad_norm": 0.38675305247306824,
      "learning_rate": 0.00029105955513035155,
      "loss": 0.1521,
      "step": 1128
    },
    {
      "epoch": 0.15823405746320954,
      "grad_norm": 0.6520252227783203,
      "learning_rate": 0.0002910452044965319,
      "loss": 0.2322,
      "step": 1129
    },
    {
      "epoch": 0.15837421163279608,
      "grad_norm": 0.5653290748596191,
      "learning_rate": 0.0002910308538627122,
      "loss": 0.1934,
      "step": 1130
    },
    {
      "epoch": 0.15851436580238262,
      "grad_norm": 0.6227279901504517,
      "learning_rate": 0.0002910165032288926,
      "loss": 0.2423,
      "step": 1131
    },
    {
      "epoch": 0.15865451997196917,
      "grad_norm": 0.5886867642402649,
      "learning_rate": 0.0002910021525950729,
      "loss": 0.186,
      "step": 1132
    },
    {
      "epoch": 0.1587946741415557,
      "grad_norm": 0.40995121002197266,
      "learning_rate": 0.00029098780196125325,
      "loss": 0.1735,
      "step": 1133
    },
    {
      "epoch": 0.15893482831114225,
      "grad_norm": 0.40254729986190796,
      "learning_rate": 0.00029097345132743363,
      "loss": 0.1902,
      "step": 1134
    },
    {
      "epoch": 0.1590749824807288,
      "grad_norm": 1.3046905994415283,
      "learning_rate": 0.00029095910069361396,
      "loss": 0.3026,
      "step": 1135
    },
    {
      "epoch": 0.15921513665031534,
      "grad_norm": 0.6232987642288208,
      "learning_rate": 0.0002909447500597943,
      "loss": 0.3754,
      "step": 1136
    },
    {
      "epoch": 0.15935529081990188,
      "grad_norm": 0.9278186559677124,
      "learning_rate": 0.0002909303994259746,
      "loss": 0.141,
      "step": 1137
    },
    {
      "epoch": 0.15949544498948845,
      "grad_norm": 0.37663301825523376,
      "learning_rate": 0.00029091604879215495,
      "loss": 0.1289,
      "step": 1138
    },
    {
      "epoch": 0.159635599159075,
      "grad_norm": 0.6425600647926331,
      "learning_rate": 0.0002909016981583353,
      "loss": 0.2836,
      "step": 1139
    },
    {
      "epoch": 0.15977575332866154,
      "grad_norm": 0.7579872608184814,
      "learning_rate": 0.00029088734752451566,
      "loss": 0.2834,
      "step": 1140
    },
    {
      "epoch": 0.15991590749824808,
      "grad_norm": 0.608771562576294,
      "learning_rate": 0.000290872996890696,
      "loss": 0.332,
      "step": 1141
    },
    {
      "epoch": 0.16005606166783462,
      "grad_norm": 0.6470134258270264,
      "learning_rate": 0.0002908586462568763,
      "loss": 0.2865,
      "step": 1142
    },
    {
      "epoch": 0.16019621583742116,
      "grad_norm": 0.6094939112663269,
      "learning_rate": 0.00029084429562305664,
      "loss": 0.1197,
      "step": 1143
    },
    {
      "epoch": 0.1603363700070077,
      "grad_norm": 0.7962201833724976,
      "learning_rate": 0.000290829944989237,
      "loss": 0.3345,
      "step": 1144
    },
    {
      "epoch": 0.16047652417659425,
      "grad_norm": 0.8785037994384766,
      "learning_rate": 0.00029081559435541735,
      "loss": 0.2335,
      "step": 1145
    },
    {
      "epoch": 0.1606166783461808,
      "grad_norm": 1.3294227123260498,
      "learning_rate": 0.0002908012437215977,
      "loss": 0.3035,
      "step": 1146
    },
    {
      "epoch": 0.16075683251576733,
      "grad_norm": 0.6238111257553101,
      "learning_rate": 0.000290786893087778,
      "loss": 0.2428,
      "step": 1147
    },
    {
      "epoch": 0.16089698668535388,
      "grad_norm": 0.8492755889892578,
      "learning_rate": 0.00029077254245395834,
      "loss": 0.3282,
      "step": 1148
    },
    {
      "epoch": 0.16103714085494045,
      "grad_norm": 0.4155972898006439,
      "learning_rate": 0.00029075819182013867,
      "loss": 0.1277,
      "step": 1149
    },
    {
      "epoch": 0.161177295024527,
      "grad_norm": 2.380148410797119,
      "learning_rate": 0.00029074384118631905,
      "loss": 0.6203,
      "step": 1150
    },
    {
      "epoch": 0.16131744919411353,
      "grad_norm": 0.677034318447113,
      "learning_rate": 0.0002907294905524994,
      "loss": 0.2896,
      "step": 1151
    },
    {
      "epoch": 0.16145760336370008,
      "grad_norm": 0.45069196820259094,
      "learning_rate": 0.0002907151399186797,
      "loss": 0.2905,
      "step": 1152
    },
    {
      "epoch": 0.16159775753328662,
      "grad_norm": 0.33286333084106445,
      "learning_rate": 0.0002907007892848601,
      "loss": 0.2217,
      "step": 1153
    },
    {
      "epoch": 0.16173791170287316,
      "grad_norm": 0.49553343653678894,
      "learning_rate": 0.0002906864386510404,
      "loss": 0.1927,
      "step": 1154
    },
    {
      "epoch": 0.1618780658724597,
      "grad_norm": 0.4819639027118683,
      "learning_rate": 0.00029067208801722075,
      "loss": 0.1897,
      "step": 1155
    },
    {
      "epoch": 0.16201822004204625,
      "grad_norm": 0.35513758659362793,
      "learning_rate": 0.0002906577373834011,
      "loss": 0.1524,
      "step": 1156
    },
    {
      "epoch": 0.1621583742116328,
      "grad_norm": 0.7157977223396301,
      "learning_rate": 0.0002906433867495814,
      "loss": 0.3223,
      "step": 1157
    },
    {
      "epoch": 0.16229852838121933,
      "grad_norm": 0.41456320881843567,
      "learning_rate": 0.00029062903611576173,
      "loss": 0.2112,
      "step": 1158
    },
    {
      "epoch": 0.16243868255080587,
      "grad_norm": 0.41751816868782043,
      "learning_rate": 0.00029061468548194206,
      "loss": 0.1673,
      "step": 1159
    },
    {
      "epoch": 0.16257883672039244,
      "grad_norm": 0.4203392267227173,
      "learning_rate": 0.00029060033484812244,
      "loss": 0.1987,
      "step": 1160
    },
    {
      "epoch": 0.162718990889979,
      "grad_norm": 0.41105687618255615,
      "learning_rate": 0.0002905859842143028,
      "loss": 0.1593,
      "step": 1161
    },
    {
      "epoch": 0.16285914505956553,
      "grad_norm": 0.8102649450302124,
      "learning_rate": 0.0002905716335804831,
      "loss": 0.276,
      "step": 1162
    },
    {
      "epoch": 0.16299929922915207,
      "grad_norm": 0.45249101519584656,
      "learning_rate": 0.0002905572829466635,
      "loss": 0.1379,
      "step": 1163
    },
    {
      "epoch": 0.16313945339873862,
      "grad_norm": 0.7759928703308105,
      "learning_rate": 0.0002905429323128438,
      "loss": 0.2219,
      "step": 1164
    },
    {
      "epoch": 0.16327960756832516,
      "grad_norm": 0.5148497819900513,
      "learning_rate": 0.00029052858167902414,
      "loss": 0.2177,
      "step": 1165
    },
    {
      "epoch": 0.1634197617379117,
      "grad_norm": 0.644158661365509,
      "learning_rate": 0.00029051423104520447,
      "loss": 0.192,
      "step": 1166
    },
    {
      "epoch": 0.16355991590749824,
      "grad_norm": 0.6915527582168579,
      "learning_rate": 0.0002904998804113848,
      "loss": 0.2209,
      "step": 1167
    },
    {
      "epoch": 0.16370007007708479,
      "grad_norm": 0.9675236344337463,
      "learning_rate": 0.00029048552977756513,
      "loss": 0.3563,
      "step": 1168
    },
    {
      "epoch": 0.16384022424667133,
      "grad_norm": 0.40057894587516785,
      "learning_rate": 0.0002904711791437455,
      "loss": 0.1739,
      "step": 1169
    },
    {
      "epoch": 0.16398037841625787,
      "grad_norm": 0.4606488049030304,
      "learning_rate": 0.00029045682850992584,
      "loss": 0.2186,
      "step": 1170
    },
    {
      "epoch": 0.16412053258584444,
      "grad_norm": 0.6207302212715149,
      "learning_rate": 0.00029044247787610617,
      "loss": 0.2411,
      "step": 1171
    },
    {
      "epoch": 0.16426068675543098,
      "grad_norm": 2.341611862182617,
      "learning_rate": 0.00029042812724228655,
      "loss": 0.1843,
      "step": 1172
    },
    {
      "epoch": 0.16440084092501753,
      "grad_norm": 0.5193768739700317,
      "learning_rate": 0.0002904137766084669,
      "loss": 0.1503,
      "step": 1173
    },
    {
      "epoch": 0.16454099509460407,
      "grad_norm": 0.40665102005004883,
      "learning_rate": 0.0002903994259746472,
      "loss": 0.1144,
      "step": 1174
    },
    {
      "epoch": 0.1646811492641906,
      "grad_norm": 0.4301244020462036,
      "learning_rate": 0.00029038507534082754,
      "loss": 0.2114,
      "step": 1175
    },
    {
      "epoch": 0.16482130343377716,
      "grad_norm": 0.33480599522590637,
      "learning_rate": 0.00029037072470700786,
      "loss": 0.1172,
      "step": 1176
    },
    {
      "epoch": 0.1649614576033637,
      "grad_norm": 0.6966586112976074,
      "learning_rate": 0.0002903563740731882,
      "loss": 0.2555,
      "step": 1177
    },
    {
      "epoch": 0.16510161177295024,
      "grad_norm": 0.29830777645111084,
      "learning_rate": 0.0002903420234393685,
      "loss": 0.1183,
      "step": 1178
    },
    {
      "epoch": 0.16524176594253678,
      "grad_norm": 0.6931130886077881,
      "learning_rate": 0.0002903276728055489,
      "loss": 0.1152,
      "step": 1179
    },
    {
      "epoch": 0.16538192011212333,
      "grad_norm": 1.156959056854248,
      "learning_rate": 0.00029031332217172923,
      "loss": 0.1983,
      "step": 1180
    },
    {
      "epoch": 0.16552207428170987,
      "grad_norm": 0.4397577941417694,
      "learning_rate": 0.00029029897153790956,
      "loss": 0.1394,
      "step": 1181
    },
    {
      "epoch": 0.16566222845129644,
      "grad_norm": 0.6719510555267334,
      "learning_rate": 0.0002902846209040899,
      "loss": 0.2182,
      "step": 1182
    },
    {
      "epoch": 0.16580238262088298,
      "grad_norm": 0.7313705086708069,
      "learning_rate": 0.0002902702702702702,
      "loss": 0.1437,
      "step": 1183
    },
    {
      "epoch": 0.16594253679046952,
      "grad_norm": 0.5166248083114624,
      "learning_rate": 0.00029025591963645055,
      "loss": 0.1476,
      "step": 1184
    },
    {
      "epoch": 0.16608269096005607,
      "grad_norm": 0.7364222407341003,
      "learning_rate": 0.00029024156900263093,
      "loss": 0.1655,
      "step": 1185
    },
    {
      "epoch": 0.1662228451296426,
      "grad_norm": 0.30972933769226074,
      "learning_rate": 0.00029022721836881126,
      "loss": 0.1021,
      "step": 1186
    },
    {
      "epoch": 0.16636299929922915,
      "grad_norm": 0.8522205352783203,
      "learning_rate": 0.0002902128677349916,
      "loss": 0.1592,
      "step": 1187
    },
    {
      "epoch": 0.1665031534688157,
      "grad_norm": 0.41604235768318176,
      "learning_rate": 0.00029019851710117197,
      "loss": 0.2006,
      "step": 1188
    },
    {
      "epoch": 0.16664330763840224,
      "grad_norm": 0.8863270282745361,
      "learning_rate": 0.0002901841664673523,
      "loss": 0.168,
      "step": 1189
    },
    {
      "epoch": 0.16678346180798878,
      "grad_norm": 0.6913481950759888,
      "learning_rate": 0.0002901698158335326,
      "loss": 0.2031,
      "step": 1190
    },
    {
      "epoch": 0.16692361597757532,
      "grad_norm": 0.5856875777244568,
      "learning_rate": 0.00029015546519971296,
      "loss": 0.1683,
      "step": 1191
    },
    {
      "epoch": 0.16706377014716187,
      "grad_norm": 1.1891721487045288,
      "learning_rate": 0.0002901411145658933,
      "loss": 0.4038,
      "step": 1192
    },
    {
      "epoch": 0.16720392431674844,
      "grad_norm": 0.7332741022109985,
      "learning_rate": 0.0002901267639320736,
      "loss": 0.3317,
      "step": 1193
    },
    {
      "epoch": 0.16734407848633498,
      "grad_norm": 0.5969712138175964,
      "learning_rate": 0.00029011241329825394,
      "loss": 0.1826,
      "step": 1194
    },
    {
      "epoch": 0.16748423265592152,
      "grad_norm": 0.69462651014328,
      "learning_rate": 0.0002900980626644343,
      "loss": 0.2139,
      "step": 1195
    },
    {
      "epoch": 0.16762438682550806,
      "grad_norm": 0.575545072555542,
      "learning_rate": 0.00029008371203061465,
      "loss": 0.1516,
      "step": 1196
    },
    {
      "epoch": 0.1677645409950946,
      "grad_norm": 0.9290174841880798,
      "learning_rate": 0.000290069361396795,
      "loss": 0.1988,
      "step": 1197
    },
    {
      "epoch": 0.16790469516468115,
      "grad_norm": 1.6319156885147095,
      "learning_rate": 0.00029005501076297536,
      "loss": 0.2739,
      "step": 1198
    },
    {
      "epoch": 0.1680448493342677,
      "grad_norm": 0.7445603609085083,
      "learning_rate": 0.0002900406601291557,
      "loss": 0.1989,
      "step": 1199
    },
    {
      "epoch": 0.16818500350385424,
      "grad_norm": 3.079102039337158,
      "learning_rate": 0.000290026309495336,
      "loss": 0.8688,
      "step": 1200
    },
    {
      "epoch": 0.16832515767344078,
      "grad_norm": 0.5677631497383118,
      "learning_rate": 0.00029001195886151635,
      "loss": 0.1762,
      "step": 1201
    },
    {
      "epoch": 0.16846531184302732,
      "grad_norm": 0.46313443779945374,
      "learning_rate": 0.0002899976082276967,
      "loss": 0.2257,
      "step": 1202
    },
    {
      "epoch": 0.16860546601261386,
      "grad_norm": 0.5117608308792114,
      "learning_rate": 0.000289983257593877,
      "loss": 0.241,
      "step": 1203
    },
    {
      "epoch": 0.16874562018220043,
      "grad_norm": 0.5307624936103821,
      "learning_rate": 0.0002899689069600574,
      "loss": 0.2225,
      "step": 1204
    },
    {
      "epoch": 0.16888577435178698,
      "grad_norm": 0.5278559923171997,
      "learning_rate": 0.0002899545563262377,
      "loss": 0.198,
      "step": 1205
    },
    {
      "epoch": 0.16902592852137352,
      "grad_norm": 0.7261818051338196,
      "learning_rate": 0.00028994020569241805,
      "loss": 0.2842,
      "step": 1206
    },
    {
      "epoch": 0.16916608269096006,
      "grad_norm": 0.6507210731506348,
      "learning_rate": 0.00028992585505859843,
      "loss": 0.3037,
      "step": 1207
    },
    {
      "epoch": 0.1693062368605466,
      "grad_norm": 0.42979133129119873,
      "learning_rate": 0.00028991150442477876,
      "loss": 0.1736,
      "step": 1208
    },
    {
      "epoch": 0.16944639103013315,
      "grad_norm": 0.5951622724533081,
      "learning_rate": 0.0002898971537909591,
      "loss": 0.1306,
      "step": 1209
    },
    {
      "epoch": 0.1695865451997197,
      "grad_norm": 0.5094631314277649,
      "learning_rate": 0.0002898828031571394,
      "loss": 0.1604,
      "step": 1210
    },
    {
      "epoch": 0.16972669936930623,
      "grad_norm": 0.5467761754989624,
      "learning_rate": 0.00028986845252331974,
      "loss": 0.276,
      "step": 1211
    },
    {
      "epoch": 0.16986685353889278,
      "grad_norm": 0.4251236021518707,
      "learning_rate": 0.00028985410188950007,
      "loss": 0.1468,
      "step": 1212
    },
    {
      "epoch": 0.17000700770847932,
      "grad_norm": 0.5952158570289612,
      "learning_rate": 0.0002898397512556804,
      "loss": 0.169,
      "step": 1213
    },
    {
      "epoch": 0.17014716187806586,
      "grad_norm": 0.4211752116680145,
      "learning_rate": 0.0002898254006218608,
      "loss": 0.0981,
      "step": 1214
    },
    {
      "epoch": 0.1702873160476524,
      "grad_norm": 0.33614712953567505,
      "learning_rate": 0.0002898110499880411,
      "loss": 0.1024,
      "step": 1215
    },
    {
      "epoch": 0.17042747021723897,
      "grad_norm": 0.40574654936790466,
      "learning_rate": 0.00028979669935422144,
      "loss": 0.1809,
      "step": 1216
    },
    {
      "epoch": 0.17056762438682552,
      "grad_norm": 0.379718154668808,
      "learning_rate": 0.0002897823487204018,
      "loss": 0.1697,
      "step": 1217
    },
    {
      "epoch": 0.17070777855641206,
      "grad_norm": 0.6298273801803589,
      "learning_rate": 0.00028976799808658215,
      "loss": 0.1259,
      "step": 1218
    },
    {
      "epoch": 0.1708479327259986,
      "grad_norm": 0.5226387977600098,
      "learning_rate": 0.0002897536474527625,
      "loss": 0.2276,
      "step": 1219
    },
    {
      "epoch": 0.17098808689558515,
      "grad_norm": 0.32652711868286133,
      "learning_rate": 0.0002897392968189428,
      "loss": 0.1374,
      "step": 1220
    },
    {
      "epoch": 0.1711282410651717,
      "grad_norm": 0.45564866065979004,
      "learning_rate": 0.00028972494618512314,
      "loss": 0.1694,
      "step": 1221
    },
    {
      "epoch": 0.17126839523475823,
      "grad_norm": 0.5506298542022705,
      "learning_rate": 0.00028971059555130347,
      "loss": 0.2138,
      "step": 1222
    },
    {
      "epoch": 0.17140854940434477,
      "grad_norm": 0.3355783224105835,
      "learning_rate": 0.00028969624491748385,
      "loss": 0.1219,
      "step": 1223
    },
    {
      "epoch": 0.17154870357393132,
      "grad_norm": 0.3820132613182068,
      "learning_rate": 0.0002896818942836642,
      "loss": 0.1469,
      "step": 1224
    },
    {
      "epoch": 0.17168885774351786,
      "grad_norm": 0.5089153051376343,
      "learning_rate": 0.0002896675436498445,
      "loss": 0.1831,
      "step": 1225
    },
    {
      "epoch": 0.1718290119131044,
      "grad_norm": 0.38970282673835754,
      "learning_rate": 0.0002896531930160249,
      "loss": 0.1031,
      "step": 1226
    },
    {
      "epoch": 0.17196916608269097,
      "grad_norm": 0.45952531695365906,
      "learning_rate": 0.0002896388423822052,
      "loss": 0.1464,
      "step": 1227
    },
    {
      "epoch": 0.17210932025227751,
      "grad_norm": 1.0034610033035278,
      "learning_rate": 0.00028962449174838555,
      "loss": 0.1909,
      "step": 1228
    },
    {
      "epoch": 0.17224947442186406,
      "grad_norm": 0.48172807693481445,
      "learning_rate": 0.0002896101411145659,
      "loss": 0.0866,
      "step": 1229
    },
    {
      "epoch": 0.1723896285914506,
      "grad_norm": 0.6868981719017029,
      "learning_rate": 0.0002895957904807462,
      "loss": 0.2065,
      "step": 1230
    },
    {
      "epoch": 0.17252978276103714,
      "grad_norm": 0.7861095070838928,
      "learning_rate": 0.00028958143984692653,
      "loss": 0.2421,
      "step": 1231
    },
    {
      "epoch": 0.17266993693062369,
      "grad_norm": 0.5128983855247498,
      "learning_rate": 0.00028956708921310686,
      "loss": 0.1752,
      "step": 1232
    },
    {
      "epoch": 0.17281009110021023,
      "grad_norm": 0.38063517212867737,
      "learning_rate": 0.00028955273857928724,
      "loss": 0.1847,
      "step": 1233
    },
    {
      "epoch": 0.17295024526979677,
      "grad_norm": 0.5560049414634705,
      "learning_rate": 0.00028953838794546757,
      "loss": 0.2814,
      "step": 1234
    },
    {
      "epoch": 0.1730903994393833,
      "grad_norm": 0.9167608618736267,
      "learning_rate": 0.0002895240373116479,
      "loss": 0.2449,
      "step": 1235
    },
    {
      "epoch": 0.17323055360896986,
      "grad_norm": 0.28223785758018494,
      "learning_rate": 0.0002895096866778283,
      "loss": 0.0955,
      "step": 1236
    },
    {
      "epoch": 0.1733707077785564,
      "grad_norm": 0.5269321799278259,
      "learning_rate": 0.0002894953360440086,
      "loss": 0.2127,
      "step": 1237
    },
    {
      "epoch": 0.17351086194814297,
      "grad_norm": 0.7515736222267151,
      "learning_rate": 0.00028948098541018894,
      "loss": 0.1773,
      "step": 1238
    },
    {
      "epoch": 0.1736510161177295,
      "grad_norm": 0.4676588773727417,
      "learning_rate": 0.00028946663477636927,
      "loss": 0.1735,
      "step": 1239
    },
    {
      "epoch": 0.17379117028731605,
      "grad_norm": 0.9766138195991516,
      "learning_rate": 0.0002894522841425496,
      "loss": 0.2785,
      "step": 1240
    },
    {
      "epoch": 0.1739313244569026,
      "grad_norm": 1.1965867280960083,
      "learning_rate": 0.0002894379335087299,
      "loss": 0.3369,
      "step": 1241
    },
    {
      "epoch": 0.17407147862648914,
      "grad_norm": 0.7254909873008728,
      "learning_rate": 0.0002894235828749103,
      "loss": 0.218,
      "step": 1242
    },
    {
      "epoch": 0.17421163279607568,
      "grad_norm": 0.4795093834400177,
      "learning_rate": 0.00028940923224109064,
      "loss": 0.0886,
      "step": 1243
    },
    {
      "epoch": 0.17435178696566223,
      "grad_norm": 0.3527316749095917,
      "learning_rate": 0.00028939488160727097,
      "loss": 0.0889,
      "step": 1244
    },
    {
      "epoch": 0.17449194113524877,
      "grad_norm": 2.0535407066345215,
      "learning_rate": 0.0002893805309734513,
      "loss": 0.3005,
      "step": 1245
    },
    {
      "epoch": 0.1746320953048353,
      "grad_norm": 1.333620309829712,
      "learning_rate": 0.0002893661803396316,
      "loss": 0.376,
      "step": 1246
    },
    {
      "epoch": 0.17477224947442185,
      "grad_norm": 1.0444930791854858,
      "learning_rate": 0.00028935182970581195,
      "loss": 0.1221,
      "step": 1247
    },
    {
      "epoch": 0.1749124036440084,
      "grad_norm": 1.4253756999969482,
      "learning_rate": 0.0002893374790719923,
      "loss": 0.2826,
      "step": 1248
    },
    {
      "epoch": 0.17505255781359497,
      "grad_norm": 3.2174179553985596,
      "learning_rate": 0.00028932312843817266,
      "loss": 0.2926,
      "step": 1249
    },
    {
      "epoch": 0.1751927119831815,
      "grad_norm": 7.869657039642334,
      "learning_rate": 0.000289308777804353,
      "loss": 0.3666,
      "step": 1250
    },
    {
      "epoch": 0.17533286615276805,
      "grad_norm": 0.5538425445556641,
      "learning_rate": 0.0002892944271705333,
      "loss": 0.1636,
      "step": 1251
    },
    {
      "epoch": 0.1754730203223546,
      "grad_norm": 0.5059335231781006,
      "learning_rate": 0.0002892800765367137,
      "loss": 0.2067,
      "step": 1252
    },
    {
      "epoch": 0.17561317449194114,
      "grad_norm": 0.5134457945823669,
      "learning_rate": 0.00028926572590289403,
      "loss": 0.2463,
      "step": 1253
    },
    {
      "epoch": 0.17575332866152768,
      "grad_norm": 0.5054817199707031,
      "learning_rate": 0.00028925137526907436,
      "loss": 0.1702,
      "step": 1254
    },
    {
      "epoch": 0.17589348283111422,
      "grad_norm": 0.555178701877594,
      "learning_rate": 0.0002892370246352547,
      "loss": 0.1271,
      "step": 1255
    },
    {
      "epoch": 0.17603363700070077,
      "grad_norm": 0.48891961574554443,
      "learning_rate": 0.000289222674001435,
      "loss": 0.1568,
      "step": 1256
    },
    {
      "epoch": 0.1761737911702873,
      "grad_norm": 0.4247031807899475,
      "learning_rate": 0.00028920832336761535,
      "loss": 0.1396,
      "step": 1257
    },
    {
      "epoch": 0.17631394533987385,
      "grad_norm": 0.49350419640541077,
      "learning_rate": 0.00028919397273379573,
      "loss": 0.1609,
      "step": 1258
    },
    {
      "epoch": 0.1764540995094604,
      "grad_norm": 0.457766592502594,
      "learning_rate": 0.00028917962209997606,
      "loss": 0.152,
      "step": 1259
    },
    {
      "epoch": 0.17659425367904696,
      "grad_norm": 0.6182723641395569,
      "learning_rate": 0.0002891652714661564,
      "loss": 0.1174,
      "step": 1260
    },
    {
      "epoch": 0.1767344078486335,
      "grad_norm": 0.41902482509613037,
      "learning_rate": 0.00028915092083233677,
      "loss": 0.1428,
      "step": 1261
    },
    {
      "epoch": 0.17687456201822005,
      "grad_norm": 0.5420647859573364,
      "learning_rate": 0.0002891365701985171,
      "loss": 0.1686,
      "step": 1262
    },
    {
      "epoch": 0.1770147161878066,
      "grad_norm": 0.7247220873832703,
      "learning_rate": 0.0002891222195646974,
      "loss": 0.1538,
      "step": 1263
    },
    {
      "epoch": 0.17715487035739313,
      "grad_norm": 0.6142184138298035,
      "learning_rate": 0.00028910786893087775,
      "loss": 0.2385,
      "step": 1264
    },
    {
      "epoch": 0.17729502452697968,
      "grad_norm": 0.3689082860946655,
      "learning_rate": 0.0002890935182970581,
      "loss": 0.153,
      "step": 1265
    },
    {
      "epoch": 0.17743517869656622,
      "grad_norm": 0.38291871547698975,
      "learning_rate": 0.0002890791676632384,
      "loss": 0.0839,
      "step": 1266
    },
    {
      "epoch": 0.17757533286615276,
      "grad_norm": 0.504447877407074,
      "learning_rate": 0.00028906481702941874,
      "loss": 0.1569,
      "step": 1267
    },
    {
      "epoch": 0.1777154870357393,
      "grad_norm": 0.8035547137260437,
      "learning_rate": 0.0002890504663955991,
      "loss": 0.1979,
      "step": 1268
    },
    {
      "epoch": 0.17785564120532585,
      "grad_norm": 0.4465544819831848,
      "learning_rate": 0.00028903611576177945,
      "loss": 0.1963,
      "step": 1269
    },
    {
      "epoch": 0.1779957953749124,
      "grad_norm": 0.5887908339500427,
      "learning_rate": 0.0002890217651279598,
      "loss": 0.2743,
      "step": 1270
    },
    {
      "epoch": 0.17813594954449896,
      "grad_norm": 0.5897113084793091,
      "learning_rate": 0.00028900741449414016,
      "loss": 0.1245,
      "step": 1271
    },
    {
      "epoch": 0.1782761037140855,
      "grad_norm": 0.626499593257904,
      "learning_rate": 0.0002889930638603205,
      "loss": 0.1179,
      "step": 1272
    },
    {
      "epoch": 0.17841625788367205,
      "grad_norm": 0.4210624694824219,
      "learning_rate": 0.0002889787132265008,
      "loss": 0.1471,
      "step": 1273
    },
    {
      "epoch": 0.1785564120532586,
      "grad_norm": 0.5204456448554993,
      "learning_rate": 0.00028896436259268115,
      "loss": 0.2015,
      "step": 1274
    },
    {
      "epoch": 0.17869656622284513,
      "grad_norm": 0.3530441224575043,
      "learning_rate": 0.0002889500119588615,
      "loss": 0.1381,
      "step": 1275
    },
    {
      "epoch": 0.17883672039243168,
      "grad_norm": 0.4907524585723877,
      "learning_rate": 0.0002889356613250418,
      "loss": 0.0779,
      "step": 1276
    },
    {
      "epoch": 0.17897687456201822,
      "grad_norm": 0.32851338386535645,
      "learning_rate": 0.0002889213106912222,
      "loss": 0.0708,
      "step": 1277
    },
    {
      "epoch": 0.17911702873160476,
      "grad_norm": 0.749977171421051,
      "learning_rate": 0.0002889069600574025,
      "loss": 0.1663,
      "step": 1278
    },
    {
      "epoch": 0.1792571829011913,
      "grad_norm": 0.7493078708648682,
      "learning_rate": 0.00028889260942358284,
      "loss": 0.3067,
      "step": 1279
    },
    {
      "epoch": 0.17939733707077785,
      "grad_norm": 1.1422078609466553,
      "learning_rate": 0.0002888782587897632,
      "loss": 0.1707,
      "step": 1280
    },
    {
      "epoch": 0.1795374912403644,
      "grad_norm": 0.5078191757202148,
      "learning_rate": 0.00028886390815594356,
      "loss": 0.1743,
      "step": 1281
    },
    {
      "epoch": 0.17967764540995096,
      "grad_norm": 0.3899461627006531,
      "learning_rate": 0.0002888495575221239,
      "loss": 0.0804,
      "step": 1282
    },
    {
      "epoch": 0.1798177995795375,
      "grad_norm": 0.4183603525161743,
      "learning_rate": 0.0002888352068883042,
      "loss": 0.1473,
      "step": 1283
    },
    {
      "epoch": 0.17995795374912404,
      "grad_norm": 0.3961012661457062,
      "learning_rate": 0.00028882085625448454,
      "loss": 0.2304,
      "step": 1284
    },
    {
      "epoch": 0.1800981079187106,
      "grad_norm": 0.5780870914459229,
      "learning_rate": 0.00028880650562066487,
      "loss": 0.1907,
      "step": 1285
    },
    {
      "epoch": 0.18023826208829713,
      "grad_norm": 0.6096864938735962,
      "learning_rate": 0.0002887921549868452,
      "loss": 0.2281,
      "step": 1286
    },
    {
      "epoch": 0.18037841625788367,
      "grad_norm": 0.3767504096031189,
      "learning_rate": 0.0002887778043530256,
      "loss": 0.1214,
      "step": 1287
    },
    {
      "epoch": 0.18051857042747022,
      "grad_norm": 0.4762822389602661,
      "learning_rate": 0.0002887634537192059,
      "loss": 0.115,
      "step": 1288
    },
    {
      "epoch": 0.18065872459705676,
      "grad_norm": 0.6894064545631409,
      "learning_rate": 0.00028874910308538624,
      "loss": 0.2195,
      "step": 1289
    },
    {
      "epoch": 0.1807988787666433,
      "grad_norm": 0.6677369475364685,
      "learning_rate": 0.0002887347524515666,
      "loss": 0.1651,
      "step": 1290
    },
    {
      "epoch": 0.18093903293622984,
      "grad_norm": 0.7982863187789917,
      "learning_rate": 0.00028872040181774695,
      "loss": 0.1831,
      "step": 1291
    },
    {
      "epoch": 0.18107918710581639,
      "grad_norm": 1.0460959672927856,
      "learning_rate": 0.0002887060511839273,
      "loss": 0.2112,
      "step": 1292
    },
    {
      "epoch": 0.18121934127540296,
      "grad_norm": 1.3564528226852417,
      "learning_rate": 0.0002886917005501076,
      "loss": 0.1927,
      "step": 1293
    },
    {
      "epoch": 0.1813594954449895,
      "grad_norm": 0.20479118824005127,
      "learning_rate": 0.00028867734991628794,
      "loss": 0.0808,
      "step": 1294
    },
    {
      "epoch": 0.18149964961457604,
      "grad_norm": 2.3068814277648926,
      "learning_rate": 0.00028866299928246826,
      "loss": 0.2418,
      "step": 1295
    },
    {
      "epoch": 0.18163980378416258,
      "grad_norm": 1.7506601810455322,
      "learning_rate": 0.00028864864864864865,
      "loss": 0.2347,
      "step": 1296
    },
    {
      "epoch": 0.18177995795374913,
      "grad_norm": 0.8877953290939331,
      "learning_rate": 0.000288634298014829,
      "loss": 0.1053,
      "step": 1297
    },
    {
      "epoch": 0.18192011212333567,
      "grad_norm": 0.4029606282711029,
      "learning_rate": 0.0002886199473810093,
      "loss": 0.0948,
      "step": 1298
    },
    {
      "epoch": 0.1820602662929222,
      "grad_norm": 2.6804661750793457,
      "learning_rate": 0.00028860559674718963,
      "loss": 0.3532,
      "step": 1299
    },
    {
      "epoch": 0.18220042046250876,
      "grad_norm": 8.153922080993652,
      "learning_rate": 0.00028859124611337,
      "loss": 0.2741,
      "step": 1300
    },
    {
      "epoch": 0.1823405746320953,
      "grad_norm": 0.6255004405975342,
      "learning_rate": 0.00028857689547955034,
      "loss": 0.2317,
      "step": 1301
    },
    {
      "epoch": 0.18248072880168184,
      "grad_norm": 0.3549646735191345,
      "learning_rate": 0.00028856254484573067,
      "loss": 0.1315,
      "step": 1302
    },
    {
      "epoch": 0.18262088297126838,
      "grad_norm": 0.6672636866569519,
      "learning_rate": 0.000288548194211911,
      "loss": 0.1785,
      "step": 1303
    },
    {
      "epoch": 0.18276103714085495,
      "grad_norm": 0.2588634490966797,
      "learning_rate": 0.00028853384357809133,
      "loss": 0.1197,
      "step": 1304
    },
    {
      "epoch": 0.1829011913104415,
      "grad_norm": 0.9322260618209839,
      "learning_rate": 0.00028851949294427166,
      "loss": 0.1204,
      "step": 1305
    },
    {
      "epoch": 0.18304134548002804,
      "grad_norm": 0.8750333786010742,
      "learning_rate": 0.00028850514231045204,
      "loss": 0.1787,
      "step": 1306
    },
    {
      "epoch": 0.18318149964961458,
      "grad_norm": 0.6682153940200806,
      "learning_rate": 0.00028849079167663237,
      "loss": 0.1977,
      "step": 1307
    },
    {
      "epoch": 0.18332165381920112,
      "grad_norm": 0.4636789858341217,
      "learning_rate": 0.0002884764410428127,
      "loss": 0.2323,
      "step": 1308
    },
    {
      "epoch": 0.18346180798878767,
      "grad_norm": 0.5990692377090454,
      "learning_rate": 0.000288462090408993,
      "loss": 0.1087,
      "step": 1309
    },
    {
      "epoch": 0.1836019621583742,
      "grad_norm": 0.7650004029273987,
      "learning_rate": 0.00028844773977517336,
      "loss": 0.2298,
      "step": 1310
    },
    {
      "epoch": 0.18374211632796075,
      "grad_norm": 0.5602273941040039,
      "learning_rate": 0.0002884333891413537,
      "loss": 0.1699,
      "step": 1311
    },
    {
      "epoch": 0.1838822704975473,
      "grad_norm": 0.5199315547943115,
      "learning_rate": 0.00028841903850753407,
      "loss": 0.2183,
      "step": 1312
    },
    {
      "epoch": 0.18402242466713384,
      "grad_norm": 0.608881950378418,
      "learning_rate": 0.0002884046878737144,
      "loss": 0.191,
      "step": 1313
    },
    {
      "epoch": 0.18416257883672038,
      "grad_norm": 0.48028749227523804,
      "learning_rate": 0.0002883903372398947,
      "loss": 0.1747,
      "step": 1314
    },
    {
      "epoch": 0.18430273300630695,
      "grad_norm": 0.4894683361053467,
      "learning_rate": 0.00028837598660607505,
      "loss": 0.1669,
      "step": 1315
    },
    {
      "epoch": 0.1844428871758935,
      "grad_norm": 0.48642149567604065,
      "learning_rate": 0.00028836163597225543,
      "loss": 0.1613,
      "step": 1316
    },
    {
      "epoch": 0.18458304134548004,
      "grad_norm": 0.9096436500549316,
      "learning_rate": 0.00028834728533843576,
      "loss": 0.1661,
      "step": 1317
    },
    {
      "epoch": 0.18472319551506658,
      "grad_norm": 0.3663023114204407,
      "learning_rate": 0.0002883329347046161,
      "loss": 0.0566,
      "step": 1318
    },
    {
      "epoch": 0.18486334968465312,
      "grad_norm": 0.34242933988571167,
      "learning_rate": 0.0002883185840707964,
      "loss": 0.1048,
      "step": 1319
    },
    {
      "epoch": 0.18500350385423966,
      "grad_norm": 0.4216375946998596,
      "learning_rate": 0.00028830423343697675,
      "loss": 0.1892,
      "step": 1320
    },
    {
      "epoch": 0.1851436580238262,
      "grad_norm": 0.46315309405326843,
      "learning_rate": 0.0002882898828031571,
      "loss": 0.1357,
      "step": 1321
    },
    {
      "epoch": 0.18528381219341275,
      "grad_norm": 0.35083702206611633,
      "learning_rate": 0.00028827553216933746,
      "loss": 0.2003,
      "step": 1322
    },
    {
      "epoch": 0.1854239663629993,
      "grad_norm": 0.3840385973453522,
      "learning_rate": 0.0002882611815355178,
      "loss": 0.0842,
      "step": 1323
    },
    {
      "epoch": 0.18556412053258584,
      "grad_norm": 0.4021473824977875,
      "learning_rate": 0.0002882468309016981,
      "loss": 0.1556,
      "step": 1324
    },
    {
      "epoch": 0.18570427470217238,
      "grad_norm": 0.5646082758903503,
      "learning_rate": 0.0002882324802678785,
      "loss": 0.1444,
      "step": 1325
    },
    {
      "epoch": 0.18584442887175892,
      "grad_norm": 0.7728922367095947,
      "learning_rate": 0.00028821812963405883,
      "loss": 0.2644,
      "step": 1326
    },
    {
      "epoch": 0.1859845830413455,
      "grad_norm": 0.4661784768104553,
      "learning_rate": 0.00028820377900023916,
      "loss": 0.2211,
      "step": 1327
    },
    {
      "epoch": 0.18612473721093203,
      "grad_norm": 0.6245259046554565,
      "learning_rate": 0.0002881894283664195,
      "loss": 0.2828,
      "step": 1328
    },
    {
      "epoch": 0.18626489138051858,
      "grad_norm": 0.4976140558719635,
      "learning_rate": 0.0002881750777325998,
      "loss": 0.1125,
      "step": 1329
    },
    {
      "epoch": 0.18640504555010512,
      "grad_norm": 0.2560175061225891,
      "learning_rate": 0.00028816072709878014,
      "loss": 0.0535,
      "step": 1330
    },
    {
      "epoch": 0.18654519971969166,
      "grad_norm": 0.3945830762386322,
      "learning_rate": 0.0002881463764649605,
      "loss": 0.0972,
      "step": 1331
    },
    {
      "epoch": 0.1866853538892782,
      "grad_norm": 0.4562370181083679,
      "learning_rate": 0.00028813202583114085,
      "loss": 0.2328,
      "step": 1332
    },
    {
      "epoch": 0.18682550805886475,
      "grad_norm": 0.5884147882461548,
      "learning_rate": 0.0002881176751973212,
      "loss": 0.1608,
      "step": 1333
    },
    {
      "epoch": 0.1869656622284513,
      "grad_norm": 0.38829508423805237,
      "learning_rate": 0.0002881033245635015,
      "loss": 0.1331,
      "step": 1334
    },
    {
      "epoch": 0.18710581639803783,
      "grad_norm": 1.426316261291504,
      "learning_rate": 0.0002880889739296819,
      "loss": 0.3349,
      "step": 1335
    },
    {
      "epoch": 0.18724597056762438,
      "grad_norm": 0.5064998269081116,
      "learning_rate": 0.0002880746232958622,
      "loss": 0.1161,
      "step": 1336
    },
    {
      "epoch": 0.18738612473721092,
      "grad_norm": 1.0166008472442627,
      "learning_rate": 0.00028806027266204255,
      "loss": 0.2141,
      "step": 1337
    },
    {
      "epoch": 0.1875262789067975,
      "grad_norm": 0.5756743550300598,
      "learning_rate": 0.0002880459220282229,
      "loss": 0.1583,
      "step": 1338
    },
    {
      "epoch": 0.18766643307638403,
      "grad_norm": 1.0782535076141357,
      "learning_rate": 0.0002880315713944032,
      "loss": 0.1775,
      "step": 1339
    },
    {
      "epoch": 0.18780658724597057,
      "grad_norm": 0.32486552000045776,
      "learning_rate": 0.00028801722076058354,
      "loss": 0.1643,
      "step": 1340
    },
    {
      "epoch": 0.18794674141555712,
      "grad_norm": 0.5556925535202026,
      "learning_rate": 0.0002880028701267639,
      "loss": 0.174,
      "step": 1341
    },
    {
      "epoch": 0.18808689558514366,
      "grad_norm": 0.811795175075531,
      "learning_rate": 0.00028798851949294425,
      "loss": 0.1511,
      "step": 1342
    },
    {
      "epoch": 0.1882270497547302,
      "grad_norm": 1.1259410381317139,
      "learning_rate": 0.0002879741688591246,
      "loss": 0.3106,
      "step": 1343
    },
    {
      "epoch": 0.18836720392431675,
      "grad_norm": 1.4740264415740967,
      "learning_rate": 0.00028795981822530496,
      "loss": 0.4054,
      "step": 1344
    },
    {
      "epoch": 0.1885073580939033,
      "grad_norm": 1.2227998971939087,
      "learning_rate": 0.0002879454675914853,
      "loss": 0.2528,
      "step": 1345
    },
    {
      "epoch": 0.18864751226348983,
      "grad_norm": 0.9776144027709961,
      "learning_rate": 0.0002879311169576656,
      "loss": 0.2043,
      "step": 1346
    },
    {
      "epoch": 0.18878766643307637,
      "grad_norm": 0.8177399635314941,
      "learning_rate": 0.00028791676632384595,
      "loss": 0.1982,
      "step": 1347
    },
    {
      "epoch": 0.18892782060266292,
      "grad_norm": 2.1901967525482178,
      "learning_rate": 0.0002879024156900263,
      "loss": 0.1518,
      "step": 1348
    },
    {
      "epoch": 0.1890679747722495,
      "grad_norm": 1.3925788402557373,
      "learning_rate": 0.0002878880650562066,
      "loss": 0.2656,
      "step": 1349
    },
    {
      "epoch": 0.18920812894183603,
      "grad_norm": 4.939290523529053,
      "learning_rate": 0.00028787371442238693,
      "loss": 0.3708,
      "step": 1350
    },
    {
      "epoch": 0.18934828311142257,
      "grad_norm": 0.5326824188232422,
      "learning_rate": 0.0002878593637885673,
      "loss": 0.2052,
      "step": 1351
    },
    {
      "epoch": 0.18948843728100911,
      "grad_norm": 0.4210324287414551,
      "learning_rate": 0.00028784501315474764,
      "loss": 0.1451,
      "step": 1352
    },
    {
      "epoch": 0.18962859145059566,
      "grad_norm": 0.49748069047927856,
      "learning_rate": 0.00028783066252092797,
      "loss": 0.1511,
      "step": 1353
    },
    {
      "epoch": 0.1897687456201822,
      "grad_norm": 0.28506726026535034,
      "learning_rate": 0.00028781631188710835,
      "loss": 0.1263,
      "step": 1354
    },
    {
      "epoch": 0.18990889978976874,
      "grad_norm": 0.4613712430000305,
      "learning_rate": 0.0002878019612532887,
      "loss": 0.2148,
      "step": 1355
    },
    {
      "epoch": 0.19004905395935529,
      "grad_norm": 0.368649959564209,
      "learning_rate": 0.000287787610619469,
      "loss": 0.1333,
      "step": 1356
    },
    {
      "epoch": 0.19018920812894183,
      "grad_norm": 1.4647291898727417,
      "learning_rate": 0.00028777325998564934,
      "loss": 0.2487,
      "step": 1357
    },
    {
      "epoch": 0.19032936229852837,
      "grad_norm": 0.3617301285266876,
      "learning_rate": 0.00028775890935182967,
      "loss": 0.1319,
      "step": 1358
    },
    {
      "epoch": 0.1904695164681149,
      "grad_norm": 0.5744082927703857,
      "learning_rate": 0.00028774455871801,
      "loss": 0.1897,
      "step": 1359
    },
    {
      "epoch": 0.19060967063770148,
      "grad_norm": 0.5286508202552795,
      "learning_rate": 0.0002877302080841904,
      "loss": 0.1503,
      "step": 1360
    },
    {
      "epoch": 0.19074982480728803,
      "grad_norm": 0.661304771900177,
      "learning_rate": 0.0002877158574503707,
      "loss": 0.1246,
      "step": 1361
    },
    {
      "epoch": 0.19088997897687457,
      "grad_norm": 0.5126602053642273,
      "learning_rate": 0.00028770150681655104,
      "loss": 0.189,
      "step": 1362
    },
    {
      "epoch": 0.1910301331464611,
      "grad_norm": 0.3035547435283661,
      "learning_rate": 0.0002876871561827314,
      "loss": 0.0818,
      "step": 1363
    },
    {
      "epoch": 0.19117028731604765,
      "grad_norm": 0.6415405869483948,
      "learning_rate": 0.00028767280554891175,
      "loss": 0.1362,
      "step": 1364
    },
    {
      "epoch": 0.1913104414856342,
      "grad_norm": 0.8118913769721985,
      "learning_rate": 0.0002876584549150921,
      "loss": 0.1395,
      "step": 1365
    },
    {
      "epoch": 0.19145059565522074,
      "grad_norm": 0.3385295271873474,
      "learning_rate": 0.0002876441042812724,
      "loss": 0.18,
      "step": 1366
    },
    {
      "epoch": 0.19159074982480728,
      "grad_norm": 0.6493491530418396,
      "learning_rate": 0.00028762975364745273,
      "loss": 0.2065,
      "step": 1367
    },
    {
      "epoch": 0.19173090399439383,
      "grad_norm": 0.7460654973983765,
      "learning_rate": 0.00028761540301363306,
      "loss": 0.1064,
      "step": 1368
    },
    {
      "epoch": 0.19187105816398037,
      "grad_norm": 0.7090083956718445,
      "learning_rate": 0.0002876010523798134,
      "loss": 0.1694,
      "step": 1369
    },
    {
      "epoch": 0.1920112123335669,
      "grad_norm": 0.34292009472846985,
      "learning_rate": 0.0002875867017459938,
      "loss": 0.1415,
      "step": 1370
    },
    {
      "epoch": 0.19215136650315348,
      "grad_norm": 0.48215794563293457,
      "learning_rate": 0.0002875723511121741,
      "loss": 0.1198,
      "step": 1371
    },
    {
      "epoch": 0.19229152067274002,
      "grad_norm": 0.5662413239479065,
      "learning_rate": 0.00028755800047835443,
      "loss": 0.1172,
      "step": 1372
    },
    {
      "epoch": 0.19243167484232657,
      "grad_norm": 0.6156641840934753,
      "learning_rate": 0.00028754364984453476,
      "loss": 0.2177,
      "step": 1373
    },
    {
      "epoch": 0.1925718290119131,
      "grad_norm": 0.4750995635986328,
      "learning_rate": 0.0002875292992107151,
      "loss": 0.1776,
      "step": 1374
    },
    {
      "epoch": 0.19271198318149965,
      "grad_norm": 0.5293868184089661,
      "learning_rate": 0.0002875149485768954,
      "loss": 0.1552,
      "step": 1375
    },
    {
      "epoch": 0.1928521373510862,
      "grad_norm": 0.9702204465866089,
      "learning_rate": 0.0002875005979430758,
      "loss": 0.1385,
      "step": 1376
    },
    {
      "epoch": 0.19299229152067274,
      "grad_norm": 0.38313114643096924,
      "learning_rate": 0.00028748624730925613,
      "loss": 0.132,
      "step": 1377
    },
    {
      "epoch": 0.19313244569025928,
      "grad_norm": 0.7046389579772949,
      "learning_rate": 0.00028747189667543646,
      "loss": 0.1788,
      "step": 1378
    },
    {
      "epoch": 0.19327259985984582,
      "grad_norm": 0.5358118414878845,
      "learning_rate": 0.00028745754604161684,
      "loss": 0.0801,
      "step": 1379
    },
    {
      "epoch": 0.19341275402943237,
      "grad_norm": 0.47918954491615295,
      "learning_rate": 0.00028744319540779717,
      "loss": 0.154,
      "step": 1380
    },
    {
      "epoch": 0.1935529081990189,
      "grad_norm": 0.44163206219673157,
      "learning_rate": 0.0002874288447739775,
      "loss": 0.089,
      "step": 1381
    },
    {
      "epoch": 0.19369306236860548,
      "grad_norm": 0.6863482594490051,
      "learning_rate": 0.0002874144941401578,
      "loss": 0.1171,
      "step": 1382
    },
    {
      "epoch": 0.19383321653819202,
      "grad_norm": 0.45869338512420654,
      "learning_rate": 0.00028740014350633815,
      "loss": 0.1539,
      "step": 1383
    },
    {
      "epoch": 0.19397337070777856,
      "grad_norm": 0.47624510526657104,
      "learning_rate": 0.0002873857928725185,
      "loss": 0.0989,
      "step": 1384
    },
    {
      "epoch": 0.1941135248773651,
      "grad_norm": 0.4153755307197571,
      "learning_rate": 0.0002873714422386988,
      "loss": 0.1437,
      "step": 1385
    },
    {
      "epoch": 0.19425367904695165,
      "grad_norm": 0.5450621843338013,
      "learning_rate": 0.0002873570916048792,
      "loss": 0.1553,
      "step": 1386
    },
    {
      "epoch": 0.1943938332165382,
      "grad_norm": 0.3404199182987213,
      "learning_rate": 0.0002873427409710595,
      "loss": 0.0683,
      "step": 1387
    },
    {
      "epoch": 0.19453398738612473,
      "grad_norm": 0.7234523892402649,
      "learning_rate": 0.00028732839033723985,
      "loss": 0.0938,
      "step": 1388
    },
    {
      "epoch": 0.19467414155571128,
      "grad_norm": 0.9329773783683777,
      "learning_rate": 0.00028731403970342023,
      "loss": 0.115,
      "step": 1389
    },
    {
      "epoch": 0.19481429572529782,
      "grad_norm": 0.9350193738937378,
      "learning_rate": 0.00028729968906960056,
      "loss": 0.1766,
      "step": 1390
    },
    {
      "epoch": 0.19495444989488436,
      "grad_norm": 0.5233431458473206,
      "learning_rate": 0.0002872853384357809,
      "loss": 0.1586,
      "step": 1391
    },
    {
      "epoch": 0.1950946040644709,
      "grad_norm": 0.683952808380127,
      "learning_rate": 0.0002872709878019612,
      "loss": 0.1786,
      "step": 1392
    },
    {
      "epoch": 0.19523475823405748,
      "grad_norm": 1.8429738283157349,
      "learning_rate": 0.00028725663716814155,
      "loss": 0.4211,
      "step": 1393
    },
    {
      "epoch": 0.19537491240364402,
      "grad_norm": 1.6243500709533691,
      "learning_rate": 0.0002872422865343219,
      "loss": 0.2126,
      "step": 1394
    },
    {
      "epoch": 0.19551506657323056,
      "grad_norm": 0.9260746240615845,
      "learning_rate": 0.00028722793590050226,
      "loss": 0.2339,
      "step": 1395
    },
    {
      "epoch": 0.1956552207428171,
      "grad_norm": 0.9837828278541565,
      "learning_rate": 0.0002872135852666826,
      "loss": 0.2341,
      "step": 1396
    },
    {
      "epoch": 0.19579537491240365,
      "grad_norm": 1.4956419467926025,
      "learning_rate": 0.0002871992346328629,
      "loss": 0.1598,
      "step": 1397
    },
    {
      "epoch": 0.1959355290819902,
      "grad_norm": 3.089427947998047,
      "learning_rate": 0.0002871848839990433,
      "loss": 0.1495,
      "step": 1398
    },
    {
      "epoch": 0.19607568325157673,
      "grad_norm": 2.4646904468536377,
      "learning_rate": 0.0002871705333652236,
      "loss": 0.467,
      "step": 1399
    },
    {
      "epoch": 0.19621583742116327,
      "grad_norm": 3.984333038330078,
      "learning_rate": 0.00028715618273140396,
      "loss": 0.4142,
      "step": 1400
    },
    {
      "epoch": 0.19635599159074982,
      "grad_norm": 0.7834465503692627,
      "learning_rate": 0.0002871418320975843,
      "loss": 0.3954,
      "step": 1401
    },
    {
      "epoch": 0.19649614576033636,
      "grad_norm": 0.5351545214653015,
      "learning_rate": 0.0002871274814637646,
      "loss": 0.1846,
      "step": 1402
    },
    {
      "epoch": 0.1966362999299229,
      "grad_norm": 0.6234658360481262,
      "learning_rate": 0.00028711313082994494,
      "loss": 0.2162,
      "step": 1403
    },
    {
      "epoch": 0.19677645409950947,
      "grad_norm": 0.414691299200058,
      "learning_rate": 0.00028709878019612527,
      "loss": 0.0986,
      "step": 1404
    },
    {
      "epoch": 0.19691660826909602,
      "grad_norm": 0.5032076239585876,
      "learning_rate": 0.00028708442956230565,
      "loss": 0.1129,
      "step": 1405
    },
    {
      "epoch": 0.19705676243868256,
      "grad_norm": 0.3615491986274719,
      "learning_rate": 0.000287070078928486,
      "loss": 0.1238,
      "step": 1406
    },
    {
      "epoch": 0.1971969166082691,
      "grad_norm": 0.4679245948791504,
      "learning_rate": 0.0002870557282946663,
      "loss": 0.2511,
      "step": 1407
    },
    {
      "epoch": 0.19733707077785564,
      "grad_norm": 0.4984559714794159,
      "learning_rate": 0.0002870413776608467,
      "loss": 0.2127,
      "step": 1408
    },
    {
      "epoch": 0.1974772249474422,
      "grad_norm": 0.46827319264411926,
      "learning_rate": 0.000287027027027027,
      "loss": 0.1863,
      "step": 1409
    },
    {
      "epoch": 0.19761737911702873,
      "grad_norm": 0.339138388633728,
      "learning_rate": 0.00028701267639320735,
      "loss": 0.1832,
      "step": 1410
    },
    {
      "epoch": 0.19775753328661527,
      "grad_norm": 0.434836208820343,
      "learning_rate": 0.0002869983257593877,
      "loss": 0.1421,
      "step": 1411
    },
    {
      "epoch": 0.19789768745620182,
      "grad_norm": 0.39442136883735657,
      "learning_rate": 0.000286983975125568,
      "loss": 0.2238,
      "step": 1412
    },
    {
      "epoch": 0.19803784162578836,
      "grad_norm": 0.5383408665657043,
      "learning_rate": 0.00028696962449174834,
      "loss": 0.1585,
      "step": 1413
    },
    {
      "epoch": 0.1981779957953749,
      "grad_norm": 0.48914170265197754,
      "learning_rate": 0.0002869552738579287,
      "loss": 0.1656,
      "step": 1414
    },
    {
      "epoch": 0.19831814996496147,
      "grad_norm": 0.3213764429092407,
      "learning_rate": 0.00028694092322410905,
      "loss": 0.0942,
      "step": 1415
    },
    {
      "epoch": 0.198458304134548,
      "grad_norm": 0.41128483414649963,
      "learning_rate": 0.0002869265725902894,
      "loss": 0.1012,
      "step": 1416
    },
    {
      "epoch": 0.19859845830413456,
      "grad_norm": 1.0505154132843018,
      "learning_rate": 0.00028691222195646976,
      "loss": 0.1758,
      "step": 1417
    },
    {
      "epoch": 0.1987386124737211,
      "grad_norm": 0.5430999994277954,
      "learning_rate": 0.0002868978713226501,
      "loss": 0.1706,
      "step": 1418
    },
    {
      "epoch": 0.19887876664330764,
      "grad_norm": 0.35043174028396606,
      "learning_rate": 0.0002868835206888304,
      "loss": 0.1575,
      "step": 1419
    },
    {
      "epoch": 0.19901892081289418,
      "grad_norm": 0.5660715699195862,
      "learning_rate": 0.00028686917005501074,
      "loss": 0.2044,
      "step": 1420
    },
    {
      "epoch": 0.19915907498248073,
      "grad_norm": 0.7045791745185852,
      "learning_rate": 0.00028685481942119107,
      "loss": 0.0895,
      "step": 1421
    },
    {
      "epoch": 0.19929922915206727,
      "grad_norm": 0.6727139949798584,
      "learning_rate": 0.0002868404687873714,
      "loss": 0.1915,
      "step": 1422
    },
    {
      "epoch": 0.1994393833216538,
      "grad_norm": 0.7017869353294373,
      "learning_rate": 0.00028682611815355173,
      "loss": 0.1363,
      "step": 1423
    },
    {
      "epoch": 0.19957953749124036,
      "grad_norm": 0.5484580993652344,
      "learning_rate": 0.0002868117675197321,
      "loss": 0.1651,
      "step": 1424
    },
    {
      "epoch": 0.1997196916608269,
      "grad_norm": 0.5813344717025757,
      "learning_rate": 0.00028679741688591244,
      "loss": 0.1533,
      "step": 1425
    },
    {
      "epoch": 0.19985984583041347,
      "grad_norm": 0.3388593792915344,
      "learning_rate": 0.00028678306625209277,
      "loss": 0.1263,
      "step": 1426
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.5175798535346985,
      "learning_rate": 0.00028676871561827315,
      "loss": 0.1913,
      "step": 1427
    },
    {
      "epoch": 0.20014015416958655,
      "grad_norm": 0.9213618636131287,
      "learning_rate": 0.0002867543649844535,
      "loss": 0.1763,
      "step": 1428
    },
    {
      "epoch": 0.2002803083391731,
      "grad_norm": 0.422063410282135,
      "learning_rate": 0.0002867400143506338,
      "loss": 0.1423,
      "step": 1429
    },
    {
      "epoch": 0.20042046250875964,
      "grad_norm": 1.2014931440353394,
      "learning_rate": 0.00028672566371681414,
      "loss": 0.1637,
      "step": 1430
    },
    {
      "epoch": 0.20056061667834618,
      "grad_norm": 0.5003225207328796,
      "learning_rate": 0.00028671131308299447,
      "loss": 0.2425,
      "step": 1431
    },
    {
      "epoch": 0.20070077084793272,
      "grad_norm": 0.42734184861183167,
      "learning_rate": 0.0002866969624491748,
      "loss": 0.1308,
      "step": 1432
    },
    {
      "epoch": 0.20084092501751927,
      "grad_norm": 0.7353927493095398,
      "learning_rate": 0.0002866826118153552,
      "loss": 0.1463,
      "step": 1433
    },
    {
      "epoch": 0.2009810791871058,
      "grad_norm": 0.4108189046382904,
      "learning_rate": 0.0002866682611815355,
      "loss": 0.091,
      "step": 1434
    },
    {
      "epoch": 0.20112123335669235,
      "grad_norm": 0.8403388261795044,
      "learning_rate": 0.00028665391054771583,
      "loss": 0.2363,
      "step": 1435
    },
    {
      "epoch": 0.2012613875262789,
      "grad_norm": 1.2731560468673706,
      "learning_rate": 0.00028663955991389616,
      "loss": 0.1101,
      "step": 1436
    },
    {
      "epoch": 0.20140154169586547,
      "grad_norm": 0.630267322063446,
      "learning_rate": 0.0002866252092800765,
      "loss": 0.1613,
      "step": 1437
    },
    {
      "epoch": 0.201541695865452,
      "grad_norm": 0.3950931429862976,
      "learning_rate": 0.0002866108586462568,
      "loss": 0.1389,
      "step": 1438
    },
    {
      "epoch": 0.20168185003503855,
      "grad_norm": 0.8729126453399658,
      "learning_rate": 0.0002865965080124372,
      "loss": 0.1055,
      "step": 1439
    },
    {
      "epoch": 0.2018220042046251,
      "grad_norm": 0.677051842212677,
      "learning_rate": 0.00028658215737861753,
      "loss": 0.1355,
      "step": 1440
    },
    {
      "epoch": 0.20196215837421164,
      "grad_norm": 0.24138006567955017,
      "learning_rate": 0.00028656780674479786,
      "loss": 0.0625,
      "step": 1441
    },
    {
      "epoch": 0.20210231254379818,
      "grad_norm": 0.36985257267951965,
      "learning_rate": 0.0002865534561109782,
      "loss": 0.0869,
      "step": 1442
    },
    {
      "epoch": 0.20224246671338472,
      "grad_norm": 0.8477890491485596,
      "learning_rate": 0.00028653910547715857,
      "loss": 0.1193,
      "step": 1443
    },
    {
      "epoch": 0.20238262088297126,
      "grad_norm": 0.593531608581543,
      "learning_rate": 0.0002865247548433389,
      "loss": 0.09,
      "step": 1444
    },
    {
      "epoch": 0.2025227750525578,
      "grad_norm": 0.7947191596031189,
      "learning_rate": 0.00028651040420951923,
      "loss": 0.1022,
      "step": 1445
    },
    {
      "epoch": 0.20266292922214435,
      "grad_norm": 0.9791816473007202,
      "learning_rate": 0.00028649605357569956,
      "loss": 0.1804,
      "step": 1446
    },
    {
      "epoch": 0.2028030833917309,
      "grad_norm": 2.58573842048645,
      "learning_rate": 0.0002864817029418799,
      "loss": 0.2212,
      "step": 1447
    },
    {
      "epoch": 0.20294323756131744,
      "grad_norm": 1.0994421243667603,
      "learning_rate": 0.0002864673523080602,
      "loss": 0.2621,
      "step": 1448
    },
    {
      "epoch": 0.203083391730904,
      "grad_norm": 4.8346734046936035,
      "learning_rate": 0.0002864530016742406,
      "loss": 0.6657,
      "step": 1449
    },
    {
      "epoch": 0.20322354590049055,
      "grad_norm": 1.9288171529769897,
      "learning_rate": 0.0002864386510404209,
      "loss": 0.1628,
      "step": 1450
    },
    {
      "epoch": 0.2033637000700771,
      "grad_norm": 0.9603410363197327,
      "learning_rate": 0.00028642430040660125,
      "loss": 0.217,
      "step": 1451
    },
    {
      "epoch": 0.20350385423966363,
      "grad_norm": 0.513755202293396,
      "learning_rate": 0.00028640994977278164,
      "loss": 0.1873,
      "step": 1452
    },
    {
      "epoch": 0.20364400840925018,
      "grad_norm": 0.3526681065559387,
      "learning_rate": 0.00028639559913896197,
      "loss": 0.0829,
      "step": 1453
    },
    {
      "epoch": 0.20378416257883672,
      "grad_norm": 0.7088523507118225,
      "learning_rate": 0.0002863812485051423,
      "loss": 0.1644,
      "step": 1454
    },
    {
      "epoch": 0.20392431674842326,
      "grad_norm": 0.420288622379303,
      "learning_rate": 0.0002863668978713226,
      "loss": 0.089,
      "step": 1455
    },
    {
      "epoch": 0.2040644709180098,
      "grad_norm": 0.5015362501144409,
      "learning_rate": 0.00028635254723750295,
      "loss": 0.179,
      "step": 1456
    },
    {
      "epoch": 0.20420462508759635,
      "grad_norm": 1.0615626573562622,
      "learning_rate": 0.0002863381966036833,
      "loss": 0.166,
      "step": 1457
    },
    {
      "epoch": 0.2043447792571829,
      "grad_norm": 0.793479859828949,
      "learning_rate": 0.0002863238459698636,
      "loss": 0.1935,
      "step": 1458
    },
    {
      "epoch": 0.20448493342676943,
      "grad_norm": 0.48837974667549133,
      "learning_rate": 0.000286309495336044,
      "loss": 0.1676,
      "step": 1459
    },
    {
      "epoch": 0.204625087596356,
      "grad_norm": 0.8710083365440369,
      "learning_rate": 0.0002862951447022243,
      "loss": 0.2354,
      "step": 1460
    },
    {
      "epoch": 0.20476524176594255,
      "grad_norm": 0.4601697325706482,
      "learning_rate": 0.00028628079406840465,
      "loss": 0.1746,
      "step": 1461
    },
    {
      "epoch": 0.2049053959355291,
      "grad_norm": 0.4005391299724579,
      "learning_rate": 0.00028626644343458503,
      "loss": 0.0803,
      "step": 1462
    },
    {
      "epoch": 0.20504555010511563,
      "grad_norm": 0.75310879945755,
      "learning_rate": 0.00028625209280076536,
      "loss": 0.1524,
      "step": 1463
    },
    {
      "epoch": 0.20518570427470217,
      "grad_norm": 0.5398361086845398,
      "learning_rate": 0.0002862377421669457,
      "loss": 0.183,
      "step": 1464
    },
    {
      "epoch": 0.20532585844428872,
      "grad_norm": 0.29483023285865784,
      "learning_rate": 0.000286223391533126,
      "loss": 0.0921,
      "step": 1465
    },
    {
      "epoch": 0.20546601261387526,
      "grad_norm": 0.7226589918136597,
      "learning_rate": 0.00028620904089930635,
      "loss": 0.146,
      "step": 1466
    },
    {
      "epoch": 0.2056061667834618,
      "grad_norm": 0.799090564250946,
      "learning_rate": 0.0002861946902654867,
      "loss": 0.2097,
      "step": 1467
    },
    {
      "epoch": 0.20574632095304835,
      "grad_norm": 0.7113801836967468,
      "learning_rate": 0.00028618033963166706,
      "loss": 0.158,
      "step": 1468
    },
    {
      "epoch": 0.2058864751226349,
      "grad_norm": 0.3885248303413391,
      "learning_rate": 0.0002861659889978474,
      "loss": 0.1376,
      "step": 1469
    },
    {
      "epoch": 0.20602662929222143,
      "grad_norm": 0.5118241310119629,
      "learning_rate": 0.0002861516383640277,
      "loss": 0.1761,
      "step": 1470
    },
    {
      "epoch": 0.206166783461808,
      "grad_norm": 0.4573477804660797,
      "learning_rate": 0.0002861372877302081,
      "loss": 0.1556,
      "step": 1471
    },
    {
      "epoch": 0.20630693763139454,
      "grad_norm": 0.5322966575622559,
      "learning_rate": 0.0002861229370963884,
      "loss": 0.194,
      "step": 1472
    },
    {
      "epoch": 0.2064470918009811,
      "grad_norm": 0.5503672957420349,
      "learning_rate": 0.00028610858646256875,
      "loss": 0.212,
      "step": 1473
    },
    {
      "epoch": 0.20658724597056763,
      "grad_norm": 0.5848712921142578,
      "learning_rate": 0.0002860942358287491,
      "loss": 0.1735,
      "step": 1474
    },
    {
      "epoch": 0.20672740014015417,
      "grad_norm": 0.8324587345123291,
      "learning_rate": 0.0002860798851949294,
      "loss": 0.1932,
      "step": 1475
    },
    {
      "epoch": 0.20686755430974071,
      "grad_norm": 0.4048483073711395,
      "learning_rate": 0.00028606553456110974,
      "loss": 0.1362,
      "step": 1476
    },
    {
      "epoch": 0.20700770847932726,
      "grad_norm": 0.46509283781051636,
      "learning_rate": 0.00028605118392729007,
      "loss": 0.0806,
      "step": 1477
    },
    {
      "epoch": 0.2071478626489138,
      "grad_norm": 0.5476142168045044,
      "learning_rate": 0.00028603683329347045,
      "loss": 0.1505,
      "step": 1478
    },
    {
      "epoch": 0.20728801681850034,
      "grad_norm": 0.7694693803787231,
      "learning_rate": 0.0002860224826596508,
      "loss": 0.1832,
      "step": 1479
    },
    {
      "epoch": 0.20742817098808689,
      "grad_norm": 0.7152474522590637,
      "learning_rate": 0.0002860081320258311,
      "loss": 0.1734,
      "step": 1480
    },
    {
      "epoch": 0.20756832515767343,
      "grad_norm": 0.8956233859062195,
      "learning_rate": 0.0002859937813920115,
      "loss": 0.2818,
      "step": 1481
    },
    {
      "epoch": 0.20770847932726,
      "grad_norm": 0.41789087653160095,
      "learning_rate": 0.0002859794307581918,
      "loss": 0.1164,
      "step": 1482
    },
    {
      "epoch": 0.20784863349684654,
      "grad_norm": 0.5818623900413513,
      "learning_rate": 0.00028596508012437215,
      "loss": 0.1244,
      "step": 1483
    },
    {
      "epoch": 0.20798878766643308,
      "grad_norm": 0.4546411335468292,
      "learning_rate": 0.0002859507294905525,
      "loss": 0.1317,
      "step": 1484
    },
    {
      "epoch": 0.20812894183601963,
      "grad_norm": 0.7050706744194031,
      "learning_rate": 0.0002859363788567328,
      "loss": 0.2242,
      "step": 1485
    },
    {
      "epoch": 0.20826909600560617,
      "grad_norm": 0.40126219391822815,
      "learning_rate": 0.00028592202822291313,
      "loss": 0.0961,
      "step": 1486
    },
    {
      "epoch": 0.2084092501751927,
      "grad_norm": 0.2570498585700989,
      "learning_rate": 0.0002859076775890935,
      "loss": 0.085,
      "step": 1487
    },
    {
      "epoch": 0.20854940434477925,
      "grad_norm": 0.48797452449798584,
      "learning_rate": 0.00028589332695527384,
      "loss": 0.1351,
      "step": 1488
    },
    {
      "epoch": 0.2086895585143658,
      "grad_norm": 0.4947861135005951,
      "learning_rate": 0.0002858789763214542,
      "loss": 0.1715,
      "step": 1489
    },
    {
      "epoch": 0.20882971268395234,
      "grad_norm": 0.5540120601654053,
      "learning_rate": 0.0002858646256876345,
      "loss": 0.1817,
      "step": 1490
    },
    {
      "epoch": 0.20896986685353888,
      "grad_norm": 0.4998747408390045,
      "learning_rate": 0.0002858502750538149,
      "loss": 0.1797,
      "step": 1491
    },
    {
      "epoch": 0.20911002102312543,
      "grad_norm": 0.3598507344722748,
      "learning_rate": 0.0002858359244199952,
      "loss": 0.0924,
      "step": 1492
    },
    {
      "epoch": 0.209250175192712,
      "grad_norm": 1.6876142024993896,
      "learning_rate": 0.00028582157378617554,
      "loss": 0.112,
      "step": 1493
    },
    {
      "epoch": 0.20939032936229854,
      "grad_norm": 0.6825133562088013,
      "learning_rate": 0.00028580722315235587,
      "loss": 0.1467,
      "step": 1494
    },
    {
      "epoch": 0.20953048353188508,
      "grad_norm": 1.0023508071899414,
      "learning_rate": 0.0002857928725185362,
      "loss": 0.2285,
      "step": 1495
    },
    {
      "epoch": 0.20967063770147162,
      "grad_norm": 0.6522983312606812,
      "learning_rate": 0.00028577852188471653,
      "loss": 0.1628,
      "step": 1496
    },
    {
      "epoch": 0.20981079187105817,
      "grad_norm": 0.8857536911964417,
      "learning_rate": 0.0002857641712508969,
      "loss": 0.2025,
      "step": 1497
    },
    {
      "epoch": 0.2099509460406447,
      "grad_norm": 1.0567398071289062,
      "learning_rate": 0.00028574982061707724,
      "loss": 0.3421,
      "step": 1498
    },
    {
      "epoch": 0.21009110021023125,
      "grad_norm": 2.0932374000549316,
      "learning_rate": 0.00028573546998325757,
      "loss": 0.2864,
      "step": 1499
    },
    {
      "epoch": 0.2102312543798178,
      "grad_norm": 1.6516791582107544,
      "learning_rate": 0.0002857211193494379,
      "loss": 0.4194,
      "step": 1500
    },
    {
      "epoch": 0.21037140854940434,
      "grad_norm": 0.5387606024742126,
      "learning_rate": 0.0002857067687156182,
      "loss": 0.1382,
      "step": 1501
    },
    {
      "epoch": 0.21051156271899088,
      "grad_norm": 0.8462368845939636,
      "learning_rate": 0.00028569241808179855,
      "loss": 0.3042,
      "step": 1502
    },
    {
      "epoch": 0.21065171688857742,
      "grad_norm": 0.3508150577545166,
      "learning_rate": 0.00028567806744797894,
      "loss": 0.1083,
      "step": 1503
    },
    {
      "epoch": 0.210791871058164,
      "grad_norm": 0.6700693368911743,
      "learning_rate": 0.00028566371681415926,
      "loss": 0.1737,
      "step": 1504
    },
    {
      "epoch": 0.21093202522775054,
      "grad_norm": 0.27587416768074036,
      "learning_rate": 0.0002856493661803396,
      "loss": 0.092,
      "step": 1505
    },
    {
      "epoch": 0.21107217939733708,
      "grad_norm": 0.5318671464920044,
      "learning_rate": 0.0002856350155465199,
      "loss": 0.2715,
      "step": 1506
    },
    {
      "epoch": 0.21121233356692362,
      "grad_norm": 0.6454751491546631,
      "learning_rate": 0.0002856206649127003,
      "loss": 0.2197,
      "step": 1507
    },
    {
      "epoch": 0.21135248773651016,
      "grad_norm": 0.33636653423309326,
      "learning_rate": 0.00028560631427888063,
      "loss": 0.1302,
      "step": 1508
    },
    {
      "epoch": 0.2114926419060967,
      "grad_norm": 0.44161441922187805,
      "learning_rate": 0.00028559196364506096,
      "loss": 0.1617,
      "step": 1509
    },
    {
      "epoch": 0.21163279607568325,
      "grad_norm": 0.2663300931453705,
      "learning_rate": 0.0002855776130112413,
      "loss": 0.0803,
      "step": 1510
    },
    {
      "epoch": 0.2117729502452698,
      "grad_norm": 0.5626079440116882,
      "learning_rate": 0.0002855632623774216,
      "loss": 0.1307,
      "step": 1511
    },
    {
      "epoch": 0.21191310441485633,
      "grad_norm": 0.5090451836585999,
      "learning_rate": 0.00028554891174360195,
      "loss": 0.1519,
      "step": 1512
    },
    {
      "epoch": 0.21205325858444288,
      "grad_norm": 0.5128536820411682,
      "learning_rate": 0.00028553456110978233,
      "loss": 0.1458,
      "step": 1513
    },
    {
      "epoch": 0.21219341275402942,
      "grad_norm": 0.43973883986473083,
      "learning_rate": 0.00028552021047596266,
      "loss": 0.1516,
      "step": 1514
    },
    {
      "epoch": 0.212333566923616,
      "grad_norm": 0.36193305253982544,
      "learning_rate": 0.000285505859842143,
      "loss": 0.0904,
      "step": 1515
    },
    {
      "epoch": 0.21247372109320253,
      "grad_norm": 0.6568478941917419,
      "learning_rate": 0.00028549150920832337,
      "loss": 0.1232,
      "step": 1516
    },
    {
      "epoch": 0.21261387526278908,
      "grad_norm": 0.5821335315704346,
      "learning_rate": 0.0002854771585745037,
      "loss": 0.1279,
      "step": 1517
    },
    {
      "epoch": 0.21275402943237562,
      "grad_norm": 0.3105257749557495,
      "learning_rate": 0.000285462807940684,
      "loss": 0.1304,
      "step": 1518
    },
    {
      "epoch": 0.21289418360196216,
      "grad_norm": 0.5969409346580505,
      "learning_rate": 0.00028544845730686436,
      "loss": 0.1366,
      "step": 1519
    },
    {
      "epoch": 0.2130343377715487,
      "grad_norm": 0.8048906326293945,
      "learning_rate": 0.0002854341066730447,
      "loss": 0.099,
      "step": 1520
    },
    {
      "epoch": 0.21317449194113525,
      "grad_norm": 0.3927711248397827,
      "learning_rate": 0.000285419756039225,
      "loss": 0.1209,
      "step": 1521
    },
    {
      "epoch": 0.2133146461107218,
      "grad_norm": 0.523062527179718,
      "learning_rate": 0.0002854054054054054,
      "loss": 0.165,
      "step": 1522
    },
    {
      "epoch": 0.21345480028030833,
      "grad_norm": 1.0338835716247559,
      "learning_rate": 0.0002853910547715857,
      "loss": 0.1858,
      "step": 1523
    },
    {
      "epoch": 0.21359495444989487,
      "grad_norm": 1.2094815969467163,
      "learning_rate": 0.00028537670413776605,
      "loss": 0.1197,
      "step": 1524
    },
    {
      "epoch": 0.21373510861948142,
      "grad_norm": 0.6911324262619019,
      "learning_rate": 0.0002853623535039464,
      "loss": 0.1792,
      "step": 1525
    },
    {
      "epoch": 0.213875262789068,
      "grad_norm": 0.8866291642189026,
      "learning_rate": 0.00028534800287012676,
      "loss": 0.2176,
      "step": 1526
    },
    {
      "epoch": 0.21401541695865453,
      "grad_norm": 0.48767244815826416,
      "learning_rate": 0.0002853336522363071,
      "loss": 0.1429,
      "step": 1527
    },
    {
      "epoch": 0.21415557112824107,
      "grad_norm": 0.27532628178596497,
      "learning_rate": 0.0002853193016024874,
      "loss": 0.0895,
      "step": 1528
    },
    {
      "epoch": 0.21429572529782762,
      "grad_norm": 0.4672256112098694,
      "learning_rate": 0.00028530495096866775,
      "loss": 0.2238,
      "step": 1529
    },
    {
      "epoch": 0.21443587946741416,
      "grad_norm": 0.3992745280265808,
      "learning_rate": 0.0002852906003348481,
      "loss": 0.1187,
      "step": 1530
    },
    {
      "epoch": 0.2145760336370007,
      "grad_norm": 0.7503077983856201,
      "learning_rate": 0.0002852762497010284,
      "loss": 0.1155,
      "step": 1531
    },
    {
      "epoch": 0.21471618780658724,
      "grad_norm": 0.47434207797050476,
      "learning_rate": 0.0002852618990672088,
      "loss": 0.1223,
      "step": 1532
    },
    {
      "epoch": 0.2148563419761738,
      "grad_norm": 0.26094192266464233,
      "learning_rate": 0.0002852475484333891,
      "loss": 0.1307,
      "step": 1533
    },
    {
      "epoch": 0.21499649614576033,
      "grad_norm": 0.8494062423706055,
      "learning_rate": 0.00028523319779956945,
      "loss": 0.1225,
      "step": 1534
    },
    {
      "epoch": 0.21513665031534687,
      "grad_norm": 0.5196310877799988,
      "learning_rate": 0.00028521884716574983,
      "loss": 0.1435,
      "step": 1535
    },
    {
      "epoch": 0.21527680448493342,
      "grad_norm": 0.6249077916145325,
      "learning_rate": 0.00028520449653193016,
      "loss": 0.141,
      "step": 1536
    },
    {
      "epoch": 0.21541695865451999,
      "grad_norm": 0.37485843896865845,
      "learning_rate": 0.0002851901458981105,
      "loss": 0.1085,
      "step": 1537
    },
    {
      "epoch": 0.21555711282410653,
      "grad_norm": 0.4761439859867096,
      "learning_rate": 0.0002851757952642908,
      "loss": 0.1515,
      "step": 1538
    },
    {
      "epoch": 0.21569726699369307,
      "grad_norm": 0.4248480200767517,
      "learning_rate": 0.00028516144463047114,
      "loss": 0.1783,
      "step": 1539
    },
    {
      "epoch": 0.2158374211632796,
      "grad_norm": 0.4296650290489197,
      "learning_rate": 0.00028514709399665147,
      "loss": 0.1289,
      "step": 1540
    },
    {
      "epoch": 0.21597757533286616,
      "grad_norm": 0.708044707775116,
      "learning_rate": 0.0002851327433628318,
      "loss": 0.1794,
      "step": 1541
    },
    {
      "epoch": 0.2161177295024527,
      "grad_norm": 1.3418065309524536,
      "learning_rate": 0.0002851183927290122,
      "loss": 0.1624,
      "step": 1542
    },
    {
      "epoch": 0.21625788367203924,
      "grad_norm": 0.4759913980960846,
      "learning_rate": 0.0002851040420951925,
      "loss": 0.1792,
      "step": 1543
    },
    {
      "epoch": 0.21639803784162578,
      "grad_norm": 0.5173408389091492,
      "learning_rate": 0.00028508969146137284,
      "loss": 0.2286,
      "step": 1544
    },
    {
      "epoch": 0.21653819201121233,
      "grad_norm": 0.9507098197937012,
      "learning_rate": 0.0002850753408275532,
      "loss": 0.3068,
      "step": 1545
    },
    {
      "epoch": 0.21667834618079887,
      "grad_norm": 0.7133569121360779,
      "learning_rate": 0.00028506099019373355,
      "loss": 0.1311,
      "step": 1546
    },
    {
      "epoch": 0.2168185003503854,
      "grad_norm": 1.594622254371643,
      "learning_rate": 0.0002850466395599139,
      "loss": 0.6437,
      "step": 1547
    },
    {
      "epoch": 0.21695865451997198,
      "grad_norm": 0.3971281349658966,
      "learning_rate": 0.0002850322889260942,
      "loss": 0.0561,
      "step": 1548
    },
    {
      "epoch": 0.21709880868955853,
      "grad_norm": 2.201596975326538,
      "learning_rate": 0.00028501793829227454,
      "loss": 0.3337,
      "step": 1549
    },
    {
      "epoch": 0.21723896285914507,
      "grad_norm": 1.44367253780365,
      "learning_rate": 0.00028500358765845487,
      "loss": 0.3225,
      "step": 1550
    },
    {
      "epoch": 0.2173791170287316,
      "grad_norm": 0.5062150359153748,
      "learning_rate": 0.00028498923702463525,
      "loss": 0.1961,
      "step": 1551
    },
    {
      "epoch": 0.21751927119831815,
      "grad_norm": 0.30372190475463867,
      "learning_rate": 0.0002849748863908156,
      "loss": 0.1103,
      "step": 1552
    },
    {
      "epoch": 0.2176594253679047,
      "grad_norm": 0.35695815086364746,
      "learning_rate": 0.0002849605357569959,
      "loss": 0.1488,
      "step": 1553
    },
    {
      "epoch": 0.21779957953749124,
      "grad_norm": 0.3762584328651428,
      "learning_rate": 0.0002849461851231763,
      "loss": 0.1434,
      "step": 1554
    },
    {
      "epoch": 0.21793973370707778,
      "grad_norm": 0.45391854643821716,
      "learning_rate": 0.0002849318344893566,
      "loss": 0.1492,
      "step": 1555
    },
    {
      "epoch": 0.21807988787666432,
      "grad_norm": 0.5545883178710938,
      "learning_rate": 0.00028491748385553695,
      "loss": 0.1336,
      "step": 1556
    },
    {
      "epoch": 0.21822004204625087,
      "grad_norm": 0.4484400451183319,
      "learning_rate": 0.0002849031332217173,
      "loss": 0.1452,
      "step": 1557
    },
    {
      "epoch": 0.2183601962158374,
      "grad_norm": 0.49088966846466064,
      "learning_rate": 0.0002848887825878976,
      "loss": 0.2148,
      "step": 1558
    },
    {
      "epoch": 0.21850035038542395,
      "grad_norm": 0.5772733092308044,
      "learning_rate": 0.00028487443195407793,
      "loss": 0.1981,
      "step": 1559
    },
    {
      "epoch": 0.21864050455501052,
      "grad_norm": 0.5382531881332397,
      "learning_rate": 0.00028486008132025826,
      "loss": 0.188,
      "step": 1560
    },
    {
      "epoch": 0.21878065872459707,
      "grad_norm": 0.2525745630264282,
      "learning_rate": 0.00028484573068643864,
      "loss": 0.0731,
      "step": 1561
    },
    {
      "epoch": 0.2189208128941836,
      "grad_norm": 0.5901613235473633,
      "learning_rate": 0.00028483138005261897,
      "loss": 0.1267,
      "step": 1562
    },
    {
      "epoch": 0.21906096706377015,
      "grad_norm": 0.38726237416267395,
      "learning_rate": 0.0002848170294187993,
      "loss": 0.077,
      "step": 1563
    },
    {
      "epoch": 0.2192011212333567,
      "grad_norm": 0.5521304607391357,
      "learning_rate": 0.00028480267878497963,
      "loss": 0.1845,
      "step": 1564
    },
    {
      "epoch": 0.21934127540294324,
      "grad_norm": 0.664147138595581,
      "learning_rate": 0.00028478832815115996,
      "loss": 0.169,
      "step": 1565
    },
    {
      "epoch": 0.21948142957252978,
      "grad_norm": 0.5293416380882263,
      "learning_rate": 0.00028477397751734034,
      "loss": 0.2059,
      "step": 1566
    },
    {
      "epoch": 0.21962158374211632,
      "grad_norm": 0.4721493124961853,
      "learning_rate": 0.00028475962688352067,
      "loss": 0.1252,
      "step": 1567
    },
    {
      "epoch": 0.21976173791170286,
      "grad_norm": 1.5911215543746948,
      "learning_rate": 0.000284745276249701,
      "loss": 0.2105,
      "step": 1568
    },
    {
      "epoch": 0.2199018920812894,
      "grad_norm": 0.4312524199485779,
      "learning_rate": 0.0002847309256158813,
      "loss": 0.1612,
      "step": 1569
    },
    {
      "epoch": 0.22004204625087595,
      "grad_norm": 0.4137956500053406,
      "learning_rate": 0.0002847165749820617,
      "loss": 0.0973,
      "step": 1570
    },
    {
      "epoch": 0.22018220042046252,
      "grad_norm": 0.568093478679657,
      "learning_rate": 0.00028470222434824204,
      "loss": 0.1812,
      "step": 1571
    },
    {
      "epoch": 0.22032235459004906,
      "grad_norm": 0.5128418803215027,
      "learning_rate": 0.00028468787371442237,
      "loss": 0.1782,
      "step": 1572
    },
    {
      "epoch": 0.2204625087596356,
      "grad_norm": 0.4304228723049164,
      "learning_rate": 0.0002846735230806027,
      "loss": 0.1027,
      "step": 1573
    },
    {
      "epoch": 0.22060266292922215,
      "grad_norm": 0.7367760539054871,
      "learning_rate": 0.000284659172446783,
      "loss": 0.2063,
      "step": 1574
    },
    {
      "epoch": 0.2207428170988087,
      "grad_norm": 0.5827558040618896,
      "learning_rate": 0.00028464482181296335,
      "loss": 0.1445,
      "step": 1575
    },
    {
      "epoch": 0.22088297126839523,
      "grad_norm": 0.4219644069671631,
      "learning_rate": 0.0002846304711791437,
      "loss": 0.1038,
      "step": 1576
    },
    {
      "epoch": 0.22102312543798178,
      "grad_norm": 0.4029349982738495,
      "learning_rate": 0.00028461612054532406,
      "loss": 0.08,
      "step": 1577
    },
    {
      "epoch": 0.22116327960756832,
      "grad_norm": 0.6098003387451172,
      "learning_rate": 0.0002846017699115044,
      "loss": 0.1589,
      "step": 1578
    },
    {
      "epoch": 0.22130343377715486,
      "grad_norm": 0.7220091223716736,
      "learning_rate": 0.0002845874192776847,
      "loss": 0.0989,
      "step": 1579
    },
    {
      "epoch": 0.2214435879467414,
      "grad_norm": 0.5958103537559509,
      "learning_rate": 0.0002845730686438651,
      "loss": 0.0983,
      "step": 1580
    },
    {
      "epoch": 0.22158374211632795,
      "grad_norm": 0.6537734866142273,
      "learning_rate": 0.00028455871801004543,
      "loss": 0.168,
      "step": 1581
    },
    {
      "epoch": 0.22172389628591452,
      "grad_norm": 0.3661618232727051,
      "learning_rate": 0.00028454436737622576,
      "loss": 0.1639,
      "step": 1582
    },
    {
      "epoch": 0.22186405045550106,
      "grad_norm": 0.5802404284477234,
      "learning_rate": 0.0002845300167424061,
      "loss": 0.1556,
      "step": 1583
    },
    {
      "epoch": 0.2220042046250876,
      "grad_norm": 0.7161227464675903,
      "learning_rate": 0.0002845156661085864,
      "loss": 0.1229,
      "step": 1584
    },
    {
      "epoch": 0.22214435879467415,
      "grad_norm": 0.32338374853134155,
      "learning_rate": 0.00028450131547476674,
      "loss": 0.1018,
      "step": 1585
    },
    {
      "epoch": 0.2222845129642607,
      "grad_norm": 0.43043947219848633,
      "learning_rate": 0.00028448696484094713,
      "loss": 0.1294,
      "step": 1586
    },
    {
      "epoch": 0.22242466713384723,
      "grad_norm": 1.0216281414031982,
      "learning_rate": 0.00028447261420712746,
      "loss": 0.2008,
      "step": 1587
    },
    {
      "epoch": 0.22256482130343377,
      "grad_norm": 0.8226951360702515,
      "learning_rate": 0.0002844582635733078,
      "loss": 0.1864,
      "step": 1588
    },
    {
      "epoch": 0.22270497547302032,
      "grad_norm": 0.5976907014846802,
      "learning_rate": 0.00028444391293948817,
      "loss": 0.1547,
      "step": 1589
    },
    {
      "epoch": 0.22284512964260686,
      "grad_norm": 0.3465006947517395,
      "learning_rate": 0.0002844295623056685,
      "loss": 0.0855,
      "step": 1590
    },
    {
      "epoch": 0.2229852838121934,
      "grad_norm": 2.0604941844940186,
      "learning_rate": 0.0002844152116718488,
      "loss": 0.2475,
      "step": 1591
    },
    {
      "epoch": 0.22312543798177994,
      "grad_norm": 0.43012574315071106,
      "learning_rate": 0.00028440086103802915,
      "loss": 0.1734,
      "step": 1592
    },
    {
      "epoch": 0.22326559215136652,
      "grad_norm": 0.5485913157463074,
      "learning_rate": 0.0002843865104042095,
      "loss": 0.1396,
      "step": 1593
    },
    {
      "epoch": 0.22340574632095306,
      "grad_norm": 0.5287312865257263,
      "learning_rate": 0.0002843721597703898,
      "loss": 0.1607,
      "step": 1594
    },
    {
      "epoch": 0.2235459004905396,
      "grad_norm": 1.0549744367599487,
      "learning_rate": 0.00028435780913657014,
      "loss": 0.1337,
      "step": 1595
    },
    {
      "epoch": 0.22368605466012614,
      "grad_norm": 0.39720407128334045,
      "learning_rate": 0.0002843434585027505,
      "loss": 0.0794,
      "step": 1596
    },
    {
      "epoch": 0.2238262088297127,
      "grad_norm": 0.4827709197998047,
      "learning_rate": 0.00028432910786893085,
      "loss": 0.1158,
      "step": 1597
    },
    {
      "epoch": 0.22396636299929923,
      "grad_norm": 0.7434794902801514,
      "learning_rate": 0.0002843147572351112,
      "loss": 0.1725,
      "step": 1598
    },
    {
      "epoch": 0.22410651716888577,
      "grad_norm": 2.4218194484710693,
      "learning_rate": 0.00028430040660129156,
      "loss": 0.197,
      "step": 1599
    },
    {
      "epoch": 0.22424667133847231,
      "grad_norm": 4.562673091888428,
      "learning_rate": 0.0002842860559674719,
      "loss": 0.7878,
      "step": 1600
    },
    {
      "epoch": 0.22438682550805886,
      "grad_norm": 0.4756203889846802,
      "learning_rate": 0.0002842717053336522,
      "loss": 0.1586,
      "step": 1601
    },
    {
      "epoch": 0.2245269796776454,
      "grad_norm": 0.28987544775009155,
      "learning_rate": 0.00028425735469983255,
      "loss": 0.0946,
      "step": 1602
    },
    {
      "epoch": 0.22466713384723194,
      "grad_norm": 0.5316409468650818,
      "learning_rate": 0.0002842430040660129,
      "loss": 0.0897,
      "step": 1603
    },
    {
      "epoch": 0.2248072880168185,
      "grad_norm": 0.5533539056777954,
      "learning_rate": 0.0002842286534321932,
      "loss": 0.1509,
      "step": 1604
    },
    {
      "epoch": 0.22494744218640506,
      "grad_norm": 0.3826991319656372,
      "learning_rate": 0.0002842143027983736,
      "loss": 0.1302,
      "step": 1605
    },
    {
      "epoch": 0.2250875963559916,
      "grad_norm": 0.6411821246147156,
      "learning_rate": 0.0002841999521645539,
      "loss": 0.1069,
      "step": 1606
    },
    {
      "epoch": 0.22522775052557814,
      "grad_norm": 1.9257643222808838,
      "learning_rate": 0.00028418560153073424,
      "loss": 0.094,
      "step": 1607
    },
    {
      "epoch": 0.22536790469516468,
      "grad_norm": 0.4175720512866974,
      "learning_rate": 0.0002841712508969146,
      "loss": 0.0768,
      "step": 1608
    },
    {
      "epoch": 0.22550805886475123,
      "grad_norm": 0.4755411446094513,
      "learning_rate": 0.00028415690026309496,
      "loss": 0.1902,
      "step": 1609
    },
    {
      "epoch": 0.22564821303433777,
      "grad_norm": 0.8655763864517212,
      "learning_rate": 0.0002841425496292753,
      "loss": 0.2774,
      "step": 1610
    },
    {
      "epoch": 0.2257883672039243,
      "grad_norm": 0.7017194628715515,
      "learning_rate": 0.0002841281989954556,
      "loss": 0.2686,
      "step": 1611
    },
    {
      "epoch": 0.22592852137351085,
      "grad_norm": 0.49989691376686096,
      "learning_rate": 0.00028411384836163594,
      "loss": 0.1321,
      "step": 1612
    },
    {
      "epoch": 0.2260686755430974,
      "grad_norm": 0.43530431389808655,
      "learning_rate": 0.00028409949772781627,
      "loss": 0.1604,
      "step": 1613
    },
    {
      "epoch": 0.22620882971268394,
      "grad_norm": 0.6284531354904175,
      "learning_rate": 0.0002840851470939966,
      "loss": 0.2665,
      "step": 1614
    },
    {
      "epoch": 0.2263489838822705,
      "grad_norm": 0.40533867478370667,
      "learning_rate": 0.000284070796460177,
      "loss": 0.1033,
      "step": 1615
    },
    {
      "epoch": 0.22648913805185705,
      "grad_norm": 1.365997314453125,
      "learning_rate": 0.0002840564458263573,
      "loss": 0.1557,
      "step": 1616
    },
    {
      "epoch": 0.2266292922214436,
      "grad_norm": 0.4368007481098175,
      "learning_rate": 0.00028404209519253764,
      "loss": 0.1064,
      "step": 1617
    },
    {
      "epoch": 0.22676944639103014,
      "grad_norm": 0.5887719392776489,
      "learning_rate": 0.000284027744558718,
      "loss": 0.275,
      "step": 1618
    },
    {
      "epoch": 0.22690960056061668,
      "grad_norm": 0.3346814215183258,
      "learning_rate": 0.00028401339392489835,
      "loss": 0.0857,
      "step": 1619
    },
    {
      "epoch": 0.22704975473020322,
      "grad_norm": 0.46464040875434875,
      "learning_rate": 0.0002839990432910787,
      "loss": 0.1826,
      "step": 1620
    },
    {
      "epoch": 0.22718990889978977,
      "grad_norm": 0.5752182602882385,
      "learning_rate": 0.000283984692657259,
      "loss": 0.2196,
      "step": 1621
    },
    {
      "epoch": 0.2273300630693763,
      "grad_norm": 0.5127241611480713,
      "learning_rate": 0.00028397034202343934,
      "loss": 0.1482,
      "step": 1622
    },
    {
      "epoch": 0.22747021723896285,
      "grad_norm": 0.7479202747344971,
      "learning_rate": 0.00028395599138961966,
      "loss": 0.1001,
      "step": 1623
    },
    {
      "epoch": 0.2276103714085494,
      "grad_norm": 0.4733017683029175,
      "learning_rate": 0.00028394164075580005,
      "loss": 0.1854,
      "step": 1624
    },
    {
      "epoch": 0.22775052557813594,
      "grad_norm": 0.6494266390800476,
      "learning_rate": 0.0002839272901219804,
      "loss": 0.0679,
      "step": 1625
    },
    {
      "epoch": 0.2278906797477225,
      "grad_norm": 0.6814122796058655,
      "learning_rate": 0.0002839129394881607,
      "loss": 0.157,
      "step": 1626
    },
    {
      "epoch": 0.22803083391730905,
      "grad_norm": 1.1917665004730225,
      "learning_rate": 0.00028389858885434103,
      "loss": 0.2648,
      "step": 1627
    },
    {
      "epoch": 0.2281709880868956,
      "grad_norm": 0.3667110502719879,
      "learning_rate": 0.00028388423822052136,
      "loss": 0.0965,
      "step": 1628
    },
    {
      "epoch": 0.22831114225648214,
      "grad_norm": 0.4370315968990326,
      "learning_rate": 0.0002838698875867017,
      "loss": 0.113,
      "step": 1629
    },
    {
      "epoch": 0.22845129642606868,
      "grad_norm": 0.2094385325908661,
      "learning_rate": 0.00028385553695288207,
      "loss": 0.0975,
      "step": 1630
    },
    {
      "epoch": 0.22859145059565522,
      "grad_norm": 0.5804761052131653,
      "learning_rate": 0.0002838411863190624,
      "loss": 0.2258,
      "step": 1631
    },
    {
      "epoch": 0.22873160476524176,
      "grad_norm": 0.632855236530304,
      "learning_rate": 0.00028382683568524273,
      "loss": 0.166,
      "step": 1632
    },
    {
      "epoch": 0.2288717589348283,
      "grad_norm": 0.4397091567516327,
      "learning_rate": 0.00028381248505142306,
      "loss": 0.1756,
      "step": 1633
    },
    {
      "epoch": 0.22901191310441485,
      "grad_norm": 0.4569563567638397,
      "learning_rate": 0.00028379813441760344,
      "loss": 0.1451,
      "step": 1634
    },
    {
      "epoch": 0.2291520672740014,
      "grad_norm": 0.2829950153827667,
      "learning_rate": 0.00028378378378378377,
      "loss": 0.0757,
      "step": 1635
    },
    {
      "epoch": 0.22929222144358793,
      "grad_norm": 0.3748777210712433,
      "learning_rate": 0.0002837694331499641,
      "loss": 0.1053,
      "step": 1636
    },
    {
      "epoch": 0.2294323756131745,
      "grad_norm": 0.6762419939041138,
      "learning_rate": 0.0002837550825161444,
      "loss": 0.1543,
      "step": 1637
    },
    {
      "epoch": 0.22957252978276105,
      "grad_norm": 0.5129497647285461,
      "learning_rate": 0.00028374073188232475,
      "loss": 0.1149,
      "step": 1638
    },
    {
      "epoch": 0.2297126839523476,
      "grad_norm": 0.8427003622055054,
      "learning_rate": 0.0002837263812485051,
      "loss": 0.2351,
      "step": 1639
    },
    {
      "epoch": 0.22985283812193413,
      "grad_norm": 0.6942387819290161,
      "learning_rate": 0.00028371203061468547,
      "loss": 0.1763,
      "step": 1640
    },
    {
      "epoch": 0.22999299229152068,
      "grad_norm": 0.6184607148170471,
      "learning_rate": 0.0002836976799808658,
      "loss": 0.1157,
      "step": 1641
    },
    {
      "epoch": 0.23013314646110722,
      "grad_norm": 0.38860252499580383,
      "learning_rate": 0.0002836833293470461,
      "loss": 0.1023,
      "step": 1642
    },
    {
      "epoch": 0.23027330063069376,
      "grad_norm": 1.0864946842193604,
      "learning_rate": 0.0002836689787132265,
      "loss": 0.2841,
      "step": 1643
    },
    {
      "epoch": 0.2304134548002803,
      "grad_norm": 0.30519142746925354,
      "learning_rate": 0.00028365462807940683,
      "loss": 0.1072,
      "step": 1644
    },
    {
      "epoch": 0.23055360896986685,
      "grad_norm": 4.428708076477051,
      "learning_rate": 0.00028364027744558716,
      "loss": 0.3774,
      "step": 1645
    },
    {
      "epoch": 0.2306937631394534,
      "grad_norm": 0.7120837569236755,
      "learning_rate": 0.0002836259268117675,
      "loss": 0.1201,
      "step": 1646
    },
    {
      "epoch": 0.23083391730903993,
      "grad_norm": 1.7699705362319946,
      "learning_rate": 0.0002836115761779478,
      "loss": 0.5443,
      "step": 1647
    },
    {
      "epoch": 0.2309740714786265,
      "grad_norm": 0.9952479004859924,
      "learning_rate": 0.00028359722554412815,
      "loss": 0.1792,
      "step": 1648
    },
    {
      "epoch": 0.23111422564821305,
      "grad_norm": 1.3076952695846558,
      "learning_rate": 0.0002835828749103085,
      "loss": 0.2112,
      "step": 1649
    },
    {
      "epoch": 0.2312543798177996,
      "grad_norm": 1.6938302516937256,
      "learning_rate": 0.00028356852427648886,
      "loss": 0.4612,
      "step": 1650
    },
    {
      "epoch": 0.23139453398738613,
      "grad_norm": 0.33778470754623413,
      "learning_rate": 0.0002835541736426692,
      "loss": 0.1282,
      "step": 1651
    },
    {
      "epoch": 0.23153468815697267,
      "grad_norm": 0.5366274118423462,
      "learning_rate": 0.0002835398230088495,
      "loss": 0.1808,
      "step": 1652
    },
    {
      "epoch": 0.23167484232655922,
      "grad_norm": 0.6916155815124512,
      "learning_rate": 0.0002835254723750299,
      "loss": 0.2167,
      "step": 1653
    },
    {
      "epoch": 0.23181499649614576,
      "grad_norm": 0.6328081488609314,
      "learning_rate": 0.00028351112174121023,
      "loss": 0.1434,
      "step": 1654
    },
    {
      "epoch": 0.2319551506657323,
      "grad_norm": 0.6392910480499268,
      "learning_rate": 0.00028349677110739056,
      "loss": 0.1037,
      "step": 1655
    },
    {
      "epoch": 0.23209530483531884,
      "grad_norm": 0.2952549457550049,
      "learning_rate": 0.0002834824204735709,
      "loss": 0.066,
      "step": 1656
    },
    {
      "epoch": 0.2322354590049054,
      "grad_norm": 0.6679016947746277,
      "learning_rate": 0.0002834680698397512,
      "loss": 0.2282,
      "step": 1657
    },
    {
      "epoch": 0.23237561317449193,
      "grad_norm": 0.354834645986557,
      "learning_rate": 0.00028345371920593154,
      "loss": 0.0842,
      "step": 1658
    },
    {
      "epoch": 0.2325157673440785,
      "grad_norm": 0.4278321862220764,
      "learning_rate": 0.0002834393685721119,
      "loss": 0.1135,
      "step": 1659
    },
    {
      "epoch": 0.23265592151366504,
      "grad_norm": 1.2079992294311523,
      "learning_rate": 0.00028342501793829225,
      "loss": 0.2546,
      "step": 1660
    },
    {
      "epoch": 0.23279607568325159,
      "grad_norm": 0.44434207677841187,
      "learning_rate": 0.0002834106673044726,
      "loss": 0.1513,
      "step": 1661
    },
    {
      "epoch": 0.23293622985283813,
      "grad_norm": 0.7500813603401184,
      "learning_rate": 0.00028339631667065297,
      "loss": 0.2029,
      "step": 1662
    },
    {
      "epoch": 0.23307638402242467,
      "grad_norm": 0.41560229659080505,
      "learning_rate": 0.0002833819660368333,
      "loss": 0.175,
      "step": 1663
    },
    {
      "epoch": 0.2332165381920112,
      "grad_norm": 0.27083083987236023,
      "learning_rate": 0.0002833676154030136,
      "loss": 0.0919,
      "step": 1664
    },
    {
      "epoch": 0.23335669236159776,
      "grad_norm": 0.47899481654167175,
      "learning_rate": 0.00028335326476919395,
      "loss": 0.0977,
      "step": 1665
    },
    {
      "epoch": 0.2334968465311843,
      "grad_norm": 0.9057362675666809,
      "learning_rate": 0.0002833389141353743,
      "loss": 0.1418,
      "step": 1666
    },
    {
      "epoch": 0.23363700070077084,
      "grad_norm": 0.868535578250885,
      "learning_rate": 0.0002833245635015546,
      "loss": 0.1956,
      "step": 1667
    },
    {
      "epoch": 0.23377715487035738,
      "grad_norm": 0.9943937063217163,
      "learning_rate": 0.00028331021286773494,
      "loss": 0.1581,
      "step": 1668
    },
    {
      "epoch": 0.23391730903994393,
      "grad_norm": 0.6929216980934143,
      "learning_rate": 0.0002832958622339153,
      "loss": 0.1216,
      "step": 1669
    },
    {
      "epoch": 0.23405746320953047,
      "grad_norm": 0.3878985643386841,
      "learning_rate": 0.00028328151160009565,
      "loss": 0.115,
      "step": 1670
    },
    {
      "epoch": 0.23419761737911704,
      "grad_norm": 0.8233346939086914,
      "learning_rate": 0.000283267160966276,
      "loss": 0.235,
      "step": 1671
    },
    {
      "epoch": 0.23433777154870358,
      "grad_norm": 0.40506336092948914,
      "learning_rate": 0.00028325281033245636,
      "loss": 0.0697,
      "step": 1672
    },
    {
      "epoch": 0.23447792571829013,
      "grad_norm": 0.8572129011154175,
      "learning_rate": 0.0002832384596986367,
      "loss": 0.0907,
      "step": 1673
    },
    {
      "epoch": 0.23461807988787667,
      "grad_norm": 0.3173794150352478,
      "learning_rate": 0.000283224109064817,
      "loss": 0.1381,
      "step": 1674
    },
    {
      "epoch": 0.2347582340574632,
      "grad_norm": 0.40838271379470825,
      "learning_rate": 0.00028320975843099735,
      "loss": 0.1155,
      "step": 1675
    },
    {
      "epoch": 0.23489838822704975,
      "grad_norm": 0.37829676270484924,
      "learning_rate": 0.0002831954077971777,
      "loss": 0.0765,
      "step": 1676
    },
    {
      "epoch": 0.2350385423966363,
      "grad_norm": 0.2388998419046402,
      "learning_rate": 0.000283181057163358,
      "loss": 0.0664,
      "step": 1677
    },
    {
      "epoch": 0.23517869656622284,
      "grad_norm": 1.0567635297775269,
      "learning_rate": 0.0002831667065295384,
      "loss": 0.104,
      "step": 1678
    },
    {
      "epoch": 0.23531885073580938,
      "grad_norm": 0.5665960311889648,
      "learning_rate": 0.0002831523558957187,
      "loss": 0.1595,
      "step": 1679
    },
    {
      "epoch": 0.23545900490539592,
      "grad_norm": 0.37826770544052124,
      "learning_rate": 0.00028313800526189904,
      "loss": 0.0732,
      "step": 1680
    },
    {
      "epoch": 0.23559915907498247,
      "grad_norm": 0.5304650664329529,
      "learning_rate": 0.00028312365462807937,
      "loss": 0.0886,
      "step": 1681
    },
    {
      "epoch": 0.23573931324456904,
      "grad_norm": 0.4689146876335144,
      "learning_rate": 0.00028310930399425975,
      "loss": 0.1095,
      "step": 1682
    },
    {
      "epoch": 0.23587946741415558,
      "grad_norm": 0.5237705707550049,
      "learning_rate": 0.0002830949533604401,
      "loss": 0.1324,
      "step": 1683
    },
    {
      "epoch": 0.23601962158374212,
      "grad_norm": 0.586449384689331,
      "learning_rate": 0.0002830806027266204,
      "loss": 0.1118,
      "step": 1684
    },
    {
      "epoch": 0.23615977575332867,
      "grad_norm": 0.8396391272544861,
      "learning_rate": 0.00028306625209280074,
      "loss": 0.209,
      "step": 1685
    },
    {
      "epoch": 0.2362999299229152,
      "grad_norm": 0.5026506185531616,
      "learning_rate": 0.00028305190145898107,
      "loss": 0.0844,
      "step": 1686
    },
    {
      "epoch": 0.23644008409250175,
      "grad_norm": 0.3826777935028076,
      "learning_rate": 0.0002830375508251614,
      "loss": 0.1025,
      "step": 1687
    },
    {
      "epoch": 0.2365802382620883,
      "grad_norm": 0.34839966893196106,
      "learning_rate": 0.0002830232001913418,
      "loss": 0.1385,
      "step": 1688
    },
    {
      "epoch": 0.23672039243167484,
      "grad_norm": 0.42290136218070984,
      "learning_rate": 0.0002830088495575221,
      "loss": 0.0731,
      "step": 1689
    },
    {
      "epoch": 0.23686054660126138,
      "grad_norm": 1.7799055576324463,
      "learning_rate": 0.00028299449892370244,
      "loss": 0.276,
      "step": 1690
    },
    {
      "epoch": 0.23700070077084792,
      "grad_norm": 1.0220530033111572,
      "learning_rate": 0.00028298014828988276,
      "loss": 0.2766,
      "step": 1691
    },
    {
      "epoch": 0.23714085494043446,
      "grad_norm": 0.5796447396278381,
      "learning_rate": 0.0002829657976560631,
      "loss": 0.1435,
      "step": 1692
    },
    {
      "epoch": 0.23728100911002104,
      "grad_norm": 0.711348295211792,
      "learning_rate": 0.0002829514470222435,
      "loss": 0.265,
      "step": 1693
    },
    {
      "epoch": 0.23742116327960758,
      "grad_norm": 1.6933900117874146,
      "learning_rate": 0.0002829370963884238,
      "loss": 0.1644,
      "step": 1694
    },
    {
      "epoch": 0.23756131744919412,
      "grad_norm": 0.5183990597724915,
      "learning_rate": 0.00028292274575460413,
      "loss": 0.096,
      "step": 1695
    },
    {
      "epoch": 0.23770147161878066,
      "grad_norm": 0.5787401795387268,
      "learning_rate": 0.00028290839512078446,
      "loss": 0.1613,
      "step": 1696
    },
    {
      "epoch": 0.2378416257883672,
      "grad_norm": 0.621388852596283,
      "learning_rate": 0.00028289404448696484,
      "loss": 0.0753,
      "step": 1697
    },
    {
      "epoch": 0.23798177995795375,
      "grad_norm": 1.5828367471694946,
      "learning_rate": 0.0002828796938531452,
      "loss": 0.3064,
      "step": 1698
    },
    {
      "epoch": 0.2381219341275403,
      "grad_norm": 0.8595340847969055,
      "learning_rate": 0.0002828653432193255,
      "loss": 0.3176,
      "step": 1699
    },
    {
      "epoch": 0.23826208829712683,
      "grad_norm": 13.367734909057617,
      "learning_rate": 0.00028285099258550583,
      "loss": 0.6157,
      "step": 1700
    },
    {
      "epoch": 0.23840224246671338,
      "grad_norm": 0.7175517678260803,
      "learning_rate": 0.00028283664195168616,
      "loss": 0.1334,
      "step": 1701
    },
    {
      "epoch": 0.23854239663629992,
      "grad_norm": 0.9889596104621887,
      "learning_rate": 0.0002828222913178665,
      "loss": 0.2325,
      "step": 1702
    },
    {
      "epoch": 0.23868255080588646,
      "grad_norm": 0.31712815165519714,
      "learning_rate": 0.0002828079406840468,
      "loss": 0.0906,
      "step": 1703
    },
    {
      "epoch": 0.23882270497547303,
      "grad_norm": 0.7756525278091431,
      "learning_rate": 0.0002827935900502272,
      "loss": 0.1938,
      "step": 1704
    },
    {
      "epoch": 0.23896285914505958,
      "grad_norm": 0.8887304663658142,
      "learning_rate": 0.00028277923941640753,
      "loss": 0.1422,
      "step": 1705
    },
    {
      "epoch": 0.23910301331464612,
      "grad_norm": 0.2697232663631439,
      "learning_rate": 0.00028276488878258786,
      "loss": 0.0939,
      "step": 1706
    },
    {
      "epoch": 0.23924316748423266,
      "grad_norm": 0.38379180431365967,
      "learning_rate": 0.00028275053814876824,
      "loss": 0.1197,
      "step": 1707
    },
    {
      "epoch": 0.2393833216538192,
      "grad_norm": 0.5237014293670654,
      "learning_rate": 0.00028273618751494857,
      "loss": 0.16,
      "step": 1708
    },
    {
      "epoch": 0.23952347582340575,
      "grad_norm": 0.8489036560058594,
      "learning_rate": 0.0002827218368811289,
      "loss": 0.1444,
      "step": 1709
    },
    {
      "epoch": 0.2396636299929923,
      "grad_norm": 0.5384160876274109,
      "learning_rate": 0.0002827074862473092,
      "loss": 0.1563,
      "step": 1710
    },
    {
      "epoch": 0.23980378416257883,
      "grad_norm": 0.8031654357910156,
      "learning_rate": 0.00028269313561348955,
      "loss": 0.1999,
      "step": 1711
    },
    {
      "epoch": 0.23994393833216537,
      "grad_norm": 0.5967836380004883,
      "learning_rate": 0.0002826787849796699,
      "loss": 0.1206,
      "step": 1712
    },
    {
      "epoch": 0.24008409250175192,
      "grad_norm": 0.6417787671089172,
      "learning_rate": 0.00028266443434585026,
      "loss": 0.2116,
      "step": 1713
    },
    {
      "epoch": 0.24022424667133846,
      "grad_norm": 0.5220252275466919,
      "learning_rate": 0.0002826500837120306,
      "loss": 0.1285,
      "step": 1714
    },
    {
      "epoch": 0.24036440084092503,
      "grad_norm": 0.5320206880569458,
      "learning_rate": 0.0002826357330782109,
      "loss": 0.1857,
      "step": 1715
    },
    {
      "epoch": 0.24050455501051157,
      "grad_norm": 0.35923105478286743,
      "learning_rate": 0.00028262138244439125,
      "loss": 0.1292,
      "step": 1716
    },
    {
      "epoch": 0.24064470918009812,
      "grad_norm": 0.7133763432502747,
      "learning_rate": 0.00028260703181057163,
      "loss": 0.1718,
      "step": 1717
    },
    {
      "epoch": 0.24078486334968466,
      "grad_norm": 0.5643021464347839,
      "learning_rate": 0.00028259268117675196,
      "loss": 0.1419,
      "step": 1718
    },
    {
      "epoch": 0.2409250175192712,
      "grad_norm": 0.27440154552459717,
      "learning_rate": 0.0002825783305429323,
      "loss": 0.0855,
      "step": 1719
    },
    {
      "epoch": 0.24106517168885774,
      "grad_norm": 0.46709468960762024,
      "learning_rate": 0.0002825639799091126,
      "loss": 0.1914,
      "step": 1720
    },
    {
      "epoch": 0.24120532585844429,
      "grad_norm": 0.2519575357437134,
      "learning_rate": 0.00028254962927529295,
      "loss": 0.0499,
      "step": 1721
    },
    {
      "epoch": 0.24134548002803083,
      "grad_norm": 0.5738582611083984,
      "learning_rate": 0.0002825352786414733,
      "loss": 0.1764,
      "step": 1722
    },
    {
      "epoch": 0.24148563419761737,
      "grad_norm": 0.43622833490371704,
      "learning_rate": 0.00028252092800765366,
      "loss": 0.1487,
      "step": 1723
    },
    {
      "epoch": 0.24162578836720391,
      "grad_norm": 0.4310140311717987,
      "learning_rate": 0.000282506577373834,
      "loss": 0.1111,
      "step": 1724
    },
    {
      "epoch": 0.24176594253679046,
      "grad_norm": 0.36717653274536133,
      "learning_rate": 0.0002824922267400143,
      "loss": 0.144,
      "step": 1725
    },
    {
      "epoch": 0.24190609670637703,
      "grad_norm": 0.34282341599464417,
      "learning_rate": 0.0002824778761061947,
      "loss": 0.0768,
      "step": 1726
    },
    {
      "epoch": 0.24204625087596357,
      "grad_norm": 0.7245255708694458,
      "learning_rate": 0.000282463525472375,
      "loss": 0.2162,
      "step": 1727
    },
    {
      "epoch": 0.2421864050455501,
      "grad_norm": 0.5724814534187317,
      "learning_rate": 0.00028244917483855536,
      "loss": 0.1002,
      "step": 1728
    },
    {
      "epoch": 0.24232655921513666,
      "grad_norm": 0.5859802961349487,
      "learning_rate": 0.0002824348242047357,
      "loss": 0.0752,
      "step": 1729
    },
    {
      "epoch": 0.2424667133847232,
      "grad_norm": 0.7630148530006409,
      "learning_rate": 0.000282420473570916,
      "loss": 0.0818,
      "step": 1730
    },
    {
      "epoch": 0.24260686755430974,
      "grad_norm": 0.41757214069366455,
      "learning_rate": 0.00028240612293709634,
      "loss": 0.109,
      "step": 1731
    },
    {
      "epoch": 0.24274702172389628,
      "grad_norm": 0.547549307346344,
      "learning_rate": 0.0002823917723032767,
      "loss": 0.126,
      "step": 1732
    },
    {
      "epoch": 0.24288717589348283,
      "grad_norm": 0.4793872833251953,
      "learning_rate": 0.00028237742166945705,
      "loss": 0.1508,
      "step": 1733
    },
    {
      "epoch": 0.24302733006306937,
      "grad_norm": 0.5726809501647949,
      "learning_rate": 0.0002823630710356374,
      "loss": 0.1467,
      "step": 1734
    },
    {
      "epoch": 0.2431674842326559,
      "grad_norm": 0.43208590149879456,
      "learning_rate": 0.0002823487204018177,
      "loss": 0.1654,
      "step": 1735
    },
    {
      "epoch": 0.24330763840224245,
      "grad_norm": 0.43308356404304504,
      "learning_rate": 0.0002823343697679981,
      "loss": 0.0821,
      "step": 1736
    },
    {
      "epoch": 0.24344779257182902,
      "grad_norm": 0.48286673426628113,
      "learning_rate": 0.0002823200191341784,
      "loss": 0.1521,
      "step": 1737
    },
    {
      "epoch": 0.24358794674141557,
      "grad_norm": 0.4979013204574585,
      "learning_rate": 0.00028230566850035875,
      "loss": 0.1914,
      "step": 1738
    },
    {
      "epoch": 0.2437281009110021,
      "grad_norm": 0.637593150138855,
      "learning_rate": 0.0002822913178665391,
      "loss": 0.1922,
      "step": 1739
    },
    {
      "epoch": 0.24386825508058865,
      "grad_norm": 0.6245976090431213,
      "learning_rate": 0.0002822769672327194,
      "loss": 0.1741,
      "step": 1740
    },
    {
      "epoch": 0.2440084092501752,
      "grad_norm": 0.548738420009613,
      "learning_rate": 0.00028226261659889973,
      "loss": 0.157,
      "step": 1741
    },
    {
      "epoch": 0.24414856341976174,
      "grad_norm": 0.5295251607894897,
      "learning_rate": 0.0002822482659650801,
      "loss": 0.2166,
      "step": 1742
    },
    {
      "epoch": 0.24428871758934828,
      "grad_norm": 0.7167810201644897,
      "learning_rate": 0.00028223391533126045,
      "loss": 0.0407,
      "step": 1743
    },
    {
      "epoch": 0.24442887175893482,
      "grad_norm": 0.5691710114479065,
      "learning_rate": 0.0002822195646974408,
      "loss": 0.1082,
      "step": 1744
    },
    {
      "epoch": 0.24456902592852137,
      "grad_norm": 0.4993327856063843,
      "learning_rate": 0.00028220521406362116,
      "loss": 0.0635,
      "step": 1745
    },
    {
      "epoch": 0.2447091800981079,
      "grad_norm": 2.1290979385375977,
      "learning_rate": 0.0002821908634298015,
      "loss": 0.2961,
      "step": 1746
    },
    {
      "epoch": 0.24484933426769445,
      "grad_norm": 1.4649319648742676,
      "learning_rate": 0.0002821765127959818,
      "loss": 0.122,
      "step": 1747
    },
    {
      "epoch": 0.24498948843728102,
      "grad_norm": 0.8563714027404785,
      "learning_rate": 0.00028216216216216214,
      "loss": 0.1444,
      "step": 1748
    },
    {
      "epoch": 0.24512964260686756,
      "grad_norm": 2.282214403152466,
      "learning_rate": 0.00028214781152834247,
      "loss": 0.3792,
      "step": 1749
    },
    {
      "epoch": 0.2452697967764541,
      "grad_norm": 1.725813627243042,
      "learning_rate": 0.0002821334608945228,
      "loss": 0.2854,
      "step": 1750
    },
    {
      "epoch": 0.24540995094604065,
      "grad_norm": 0.451020747423172,
      "learning_rate": 0.00028211911026070313,
      "loss": 0.1077,
      "step": 1751
    },
    {
      "epoch": 0.2455501051156272,
      "grad_norm": 0.7115446925163269,
      "learning_rate": 0.0002821047596268835,
      "loss": 0.1016,
      "step": 1752
    },
    {
      "epoch": 0.24569025928521374,
      "grad_norm": 0.6940094232559204,
      "learning_rate": 0.00028209040899306384,
      "loss": 0.2974,
      "step": 1753
    },
    {
      "epoch": 0.24583041345480028,
      "grad_norm": 0.4503176212310791,
      "learning_rate": 0.00028207605835924417,
      "loss": 0.1457,
      "step": 1754
    },
    {
      "epoch": 0.24597056762438682,
      "grad_norm": 0.2638428509235382,
      "learning_rate": 0.0002820617077254245,
      "loss": 0.1091,
      "step": 1755
    },
    {
      "epoch": 0.24611072179397336,
      "grad_norm": 0.4660966694355011,
      "learning_rate": 0.0002820473570916048,
      "loss": 0.1168,
      "step": 1756
    },
    {
      "epoch": 0.2462508759635599,
      "grad_norm": 0.3073606491088867,
      "learning_rate": 0.0002820330064577852,
      "loss": 0.0645,
      "step": 1757
    },
    {
      "epoch": 0.24639103013314645,
      "grad_norm": 0.2171139121055603,
      "learning_rate": 0.00028201865582396554,
      "loss": 0.0749,
      "step": 1758
    },
    {
      "epoch": 0.24653118430273302,
      "grad_norm": 0.5129942297935486,
      "learning_rate": 0.00028200430519014587,
      "loss": 0.2485,
      "step": 1759
    },
    {
      "epoch": 0.24667133847231956,
      "grad_norm": 0.329819917678833,
      "learning_rate": 0.0002819899545563262,
      "loss": 0.1029,
      "step": 1760
    },
    {
      "epoch": 0.2468114926419061,
      "grad_norm": 0.13715021312236786,
      "learning_rate": 0.0002819756039225066,
      "loss": 0.0382,
      "step": 1761
    },
    {
      "epoch": 0.24695164681149265,
      "grad_norm": 1.19892156124115,
      "learning_rate": 0.0002819612532886869,
      "loss": 0.2696,
      "step": 1762
    },
    {
      "epoch": 0.2470918009810792,
      "grad_norm": 0.6392252445220947,
      "learning_rate": 0.00028194690265486723,
      "loss": 0.1501,
      "step": 1763
    },
    {
      "epoch": 0.24723195515066573,
      "grad_norm": 0.36968758702278137,
      "learning_rate": 0.00028193255202104756,
      "loss": 0.1253,
      "step": 1764
    },
    {
      "epoch": 0.24737210932025228,
      "grad_norm": 0.5530754923820496,
      "learning_rate": 0.0002819182013872279,
      "loss": 0.1945,
      "step": 1765
    },
    {
      "epoch": 0.24751226348983882,
      "grad_norm": 0.4176405668258667,
      "learning_rate": 0.0002819038507534082,
      "loss": 0.1087,
      "step": 1766
    },
    {
      "epoch": 0.24765241765942536,
      "grad_norm": 0.7563183903694153,
      "learning_rate": 0.00028188950011958855,
      "loss": 0.2602,
      "step": 1767
    },
    {
      "epoch": 0.2477925718290119,
      "grad_norm": 0.455348402261734,
      "learning_rate": 0.00028187514948576893,
      "loss": 0.0768,
      "step": 1768
    },
    {
      "epoch": 0.24793272599859845,
      "grad_norm": 0.4709351360797882,
      "learning_rate": 0.00028186079885194926,
      "loss": 0.1149,
      "step": 1769
    },
    {
      "epoch": 0.24807288016818502,
      "grad_norm": 0.4431311786174774,
      "learning_rate": 0.0002818464482181296,
      "loss": 0.1066,
      "step": 1770
    },
    {
      "epoch": 0.24821303433777156,
      "grad_norm": 0.29459160566329956,
      "learning_rate": 0.00028183209758430997,
      "loss": 0.0773,
      "step": 1771
    },
    {
      "epoch": 0.2483531885073581,
      "grad_norm": 0.6708833575248718,
      "learning_rate": 0.0002818177469504903,
      "loss": 0.1052,
      "step": 1772
    },
    {
      "epoch": 0.24849334267694465,
      "grad_norm": 0.3578279912471771,
      "learning_rate": 0.00028180339631667063,
      "loss": 0.0929,
      "step": 1773
    },
    {
      "epoch": 0.2486334968465312,
      "grad_norm": 0.35013461112976074,
      "learning_rate": 0.00028178904568285096,
      "loss": 0.1361,
      "step": 1774
    },
    {
      "epoch": 0.24877365101611773,
      "grad_norm": 1.1069989204406738,
      "learning_rate": 0.0002817746950490313,
      "loss": 0.2124,
      "step": 1775
    },
    {
      "epoch": 0.24891380518570427,
      "grad_norm": 0.41144025325775146,
      "learning_rate": 0.0002817603444152116,
      "loss": 0.1135,
      "step": 1776
    },
    {
      "epoch": 0.24905395935529082,
      "grad_norm": 0.3035569190979004,
      "learning_rate": 0.000281745993781392,
      "loss": 0.091,
      "step": 1777
    },
    {
      "epoch": 0.24919411352487736,
      "grad_norm": 0.4869634807109833,
      "learning_rate": 0.0002817316431475723,
      "loss": 0.1446,
      "step": 1778
    },
    {
      "epoch": 0.2493342676944639,
      "grad_norm": 0.46246063709259033,
      "learning_rate": 0.00028171729251375265,
      "loss": 0.1514,
      "step": 1779
    },
    {
      "epoch": 0.24947442186405044,
      "grad_norm": 0.48864373564720154,
      "learning_rate": 0.00028170294187993304,
      "loss": 0.1294,
      "step": 1780
    },
    {
      "epoch": 0.24961457603363701,
      "grad_norm": 0.7949655652046204,
      "learning_rate": 0.00028168859124611337,
      "loss": 0.1426,
      "step": 1781
    },
    {
      "epoch": 0.24975473020322356,
      "grad_norm": 0.3172481656074524,
      "learning_rate": 0.0002816742406122937,
      "loss": 0.124,
      "step": 1782
    },
    {
      "epoch": 0.2498948843728101,
      "grad_norm": 0.5989638566970825,
      "learning_rate": 0.000281659889978474,
      "loss": 0.1502,
      "step": 1783
    },
    {
      "epoch": 0.2500350385423966,
      "grad_norm": 0.3476879298686981,
      "learning_rate": 0.00028164553934465435,
      "loss": 0.0561,
      "step": 1784
    },
    {
      "epoch": 0.2501751927119832,
      "grad_norm": 0.3938239812850952,
      "learning_rate": 0.0002816311887108347,
      "loss": 0.0708,
      "step": 1785
    },
    {
      "epoch": 0.2503153468815697,
      "grad_norm": 0.42315611243247986,
      "learning_rate": 0.000281616838077015,
      "loss": 0.1009,
      "step": 1786
    },
    {
      "epoch": 0.25045550105115627,
      "grad_norm": 0.7514938116073608,
      "learning_rate": 0.0002816024874431954,
      "loss": 0.2724,
      "step": 1787
    },
    {
      "epoch": 0.25059565522074284,
      "grad_norm": 0.6184118390083313,
      "learning_rate": 0.0002815881368093757,
      "loss": 0.1858,
      "step": 1788
    },
    {
      "epoch": 0.25073580939032936,
      "grad_norm": 0.8807806968688965,
      "learning_rate": 0.00028157378617555605,
      "loss": 0.2336,
      "step": 1789
    },
    {
      "epoch": 0.2508759635599159,
      "grad_norm": 0.9115694761276245,
      "learning_rate": 0.00028155943554173643,
      "loss": 0.1494,
      "step": 1790
    },
    {
      "epoch": 0.25101611772950244,
      "grad_norm": 0.9098187685012817,
      "learning_rate": 0.00028154508490791676,
      "loss": 0.1941,
      "step": 1791
    },
    {
      "epoch": 0.251156271899089,
      "grad_norm": 0.44079074263572693,
      "learning_rate": 0.0002815307342740971,
      "loss": 0.143,
      "step": 1792
    },
    {
      "epoch": 0.2512964260686755,
      "grad_norm": 0.6528348326683044,
      "learning_rate": 0.0002815163836402774,
      "loss": 0.1396,
      "step": 1793
    },
    {
      "epoch": 0.2514365802382621,
      "grad_norm": 0.6168358325958252,
      "learning_rate": 0.00028150203300645774,
      "loss": 0.135,
      "step": 1794
    },
    {
      "epoch": 0.2515767344078486,
      "grad_norm": 0.700537919998169,
      "learning_rate": 0.0002814876823726381,
      "loss": 0.222,
      "step": 1795
    },
    {
      "epoch": 0.2517168885774352,
      "grad_norm": 0.21935659646987915,
      "learning_rate": 0.00028147333173881846,
      "loss": 0.03,
      "step": 1796
    },
    {
      "epoch": 0.2518570427470217,
      "grad_norm": 1.1119674444198608,
      "learning_rate": 0.0002814589811049988,
      "loss": 0.2208,
      "step": 1797
    },
    {
      "epoch": 0.25199719691660827,
      "grad_norm": 0.7732111811637878,
      "learning_rate": 0.0002814446304711791,
      "loss": 0.2581,
      "step": 1798
    },
    {
      "epoch": 0.25213735108619484,
      "grad_norm": 1.658157229423523,
      "learning_rate": 0.0002814302798373595,
      "loss": 0.1683,
      "step": 1799
    },
    {
      "epoch": 0.25227750525578135,
      "grad_norm": 0.7387439608573914,
      "learning_rate": 0.0002814159292035398,
      "loss": 0.1923,
      "step": 1800
    },
    {
      "epoch": 0.2524176594253679,
      "grad_norm": 0.43980246782302856,
      "learning_rate": 0.00028140157856972015,
      "loss": 0.1267,
      "step": 1801
    },
    {
      "epoch": 0.25255781359495444,
      "grad_norm": 0.5206082463264465,
      "learning_rate": 0.0002813872279359005,
      "loss": 0.1216,
      "step": 1802
    },
    {
      "epoch": 0.252697967764541,
      "grad_norm": 0.8885549306869507,
      "learning_rate": 0.0002813728773020808,
      "loss": 0.2284,
      "step": 1803
    },
    {
      "epoch": 0.2528381219341275,
      "grad_norm": 0.3616775870323181,
      "learning_rate": 0.00028135852666826114,
      "loss": 0.0819,
      "step": 1804
    },
    {
      "epoch": 0.2529782761037141,
      "grad_norm": 0.3501027226448059,
      "learning_rate": 0.00028134417603444147,
      "loss": 0.1188,
      "step": 1805
    },
    {
      "epoch": 0.2531184302733006,
      "grad_norm": 0.4040818214416504,
      "learning_rate": 0.00028132982540062185,
      "loss": 0.1001,
      "step": 1806
    },
    {
      "epoch": 0.2532585844428872,
      "grad_norm": 0.6182248592376709,
      "learning_rate": 0.0002813154747668022,
      "loss": 0.2516,
      "step": 1807
    },
    {
      "epoch": 0.2533987386124737,
      "grad_norm": 0.7381235361099243,
      "learning_rate": 0.0002813011241329825,
      "loss": 0.2194,
      "step": 1808
    },
    {
      "epoch": 0.25353889278206027,
      "grad_norm": 0.48105478286743164,
      "learning_rate": 0.0002812867734991629,
      "loss": 0.1115,
      "step": 1809
    },
    {
      "epoch": 0.25367904695164684,
      "grad_norm": 0.5309714674949646,
      "learning_rate": 0.0002812724228653432,
      "loss": 0.1236,
      "step": 1810
    },
    {
      "epoch": 0.25381920112123335,
      "grad_norm": 0.5463789105415344,
      "learning_rate": 0.00028125807223152355,
      "loss": 0.0989,
      "step": 1811
    },
    {
      "epoch": 0.2539593552908199,
      "grad_norm": 0.5178471803665161,
      "learning_rate": 0.0002812437215977039,
      "loss": 0.1162,
      "step": 1812
    },
    {
      "epoch": 0.25409950946040644,
      "grad_norm": 0.5952447056770325,
      "learning_rate": 0.0002812293709638842,
      "loss": 0.172,
      "step": 1813
    },
    {
      "epoch": 0.254239663629993,
      "grad_norm": 0.27206680178642273,
      "learning_rate": 0.00028121502033006453,
      "loss": 0.0503,
      "step": 1814
    },
    {
      "epoch": 0.2543798177995795,
      "grad_norm": 0.4301283657550812,
      "learning_rate": 0.0002812006696962449,
      "loss": 0.096,
      "step": 1815
    },
    {
      "epoch": 0.2545199719691661,
      "grad_norm": 0.4496886134147644,
      "learning_rate": 0.00028118631906242524,
      "loss": 0.1711,
      "step": 1816
    },
    {
      "epoch": 0.2546601261387526,
      "grad_norm": 0.9504410624504089,
      "learning_rate": 0.00028117196842860557,
      "loss": 0.1346,
      "step": 1817
    },
    {
      "epoch": 0.2548002803083392,
      "grad_norm": 0.47178420424461365,
      "learning_rate": 0.0002811576177947859,
      "loss": 0.0894,
      "step": 1818
    },
    {
      "epoch": 0.2549404344779257,
      "grad_norm": 0.4664304852485657,
      "learning_rate": 0.00028114326716096623,
      "loss": 0.1447,
      "step": 1819
    },
    {
      "epoch": 0.25508058864751226,
      "grad_norm": 0.7794228196144104,
      "learning_rate": 0.0002811289165271466,
      "loss": 0.1076,
      "step": 1820
    },
    {
      "epoch": 0.25522074281709883,
      "grad_norm": 0.41424092650413513,
      "learning_rate": 0.00028111456589332694,
      "loss": 0.1063,
      "step": 1821
    },
    {
      "epoch": 0.25536089698668535,
      "grad_norm": 0.3069503903388977,
      "learning_rate": 0.00028110021525950727,
      "loss": 0.1064,
      "step": 1822
    },
    {
      "epoch": 0.2555010511562719,
      "grad_norm": 0.6126590371131897,
      "learning_rate": 0.0002810858646256876,
      "loss": 0.1543,
      "step": 1823
    },
    {
      "epoch": 0.25564120532585843,
      "grad_norm": 0.8415027856826782,
      "learning_rate": 0.0002810715139918679,
      "loss": 0.1141,
      "step": 1824
    },
    {
      "epoch": 0.255781359495445,
      "grad_norm": 0.5680672526359558,
      "learning_rate": 0.0002810571633580483,
      "loss": 0.1598,
      "step": 1825
    },
    {
      "epoch": 0.2559215136650315,
      "grad_norm": 0.38887882232666016,
      "learning_rate": 0.00028104281272422864,
      "loss": 0.1737,
      "step": 1826
    },
    {
      "epoch": 0.2560616678346181,
      "grad_norm": 0.634729266166687,
      "learning_rate": 0.00028102846209040897,
      "loss": 0.1353,
      "step": 1827
    },
    {
      "epoch": 0.2562018220042046,
      "grad_norm": 0.3087935745716095,
      "learning_rate": 0.0002810141114565893,
      "loss": 0.0708,
      "step": 1828
    },
    {
      "epoch": 0.2563419761737912,
      "grad_norm": 0.6217717528343201,
      "learning_rate": 0.0002809997608227696,
      "loss": 0.1578,
      "step": 1829
    },
    {
      "epoch": 0.2564821303433777,
      "grad_norm": 0.8649902939796448,
      "learning_rate": 0.00028098541018894995,
      "loss": 0.1094,
      "step": 1830
    },
    {
      "epoch": 0.25662228451296426,
      "grad_norm": 0.6316784024238586,
      "learning_rate": 0.00028097105955513034,
      "loss": 0.2179,
      "step": 1831
    },
    {
      "epoch": 0.25676243868255083,
      "grad_norm": 0.9165083765983582,
      "learning_rate": 0.00028095670892131066,
      "loss": 0.286,
      "step": 1832
    },
    {
      "epoch": 0.25690259285213735,
      "grad_norm": 0.39727166295051575,
      "learning_rate": 0.000280942358287491,
      "loss": 0.0984,
      "step": 1833
    },
    {
      "epoch": 0.2570427470217239,
      "grad_norm": 0.3837035298347473,
      "learning_rate": 0.0002809280076536714,
      "loss": 0.0675,
      "step": 1834
    },
    {
      "epoch": 0.25718290119131043,
      "grad_norm": 0.7482810020446777,
      "learning_rate": 0.0002809136570198517,
      "loss": 0.2355,
      "step": 1835
    },
    {
      "epoch": 0.257323055360897,
      "grad_norm": 0.4094725549221039,
      "learning_rate": 0.00028089930638603203,
      "loss": 0.182,
      "step": 1836
    },
    {
      "epoch": 0.2574632095304835,
      "grad_norm": 0.3261123299598694,
      "learning_rate": 0.00028088495575221236,
      "loss": 0.1948,
      "step": 1837
    },
    {
      "epoch": 0.2576033637000701,
      "grad_norm": 0.8239557147026062,
      "learning_rate": 0.0002808706051183927,
      "loss": 0.0818,
      "step": 1838
    },
    {
      "epoch": 0.2577435178696566,
      "grad_norm": 0.2915264368057251,
      "learning_rate": 0.000280856254484573,
      "loss": 0.0542,
      "step": 1839
    },
    {
      "epoch": 0.2578836720392432,
      "grad_norm": 0.8499298095703125,
      "learning_rate": 0.00028084190385075335,
      "loss": 0.1301,
      "step": 1840
    },
    {
      "epoch": 0.2580238262088297,
      "grad_norm": 0.37676772475242615,
      "learning_rate": 0.00028082755321693373,
      "loss": 0.1291,
      "step": 1841
    },
    {
      "epoch": 0.25816398037841626,
      "grad_norm": 0.8743988275527954,
      "learning_rate": 0.00028081320258311406,
      "loss": 0.1639,
      "step": 1842
    },
    {
      "epoch": 0.25830413454800283,
      "grad_norm": 0.7587233781814575,
      "learning_rate": 0.0002807988519492944,
      "loss": 0.1182,
      "step": 1843
    },
    {
      "epoch": 0.25844428871758934,
      "grad_norm": 0.36376047134399414,
      "learning_rate": 0.00028078450131547477,
      "loss": 0.0992,
      "step": 1844
    },
    {
      "epoch": 0.2585844428871759,
      "grad_norm": 0.5320486426353455,
      "learning_rate": 0.0002807701506816551,
      "loss": 0.0972,
      "step": 1845
    },
    {
      "epoch": 0.25872459705676243,
      "grad_norm": 0.6233716011047363,
      "learning_rate": 0.0002807558000478354,
      "loss": 0.0923,
      "step": 1846
    },
    {
      "epoch": 0.258864751226349,
      "grad_norm": 0.6135765910148621,
      "learning_rate": 0.00028074144941401575,
      "loss": 0.1878,
      "step": 1847
    },
    {
      "epoch": 0.2590049053959355,
      "grad_norm": 1.0948060750961304,
      "learning_rate": 0.0002807270987801961,
      "loss": 0.1315,
      "step": 1848
    },
    {
      "epoch": 0.2591450595655221,
      "grad_norm": 0.6060850620269775,
      "learning_rate": 0.0002807127481463764,
      "loss": 0.0968,
      "step": 1849
    },
    {
      "epoch": 0.2592852137351086,
      "grad_norm": 0.9469805955886841,
      "learning_rate": 0.0002806983975125568,
      "loss": 0.1142,
      "step": 1850
    },
    {
      "epoch": 0.25942536790469517,
      "grad_norm": 0.6662535667419434,
      "learning_rate": 0.0002806840468787371,
      "loss": 0.2374,
      "step": 1851
    },
    {
      "epoch": 0.2595655220742817,
      "grad_norm": 0.25526994466781616,
      "learning_rate": 0.00028066969624491745,
      "loss": 0.0783,
      "step": 1852
    },
    {
      "epoch": 0.25970567624386826,
      "grad_norm": 0.3429538309574127,
      "learning_rate": 0.00028065534561109783,
      "loss": 0.076,
      "step": 1853
    },
    {
      "epoch": 0.2598458304134548,
      "grad_norm": 0.6768313646316528,
      "learning_rate": 0.00028064099497727816,
      "loss": 0.1519,
      "step": 1854
    },
    {
      "epoch": 0.25998598458304134,
      "grad_norm": 0.36945641040802,
      "learning_rate": 0.0002806266443434585,
      "loss": 0.0899,
      "step": 1855
    },
    {
      "epoch": 0.2601261387526279,
      "grad_norm": 0.5451635718345642,
      "learning_rate": 0.0002806122937096388,
      "loss": 0.1678,
      "step": 1856
    },
    {
      "epoch": 0.2602662929222144,
      "grad_norm": 0.9854544997215271,
      "learning_rate": 0.00028059794307581915,
      "loss": 0.1791,
      "step": 1857
    },
    {
      "epoch": 0.260406447091801,
      "grad_norm": 0.576852560043335,
      "learning_rate": 0.0002805835924419995,
      "loss": 0.2015,
      "step": 1858
    },
    {
      "epoch": 0.2605466012613875,
      "grad_norm": 0.4926300048828125,
      "learning_rate": 0.0002805692418081798,
      "loss": 0.1634,
      "step": 1859
    },
    {
      "epoch": 0.2606867554309741,
      "grad_norm": 0.6340547800064087,
      "learning_rate": 0.0002805548911743602,
      "loss": 0.1112,
      "step": 1860
    },
    {
      "epoch": 0.2608269096005606,
      "grad_norm": 0.479902058839798,
      "learning_rate": 0.0002805405405405405,
      "loss": 0.1006,
      "step": 1861
    },
    {
      "epoch": 0.26096706377014717,
      "grad_norm": 0.4118190407752991,
      "learning_rate": 0.00028052618990672085,
      "loss": 0.066,
      "step": 1862
    },
    {
      "epoch": 0.2611072179397337,
      "grad_norm": 0.9400238990783691,
      "learning_rate": 0.00028051183927290123,
      "loss": 0.1515,
      "step": 1863
    },
    {
      "epoch": 0.26124737210932025,
      "grad_norm": 0.33944255113601685,
      "learning_rate": 0.00028049748863908156,
      "loss": 0.1046,
      "step": 1864
    },
    {
      "epoch": 0.2613875262789068,
      "grad_norm": 0.34444308280944824,
      "learning_rate": 0.0002804831380052619,
      "loss": 0.1608,
      "step": 1865
    },
    {
      "epoch": 0.26152768044849334,
      "grad_norm": 0.3691379129886627,
      "learning_rate": 0.0002804687873714422,
      "loss": 0.13,
      "step": 1866
    },
    {
      "epoch": 0.2616678346180799,
      "grad_norm": 0.24631360173225403,
      "learning_rate": 0.00028045443673762254,
      "loss": 0.0661,
      "step": 1867
    },
    {
      "epoch": 0.2618079887876664,
      "grad_norm": 1.240393877029419,
      "learning_rate": 0.00028044008610380287,
      "loss": 0.163,
      "step": 1868
    },
    {
      "epoch": 0.261948142957253,
      "grad_norm": 0.6703735589981079,
      "learning_rate": 0.00028042573546998325,
      "loss": 0.0774,
      "step": 1869
    },
    {
      "epoch": 0.2620882971268395,
      "grad_norm": 0.5705479979515076,
      "learning_rate": 0.0002804113848361636,
      "loss": 0.1566,
      "step": 1870
    },
    {
      "epoch": 0.2622284512964261,
      "grad_norm": 0.5477330684661865,
      "learning_rate": 0.0002803970342023439,
      "loss": 0.1033,
      "step": 1871
    },
    {
      "epoch": 0.2623686054660126,
      "grad_norm": 0.23318515717983246,
      "learning_rate": 0.00028038268356852424,
      "loss": 0.0696,
      "step": 1872
    },
    {
      "epoch": 0.26250875963559916,
      "grad_norm": 0.5640891194343567,
      "learning_rate": 0.0002803683329347046,
      "loss": 0.1741,
      "step": 1873
    },
    {
      "epoch": 0.2626489138051857,
      "grad_norm": 0.24505697190761566,
      "learning_rate": 0.00028035398230088495,
      "loss": 0.0481,
      "step": 1874
    },
    {
      "epoch": 0.26278906797477225,
      "grad_norm": 0.5220059156417847,
      "learning_rate": 0.0002803396316670653,
      "loss": 0.0798,
      "step": 1875
    },
    {
      "epoch": 0.2629292221443588,
      "grad_norm": 0.47825273871421814,
      "learning_rate": 0.0002803252810332456,
      "loss": 0.1447,
      "step": 1876
    },
    {
      "epoch": 0.26306937631394534,
      "grad_norm": 0.5538408756256104,
      "learning_rate": 0.00028031093039942594,
      "loss": 0.2263,
      "step": 1877
    },
    {
      "epoch": 0.2632095304835319,
      "grad_norm": 0.7235707640647888,
      "learning_rate": 0.00028029657976560627,
      "loss": 0.1822,
      "step": 1878
    },
    {
      "epoch": 0.2633496846531184,
      "grad_norm": 0.41711851954460144,
      "learning_rate": 0.00028028222913178665,
      "loss": 0.1532,
      "step": 1879
    },
    {
      "epoch": 0.263489838822705,
      "grad_norm": 0.48717182874679565,
      "learning_rate": 0.000280267878497967,
      "loss": 0.1238,
      "step": 1880
    },
    {
      "epoch": 0.2636299929922915,
      "grad_norm": 0.3388001024723053,
      "learning_rate": 0.0002802535278641473,
      "loss": 0.1096,
      "step": 1881
    },
    {
      "epoch": 0.2637701471618781,
      "grad_norm": 0.7721616625785828,
      "learning_rate": 0.00028023917723032763,
      "loss": 0.1976,
      "step": 1882
    },
    {
      "epoch": 0.2639103013314646,
      "grad_norm": 0.6086646318435669,
      "learning_rate": 0.000280224826596508,
      "loss": 0.1601,
      "step": 1883
    },
    {
      "epoch": 0.26405045550105116,
      "grad_norm": 0.4716200828552246,
      "learning_rate": 0.00028021047596268835,
      "loss": 0.1791,
      "step": 1884
    },
    {
      "epoch": 0.2641906096706377,
      "grad_norm": 0.6042973399162292,
      "learning_rate": 0.0002801961253288687,
      "loss": 0.1304,
      "step": 1885
    },
    {
      "epoch": 0.26433076384022425,
      "grad_norm": 0.43370291590690613,
      "learning_rate": 0.000280181774695049,
      "loss": 0.1164,
      "step": 1886
    },
    {
      "epoch": 0.2644709180098108,
      "grad_norm": 0.7117361426353455,
      "learning_rate": 0.00028016742406122933,
      "loss": 0.1952,
      "step": 1887
    },
    {
      "epoch": 0.26461107217939733,
      "grad_norm": 0.4921676218509674,
      "learning_rate": 0.0002801530734274097,
      "loss": 0.0949,
      "step": 1888
    },
    {
      "epoch": 0.2647512263489839,
      "grad_norm": 0.5655998587608337,
      "learning_rate": 0.00028013872279359004,
      "loss": 0.1588,
      "step": 1889
    },
    {
      "epoch": 0.2648913805185704,
      "grad_norm": 0.649277925491333,
      "learning_rate": 0.00028012437215977037,
      "loss": 0.2138,
      "step": 1890
    },
    {
      "epoch": 0.265031534688157,
      "grad_norm": 0.29094550013542175,
      "learning_rate": 0.0002801100215259507,
      "loss": 0.0561,
      "step": 1891
    },
    {
      "epoch": 0.2651716888577435,
      "grad_norm": 0.42921003699302673,
      "learning_rate": 0.00028009567089213103,
      "loss": 0.1148,
      "step": 1892
    },
    {
      "epoch": 0.2653118430273301,
      "grad_norm": 0.5670558214187622,
      "learning_rate": 0.00028008132025831136,
      "loss": 0.1614,
      "step": 1893
    },
    {
      "epoch": 0.2654519971969166,
      "grad_norm": 0.5095421075820923,
      "learning_rate": 0.0002800669696244917,
      "loss": 0.1247,
      "step": 1894
    },
    {
      "epoch": 0.26559215136650316,
      "grad_norm": 0.8093964457511902,
      "learning_rate": 0.00028005261899067207,
      "loss": 0.2039,
      "step": 1895
    },
    {
      "epoch": 0.2657323055360897,
      "grad_norm": 0.5574644207954407,
      "learning_rate": 0.0002800382683568524,
      "loss": 0.1065,
      "step": 1896
    },
    {
      "epoch": 0.26587245970567625,
      "grad_norm": 1.0862740278244019,
      "learning_rate": 0.0002800239177230327,
      "loss": 0.195,
      "step": 1897
    },
    {
      "epoch": 0.2660126138752628,
      "grad_norm": 0.887651264667511,
      "learning_rate": 0.0002800095670892131,
      "loss": 0.1834,
      "step": 1898
    },
    {
      "epoch": 0.26615276804484933,
      "grad_norm": 1.5075024366378784,
      "learning_rate": 0.00027999521645539344,
      "loss": 0.3211,
      "step": 1899
    },
    {
      "epoch": 0.2662929222144359,
      "grad_norm": 1.5829147100448608,
      "learning_rate": 0.00027998086582157376,
      "loss": 0.3588,
      "step": 1900
    },
    {
      "epoch": 0.2664330763840224,
      "grad_norm": 0.47432902455329895,
      "learning_rate": 0.0002799665151877541,
      "loss": 0.1559,
      "step": 1901
    },
    {
      "epoch": 0.266573230553609,
      "grad_norm": 0.5671768188476562,
      "learning_rate": 0.0002799521645539344,
      "loss": 0.1576,
      "step": 1902
    },
    {
      "epoch": 0.2667133847231955,
      "grad_norm": 0.6967267990112305,
      "learning_rate": 0.00027993781392011475,
      "loss": 0.1389,
      "step": 1903
    },
    {
      "epoch": 0.26685353889278207,
      "grad_norm": 0.7033321261405945,
      "learning_rate": 0.00027992346328629513,
      "loss": 0.1995,
      "step": 1904
    },
    {
      "epoch": 0.2669936930623686,
      "grad_norm": 0.4978240728378296,
      "learning_rate": 0.00027990911265247546,
      "loss": 0.1443,
      "step": 1905
    },
    {
      "epoch": 0.26713384723195516,
      "grad_norm": 0.37465062737464905,
      "learning_rate": 0.0002798947620186558,
      "loss": 0.0831,
      "step": 1906
    },
    {
      "epoch": 0.26727400140154167,
      "grad_norm": 0.25860920548439026,
      "learning_rate": 0.0002798804113848361,
      "loss": 0.0893,
      "step": 1907
    },
    {
      "epoch": 0.26741415557112824,
      "grad_norm": 0.30216631293296814,
      "learning_rate": 0.0002798660607510165,
      "loss": 0.0761,
      "step": 1908
    },
    {
      "epoch": 0.2675543097407148,
      "grad_norm": 0.4658152759075165,
      "learning_rate": 0.00027985171011719683,
      "loss": 0.1566,
      "step": 1909
    },
    {
      "epoch": 0.26769446391030133,
      "grad_norm": 0.32072290778160095,
      "learning_rate": 0.00027983735948337716,
      "loss": 0.1108,
      "step": 1910
    },
    {
      "epoch": 0.2678346180798879,
      "grad_norm": 0.8918739557266235,
      "learning_rate": 0.0002798230088495575,
      "loss": 0.145,
      "step": 1911
    },
    {
      "epoch": 0.2679747722494744,
      "grad_norm": 0.6238973140716553,
      "learning_rate": 0.0002798086582157378,
      "loss": 0.1087,
      "step": 1912
    },
    {
      "epoch": 0.268114926419061,
      "grad_norm": 0.5043307542800903,
      "learning_rate": 0.00027979430758191814,
      "loss": 0.1154,
      "step": 1913
    },
    {
      "epoch": 0.2682550805886475,
      "grad_norm": 0.27551767230033875,
      "learning_rate": 0.00027977995694809853,
      "loss": 0.0949,
      "step": 1914
    },
    {
      "epoch": 0.26839523475823407,
      "grad_norm": 0.43672341108322144,
      "learning_rate": 0.00027976560631427886,
      "loss": 0.1264,
      "step": 1915
    },
    {
      "epoch": 0.2685353889278206,
      "grad_norm": 0.4689143896102905,
      "learning_rate": 0.0002797512556804592,
      "loss": 0.0713,
      "step": 1916
    },
    {
      "epoch": 0.26867554309740715,
      "grad_norm": 0.6661012768745422,
      "learning_rate": 0.00027973690504663957,
      "loss": 0.1595,
      "step": 1917
    },
    {
      "epoch": 0.26881569726699367,
      "grad_norm": 0.30119356513023376,
      "learning_rate": 0.0002797225544128199,
      "loss": 0.0826,
      "step": 1918
    },
    {
      "epoch": 0.26895585143658024,
      "grad_norm": 0.6648711562156677,
      "learning_rate": 0.0002797082037790002,
      "loss": 0.1563,
      "step": 1919
    },
    {
      "epoch": 0.2690960056061668,
      "grad_norm": 0.5036062002182007,
      "learning_rate": 0.00027969385314518055,
      "loss": 0.203,
      "step": 1920
    },
    {
      "epoch": 0.2692361597757533,
      "grad_norm": 0.7475723028182983,
      "learning_rate": 0.0002796795025113609,
      "loss": 0.1058,
      "step": 1921
    },
    {
      "epoch": 0.2693763139453399,
      "grad_norm": 0.5780730843544006,
      "learning_rate": 0.0002796651518775412,
      "loss": 0.0567,
      "step": 1922
    },
    {
      "epoch": 0.2695164681149264,
      "grad_norm": 0.5885787606239319,
      "learning_rate": 0.0002796508012437216,
      "loss": 0.1979,
      "step": 1923
    },
    {
      "epoch": 0.269656622284513,
      "grad_norm": 0.590232789516449,
      "learning_rate": 0.0002796364506099019,
      "loss": 0.1878,
      "step": 1924
    },
    {
      "epoch": 0.2697967764540995,
      "grad_norm": 0.42100223898887634,
      "learning_rate": 0.00027962209997608225,
      "loss": 0.1478,
      "step": 1925
    },
    {
      "epoch": 0.26993693062368607,
      "grad_norm": 0.9727389216423035,
      "learning_rate": 0.0002796077493422626,
      "loss": 0.1672,
      "step": 1926
    },
    {
      "epoch": 0.2700770847932726,
      "grad_norm": 0.20500816404819489,
      "learning_rate": 0.00027959339870844296,
      "loss": 0.0528,
      "step": 1927
    },
    {
      "epoch": 0.27021723896285915,
      "grad_norm": 0.3219503164291382,
      "learning_rate": 0.0002795790480746233,
      "loss": 0.0874,
      "step": 1928
    },
    {
      "epoch": 0.27035739313244567,
      "grad_norm": 0.7591564059257507,
      "learning_rate": 0.0002795646974408036,
      "loss": 0.2104,
      "step": 1929
    },
    {
      "epoch": 0.27049754730203224,
      "grad_norm": 0.48872920870780945,
      "learning_rate": 0.00027955034680698395,
      "loss": 0.0533,
      "step": 1930
    },
    {
      "epoch": 0.2706377014716188,
      "grad_norm": 0.6220650672912598,
      "learning_rate": 0.0002795359961731643,
      "loss": 0.1051,
      "step": 1931
    },
    {
      "epoch": 0.2707778556412053,
      "grad_norm": 0.8095846772193909,
      "learning_rate": 0.0002795216455393446,
      "loss": 0.421,
      "step": 1932
    },
    {
      "epoch": 0.2709180098107919,
      "grad_norm": 0.3527679741382599,
      "learning_rate": 0.000279507294905525,
      "loss": 0.2359,
      "step": 1933
    },
    {
      "epoch": 0.2710581639803784,
      "grad_norm": 0.40157654881477356,
      "learning_rate": 0.0002794929442717053,
      "loss": 0.075,
      "step": 1934
    },
    {
      "epoch": 0.271198318149965,
      "grad_norm": 0.25354140996932983,
      "learning_rate": 0.00027947859363788564,
      "loss": 0.1114,
      "step": 1935
    },
    {
      "epoch": 0.2713384723195515,
      "grad_norm": 0.4825376868247986,
      "learning_rate": 0.000279464243004066,
      "loss": 0.1007,
      "step": 1936
    },
    {
      "epoch": 0.27147862648913806,
      "grad_norm": 0.4860278069972992,
      "learning_rate": 0.00027944989237024636,
      "loss": 0.1568,
      "step": 1937
    },
    {
      "epoch": 0.2716187806587246,
      "grad_norm": 1.5211390256881714,
      "learning_rate": 0.0002794355417364267,
      "loss": 0.109,
      "step": 1938
    },
    {
      "epoch": 0.27175893482831115,
      "grad_norm": 0.7738736271858215,
      "learning_rate": 0.000279421191102607,
      "loss": 0.1209,
      "step": 1939
    },
    {
      "epoch": 0.27189908899789766,
      "grad_norm": 0.7095392346382141,
      "learning_rate": 0.00027940684046878734,
      "loss": 0.1307,
      "step": 1940
    },
    {
      "epoch": 0.27203924316748423,
      "grad_norm": 0.5415706634521484,
      "learning_rate": 0.00027939248983496767,
      "loss": 0.1052,
      "step": 1941
    },
    {
      "epoch": 0.2721793973370708,
      "grad_norm": 1.4570891857147217,
      "learning_rate": 0.000279378139201148,
      "loss": 0.1442,
      "step": 1942
    },
    {
      "epoch": 0.2723195515066573,
      "grad_norm": 0.6593202948570251,
      "learning_rate": 0.0002793637885673284,
      "loss": 0.1015,
      "step": 1943
    },
    {
      "epoch": 0.2724597056762439,
      "grad_norm": 0.48740774393081665,
      "learning_rate": 0.0002793494379335087,
      "loss": 0.1103,
      "step": 1944
    },
    {
      "epoch": 0.2725998598458304,
      "grad_norm": 0.61367267370224,
      "learning_rate": 0.00027933508729968904,
      "loss": 0.1874,
      "step": 1945
    },
    {
      "epoch": 0.272740014015417,
      "grad_norm": 0.623350977897644,
      "learning_rate": 0.00027932073666586937,
      "loss": 0.1305,
      "step": 1946
    },
    {
      "epoch": 0.2728801681850035,
      "grad_norm": 0.7416171431541443,
      "learning_rate": 0.00027930638603204975,
      "loss": 0.1333,
      "step": 1947
    },
    {
      "epoch": 0.27302032235459006,
      "grad_norm": 1.5345262289047241,
      "learning_rate": 0.0002792920353982301,
      "loss": 0.3047,
      "step": 1948
    },
    {
      "epoch": 0.2731604765241766,
      "grad_norm": 0.9212145209312439,
      "learning_rate": 0.0002792776847644104,
      "loss": 0.2816,
      "step": 1949
    },
    {
      "epoch": 0.27330063069376315,
      "grad_norm": 1.7418906688690186,
      "learning_rate": 0.00027926333413059074,
      "loss": 0.2853,
      "step": 1950
    },
    {
      "epoch": 0.27344078486334966,
      "grad_norm": 0.3374670743942261,
      "learning_rate": 0.00027924898349677106,
      "loss": 0.1446,
      "step": 1951
    },
    {
      "epoch": 0.27358093903293623,
      "grad_norm": 0.2873058617115021,
      "learning_rate": 0.00027923463286295145,
      "loss": 0.0837,
      "step": 1952
    },
    {
      "epoch": 0.2737210932025228,
      "grad_norm": 0.4602125287055969,
      "learning_rate": 0.0002792202822291318,
      "loss": 0.1722,
      "step": 1953
    },
    {
      "epoch": 0.2738612473721093,
      "grad_norm": 0.45416730642318726,
      "learning_rate": 0.0002792059315953121,
      "loss": 0.1651,
      "step": 1954
    },
    {
      "epoch": 0.2740014015416959,
      "grad_norm": 0.3982296884059906,
      "learning_rate": 0.00027919158096149243,
      "loss": 0.1645,
      "step": 1955
    },
    {
      "epoch": 0.2741415557112824,
      "grad_norm": 0.48279327154159546,
      "learning_rate": 0.00027917723032767276,
      "loss": 0.1068,
      "step": 1956
    },
    {
      "epoch": 0.274281709880869,
      "grad_norm": 0.7379068732261658,
      "learning_rate": 0.0002791628796938531,
      "loss": 0.2246,
      "step": 1957
    },
    {
      "epoch": 0.2744218640504555,
      "grad_norm": 0.45647430419921875,
      "learning_rate": 0.00027914852906003347,
      "loss": 0.1772,
      "step": 1958
    },
    {
      "epoch": 0.27456201822004206,
      "grad_norm": 0.3247532546520233,
      "learning_rate": 0.0002791341784262138,
      "loss": 0.0913,
      "step": 1959
    },
    {
      "epoch": 0.2747021723896286,
      "grad_norm": 0.5176326632499695,
      "learning_rate": 0.00027911982779239413,
      "loss": 0.1146,
      "step": 1960
    },
    {
      "epoch": 0.27484232655921514,
      "grad_norm": 0.31669044494628906,
      "learning_rate": 0.00027910547715857446,
      "loss": 0.088,
      "step": 1961
    },
    {
      "epoch": 0.27498248072880166,
      "grad_norm": 0.40046054124832153,
      "learning_rate": 0.00027909112652475484,
      "loss": 0.069,
      "step": 1962
    },
    {
      "epoch": 0.27512263489838823,
      "grad_norm": 1.0397855043411255,
      "learning_rate": 0.00027907677589093517,
      "loss": 0.185,
      "step": 1963
    },
    {
      "epoch": 0.27526278906797474,
      "grad_norm": 0.7936228513717651,
      "learning_rate": 0.0002790624252571155,
      "loss": 0.1904,
      "step": 1964
    },
    {
      "epoch": 0.2754029432375613,
      "grad_norm": 0.31362998485565186,
      "learning_rate": 0.0002790480746232958,
      "loss": 0.0623,
      "step": 1965
    },
    {
      "epoch": 0.2755430974071479,
      "grad_norm": 0.5750884413719177,
      "learning_rate": 0.00027903372398947615,
      "loss": 0.17,
      "step": 1966
    },
    {
      "epoch": 0.2756832515767344,
      "grad_norm": 0.205519899725914,
      "learning_rate": 0.0002790193733556565,
      "loss": 0.0783,
      "step": 1967
    },
    {
      "epoch": 0.27582340574632097,
      "grad_norm": 0.9362422823905945,
      "learning_rate": 0.00027900502272183687,
      "loss": 0.1199,
      "step": 1968
    },
    {
      "epoch": 0.2759635599159075,
      "grad_norm": 0.5203693509101868,
      "learning_rate": 0.0002789906720880172,
      "loss": 0.1317,
      "step": 1969
    },
    {
      "epoch": 0.27610371408549406,
      "grad_norm": 0.37986159324645996,
      "learning_rate": 0.0002789763214541975,
      "loss": 0.1756,
      "step": 1970
    },
    {
      "epoch": 0.27624386825508057,
      "grad_norm": 0.5829340219497681,
      "learning_rate": 0.0002789619708203779,
      "loss": 0.1251,
      "step": 1971
    },
    {
      "epoch": 0.27638402242466714,
      "grad_norm": 0.2515009343624115,
      "learning_rate": 0.00027894762018655823,
      "loss": 0.0744,
      "step": 1972
    },
    {
      "epoch": 0.27652417659425366,
      "grad_norm": 0.6552347540855408,
      "learning_rate": 0.00027893326955273856,
      "loss": 0.1315,
      "step": 1973
    },
    {
      "epoch": 0.2766643307638402,
      "grad_norm": 0.288085401058197,
      "learning_rate": 0.0002789189189189189,
      "loss": 0.123,
      "step": 1974
    },
    {
      "epoch": 0.27680448493342674,
      "grad_norm": 0.45238152146339417,
      "learning_rate": 0.0002789045682850992,
      "loss": 0.1116,
      "step": 1975
    },
    {
      "epoch": 0.2769446391030133,
      "grad_norm": 0.47496235370635986,
      "learning_rate": 0.00027889021765127955,
      "loss": 0.1037,
      "step": 1976
    },
    {
      "epoch": 0.2770847932725999,
      "grad_norm": 0.8938223719596863,
      "learning_rate": 0.0002788758670174599,
      "loss": 0.1688,
      "step": 1977
    },
    {
      "epoch": 0.2772249474421864,
      "grad_norm": 0.6534609794616699,
      "learning_rate": 0.00027886151638364026,
      "loss": 0.181,
      "step": 1978
    },
    {
      "epoch": 0.27736510161177297,
      "grad_norm": 0.5342714190483093,
      "learning_rate": 0.0002788471657498206,
      "loss": 0.1323,
      "step": 1979
    },
    {
      "epoch": 0.2775052557813595,
      "grad_norm": 0.30895674228668213,
      "learning_rate": 0.0002788328151160009,
      "loss": 0.0859,
      "step": 1980
    },
    {
      "epoch": 0.27764540995094605,
      "grad_norm": 0.39004939794540405,
      "learning_rate": 0.0002788184644821813,
      "loss": 0.1137,
      "step": 1981
    },
    {
      "epoch": 0.27778556412053257,
      "grad_norm": 0.5131301283836365,
      "learning_rate": 0.00027880411384836163,
      "loss": 0.1123,
      "step": 1982
    },
    {
      "epoch": 0.27792571829011914,
      "grad_norm": 0.9090350270271301,
      "learning_rate": 0.00027878976321454196,
      "loss": 0.1331,
      "step": 1983
    },
    {
      "epoch": 0.27806587245970565,
      "grad_norm": 0.2680957615375519,
      "learning_rate": 0.0002787754125807223,
      "loss": 0.057,
      "step": 1984
    },
    {
      "epoch": 0.2782060266292922,
      "grad_norm": 0.3268495500087738,
      "learning_rate": 0.0002787610619469026,
      "loss": 0.0908,
      "step": 1985
    },
    {
      "epoch": 0.27834618079887874,
      "grad_norm": 0.9438428282737732,
      "learning_rate": 0.00027874671131308294,
      "loss": 0.1241,
      "step": 1986
    },
    {
      "epoch": 0.2784863349684653,
      "grad_norm": 0.8221608996391296,
      "learning_rate": 0.0002787323606792633,
      "loss": 0.1307,
      "step": 1987
    },
    {
      "epoch": 0.2786264891380519,
      "grad_norm": 0.25961288809776306,
      "learning_rate": 0.00027871801004544365,
      "loss": 0.0677,
      "step": 1988
    },
    {
      "epoch": 0.2787666433076384,
      "grad_norm": 0.3286683261394501,
      "learning_rate": 0.000278703659411624,
      "loss": 0.1018,
      "step": 1989
    },
    {
      "epoch": 0.27890679747722497,
      "grad_norm": 0.3367217779159546,
      "learning_rate": 0.00027868930877780437,
      "loss": 0.0607,
      "step": 1990
    },
    {
      "epoch": 0.2790469516468115,
      "grad_norm": 0.36401456594467163,
      "learning_rate": 0.0002786749581439847,
      "loss": 0.099,
      "step": 1991
    },
    {
      "epoch": 0.27918710581639805,
      "grad_norm": 1.0048301219940186,
      "learning_rate": 0.000278660607510165,
      "loss": 0.1441,
      "step": 1992
    },
    {
      "epoch": 0.27932725998598457,
      "grad_norm": 0.8426941633224487,
      "learning_rate": 0.00027864625687634535,
      "loss": 0.1515,
      "step": 1993
    },
    {
      "epoch": 0.27946741415557114,
      "grad_norm": 0.6565719246864319,
      "learning_rate": 0.0002786319062425257,
      "loss": 0.2035,
      "step": 1994
    },
    {
      "epoch": 0.27960756832515765,
      "grad_norm": 1.0536584854125977,
      "learning_rate": 0.000278617555608706,
      "loss": 0.3281,
      "step": 1995
    },
    {
      "epoch": 0.2797477224947442,
      "grad_norm": 0.9979150891304016,
      "learning_rate": 0.00027860320497488634,
      "loss": 0.1591,
      "step": 1996
    },
    {
      "epoch": 0.27988787666433074,
      "grad_norm": 1.3841400146484375,
      "learning_rate": 0.0002785888543410667,
      "loss": 0.262,
      "step": 1997
    },
    {
      "epoch": 0.2800280308339173,
      "grad_norm": 0.6444082856178284,
      "learning_rate": 0.00027857450370724705,
      "loss": 0.1298,
      "step": 1998
    },
    {
      "epoch": 0.2801681850035039,
      "grad_norm": 1.870656967163086,
      "learning_rate": 0.0002785601530734274,
      "loss": 0.2845,
      "step": 1999
    },
    {
      "epoch": 0.2803083391730904,
      "grad_norm": 2.030590057373047,
      "learning_rate": 0.00027854580243960776,
      "loss": 0.2723,
      "step": 2000
    },
    {
      "epoch": 0.28044849334267696,
      "grad_norm": 0.5691972970962524,
      "learning_rate": 0.0002785314518057881,
      "loss": 0.2209,
      "step": 2001
    },
    {
      "epoch": 0.2805886475122635,
      "grad_norm": 0.8753064870834351,
      "learning_rate": 0.0002785171011719684,
      "loss": 0.0718,
      "step": 2002
    },
    {
      "epoch": 0.28072880168185005,
      "grad_norm": 0.178070068359375,
      "learning_rate": 0.00027850275053814875,
      "loss": 0.0348,
      "step": 2003
    },
    {
      "epoch": 0.28086895585143656,
      "grad_norm": 0.2784869074821472,
      "learning_rate": 0.0002784883999043291,
      "loss": 0.0897,
      "step": 2004
    },
    {
      "epoch": 0.28100911002102313,
      "grad_norm": 0.6817112565040588,
      "learning_rate": 0.0002784740492705094,
      "loss": 0.0816,
      "step": 2005
    },
    {
      "epoch": 0.28114926419060965,
      "grad_norm": 0.4369611144065857,
      "learning_rate": 0.0002784596986366898,
      "loss": 0.114,
      "step": 2006
    },
    {
      "epoch": 0.2812894183601962,
      "grad_norm": 0.473345011472702,
      "learning_rate": 0.0002784453480028701,
      "loss": 0.1443,
      "step": 2007
    },
    {
      "epoch": 0.28142957252978273,
      "grad_norm": 0.41427335143089294,
      "learning_rate": 0.00027843099736905044,
      "loss": 0.1758,
      "step": 2008
    },
    {
      "epoch": 0.2815697266993693,
      "grad_norm": 0.40301406383514404,
      "learning_rate": 0.00027841664673523077,
      "loss": 0.0994,
      "step": 2009
    },
    {
      "epoch": 0.2817098808689559,
      "grad_norm": 0.44364669919013977,
      "learning_rate": 0.00027840229610141115,
      "loss": 0.1118,
      "step": 2010
    },
    {
      "epoch": 0.2818500350385424,
      "grad_norm": 0.48147669434547424,
      "learning_rate": 0.0002783879454675915,
      "loss": 0.1248,
      "step": 2011
    },
    {
      "epoch": 0.28199018920812896,
      "grad_norm": 0.293087363243103,
      "learning_rate": 0.0002783735948337718,
      "loss": 0.1014,
      "step": 2012
    },
    {
      "epoch": 0.2821303433777155,
      "grad_norm": 0.6260691285133362,
      "learning_rate": 0.00027835924419995214,
      "loss": 0.1585,
      "step": 2013
    },
    {
      "epoch": 0.28227049754730205,
      "grad_norm": 0.28251245617866516,
      "learning_rate": 0.00027834489356613247,
      "loss": 0.0681,
      "step": 2014
    },
    {
      "epoch": 0.28241065171688856,
      "grad_norm": 0.4360058605670929,
      "learning_rate": 0.0002783305429323128,
      "loss": 0.1168,
      "step": 2015
    },
    {
      "epoch": 0.28255080588647513,
      "grad_norm": 0.7703859806060791,
      "learning_rate": 0.0002783161922984932,
      "loss": 0.0812,
      "step": 2016
    },
    {
      "epoch": 0.28269096005606165,
      "grad_norm": 0.7642639875411987,
      "learning_rate": 0.0002783018416646735,
      "loss": 0.0815,
      "step": 2017
    },
    {
      "epoch": 0.2828311142256482,
      "grad_norm": 1.523699164390564,
      "learning_rate": 0.00027828749103085384,
      "loss": 0.1542,
      "step": 2018
    },
    {
      "epoch": 0.28297126839523473,
      "grad_norm": 0.9008207321166992,
      "learning_rate": 0.00027827314039703416,
      "loss": 0.1171,
      "step": 2019
    },
    {
      "epoch": 0.2831114225648213,
      "grad_norm": 0.40990594029426575,
      "learning_rate": 0.0002782587897632145,
      "loss": 0.1253,
      "step": 2020
    },
    {
      "epoch": 0.2832515767344079,
      "grad_norm": 0.33235472440719604,
      "learning_rate": 0.0002782444391293948,
      "loss": 0.1417,
      "step": 2021
    },
    {
      "epoch": 0.2833917309039944,
      "grad_norm": 0.7677909135818481,
      "learning_rate": 0.0002782300884955752,
      "loss": 0.1365,
      "step": 2022
    },
    {
      "epoch": 0.28353188507358096,
      "grad_norm": 0.34211885929107666,
      "learning_rate": 0.00027821573786175553,
      "loss": 0.0595,
      "step": 2023
    },
    {
      "epoch": 0.2836720392431675,
      "grad_norm": 0.6408612132072449,
      "learning_rate": 0.00027820138722793586,
      "loss": 0.1728,
      "step": 2024
    },
    {
      "epoch": 0.28381219341275404,
      "grad_norm": 0.5557684302330017,
      "learning_rate": 0.00027818703659411624,
      "loss": 0.1429,
      "step": 2025
    },
    {
      "epoch": 0.28395234758234056,
      "grad_norm": 0.45655104517936707,
      "learning_rate": 0.0002781726859602966,
      "loss": 0.0908,
      "step": 2026
    },
    {
      "epoch": 0.28409250175192713,
      "grad_norm": 0.8670726418495178,
      "learning_rate": 0.0002781583353264769,
      "loss": 0.0976,
      "step": 2027
    },
    {
      "epoch": 0.28423265592151364,
      "grad_norm": 0.7119932770729065,
      "learning_rate": 0.00027814398469265723,
      "loss": 0.1147,
      "step": 2028
    },
    {
      "epoch": 0.2843728100911002,
      "grad_norm": 0.45708343386650085,
      "learning_rate": 0.00027812963405883756,
      "loss": 0.1077,
      "step": 2029
    },
    {
      "epoch": 0.28451296426068673,
      "grad_norm": 0.26318418979644775,
      "learning_rate": 0.0002781152834250179,
      "loss": 0.0616,
      "step": 2030
    },
    {
      "epoch": 0.2846531184302733,
      "grad_norm": 0.7760747075080872,
      "learning_rate": 0.0002781009327911982,
      "loss": 0.086,
      "step": 2031
    },
    {
      "epoch": 0.28479327259985987,
      "grad_norm": 1.0396136045455933,
      "learning_rate": 0.0002780865821573786,
      "loss": 0.1421,
      "step": 2032
    },
    {
      "epoch": 0.2849334267694464,
      "grad_norm": 0.6434550285339355,
      "learning_rate": 0.00027807223152355893,
      "loss": 0.1136,
      "step": 2033
    },
    {
      "epoch": 0.28507358093903296,
      "grad_norm": 0.3558732867240906,
      "learning_rate": 0.00027805788088973926,
      "loss": 0.0528,
      "step": 2034
    },
    {
      "epoch": 0.28521373510861947,
      "grad_norm": 0.6915003657341003,
      "learning_rate": 0.00027804353025591964,
      "loss": 0.2294,
      "step": 2035
    },
    {
      "epoch": 0.28535388927820604,
      "grad_norm": 0.6733664274215698,
      "learning_rate": 0.00027802917962209997,
      "loss": 0.1216,
      "step": 2036
    },
    {
      "epoch": 0.28549404344779256,
      "grad_norm": 0.9360905885696411,
      "learning_rate": 0.0002780148289882803,
      "loss": 0.1214,
      "step": 2037
    },
    {
      "epoch": 0.2856341976173791,
      "grad_norm": 1.0321345329284668,
      "learning_rate": 0.0002780004783544606,
      "loss": 0.2451,
      "step": 2038
    },
    {
      "epoch": 0.28577435178696564,
      "grad_norm": 0.775350034236908,
      "learning_rate": 0.00027798612772064095,
      "loss": 0.0924,
      "step": 2039
    },
    {
      "epoch": 0.2859145059565522,
      "grad_norm": 2.416033983230591,
      "learning_rate": 0.0002779717770868213,
      "loss": 0.3264,
      "step": 2040
    },
    {
      "epoch": 0.2860546601261387,
      "grad_norm": 0.5261764526367188,
      "learning_rate": 0.00027795742645300166,
      "loss": 0.0972,
      "step": 2041
    },
    {
      "epoch": 0.2861948142957253,
      "grad_norm": 0.5157500505447388,
      "learning_rate": 0.000277943075819182,
      "loss": 0.1498,
      "step": 2042
    },
    {
      "epoch": 0.28633496846531187,
      "grad_norm": 0.8179082274436951,
      "learning_rate": 0.0002779287251853623,
      "loss": 0.176,
      "step": 2043
    },
    {
      "epoch": 0.2864751226348984,
      "grad_norm": 0.9213723540306091,
      "learning_rate": 0.0002779143745515427,
      "loss": 0.179,
      "step": 2044
    },
    {
      "epoch": 0.28661527680448495,
      "grad_norm": 0.7316744327545166,
      "learning_rate": 0.00027790002391772303,
      "loss": 0.2004,
      "step": 2045
    },
    {
      "epoch": 0.28675543097407147,
      "grad_norm": 0.3566551208496094,
      "learning_rate": 0.00027788567328390336,
      "loss": 0.0421,
      "step": 2046
    },
    {
      "epoch": 0.28689558514365804,
      "grad_norm": 1.1435343027114868,
      "learning_rate": 0.0002778713226500837,
      "loss": 0.259,
      "step": 2047
    },
    {
      "epoch": 0.28703573931324455,
      "grad_norm": 1.2187875509262085,
      "learning_rate": 0.000277856972016264,
      "loss": 0.1821,
      "step": 2048
    },
    {
      "epoch": 0.2871758934828311,
      "grad_norm": 2.9546449184417725,
      "learning_rate": 0.00027784262138244435,
      "loss": 0.2565,
      "step": 2049
    },
    {
      "epoch": 0.28731604765241764,
      "grad_norm": 3.7390589714050293,
      "learning_rate": 0.0002778282707486247,
      "loss": 0.4227,
      "step": 2050
    },
    {
      "epoch": 0.2874562018220042,
      "grad_norm": 0.783820629119873,
      "learning_rate": 0.00027781392011480506,
      "loss": 0.2159,
      "step": 2051
    },
    {
      "epoch": 0.2875963559915907,
      "grad_norm": 0.3262910544872284,
      "learning_rate": 0.0002777995694809854,
      "loss": 0.07,
      "step": 2052
    },
    {
      "epoch": 0.2877365101611773,
      "grad_norm": 0.5076556205749512,
      "learning_rate": 0.0002777852188471657,
      "loss": 0.0809,
      "step": 2053
    },
    {
      "epoch": 0.28787666433076387,
      "grad_norm": 0.4554128050804138,
      "learning_rate": 0.0002777708682133461,
      "loss": 0.1281,
      "step": 2054
    },
    {
      "epoch": 0.2880168185003504,
      "grad_norm": 0.6828322410583496,
      "learning_rate": 0.0002777565175795264,
      "loss": 0.1742,
      "step": 2055
    },
    {
      "epoch": 0.28815697266993695,
      "grad_norm": 0.45777180790901184,
      "learning_rate": 0.00027774216694570676,
      "loss": 0.1594,
      "step": 2056
    },
    {
      "epoch": 0.28829712683952347,
      "grad_norm": 0.8016679883003235,
      "learning_rate": 0.0002777278163118871,
      "loss": 0.0888,
      "step": 2057
    },
    {
      "epoch": 0.28843728100911004,
      "grad_norm": 0.48551881313323975,
      "learning_rate": 0.0002777134656780674,
      "loss": 0.1238,
      "step": 2058
    },
    {
      "epoch": 0.28857743517869655,
      "grad_norm": 0.399253249168396,
      "learning_rate": 0.00027769911504424774,
      "loss": 0.1403,
      "step": 2059
    },
    {
      "epoch": 0.2887175893482831,
      "grad_norm": 1.1162843704223633,
      "learning_rate": 0.0002776847644104281,
      "loss": 0.096,
      "step": 2060
    },
    {
      "epoch": 0.28885774351786964,
      "grad_norm": 0.4744342267513275,
      "learning_rate": 0.00027767041377660845,
      "loss": 0.1264,
      "step": 2061
    },
    {
      "epoch": 0.2889978976874562,
      "grad_norm": 0.4353397786617279,
      "learning_rate": 0.0002776560631427888,
      "loss": 0.0911,
      "step": 2062
    },
    {
      "epoch": 0.2891380518570427,
      "grad_norm": 0.5171084403991699,
      "learning_rate": 0.0002776417125089691,
      "loss": 0.0827,
      "step": 2063
    },
    {
      "epoch": 0.2892782060266293,
      "grad_norm": 0.5962603092193604,
      "learning_rate": 0.0002776273618751495,
      "loss": 0.0975,
      "step": 2064
    },
    {
      "epoch": 0.28941836019621586,
      "grad_norm": 0.884111762046814,
      "learning_rate": 0.0002776130112413298,
      "loss": 0.1641,
      "step": 2065
    },
    {
      "epoch": 0.2895585143658024,
      "grad_norm": 0.4736906886100769,
      "learning_rate": 0.00027759866060751015,
      "loss": 0.1806,
      "step": 2066
    },
    {
      "epoch": 0.28969866853538895,
      "grad_norm": 0.5883544683456421,
      "learning_rate": 0.0002775843099736905,
      "loss": 0.1239,
      "step": 2067
    },
    {
      "epoch": 0.28983882270497546,
      "grad_norm": 0.6384047269821167,
      "learning_rate": 0.0002775699593398708,
      "loss": 0.1671,
      "step": 2068
    },
    {
      "epoch": 0.28997897687456203,
      "grad_norm": 0.7631522417068481,
      "learning_rate": 0.00027755560870605113,
      "loss": 0.0713,
      "step": 2069
    },
    {
      "epoch": 0.29011913104414855,
      "grad_norm": 0.6706103086471558,
      "learning_rate": 0.0002775412580722315,
      "loss": 0.1271,
      "step": 2070
    },
    {
      "epoch": 0.2902592852137351,
      "grad_norm": 0.3679930567741394,
      "learning_rate": 0.00027752690743841185,
      "loss": 0.1174,
      "step": 2071
    },
    {
      "epoch": 0.29039943938332163,
      "grad_norm": 0.416925311088562,
      "learning_rate": 0.0002775125568045922,
      "loss": 0.1083,
      "step": 2072
    },
    {
      "epoch": 0.2905395935529082,
      "grad_norm": 0.6641778349876404,
      "learning_rate": 0.0002774982061707725,
      "loss": 0.1367,
      "step": 2073
    },
    {
      "epoch": 0.2906797477224947,
      "grad_norm": 0.5840435028076172,
      "learning_rate": 0.0002774838555369529,
      "loss": 0.0973,
      "step": 2074
    },
    {
      "epoch": 0.2908199018920813,
      "grad_norm": 0.8699474930763245,
      "learning_rate": 0.0002774695049031332,
      "loss": 0.1856,
      "step": 2075
    },
    {
      "epoch": 0.29096005606166786,
      "grad_norm": 0.5407617688179016,
      "learning_rate": 0.00027745515426931354,
      "loss": 0.0798,
      "step": 2076
    },
    {
      "epoch": 0.2911002102312544,
      "grad_norm": 0.4400215148925781,
      "learning_rate": 0.00027744080363549387,
      "loss": 0.1526,
      "step": 2077
    },
    {
      "epoch": 0.29124036440084095,
      "grad_norm": 0.6826701760292053,
      "learning_rate": 0.0002774264530016742,
      "loss": 0.1006,
      "step": 2078
    },
    {
      "epoch": 0.29138051857042746,
      "grad_norm": 0.33418065309524536,
      "learning_rate": 0.0002774121023678546,
      "loss": 0.0771,
      "step": 2079
    },
    {
      "epoch": 0.29152067274001403,
      "grad_norm": 0.46389448642730713,
      "learning_rate": 0.0002773977517340349,
      "loss": 0.1087,
      "step": 2080
    },
    {
      "epoch": 0.29166082690960055,
      "grad_norm": 0.80023592710495,
      "learning_rate": 0.00027738340110021524,
      "loss": 0.115,
      "step": 2081
    },
    {
      "epoch": 0.2918009810791871,
      "grad_norm": 0.9113709926605225,
      "learning_rate": 0.00027736905046639557,
      "loss": 0.106,
      "step": 2082
    },
    {
      "epoch": 0.29194113524877363,
      "grad_norm": 0.8177204728126526,
      "learning_rate": 0.0002773546998325759,
      "loss": 0.1477,
      "step": 2083
    },
    {
      "epoch": 0.2920812894183602,
      "grad_norm": 0.9234193563461304,
      "learning_rate": 0.0002773403491987562,
      "loss": 0.2335,
      "step": 2084
    },
    {
      "epoch": 0.2922214435879467,
      "grad_norm": 0.30937492847442627,
      "learning_rate": 0.00027732599856493655,
      "loss": 0.1049,
      "step": 2085
    },
    {
      "epoch": 0.2923615977575333,
      "grad_norm": 0.6365763545036316,
      "learning_rate": 0.00027731164793111694,
      "loss": 0.094,
      "step": 2086
    },
    {
      "epoch": 0.29250175192711986,
      "grad_norm": 0.7770087718963623,
      "learning_rate": 0.00027729729729729727,
      "loss": 0.1203,
      "step": 2087
    },
    {
      "epoch": 0.2926419060967064,
      "grad_norm": 1.9098001718521118,
      "learning_rate": 0.0002772829466634776,
      "loss": 0.2285,
      "step": 2088
    },
    {
      "epoch": 0.29278206026629294,
      "grad_norm": 0.9815558791160583,
      "learning_rate": 0.000277268596029658,
      "loss": 0.1604,
      "step": 2089
    },
    {
      "epoch": 0.29292221443587946,
      "grad_norm": 0.6442099809646606,
      "learning_rate": 0.0002772542453958383,
      "loss": 0.1251,
      "step": 2090
    },
    {
      "epoch": 0.29306236860546603,
      "grad_norm": 0.6111890077590942,
      "learning_rate": 0.00027723989476201863,
      "loss": 0.0643,
      "step": 2091
    },
    {
      "epoch": 0.29320252277505254,
      "grad_norm": 1.4717985391616821,
      "learning_rate": 0.00027722554412819896,
      "loss": 0.1283,
      "step": 2092
    },
    {
      "epoch": 0.2933426769446391,
      "grad_norm": 0.5396863222122192,
      "learning_rate": 0.0002772111934943793,
      "loss": 0.1471,
      "step": 2093
    },
    {
      "epoch": 0.29348283111422563,
      "grad_norm": 2.047018051147461,
      "learning_rate": 0.0002771968428605596,
      "loss": 0.2377,
      "step": 2094
    },
    {
      "epoch": 0.2936229852838122,
      "grad_norm": 0.45205017924308777,
      "learning_rate": 0.00027718249222674,
      "loss": 0.1273,
      "step": 2095
    },
    {
      "epoch": 0.2937631394533987,
      "grad_norm": 0.9719533920288086,
      "learning_rate": 0.00027716814159292033,
      "loss": 0.2856,
      "step": 2096
    },
    {
      "epoch": 0.2939032936229853,
      "grad_norm": 0.8585732579231262,
      "learning_rate": 0.00027715379095910066,
      "loss": 0.1802,
      "step": 2097
    },
    {
      "epoch": 0.29404344779257185,
      "grad_norm": 1.4114614725112915,
      "learning_rate": 0.000277139440325281,
      "loss": 0.3464,
      "step": 2098
    },
    {
      "epoch": 0.29418360196215837,
      "grad_norm": 1.4573348760604858,
      "learning_rate": 0.00027712508969146137,
      "loss": 0.362,
      "step": 2099
    },
    {
      "epoch": 0.29432375613174494,
      "grad_norm": 1.972069263458252,
      "learning_rate": 0.0002771107390576417,
      "loss": 0.1895,
      "step": 2100
    },
    {
      "epoch": 0.29446391030133146,
      "grad_norm": 0.515295147895813,
      "learning_rate": 0.00027709638842382203,
      "loss": 0.1237,
      "step": 2101
    },
    {
      "epoch": 0.294604064470918,
      "grad_norm": 0.3586462736129761,
      "learning_rate": 0.00027708203779000236,
      "loss": 0.1471,
      "step": 2102
    },
    {
      "epoch": 0.29474421864050454,
      "grad_norm": 0.5159236192703247,
      "learning_rate": 0.0002770676871561827,
      "loss": 0.082,
      "step": 2103
    },
    {
      "epoch": 0.2948843728100911,
      "grad_norm": 0.5777571797370911,
      "learning_rate": 0.000277053336522363,
      "loss": 0.1718,
      "step": 2104
    },
    {
      "epoch": 0.2950245269796776,
      "grad_norm": 0.44277429580688477,
      "learning_rate": 0.0002770389858885434,
      "loss": 0.1499,
      "step": 2105
    },
    {
      "epoch": 0.2951646811492642,
      "grad_norm": 0.757902979850769,
      "learning_rate": 0.0002770246352547237,
      "loss": 0.1492,
      "step": 2106
    },
    {
      "epoch": 0.2953048353188507,
      "grad_norm": 0.5982668399810791,
      "learning_rate": 0.00027701028462090405,
      "loss": 0.0845,
      "step": 2107
    },
    {
      "epoch": 0.2954449894884373,
      "grad_norm": 0.5739657878875732,
      "learning_rate": 0.00027699593398708444,
      "loss": 0.1285,
      "step": 2108
    },
    {
      "epoch": 0.29558514365802385,
      "grad_norm": 0.6959678530693054,
      "learning_rate": 0.00027698158335326477,
      "loss": 0.1415,
      "step": 2109
    },
    {
      "epoch": 0.29572529782761037,
      "grad_norm": 0.630250871181488,
      "learning_rate": 0.0002769672327194451,
      "loss": 0.1056,
      "step": 2110
    },
    {
      "epoch": 0.29586545199719694,
      "grad_norm": 0.36131590604782104,
      "learning_rate": 0.0002769528820856254,
      "loss": 0.092,
      "step": 2111
    },
    {
      "epoch": 0.29600560616678345,
      "grad_norm": 0.7285513877868652,
      "learning_rate": 0.00027693853145180575,
      "loss": 0.1308,
      "step": 2112
    },
    {
      "epoch": 0.29614576033637,
      "grad_norm": 0.640658974647522,
      "learning_rate": 0.0002769241808179861,
      "loss": 0.2182,
      "step": 2113
    },
    {
      "epoch": 0.29628591450595654,
      "grad_norm": 0.43515151739120483,
      "learning_rate": 0.00027690983018416646,
      "loss": 0.1212,
      "step": 2114
    },
    {
      "epoch": 0.2964260686755431,
      "grad_norm": 0.4214324951171875,
      "learning_rate": 0.0002768954795503468,
      "loss": 0.0648,
      "step": 2115
    },
    {
      "epoch": 0.2965662228451296,
      "grad_norm": 0.31286269426345825,
      "learning_rate": 0.0002768811289165271,
      "loss": 0.094,
      "step": 2116
    },
    {
      "epoch": 0.2967063770147162,
      "grad_norm": 0.657855749130249,
      "learning_rate": 0.00027686677828270745,
      "loss": 0.1252,
      "step": 2117
    },
    {
      "epoch": 0.2968465311843027,
      "grad_norm": 0.5995632410049438,
      "learning_rate": 0.00027685242764888783,
      "loss": 0.1427,
      "step": 2118
    },
    {
      "epoch": 0.2969866853538893,
      "grad_norm": 0.6053673028945923,
      "learning_rate": 0.00027683807701506816,
      "loss": 0.1685,
      "step": 2119
    },
    {
      "epoch": 0.29712683952347585,
      "grad_norm": 0.2206595540046692,
      "learning_rate": 0.0002768237263812485,
      "loss": 0.0544,
      "step": 2120
    },
    {
      "epoch": 0.29726699369306236,
      "grad_norm": 0.4044966995716095,
      "learning_rate": 0.0002768093757474288,
      "loss": 0.1102,
      "step": 2121
    },
    {
      "epoch": 0.29740714786264894,
      "grad_norm": 0.6795877814292908,
      "learning_rate": 0.00027679502511360914,
      "loss": 0.1473,
      "step": 2122
    },
    {
      "epoch": 0.29754730203223545,
      "grad_norm": 0.3030228614807129,
      "learning_rate": 0.0002767806744797895,
      "loss": 0.0765,
      "step": 2123
    },
    {
      "epoch": 0.297687456201822,
      "grad_norm": 0.41661736369132996,
      "learning_rate": 0.00027676632384596986,
      "loss": 0.1451,
      "step": 2124
    },
    {
      "epoch": 0.29782761037140854,
      "grad_norm": 0.26537394523620605,
      "learning_rate": 0.0002767519732121502,
      "loss": 0.0769,
      "step": 2125
    },
    {
      "epoch": 0.2979677645409951,
      "grad_norm": 0.6869179010391235,
      "learning_rate": 0.0002767376225783305,
      "loss": 0.1007,
      "step": 2126
    },
    {
      "epoch": 0.2981079187105816,
      "grad_norm": 0.3951539993286133,
      "learning_rate": 0.0002767232719445109,
      "loss": 0.1099,
      "step": 2127
    },
    {
      "epoch": 0.2982480728801682,
      "grad_norm": 0.812346339225769,
      "learning_rate": 0.0002767089213106912,
      "loss": 0.1996,
      "step": 2128
    },
    {
      "epoch": 0.2983882270497547,
      "grad_norm": 0.5661276578903198,
      "learning_rate": 0.00027669457067687155,
      "loss": 0.079,
      "step": 2129
    },
    {
      "epoch": 0.2985283812193413,
      "grad_norm": 0.6014966368675232,
      "learning_rate": 0.0002766802200430519,
      "loss": 0.119,
      "step": 2130
    },
    {
      "epoch": 0.29866853538892785,
      "grad_norm": 0.6107602119445801,
      "learning_rate": 0.0002766658694092322,
      "loss": 0.1715,
      "step": 2131
    },
    {
      "epoch": 0.29880868955851436,
      "grad_norm": 0.4255887269973755,
      "learning_rate": 0.00027665151877541254,
      "loss": 0.1627,
      "step": 2132
    },
    {
      "epoch": 0.29894884372810093,
      "grad_norm": 0.9967101216316223,
      "learning_rate": 0.00027663716814159287,
      "loss": 0.1354,
      "step": 2133
    },
    {
      "epoch": 0.29908899789768745,
      "grad_norm": 0.6783173680305481,
      "learning_rate": 0.00027662281750777325,
      "loss": 0.1371,
      "step": 2134
    },
    {
      "epoch": 0.299229152067274,
      "grad_norm": 0.5224384665489197,
      "learning_rate": 0.0002766084668739536,
      "loss": 0.1273,
      "step": 2135
    },
    {
      "epoch": 0.29936930623686053,
      "grad_norm": 1.199905514717102,
      "learning_rate": 0.0002765941162401339,
      "loss": 0.1542,
      "step": 2136
    },
    {
      "epoch": 0.2995094604064471,
      "grad_norm": 0.5567576289176941,
      "learning_rate": 0.0002765797656063143,
      "loss": 0.1095,
      "step": 2137
    },
    {
      "epoch": 0.2996496145760336,
      "grad_norm": 0.2655661404132843,
      "learning_rate": 0.0002765654149724946,
      "loss": 0.0691,
      "step": 2138
    },
    {
      "epoch": 0.2997897687456202,
      "grad_norm": 0.6687948107719421,
      "learning_rate": 0.00027655106433867495,
      "loss": 0.1734,
      "step": 2139
    },
    {
      "epoch": 0.2999299229152067,
      "grad_norm": 0.6922125220298767,
      "learning_rate": 0.0002765367137048553,
      "loss": 0.1694,
      "step": 2140
    },
    {
      "epoch": 0.3000700770847933,
      "grad_norm": 0.2712704837322235,
      "learning_rate": 0.0002765223630710356,
      "loss": 0.0718,
      "step": 2141
    },
    {
      "epoch": 0.30021023125437984,
      "grad_norm": 0.9073896408081055,
      "learning_rate": 0.00027650801243721593,
      "loss": 0.1681,
      "step": 2142
    },
    {
      "epoch": 0.30035038542396636,
      "grad_norm": 0.5596049427986145,
      "learning_rate": 0.0002764936618033963,
      "loss": 0.157,
      "step": 2143
    },
    {
      "epoch": 0.30049053959355293,
      "grad_norm": 0.5209670066833496,
      "learning_rate": 0.00027647931116957664,
      "loss": 0.1109,
      "step": 2144
    },
    {
      "epoch": 0.30063069376313944,
      "grad_norm": 0.7527725696563721,
      "learning_rate": 0.00027646496053575697,
      "loss": 0.2062,
      "step": 2145
    },
    {
      "epoch": 0.300770847932726,
      "grad_norm": 0.6075131297111511,
      "learning_rate": 0.0002764506099019373,
      "loss": 0.222,
      "step": 2146
    },
    {
      "epoch": 0.30091100210231253,
      "grad_norm": 1.0059218406677246,
      "learning_rate": 0.00027643625926811763,
      "loss": 0.1261,
      "step": 2147
    },
    {
      "epoch": 0.3010511562718991,
      "grad_norm": 2.278404474258423,
      "learning_rate": 0.00027642190863429796,
      "loss": 0.3453,
      "step": 2148
    },
    {
      "epoch": 0.3011913104414856,
      "grad_norm": 2.37385892868042,
      "learning_rate": 0.00027640755800047834,
      "loss": 0.0978,
      "step": 2149
    },
    {
      "epoch": 0.3013314646110722,
      "grad_norm": 2.6766862869262695,
      "learning_rate": 0.00027639320736665867,
      "loss": 0.1361,
      "step": 2150
    },
    {
      "epoch": 0.3014716187806587,
      "grad_norm": 0.46212106943130493,
      "learning_rate": 0.000276378856732839,
      "loss": 0.1649,
      "step": 2151
    },
    {
      "epoch": 0.30161177295024527,
      "grad_norm": 0.6219832301139832,
      "learning_rate": 0.0002763645060990193,
      "loss": 0.1395,
      "step": 2152
    },
    {
      "epoch": 0.30175192711983184,
      "grad_norm": 0.4304855465888977,
      "learning_rate": 0.0002763501554651997,
      "loss": 0.1202,
      "step": 2153
    },
    {
      "epoch": 0.30189208128941836,
      "grad_norm": 0.7878482341766357,
      "learning_rate": 0.00027633580483138004,
      "loss": 0.1496,
      "step": 2154
    },
    {
      "epoch": 0.3020322354590049,
      "grad_norm": 0.48217785358428955,
      "learning_rate": 0.00027632145419756037,
      "loss": 0.1022,
      "step": 2155
    },
    {
      "epoch": 0.30217238962859144,
      "grad_norm": 0.4914371371269226,
      "learning_rate": 0.0002763071035637407,
      "loss": 0.0931,
      "step": 2156
    },
    {
      "epoch": 0.302312543798178,
      "grad_norm": 0.46999767422676086,
      "learning_rate": 0.000276292752929921,
      "loss": 0.1325,
      "step": 2157
    },
    {
      "epoch": 0.3024526979677645,
      "grad_norm": 0.328476220369339,
      "learning_rate": 0.00027627840229610135,
      "loss": 0.1263,
      "step": 2158
    },
    {
      "epoch": 0.3025928521373511,
      "grad_norm": 0.729601263999939,
      "learning_rate": 0.00027626405166228174,
      "loss": 0.1066,
      "step": 2159
    },
    {
      "epoch": 0.3027330063069376,
      "grad_norm": 0.3119988739490509,
      "learning_rate": 0.00027624970102846206,
      "loss": 0.0844,
      "step": 2160
    },
    {
      "epoch": 0.3028731604765242,
      "grad_norm": 0.7414813041687012,
      "learning_rate": 0.0002762353503946424,
      "loss": 0.2207,
      "step": 2161
    },
    {
      "epoch": 0.3030133146461107,
      "grad_norm": 0.469025194644928,
      "learning_rate": 0.0002762209997608228,
      "loss": 0.1302,
      "step": 2162
    },
    {
      "epoch": 0.30315346881569727,
      "grad_norm": 0.684956967830658,
      "learning_rate": 0.0002762066491270031,
      "loss": 0.2083,
      "step": 2163
    },
    {
      "epoch": 0.30329362298528384,
      "grad_norm": 0.4641209542751312,
      "learning_rate": 0.00027619229849318343,
      "loss": 0.1101,
      "step": 2164
    },
    {
      "epoch": 0.30343377715487035,
      "grad_norm": 0.29255983233451843,
      "learning_rate": 0.00027617794785936376,
      "loss": 0.108,
      "step": 2165
    },
    {
      "epoch": 0.3035739313244569,
      "grad_norm": 1.2404125928878784,
      "learning_rate": 0.0002761635972255441,
      "loss": 0.2365,
      "step": 2166
    },
    {
      "epoch": 0.30371408549404344,
      "grad_norm": 0.3747747540473938,
      "learning_rate": 0.0002761492465917244,
      "loss": 0.1346,
      "step": 2167
    },
    {
      "epoch": 0.30385423966363,
      "grad_norm": 0.41646280884742737,
      "learning_rate": 0.00027613489595790475,
      "loss": 0.1159,
      "step": 2168
    },
    {
      "epoch": 0.3039943938332165,
      "grad_norm": 0.5553839802742004,
      "learning_rate": 0.00027612054532408513,
      "loss": 0.0909,
      "step": 2169
    },
    {
      "epoch": 0.3041345480028031,
      "grad_norm": 0.20594193041324615,
      "learning_rate": 0.00027610619469026546,
      "loss": 0.0503,
      "step": 2170
    },
    {
      "epoch": 0.3042747021723896,
      "grad_norm": 0.503252387046814,
      "learning_rate": 0.0002760918440564458,
      "loss": 0.115,
      "step": 2171
    },
    {
      "epoch": 0.3044148563419762,
      "grad_norm": 0.645164966583252,
      "learning_rate": 0.00027607749342262617,
      "loss": 0.1806,
      "step": 2172
    },
    {
      "epoch": 0.3045550105115627,
      "grad_norm": 0.5107470750808716,
      "learning_rate": 0.0002760631427888065,
      "loss": 0.1072,
      "step": 2173
    },
    {
      "epoch": 0.30469516468114927,
      "grad_norm": 0.41755470633506775,
      "learning_rate": 0.0002760487921549868,
      "loss": 0.2055,
      "step": 2174
    },
    {
      "epoch": 0.30483531885073584,
      "grad_norm": 0.3775356411933899,
      "learning_rate": 0.00027603444152116715,
      "loss": 0.1298,
      "step": 2175
    },
    {
      "epoch": 0.30497547302032235,
      "grad_norm": 0.6275502443313599,
      "learning_rate": 0.0002760200908873475,
      "loss": 0.1854,
      "step": 2176
    },
    {
      "epoch": 0.3051156271899089,
      "grad_norm": 0.17002259194850922,
      "learning_rate": 0.0002760057402535278,
      "loss": 0.0361,
      "step": 2177
    },
    {
      "epoch": 0.30525578135949544,
      "grad_norm": 0.5578188300132751,
      "learning_rate": 0.0002759913896197082,
      "loss": 0.1732,
      "step": 2178
    },
    {
      "epoch": 0.305395935529082,
      "grad_norm": 0.4350545406341553,
      "learning_rate": 0.0002759770389858885,
      "loss": 0.1123,
      "step": 2179
    },
    {
      "epoch": 0.3055360896986685,
      "grad_norm": 0.6175668239593506,
      "learning_rate": 0.00027596268835206885,
      "loss": 0.1147,
      "step": 2180
    },
    {
      "epoch": 0.3056762438682551,
      "grad_norm": 0.4442678689956665,
      "learning_rate": 0.00027594833771824923,
      "loss": 0.1469,
      "step": 2181
    },
    {
      "epoch": 0.3058163980378416,
      "grad_norm": 0.4055914580821991,
      "learning_rate": 0.00027593398708442956,
      "loss": 0.101,
      "step": 2182
    },
    {
      "epoch": 0.3059565522074282,
      "grad_norm": 0.4393932521343231,
      "learning_rate": 0.0002759196364506099,
      "loss": 0.1427,
      "step": 2183
    },
    {
      "epoch": 0.3060967063770147,
      "grad_norm": 0.41718894243240356,
      "learning_rate": 0.0002759052858167902,
      "loss": 0.0872,
      "step": 2184
    },
    {
      "epoch": 0.30623686054660126,
      "grad_norm": 0.5062853693962097,
      "learning_rate": 0.00027589093518297055,
      "loss": 0.1061,
      "step": 2185
    },
    {
      "epoch": 0.30637701471618783,
      "grad_norm": 0.4873693585395813,
      "learning_rate": 0.0002758765845491509,
      "loss": 0.0687,
      "step": 2186
    },
    {
      "epoch": 0.30651716888577435,
      "grad_norm": 0.2480587512254715,
      "learning_rate": 0.0002758622339153312,
      "loss": 0.046,
      "step": 2187
    },
    {
      "epoch": 0.3066573230553609,
      "grad_norm": 0.6864371299743652,
      "learning_rate": 0.0002758478832815116,
      "loss": 0.0848,
      "step": 2188
    },
    {
      "epoch": 0.30679747722494743,
      "grad_norm": 0.4345424473285675,
      "learning_rate": 0.0002758335326476919,
      "loss": 0.061,
      "step": 2189
    },
    {
      "epoch": 0.306937631394534,
      "grad_norm": 1.7501213550567627,
      "learning_rate": 0.00027581918201387225,
      "loss": 0.2407,
      "step": 2190
    },
    {
      "epoch": 0.3070777855641205,
      "grad_norm": 0.7332161068916321,
      "learning_rate": 0.00027580483138005263,
      "loss": 0.3168,
      "step": 2191
    },
    {
      "epoch": 0.3072179397337071,
      "grad_norm": 0.611071765422821,
      "learning_rate": 0.00027579048074623296,
      "loss": 0.0883,
      "step": 2192
    },
    {
      "epoch": 0.3073580939032936,
      "grad_norm": 0.386831670999527,
      "learning_rate": 0.0002757761301124133,
      "loss": 0.0785,
      "step": 2193
    },
    {
      "epoch": 0.3074982480728802,
      "grad_norm": 1.0226197242736816,
      "learning_rate": 0.0002757617794785936,
      "loss": 0.294,
      "step": 2194
    },
    {
      "epoch": 0.3076384022424667,
      "grad_norm": 0.9176251888275146,
      "learning_rate": 0.00027574742884477394,
      "loss": 0.3836,
      "step": 2195
    },
    {
      "epoch": 0.30777855641205326,
      "grad_norm": 2.0792934894561768,
      "learning_rate": 0.00027573307821095427,
      "loss": 0.2776,
      "step": 2196
    },
    {
      "epoch": 0.3079187105816398,
      "grad_norm": 1.2955394983291626,
      "learning_rate": 0.00027571872757713465,
      "loss": 0.5569,
      "step": 2197
    },
    {
      "epoch": 0.30805886475122635,
      "grad_norm": 1.8353787660598755,
      "learning_rate": 0.000275704376943315,
      "loss": 0.3247,
      "step": 2198
    },
    {
      "epoch": 0.3081990189208129,
      "grad_norm": 0.8331082463264465,
      "learning_rate": 0.0002756900263094953,
      "loss": 0.2205,
      "step": 2199
    },
    {
      "epoch": 0.30833917309039943,
      "grad_norm": 1.3991813659667969,
      "learning_rate": 0.00027567567567567564,
      "loss": 0.1944,
      "step": 2200
    },
    {
      "epoch": 0.308479327259986,
      "grad_norm": 0.2933048605918884,
      "learning_rate": 0.000275661325041856,
      "loss": 0.1416,
      "step": 2201
    },
    {
      "epoch": 0.3086194814295725,
      "grad_norm": 0.3607500493526459,
      "learning_rate": 0.00027564697440803635,
      "loss": 0.1689,
      "step": 2202
    },
    {
      "epoch": 0.3087596355991591,
      "grad_norm": 0.5643554925918579,
      "learning_rate": 0.0002756326237742167,
      "loss": 0.2199,
      "step": 2203
    },
    {
      "epoch": 0.3088997897687456,
      "grad_norm": 0.36085689067840576,
      "learning_rate": 0.000275618273140397,
      "loss": 0.1228,
      "step": 2204
    },
    {
      "epoch": 0.3090399439383322,
      "grad_norm": 0.39612624049186707,
      "learning_rate": 0.00027560392250657734,
      "loss": 0.1684,
      "step": 2205
    },
    {
      "epoch": 0.3091800981079187,
      "grad_norm": 0.6036938428878784,
      "learning_rate": 0.00027558957187275767,
      "loss": 0.1474,
      "step": 2206
    },
    {
      "epoch": 0.30932025227750526,
      "grad_norm": 0.37125927209854126,
      "learning_rate": 0.00027557522123893805,
      "loss": 0.1597,
      "step": 2207
    },
    {
      "epoch": 0.3094604064470918,
      "grad_norm": 0.42839667201042175,
      "learning_rate": 0.0002755608706051184,
      "loss": 0.1155,
      "step": 2208
    },
    {
      "epoch": 0.30960056061667834,
      "grad_norm": 0.2159223109483719,
      "learning_rate": 0.0002755465199712987,
      "loss": 0.0748,
      "step": 2209
    },
    {
      "epoch": 0.3097407147862649,
      "grad_norm": 0.27447277307510376,
      "learning_rate": 0.00027553216933747903,
      "loss": 0.0879,
      "step": 2210
    },
    {
      "epoch": 0.30988086895585143,
      "grad_norm": 0.7030569314956665,
      "learning_rate": 0.00027551781870365936,
      "loss": 0.1659,
      "step": 2211
    },
    {
      "epoch": 0.310021023125438,
      "grad_norm": 0.5806667804718018,
      "learning_rate": 0.0002755034680698397,
      "loss": 0.1167,
      "step": 2212
    },
    {
      "epoch": 0.3101611772950245,
      "grad_norm": 0.44887638092041016,
      "learning_rate": 0.0002754891174360201,
      "loss": 0.1281,
      "step": 2213
    },
    {
      "epoch": 0.3103013314646111,
      "grad_norm": 0.8567746877670288,
      "learning_rate": 0.0002754747668022004,
      "loss": 0.2058,
      "step": 2214
    },
    {
      "epoch": 0.3104414856341976,
      "grad_norm": 0.7291098237037659,
      "learning_rate": 0.00027546041616838073,
      "loss": 0.1576,
      "step": 2215
    },
    {
      "epoch": 0.31058163980378417,
      "grad_norm": 0.3068133294582367,
      "learning_rate": 0.0002754460655345611,
      "loss": 0.0861,
      "step": 2216
    },
    {
      "epoch": 0.3107217939733707,
      "grad_norm": 0.7029969692230225,
      "learning_rate": 0.00027543171490074144,
      "loss": 0.0961,
      "step": 2217
    },
    {
      "epoch": 0.31086194814295726,
      "grad_norm": 0.6289925575256348,
      "learning_rate": 0.00027541736426692177,
      "loss": 0.0877,
      "step": 2218
    },
    {
      "epoch": 0.31100210231254377,
      "grad_norm": 0.8315983414649963,
      "learning_rate": 0.0002754030136331021,
      "loss": 0.1617,
      "step": 2219
    },
    {
      "epoch": 0.31114225648213034,
      "grad_norm": 0.3475262224674225,
      "learning_rate": 0.00027538866299928243,
      "loss": 0.0862,
      "step": 2220
    },
    {
      "epoch": 0.3112824106517169,
      "grad_norm": 0.4499328136444092,
      "learning_rate": 0.00027537431236546276,
      "loss": 0.1697,
      "step": 2221
    },
    {
      "epoch": 0.3114225648213034,
      "grad_norm": 0.3628060221672058,
      "learning_rate": 0.0002753599617316431,
      "loss": 0.1003,
      "step": 2222
    },
    {
      "epoch": 0.31156271899089,
      "grad_norm": 0.4234078526496887,
      "learning_rate": 0.00027534561109782347,
      "loss": 0.0853,
      "step": 2223
    },
    {
      "epoch": 0.3117028731604765,
      "grad_norm": 0.6662800312042236,
      "learning_rate": 0.0002753312604640038,
      "loss": 0.1542,
      "step": 2224
    },
    {
      "epoch": 0.3118430273300631,
      "grad_norm": 0.5647245049476624,
      "learning_rate": 0.0002753169098301841,
      "loss": 0.0793,
      "step": 2225
    },
    {
      "epoch": 0.3119831814996496,
      "grad_norm": 1.3708993196487427,
      "learning_rate": 0.0002753025591963645,
      "loss": 0.0948,
      "step": 2226
    },
    {
      "epoch": 0.31212333566923617,
      "grad_norm": 0.4312724471092224,
      "learning_rate": 0.00027528820856254484,
      "loss": 0.1168,
      "step": 2227
    },
    {
      "epoch": 0.3122634898388227,
      "grad_norm": 0.5120587944984436,
      "learning_rate": 0.00027527385792872516,
      "loss": 0.0557,
      "step": 2228
    },
    {
      "epoch": 0.31240364400840925,
      "grad_norm": 0.3675335645675659,
      "learning_rate": 0.0002752595072949055,
      "loss": 0.0964,
      "step": 2229
    },
    {
      "epoch": 0.31254379817799577,
      "grad_norm": 0.6422421336174011,
      "learning_rate": 0.0002752451566610858,
      "loss": 0.1545,
      "step": 2230
    },
    {
      "epoch": 0.31268395234758234,
      "grad_norm": 0.33705493807792664,
      "learning_rate": 0.00027523080602726615,
      "loss": 0.0765,
      "step": 2231
    },
    {
      "epoch": 0.3128241065171689,
      "grad_norm": 0.3716358542442322,
      "learning_rate": 0.00027521645539344653,
      "loss": 0.0688,
      "step": 2232
    },
    {
      "epoch": 0.3129642606867554,
      "grad_norm": 0.8271526098251343,
      "learning_rate": 0.00027520210475962686,
      "loss": 0.1556,
      "step": 2233
    },
    {
      "epoch": 0.313104414856342,
      "grad_norm": 0.5413151383399963,
      "learning_rate": 0.0002751877541258072,
      "loss": 0.0599,
      "step": 2234
    },
    {
      "epoch": 0.3132445690259285,
      "grad_norm": 0.6147760152816772,
      "learning_rate": 0.0002751734034919876,
      "loss": 0.1059,
      "step": 2235
    },
    {
      "epoch": 0.3133847231955151,
      "grad_norm": 0.5133717060089111,
      "learning_rate": 0.0002751590528581679,
      "loss": 0.137,
      "step": 2236
    },
    {
      "epoch": 0.3135248773651016,
      "grad_norm": 0.2922382056713104,
      "learning_rate": 0.00027514470222434823,
      "loss": 0.0415,
      "step": 2237
    },
    {
      "epoch": 0.31366503153468817,
      "grad_norm": 0.6263818144798279,
      "learning_rate": 0.00027513035159052856,
      "loss": 0.0994,
      "step": 2238
    },
    {
      "epoch": 0.3138051857042747,
      "grad_norm": 0.3334684371948242,
      "learning_rate": 0.0002751160009567089,
      "loss": 0.0806,
      "step": 2239
    },
    {
      "epoch": 0.31394533987386125,
      "grad_norm": 0.4657023549079895,
      "learning_rate": 0.0002751016503228892,
      "loss": 0.2023,
      "step": 2240
    },
    {
      "epoch": 0.31408549404344777,
      "grad_norm": 0.43214061856269836,
      "learning_rate": 0.00027508729968906954,
      "loss": 0.0662,
      "step": 2241
    },
    {
      "epoch": 0.31422564821303434,
      "grad_norm": 0.6005951166152954,
      "learning_rate": 0.00027507294905524993,
      "loss": 0.2053,
      "step": 2242
    },
    {
      "epoch": 0.3143658023826209,
      "grad_norm": 0.6431288719177246,
      "learning_rate": 0.00027505859842143026,
      "loss": 0.1224,
      "step": 2243
    },
    {
      "epoch": 0.3145059565522074,
      "grad_norm": 0.7339452505111694,
      "learning_rate": 0.0002750442477876106,
      "loss": 0.2113,
      "step": 2244
    },
    {
      "epoch": 0.314646110721794,
      "grad_norm": 1.1418890953063965,
      "learning_rate": 0.00027502989715379097,
      "loss": 0.2087,
      "step": 2245
    },
    {
      "epoch": 0.3147862648913805,
      "grad_norm": 0.8680152893066406,
      "learning_rate": 0.0002750155465199713,
      "loss": 0.2276,
      "step": 2246
    },
    {
      "epoch": 0.3149264190609671,
      "grad_norm": 1.125942587852478,
      "learning_rate": 0.0002750011958861516,
      "loss": 0.1291,
      "step": 2247
    },
    {
      "epoch": 0.3150665732305536,
      "grad_norm": 0.7187932133674622,
      "learning_rate": 0.00027498684525233195,
      "loss": 0.18,
      "step": 2248
    },
    {
      "epoch": 0.31520672740014016,
      "grad_norm": 4.580008029937744,
      "learning_rate": 0.0002749724946185123,
      "loss": 0.2097,
      "step": 2249
    },
    {
      "epoch": 0.3153468815697267,
      "grad_norm": 3.5999698638916016,
      "learning_rate": 0.0002749581439846926,
      "loss": 0.1976,
      "step": 2250
    },
    {
      "epoch": 0.31548703573931325,
      "grad_norm": 0.9778472781181335,
      "learning_rate": 0.000274943793350873,
      "loss": 0.1211,
      "step": 2251
    },
    {
      "epoch": 0.31562718990889976,
      "grad_norm": 0.5670143365859985,
      "learning_rate": 0.0002749294427170533,
      "loss": 0.1103,
      "step": 2252
    },
    {
      "epoch": 0.31576734407848633,
      "grad_norm": 0.3790888488292694,
      "learning_rate": 0.00027491509208323365,
      "loss": 0.1399,
      "step": 2253
    },
    {
      "epoch": 0.3159074982480729,
      "grad_norm": 0.3025812804698944,
      "learning_rate": 0.00027490074144941403,
      "loss": 0.083,
      "step": 2254
    },
    {
      "epoch": 0.3160476524176594,
      "grad_norm": 0.3784865438938141,
      "learning_rate": 0.00027488639081559436,
      "loss": 0.0829,
      "step": 2255
    },
    {
      "epoch": 0.316187806587246,
      "grad_norm": 0.35081565380096436,
      "learning_rate": 0.0002748720401817747,
      "loss": 0.0657,
      "step": 2256
    },
    {
      "epoch": 0.3163279607568325,
      "grad_norm": 0.4120871126651764,
      "learning_rate": 0.000274857689547955,
      "loss": 0.1125,
      "step": 2257
    },
    {
      "epoch": 0.3164681149264191,
      "grad_norm": 1.0590656995773315,
      "learning_rate": 0.00027484333891413535,
      "loss": 0.0828,
      "step": 2258
    },
    {
      "epoch": 0.3166082690960056,
      "grad_norm": 0.5151995420455933,
      "learning_rate": 0.0002748289882803157,
      "loss": 0.1685,
      "step": 2259
    },
    {
      "epoch": 0.31674842326559216,
      "grad_norm": 0.416642963886261,
      "learning_rate": 0.000274814637646496,
      "loss": 0.1337,
      "step": 2260
    },
    {
      "epoch": 0.3168885774351787,
      "grad_norm": 0.2912634015083313,
      "learning_rate": 0.0002748002870126764,
      "loss": 0.052,
      "step": 2261
    },
    {
      "epoch": 0.31702873160476525,
      "grad_norm": 0.43086275458335876,
      "learning_rate": 0.0002747859363788567,
      "loss": 0.1773,
      "step": 2262
    },
    {
      "epoch": 0.31716888577435176,
      "grad_norm": 0.6742277145385742,
      "learning_rate": 0.00027477158574503704,
      "loss": 0.1893,
      "step": 2263
    },
    {
      "epoch": 0.31730903994393833,
      "grad_norm": 0.21309377253055573,
      "learning_rate": 0.0002747572351112174,
      "loss": 0.0588,
      "step": 2264
    },
    {
      "epoch": 0.3174491941135249,
      "grad_norm": 0.30748504400253296,
      "learning_rate": 0.00027474288447739776,
      "loss": 0.0832,
      "step": 2265
    },
    {
      "epoch": 0.3175893482831114,
      "grad_norm": 0.4409842789173126,
      "learning_rate": 0.0002747285338435781,
      "loss": 0.1029,
      "step": 2266
    },
    {
      "epoch": 0.317729502452698,
      "grad_norm": 0.4778917729854584,
      "learning_rate": 0.0002747141832097584,
      "loss": 0.0937,
      "step": 2267
    },
    {
      "epoch": 0.3178696566222845,
      "grad_norm": 0.31895700097084045,
      "learning_rate": 0.00027469983257593874,
      "loss": 0.1028,
      "step": 2268
    },
    {
      "epoch": 0.3180098107918711,
      "grad_norm": 0.4421759247779846,
      "learning_rate": 0.00027468548194211907,
      "loss": 0.1301,
      "step": 2269
    },
    {
      "epoch": 0.3181499649614576,
      "grad_norm": 0.5262652635574341,
      "learning_rate": 0.00027467113130829945,
      "loss": 0.0738,
      "step": 2270
    },
    {
      "epoch": 0.31829011913104416,
      "grad_norm": 1.1306207180023193,
      "learning_rate": 0.0002746567806744798,
      "loss": 0.1503,
      "step": 2271
    },
    {
      "epoch": 0.3184302733006307,
      "grad_norm": 0.4872584342956543,
      "learning_rate": 0.0002746424300406601,
      "loss": 0.1322,
      "step": 2272
    },
    {
      "epoch": 0.31857042747021724,
      "grad_norm": 0.6786035895347595,
      "learning_rate": 0.00027462807940684044,
      "loss": 0.1121,
      "step": 2273
    },
    {
      "epoch": 0.31871058163980376,
      "grad_norm": 0.7159054279327393,
      "learning_rate": 0.00027461372877302077,
      "loss": 0.1216,
      "step": 2274
    },
    {
      "epoch": 0.31885073580939033,
      "grad_norm": 0.6088541746139526,
      "learning_rate": 0.0002745993781392011,
      "loss": 0.0756,
      "step": 2275
    },
    {
      "epoch": 0.3189908899789769,
      "grad_norm": 0.41552990674972534,
      "learning_rate": 0.0002745850275053814,
      "loss": 0.1689,
      "step": 2276
    },
    {
      "epoch": 0.3191310441485634,
      "grad_norm": 0.5906903147697449,
      "learning_rate": 0.0002745706768715618,
      "loss": 0.1266,
      "step": 2277
    },
    {
      "epoch": 0.31927119831815,
      "grad_norm": 0.5163956880569458,
      "learning_rate": 0.00027455632623774213,
      "loss": 0.1437,
      "step": 2278
    },
    {
      "epoch": 0.3194113524877365,
      "grad_norm": 0.5164892673492432,
      "learning_rate": 0.00027454197560392246,
      "loss": 0.1044,
      "step": 2279
    },
    {
      "epoch": 0.31955150665732307,
      "grad_norm": 0.6598469018936157,
      "learning_rate": 0.00027452762497010285,
      "loss": 0.149,
      "step": 2280
    },
    {
      "epoch": 0.3196916608269096,
      "grad_norm": 0.4031849503517151,
      "learning_rate": 0.0002745132743362832,
      "loss": 0.0607,
      "step": 2281
    },
    {
      "epoch": 0.31983181499649616,
      "grad_norm": 0.42672184109687805,
      "learning_rate": 0.0002744989237024635,
      "loss": 0.1155,
      "step": 2282
    },
    {
      "epoch": 0.31997196916608267,
      "grad_norm": 0.3876674771308899,
      "learning_rate": 0.00027448457306864383,
      "loss": 0.1112,
      "step": 2283
    },
    {
      "epoch": 0.32011212333566924,
      "grad_norm": 0.37237003445625305,
      "learning_rate": 0.00027447022243482416,
      "loss": 0.1216,
      "step": 2284
    },
    {
      "epoch": 0.32025227750525576,
      "grad_norm": 0.2973545789718628,
      "learning_rate": 0.0002744558718010045,
      "loss": 0.1295,
      "step": 2285
    },
    {
      "epoch": 0.3203924316748423,
      "grad_norm": 0.40184876322746277,
      "learning_rate": 0.00027444152116718487,
      "loss": 0.0885,
      "step": 2286
    },
    {
      "epoch": 0.3205325858444289,
      "grad_norm": 0.35069435834884644,
      "learning_rate": 0.0002744271705333652,
      "loss": 0.0852,
      "step": 2287
    },
    {
      "epoch": 0.3206727400140154,
      "grad_norm": 0.854923665523529,
      "learning_rate": 0.00027441281989954553,
      "loss": 0.1985,
      "step": 2288
    },
    {
      "epoch": 0.320812894183602,
      "grad_norm": 0.4210989773273468,
      "learning_rate": 0.0002743984692657259,
      "loss": 0.1204,
      "step": 2289
    },
    {
      "epoch": 0.3209530483531885,
      "grad_norm": 0.5041469931602478,
      "learning_rate": 0.00027438411863190624,
      "loss": 0.0988,
      "step": 2290
    },
    {
      "epoch": 0.32109320252277507,
      "grad_norm": 0.4445111155509949,
      "learning_rate": 0.00027436976799808657,
      "loss": 0.1457,
      "step": 2291
    },
    {
      "epoch": 0.3212333566923616,
      "grad_norm": 1.157723307609558,
      "learning_rate": 0.0002743554173642669,
      "loss": 0.1566,
      "step": 2292
    },
    {
      "epoch": 0.32137351086194815,
      "grad_norm": 0.4383234977722168,
      "learning_rate": 0.0002743410667304472,
      "loss": 0.0529,
      "step": 2293
    },
    {
      "epoch": 0.32151366503153467,
      "grad_norm": 1.1336066722869873,
      "learning_rate": 0.00027432671609662755,
      "loss": 0.0918,
      "step": 2294
    },
    {
      "epoch": 0.32165381920112124,
      "grad_norm": 0.6418235301971436,
      "learning_rate": 0.0002743123654628079,
      "loss": 0.1019,
      "step": 2295
    },
    {
      "epoch": 0.32179397337070775,
      "grad_norm": 0.8508378863334656,
      "learning_rate": 0.00027429801482898827,
      "loss": 0.1527,
      "step": 2296
    },
    {
      "epoch": 0.3219341275402943,
      "grad_norm": 2.3819377422332764,
      "learning_rate": 0.0002742836641951686,
      "loss": 0.2723,
      "step": 2297
    },
    {
      "epoch": 0.3220742817098809,
      "grad_norm": 1.8200066089630127,
      "learning_rate": 0.0002742693135613489,
      "loss": 0.2375,
      "step": 2298
    },
    {
      "epoch": 0.3222144358794674,
      "grad_norm": 0.1963467001914978,
      "learning_rate": 0.0002742549629275293,
      "loss": 0.0159,
      "step": 2299
    },
    {
      "epoch": 0.322354590049054,
      "grad_norm": 1.9495701789855957,
      "learning_rate": 0.00027424061229370963,
      "loss": 0.3231,
      "step": 2300
    },
    {
      "epoch": 0.3224947442186405,
      "grad_norm": 0.4890349507331848,
      "learning_rate": 0.00027422626165988996,
      "loss": 0.2164,
      "step": 2301
    },
    {
      "epoch": 0.32263489838822706,
      "grad_norm": 0.28434503078460693,
      "learning_rate": 0.0002742119110260703,
      "loss": 0.0338,
      "step": 2302
    },
    {
      "epoch": 0.3227750525578136,
      "grad_norm": 0.46354666352272034,
      "learning_rate": 0.0002741975603922506,
      "loss": 0.1001,
      "step": 2303
    },
    {
      "epoch": 0.32291520672740015,
      "grad_norm": 0.34605783224105835,
      "learning_rate": 0.00027418320975843095,
      "loss": 0.102,
      "step": 2304
    },
    {
      "epoch": 0.32305536089698667,
      "grad_norm": 0.30339422821998596,
      "learning_rate": 0.00027416885912461133,
      "loss": 0.0734,
      "step": 2305
    },
    {
      "epoch": 0.32319551506657324,
      "grad_norm": 1.1335664987564087,
      "learning_rate": 0.00027415450849079166,
      "loss": 0.1667,
      "step": 2306
    },
    {
      "epoch": 0.32333566923615975,
      "grad_norm": 0.4269837439060211,
      "learning_rate": 0.000274140157856972,
      "loss": 0.1496,
      "step": 2307
    },
    {
      "epoch": 0.3234758234057463,
      "grad_norm": 0.3875157833099365,
      "learning_rate": 0.0002741258072231523,
      "loss": 0.1133,
      "step": 2308
    },
    {
      "epoch": 0.3236159775753329,
      "grad_norm": 0.45066824555397034,
      "learning_rate": 0.0002741114565893327,
      "loss": 0.0993,
      "step": 2309
    },
    {
      "epoch": 0.3237561317449194,
      "grad_norm": 0.450717568397522,
      "learning_rate": 0.00027409710595551303,
      "loss": 0.0484,
      "step": 2310
    },
    {
      "epoch": 0.323896285914506,
      "grad_norm": 0.46739262342453003,
      "learning_rate": 0.00027408275532169336,
      "loss": 0.1357,
      "step": 2311
    },
    {
      "epoch": 0.3240364400840925,
      "grad_norm": 0.5234241485595703,
      "learning_rate": 0.0002740684046878737,
      "loss": 0.1473,
      "step": 2312
    },
    {
      "epoch": 0.32417659425367906,
      "grad_norm": 0.478306382894516,
      "learning_rate": 0.000274054054054054,
      "loss": 0.159,
      "step": 2313
    },
    {
      "epoch": 0.3243167484232656,
      "grad_norm": 0.48222991824150085,
      "learning_rate": 0.00027403970342023434,
      "loss": 0.1308,
      "step": 2314
    },
    {
      "epoch": 0.32445690259285215,
      "grad_norm": 0.2981216013431549,
      "learning_rate": 0.0002740253527864147,
      "loss": 0.1422,
      "step": 2315
    },
    {
      "epoch": 0.32459705676243866,
      "grad_norm": 0.2919451892375946,
      "learning_rate": 0.00027401100215259505,
      "loss": 0.1156,
      "step": 2316
    },
    {
      "epoch": 0.32473721093202523,
      "grad_norm": 0.3588232100009918,
      "learning_rate": 0.0002739966515187754,
      "loss": 0.0591,
      "step": 2317
    },
    {
      "epoch": 0.32487736510161175,
      "grad_norm": 0.4017181992530823,
      "learning_rate": 0.00027398230088495577,
      "loss": 0.1042,
      "step": 2318
    },
    {
      "epoch": 0.3250175192711983,
      "grad_norm": 0.5259623527526855,
      "learning_rate": 0.0002739679502511361,
      "loss": 0.0716,
      "step": 2319
    },
    {
      "epoch": 0.3251576734407849,
      "grad_norm": 0.59149169921875,
      "learning_rate": 0.0002739535996173164,
      "loss": 0.1199,
      "step": 2320
    },
    {
      "epoch": 0.3252978276103714,
      "grad_norm": 0.8077289462089539,
      "learning_rate": 0.00027393924898349675,
      "loss": 0.0706,
      "step": 2321
    },
    {
      "epoch": 0.325437981779958,
      "grad_norm": 0.2897542715072632,
      "learning_rate": 0.0002739248983496771,
      "loss": 0.0562,
      "step": 2322
    },
    {
      "epoch": 0.3255781359495445,
      "grad_norm": 0.5956229567527771,
      "learning_rate": 0.0002739105477158574,
      "loss": 0.1163,
      "step": 2323
    },
    {
      "epoch": 0.32571829011913106,
      "grad_norm": 0.6491693258285522,
      "learning_rate": 0.00027389619708203774,
      "loss": 0.1054,
      "step": 2324
    },
    {
      "epoch": 0.3258584442887176,
      "grad_norm": 0.4410995841026306,
      "learning_rate": 0.0002738818464482181,
      "loss": 0.0892,
      "step": 2325
    },
    {
      "epoch": 0.32599859845830415,
      "grad_norm": 0.5765685439109802,
      "learning_rate": 0.00027386749581439845,
      "loss": 0.1489,
      "step": 2326
    },
    {
      "epoch": 0.32613875262789066,
      "grad_norm": 0.2881010174751282,
      "learning_rate": 0.0002738531451805788,
      "loss": 0.1369,
      "step": 2327
    },
    {
      "epoch": 0.32627890679747723,
      "grad_norm": 0.27360406517982483,
      "learning_rate": 0.00027383879454675916,
      "loss": 0.0889,
      "step": 2328
    },
    {
      "epoch": 0.32641906096706375,
      "grad_norm": 0.4190860390663147,
      "learning_rate": 0.0002738244439129395,
      "loss": 0.1312,
      "step": 2329
    },
    {
      "epoch": 0.3265592151366503,
      "grad_norm": 0.31932252645492554,
      "learning_rate": 0.0002738100932791198,
      "loss": 0.0836,
      "step": 2330
    },
    {
      "epoch": 0.3266993693062369,
      "grad_norm": 0.4889891743659973,
      "learning_rate": 0.00027379574264530014,
      "loss": 0.0817,
      "step": 2331
    },
    {
      "epoch": 0.3268395234758234,
      "grad_norm": 0.653013288974762,
      "learning_rate": 0.0002737813920114805,
      "loss": 0.0734,
      "step": 2332
    },
    {
      "epoch": 0.32697967764540997,
      "grad_norm": 0.20587152242660522,
      "learning_rate": 0.0002737670413776608,
      "loss": 0.0331,
      "step": 2333
    },
    {
      "epoch": 0.3271198318149965,
      "grad_norm": 0.41813984513282776,
      "learning_rate": 0.0002737526907438412,
      "loss": 0.0714,
      "step": 2334
    },
    {
      "epoch": 0.32725998598458306,
      "grad_norm": 0.4206618368625641,
      "learning_rate": 0.0002737383401100215,
      "loss": 0.0963,
      "step": 2335
    },
    {
      "epoch": 0.32740014015416957,
      "grad_norm": 0.5330125093460083,
      "learning_rate": 0.00027372398947620184,
      "loss": 0.1573,
      "step": 2336
    },
    {
      "epoch": 0.32754029432375614,
      "grad_norm": 0.6578435301780701,
      "learning_rate": 0.00027370963884238217,
      "loss": 0.1191,
      "step": 2337
    },
    {
      "epoch": 0.32768044849334266,
      "grad_norm": 0.7211946249008179,
      "learning_rate": 0.0002736952882085625,
      "loss": 0.1443,
      "step": 2338
    },
    {
      "epoch": 0.32782060266292923,
      "grad_norm": 0.5612217783927917,
      "learning_rate": 0.00027368093757474283,
      "loss": 0.045,
      "step": 2339
    },
    {
      "epoch": 0.32796075683251574,
      "grad_norm": 0.20903946459293365,
      "learning_rate": 0.0002736665869409232,
      "loss": 0.0442,
      "step": 2340
    },
    {
      "epoch": 0.3281009110021023,
      "grad_norm": 0.5159934163093567,
      "learning_rate": 0.00027365223630710354,
      "loss": 0.1769,
      "step": 2341
    },
    {
      "epoch": 0.3282410651716889,
      "grad_norm": 1.0761363506317139,
      "learning_rate": 0.00027363788567328387,
      "loss": 0.0786,
      "step": 2342
    },
    {
      "epoch": 0.3283812193412754,
      "grad_norm": 0.5273336172103882,
      "learning_rate": 0.0002736235350394642,
      "loss": 0.0591,
      "step": 2343
    },
    {
      "epoch": 0.32852137351086197,
      "grad_norm": 0.9999638199806213,
      "learning_rate": 0.0002736091844056446,
      "loss": 0.0878,
      "step": 2344
    },
    {
      "epoch": 0.3286615276804485,
      "grad_norm": 0.7054980397224426,
      "learning_rate": 0.0002735948337718249,
      "loss": 0.1423,
      "step": 2345
    },
    {
      "epoch": 0.32880168185003505,
      "grad_norm": 1.0636564493179321,
      "learning_rate": 0.00027358048313800524,
      "loss": 0.1585,
      "step": 2346
    },
    {
      "epoch": 0.32894183601962157,
      "grad_norm": 2.299084186553955,
      "learning_rate": 0.00027356613250418556,
      "loss": 0.2647,
      "step": 2347
    },
    {
      "epoch": 0.32908199018920814,
      "grad_norm": 1.3058886528015137,
      "learning_rate": 0.0002735517818703659,
      "loss": 0.4329,
      "step": 2348
    },
    {
      "epoch": 0.32922214435879465,
      "grad_norm": 0.9073130488395691,
      "learning_rate": 0.0002735374312365462,
      "loss": 0.111,
      "step": 2349
    },
    {
      "epoch": 0.3293622985283812,
      "grad_norm": 1.7146743535995483,
      "learning_rate": 0.0002735230806027266,
      "loss": 0.4954,
      "step": 2350
    },
    {
      "epoch": 0.32950245269796774,
      "grad_norm": 0.5724501013755798,
      "learning_rate": 0.00027350872996890693,
      "loss": 0.1379,
      "step": 2351
    },
    {
      "epoch": 0.3296426068675543,
      "grad_norm": 0.7578223943710327,
      "learning_rate": 0.00027349437933508726,
      "loss": 0.1376,
      "step": 2352
    },
    {
      "epoch": 0.3297827610371409,
      "grad_norm": 0.4114321768283844,
      "learning_rate": 0.00027348002870126764,
      "loss": 0.1206,
      "step": 2353
    },
    {
      "epoch": 0.3299229152067274,
      "grad_norm": 0.4495943486690521,
      "learning_rate": 0.00027346567806744797,
      "loss": 0.1106,
      "step": 2354
    },
    {
      "epoch": 0.33006306937631397,
      "grad_norm": 0.42052367329597473,
      "learning_rate": 0.0002734513274336283,
      "loss": 0.1106,
      "step": 2355
    },
    {
      "epoch": 0.3302032235459005,
      "grad_norm": 0.5298296809196472,
      "learning_rate": 0.00027343697679980863,
      "loss": 0.1954,
      "step": 2356
    },
    {
      "epoch": 0.33034337771548705,
      "grad_norm": 0.40673714876174927,
      "learning_rate": 0.00027342262616598896,
      "loss": 0.1229,
      "step": 2357
    },
    {
      "epoch": 0.33048353188507357,
      "grad_norm": 0.4631592035293579,
      "learning_rate": 0.0002734082755321693,
      "loss": 0.1185,
      "step": 2358
    },
    {
      "epoch": 0.33062368605466014,
      "grad_norm": 0.29532358050346375,
      "learning_rate": 0.0002733939248983496,
      "loss": 0.1155,
      "step": 2359
    },
    {
      "epoch": 0.33076384022424665,
      "grad_norm": 0.15663360059261322,
      "learning_rate": 0.00027337957426453,
      "loss": 0.041,
      "step": 2360
    },
    {
      "epoch": 0.3309039943938332,
      "grad_norm": 0.30265161395072937,
      "learning_rate": 0.0002733652236307103,
      "loss": 0.1252,
      "step": 2361
    },
    {
      "epoch": 0.33104414856341974,
      "grad_norm": 0.2407407909631729,
      "learning_rate": 0.00027335087299689066,
      "loss": 0.0421,
      "step": 2362
    },
    {
      "epoch": 0.3311843027330063,
      "grad_norm": 0.24498705565929413,
      "learning_rate": 0.00027333652236307104,
      "loss": 0.0686,
      "step": 2363
    },
    {
      "epoch": 0.3313244569025929,
      "grad_norm": 0.3211524784564972,
      "learning_rate": 0.00027332217172925137,
      "loss": 0.1107,
      "step": 2364
    },
    {
      "epoch": 0.3314646110721794,
      "grad_norm": 0.4031507074832916,
      "learning_rate": 0.0002733078210954317,
      "loss": 0.1145,
      "step": 2365
    },
    {
      "epoch": 0.33160476524176596,
      "grad_norm": 0.35542893409729004,
      "learning_rate": 0.000273293470461612,
      "loss": 0.1093,
      "step": 2366
    },
    {
      "epoch": 0.3317449194113525,
      "grad_norm": 0.8909142017364502,
      "learning_rate": 0.00027327911982779235,
      "loss": 0.1011,
      "step": 2367
    },
    {
      "epoch": 0.33188507358093905,
      "grad_norm": 0.6860010623931885,
      "learning_rate": 0.0002732647691939727,
      "loss": 0.1648,
      "step": 2368
    },
    {
      "epoch": 0.33202522775052556,
      "grad_norm": 0.6805834770202637,
      "learning_rate": 0.00027325041856015306,
      "loss": 0.0957,
      "step": 2369
    },
    {
      "epoch": 0.33216538192011213,
      "grad_norm": 0.31497296690940857,
      "learning_rate": 0.0002732360679263334,
      "loss": 0.0735,
      "step": 2370
    },
    {
      "epoch": 0.33230553608969865,
      "grad_norm": 0.2355368733406067,
      "learning_rate": 0.0002732217172925137,
      "loss": 0.0729,
      "step": 2371
    },
    {
      "epoch": 0.3324456902592852,
      "grad_norm": 0.2858748137950897,
      "learning_rate": 0.0002732073666586941,
      "loss": 0.0703,
      "step": 2372
    },
    {
      "epoch": 0.33258584442887174,
      "grad_norm": 0.9212529063224792,
      "learning_rate": 0.00027319301602487443,
      "loss": 0.1264,
      "step": 2373
    },
    {
      "epoch": 0.3327259985984583,
      "grad_norm": 0.4797137379646301,
      "learning_rate": 0.00027317866539105476,
      "loss": 0.0964,
      "step": 2374
    },
    {
      "epoch": 0.3328661527680449,
      "grad_norm": 0.5888754725456238,
      "learning_rate": 0.0002731643147572351,
      "loss": 0.1047,
      "step": 2375
    },
    {
      "epoch": 0.3330063069376314,
      "grad_norm": 1.2329363822937012,
      "learning_rate": 0.0002731499641234154,
      "loss": 0.0467,
      "step": 2376
    },
    {
      "epoch": 0.33314646110721796,
      "grad_norm": 0.4240868091583252,
      "learning_rate": 0.00027313561348959575,
      "loss": 0.1885,
      "step": 2377
    },
    {
      "epoch": 0.3332866152768045,
      "grad_norm": 0.5215107798576355,
      "learning_rate": 0.0002731212628557761,
      "loss": 0.16,
      "step": 2378
    },
    {
      "epoch": 0.33342676944639105,
      "grad_norm": 0.7044126391410828,
      "learning_rate": 0.00027310691222195646,
      "loss": 0.1282,
      "step": 2379
    },
    {
      "epoch": 0.33356692361597756,
      "grad_norm": 0.4650145173072815,
      "learning_rate": 0.0002730925615881368,
      "loss": 0.0771,
      "step": 2380
    },
    {
      "epoch": 0.33370707778556413,
      "grad_norm": 0.7449501752853394,
      "learning_rate": 0.0002730782109543171,
      "loss": 0.1764,
      "step": 2381
    },
    {
      "epoch": 0.33384723195515065,
      "grad_norm": 0.53734290599823,
      "learning_rate": 0.0002730638603204975,
      "loss": 0.1278,
      "step": 2382
    },
    {
      "epoch": 0.3339873861247372,
      "grad_norm": 0.822650134563446,
      "learning_rate": 0.0002730495096866778,
      "loss": 0.0721,
      "step": 2383
    },
    {
      "epoch": 0.33412754029432373,
      "grad_norm": 0.2684287428855896,
      "learning_rate": 0.00027303515905285815,
      "loss": 0.127,
      "step": 2384
    },
    {
      "epoch": 0.3342676944639103,
      "grad_norm": 0.2761191427707672,
      "learning_rate": 0.0002730208084190385,
      "loss": 0.0441,
      "step": 2385
    },
    {
      "epoch": 0.3344078486334969,
      "grad_norm": 0.39176127314567566,
      "learning_rate": 0.0002730064577852188,
      "loss": 0.1067,
      "step": 2386
    },
    {
      "epoch": 0.3345480028030834,
      "grad_norm": 0.21785077452659607,
      "learning_rate": 0.00027299210715139914,
      "loss": 0.0578,
      "step": 2387
    },
    {
      "epoch": 0.33468815697266996,
      "grad_norm": 0.266147643327713,
      "learning_rate": 0.0002729777565175795,
      "loss": 0.0525,
      "step": 2388
    },
    {
      "epoch": 0.3348283111422565,
      "grad_norm": 0.35395312309265137,
      "learning_rate": 0.00027296340588375985,
      "loss": 0.1046,
      "step": 2389
    },
    {
      "epoch": 0.33496846531184304,
      "grad_norm": 0.31794142723083496,
      "learning_rate": 0.0002729490552499402,
      "loss": 0.0561,
      "step": 2390
    },
    {
      "epoch": 0.33510861948142956,
      "grad_norm": 0.5520455241203308,
      "learning_rate": 0.00027293470461612056,
      "loss": 0.1263,
      "step": 2391
    },
    {
      "epoch": 0.33524877365101613,
      "grad_norm": 0.3143710494041443,
      "learning_rate": 0.0002729203539823009,
      "loss": 0.0439,
      "step": 2392
    },
    {
      "epoch": 0.33538892782060264,
      "grad_norm": 0.5282881855964661,
      "learning_rate": 0.0002729060033484812,
      "loss": 0.2055,
      "step": 2393
    },
    {
      "epoch": 0.3355290819901892,
      "grad_norm": 0.3690708577632904,
      "learning_rate": 0.00027289165271466155,
      "loss": 0.0657,
      "step": 2394
    },
    {
      "epoch": 0.33566923615977573,
      "grad_norm": 0.34283822774887085,
      "learning_rate": 0.0002728773020808419,
      "loss": 0.1089,
      "step": 2395
    },
    {
      "epoch": 0.3358093903293623,
      "grad_norm": 0.6684614419937134,
      "learning_rate": 0.0002728629514470222,
      "loss": 0.1495,
      "step": 2396
    },
    {
      "epoch": 0.33594954449894887,
      "grad_norm": 0.2404477596282959,
      "learning_rate": 0.00027284860081320253,
      "loss": 0.081,
      "step": 2397
    },
    {
      "epoch": 0.3360896986685354,
      "grad_norm": 0.7326449751853943,
      "learning_rate": 0.0002728342501793829,
      "loss": 0.1415,
      "step": 2398
    },
    {
      "epoch": 0.33622985283812196,
      "grad_norm": 6.717862606048584,
      "learning_rate": 0.00027281989954556325,
      "loss": 0.3853,
      "step": 2399
    },
    {
      "epoch": 0.33637000700770847,
      "grad_norm": 0.7335455417633057,
      "learning_rate": 0.0002728055489117436,
      "loss": 0.1097,
      "step": 2400
    },
    {
      "epoch": 0.33651016117729504,
      "grad_norm": 0.5232634544372559,
      "learning_rate": 0.0002727911982779239,
      "loss": 0.1818,
      "step": 2401
    },
    {
      "epoch": 0.33665031534688156,
      "grad_norm": 0.28279921412467957,
      "learning_rate": 0.00027277684764410423,
      "loss": 0.0704,
      "step": 2402
    },
    {
      "epoch": 0.3367904695164681,
      "grad_norm": 1.2273950576782227,
      "learning_rate": 0.00027276249701028456,
      "loss": 0.0747,
      "step": 2403
    },
    {
      "epoch": 0.33693062368605464,
      "grad_norm": 0.27787306904792786,
      "learning_rate": 0.00027274814637646494,
      "loss": 0.0369,
      "step": 2404
    },
    {
      "epoch": 0.3370707778556412,
      "grad_norm": 0.42845991253852844,
      "learning_rate": 0.00027273379574264527,
      "loss": 0.1509,
      "step": 2405
    },
    {
      "epoch": 0.3372109320252277,
      "grad_norm": 0.21993932127952576,
      "learning_rate": 0.0002727194451088256,
      "loss": 0.0511,
      "step": 2406
    },
    {
      "epoch": 0.3373510861948143,
      "grad_norm": 0.4367145001888275,
      "learning_rate": 0.000272705094475006,
      "loss": 0.1215,
      "step": 2407
    },
    {
      "epoch": 0.33749124036440087,
      "grad_norm": 1.1007795333862305,
      "learning_rate": 0.0002726907438411863,
      "loss": 0.1673,
      "step": 2408
    },
    {
      "epoch": 0.3376313945339874,
      "grad_norm": 0.5768246650695801,
      "learning_rate": 0.00027267639320736664,
      "loss": 0.1904,
      "step": 2409
    },
    {
      "epoch": 0.33777154870357395,
      "grad_norm": 0.6666380167007446,
      "learning_rate": 0.00027266204257354697,
      "loss": 0.1605,
      "step": 2410
    },
    {
      "epoch": 0.33791170287316047,
      "grad_norm": 0.2841531038284302,
      "learning_rate": 0.0002726476919397273,
      "loss": 0.0635,
      "step": 2411
    },
    {
      "epoch": 0.33805185704274704,
      "grad_norm": 0.3090771734714508,
      "learning_rate": 0.0002726333413059076,
      "loss": 0.082,
      "step": 2412
    },
    {
      "epoch": 0.33819201121233355,
      "grad_norm": 0.37035658955574036,
      "learning_rate": 0.00027261899067208795,
      "loss": 0.0988,
      "step": 2413
    },
    {
      "epoch": 0.3383321653819201,
      "grad_norm": 0.19008462131023407,
      "learning_rate": 0.00027260464003826834,
      "loss": 0.0361,
      "step": 2414
    },
    {
      "epoch": 0.33847231955150664,
      "grad_norm": 0.36050063371658325,
      "learning_rate": 0.00027259028940444867,
      "loss": 0.0656,
      "step": 2415
    },
    {
      "epoch": 0.3386124737210932,
      "grad_norm": 0.39316341280937195,
      "learning_rate": 0.000272575938770629,
      "loss": 0.1009,
      "step": 2416
    },
    {
      "epoch": 0.3387526278906797,
      "grad_norm": 0.5990994572639465,
      "learning_rate": 0.0002725615881368094,
      "loss": 0.1116,
      "step": 2417
    },
    {
      "epoch": 0.3388927820602663,
      "grad_norm": 0.3368386924266815,
      "learning_rate": 0.0002725472375029897,
      "loss": 0.1009,
      "step": 2418
    },
    {
      "epoch": 0.3390329362298528,
      "grad_norm": 0.3756292164325714,
      "learning_rate": 0.00027253288686917003,
      "loss": 0.1382,
      "step": 2419
    },
    {
      "epoch": 0.3391730903994394,
      "grad_norm": 0.5949392318725586,
      "learning_rate": 0.00027251853623535036,
      "loss": 0.1747,
      "step": 2420
    },
    {
      "epoch": 0.33931324456902595,
      "grad_norm": 0.36781832575798035,
      "learning_rate": 0.0002725041856015307,
      "loss": 0.0894,
      "step": 2421
    },
    {
      "epoch": 0.33945339873861247,
      "grad_norm": 0.7466620802879333,
      "learning_rate": 0.000272489834967711,
      "loss": 0.1488,
      "step": 2422
    },
    {
      "epoch": 0.33959355290819904,
      "grad_norm": 0.4443918466567993,
      "learning_rate": 0.0002724754843338914,
      "loss": 0.0925,
      "step": 2423
    },
    {
      "epoch": 0.33973370707778555,
      "grad_norm": 0.43173474073410034,
      "learning_rate": 0.00027246113370007173,
      "loss": 0.1676,
      "step": 2424
    },
    {
      "epoch": 0.3398738612473721,
      "grad_norm": 0.37713611125946045,
      "learning_rate": 0.00027244678306625206,
      "loss": 0.0608,
      "step": 2425
    },
    {
      "epoch": 0.34001401541695864,
      "grad_norm": 0.42651447653770447,
      "learning_rate": 0.00027243243243243244,
      "loss": 0.1129,
      "step": 2426
    },
    {
      "epoch": 0.3401541695865452,
      "grad_norm": 3.156534194946289,
      "learning_rate": 0.00027241808179861277,
      "loss": 0.1655,
      "step": 2427
    },
    {
      "epoch": 0.3402943237561317,
      "grad_norm": 0.6820510625839233,
      "learning_rate": 0.0002724037311647931,
      "loss": 0.1953,
      "step": 2428
    },
    {
      "epoch": 0.3404344779257183,
      "grad_norm": 0.8211063146591187,
      "learning_rate": 0.00027238938053097343,
      "loss": 0.0797,
      "step": 2429
    },
    {
      "epoch": 0.3405746320953048,
      "grad_norm": 0.643652081489563,
      "learning_rate": 0.00027237502989715376,
      "loss": 0.1635,
      "step": 2430
    },
    {
      "epoch": 0.3407147862648914,
      "grad_norm": 0.4518626928329468,
      "learning_rate": 0.0002723606792633341,
      "loss": 0.0927,
      "step": 2431
    },
    {
      "epoch": 0.34085494043447795,
      "grad_norm": 0.7045007944107056,
      "learning_rate": 0.0002723463286295144,
      "loss": 0.1078,
      "step": 2432
    },
    {
      "epoch": 0.34099509460406446,
      "grad_norm": 0.32810983061790466,
      "learning_rate": 0.0002723319779956948,
      "loss": 0.1296,
      "step": 2433
    },
    {
      "epoch": 0.34113524877365103,
      "grad_norm": 0.28168684244155884,
      "learning_rate": 0.0002723176273618751,
      "loss": 0.0869,
      "step": 2434
    },
    {
      "epoch": 0.34127540294323755,
      "grad_norm": 0.6390880346298218,
      "learning_rate": 0.00027230327672805545,
      "loss": 0.0685,
      "step": 2435
    },
    {
      "epoch": 0.3414155571128241,
      "grad_norm": 0.569926917552948,
      "learning_rate": 0.00027228892609423584,
      "loss": 0.0687,
      "step": 2436
    },
    {
      "epoch": 0.34155571128241063,
      "grad_norm": 1.0435515642166138,
      "learning_rate": 0.00027227457546041616,
      "loss": 0.1286,
      "step": 2437
    },
    {
      "epoch": 0.3416958654519972,
      "grad_norm": 0.20368793606758118,
      "learning_rate": 0.0002722602248265965,
      "loss": 0.0492,
      "step": 2438
    },
    {
      "epoch": 0.3418360196215837,
      "grad_norm": 0.23373551666736603,
      "learning_rate": 0.0002722458741927768,
      "loss": 0.0376,
      "step": 2439
    },
    {
      "epoch": 0.3419761737911703,
      "grad_norm": 0.47204384207725525,
      "learning_rate": 0.00027223152355895715,
      "loss": 0.0353,
      "step": 2440
    },
    {
      "epoch": 0.3421163279607568,
      "grad_norm": 0.4461885392665863,
      "learning_rate": 0.0002722171729251375,
      "loss": 0.1415,
      "step": 2441
    },
    {
      "epoch": 0.3422564821303434,
      "grad_norm": 0.9008255004882812,
      "learning_rate": 0.00027220282229131786,
      "loss": 0.148,
      "step": 2442
    },
    {
      "epoch": 0.34239663629992995,
      "grad_norm": 0.5254664421081543,
      "learning_rate": 0.0002721884716574982,
      "loss": 0.0916,
      "step": 2443
    },
    {
      "epoch": 0.34253679046951646,
      "grad_norm": 1.1218018531799316,
      "learning_rate": 0.0002721741210236785,
      "loss": 0.1065,
      "step": 2444
    },
    {
      "epoch": 0.34267694463910303,
      "grad_norm": 2.630772590637207,
      "learning_rate": 0.0002721597703898589,
      "loss": 0.1218,
      "step": 2445
    },
    {
      "epoch": 0.34281709880868955,
      "grad_norm": 0.7606246471405029,
      "learning_rate": 0.00027214541975603923,
      "loss": 0.2005,
      "step": 2446
    },
    {
      "epoch": 0.3429572529782761,
      "grad_norm": 1.449053406715393,
      "learning_rate": 0.00027213106912221956,
      "loss": 0.199,
      "step": 2447
    },
    {
      "epoch": 0.34309740714786263,
      "grad_norm": 1.8010755777359009,
      "learning_rate": 0.0002721167184883999,
      "loss": 0.2187,
      "step": 2448
    },
    {
      "epoch": 0.3432375613174492,
      "grad_norm": 0.8930305242538452,
      "learning_rate": 0.0002721023678545802,
      "loss": 0.1005,
      "step": 2449
    },
    {
      "epoch": 0.3433777154870357,
      "grad_norm": 0.7936392426490784,
      "learning_rate": 0.00027208801722076054,
      "loss": 0.1686,
      "step": 2450
    },
    {
      "epoch": 0.3435178696566223,
      "grad_norm": 0.6133139729499817,
      "learning_rate": 0.0002720736665869409,
      "loss": 0.1502,
      "step": 2451
    },
    {
      "epoch": 0.3436580238262088,
      "grad_norm": 0.7587975263595581,
      "learning_rate": 0.00027205931595312126,
      "loss": 0.1817,
      "step": 2452
    },
    {
      "epoch": 0.3437981779957954,
      "grad_norm": 0.4359076917171478,
      "learning_rate": 0.0002720449653193016,
      "loss": 0.0672,
      "step": 2453
    },
    {
      "epoch": 0.34393833216538194,
      "grad_norm": 0.46345996856689453,
      "learning_rate": 0.0002720306146854819,
      "loss": 0.0888,
      "step": 2454
    },
    {
      "epoch": 0.34407848633496846,
      "grad_norm": 0.4289470613002777,
      "learning_rate": 0.0002720162640516623,
      "loss": 0.0948,
      "step": 2455
    },
    {
      "epoch": 0.34421864050455503,
      "grad_norm": 1.0907361507415771,
      "learning_rate": 0.0002720019134178426,
      "loss": 0.1187,
      "step": 2456
    },
    {
      "epoch": 0.34435879467414154,
      "grad_norm": 0.43418994545936584,
      "learning_rate": 0.00027198756278402295,
      "loss": 0.1056,
      "step": 2457
    },
    {
      "epoch": 0.3444989488437281,
      "grad_norm": 0.33865877985954285,
      "learning_rate": 0.0002719732121502033,
      "loss": 0.0924,
      "step": 2458
    },
    {
      "epoch": 0.34463910301331463,
      "grad_norm": 0.29642388224601746,
      "learning_rate": 0.0002719588615163836,
      "loss": 0.0834,
      "step": 2459
    },
    {
      "epoch": 0.3447792571829012,
      "grad_norm": 0.28908008337020874,
      "learning_rate": 0.00027194451088256394,
      "loss": 0.1045,
      "step": 2460
    },
    {
      "epoch": 0.3449194113524877,
      "grad_norm": 0.4186856150627136,
      "learning_rate": 0.0002719301602487443,
      "loss": 0.0856,
      "step": 2461
    },
    {
      "epoch": 0.3450595655220743,
      "grad_norm": 0.4745382070541382,
      "learning_rate": 0.00027191580961492465,
      "loss": 0.1195,
      "step": 2462
    },
    {
      "epoch": 0.3451997196916608,
      "grad_norm": 0.7490723729133606,
      "learning_rate": 0.000271901458981105,
      "loss": 0.1232,
      "step": 2463
    },
    {
      "epoch": 0.34533987386124737,
      "grad_norm": 1.2677481174468994,
      "learning_rate": 0.0002718871083472853,
      "loss": 0.1933,
      "step": 2464
    },
    {
      "epoch": 0.34548002803083394,
      "grad_norm": 0.7623156905174255,
      "learning_rate": 0.00027187275771346564,
      "loss": 0.1704,
      "step": 2465
    },
    {
      "epoch": 0.34562018220042046,
      "grad_norm": 0.5481401085853577,
      "learning_rate": 0.00027185840707964596,
      "loss": 0.2198,
      "step": 2466
    },
    {
      "epoch": 0.345760336370007,
      "grad_norm": 0.6694737672805786,
      "learning_rate": 0.0002718440564458263,
      "loss": 0.1672,
      "step": 2467
    },
    {
      "epoch": 0.34590049053959354,
      "grad_norm": 0.6540376543998718,
      "learning_rate": 0.0002718297058120067,
      "loss": 0.238,
      "step": 2468
    },
    {
      "epoch": 0.3460406447091801,
      "grad_norm": 0.6377232074737549,
      "learning_rate": 0.000271815355178187,
      "loss": 0.2207,
      "step": 2469
    },
    {
      "epoch": 0.3461807988787666,
      "grad_norm": 0.4143263101577759,
      "learning_rate": 0.00027180100454436733,
      "loss": 0.0804,
      "step": 2470
    },
    {
      "epoch": 0.3463209530483532,
      "grad_norm": 0.4887005090713501,
      "learning_rate": 0.0002717866539105477,
      "loss": 0.1923,
      "step": 2471
    },
    {
      "epoch": 0.3464611072179397,
      "grad_norm": 0.4390082359313965,
      "learning_rate": 0.00027177230327672804,
      "loss": 0.0777,
      "step": 2472
    },
    {
      "epoch": 0.3466012613875263,
      "grad_norm": 0.22069427371025085,
      "learning_rate": 0.00027175795264290837,
      "loss": 0.0511,
      "step": 2473
    },
    {
      "epoch": 0.3467414155571128,
      "grad_norm": 0.417019784450531,
      "learning_rate": 0.0002717436020090887,
      "loss": 0.0865,
      "step": 2474
    },
    {
      "epoch": 0.34688156972669937,
      "grad_norm": 0.6320773959159851,
      "learning_rate": 0.00027172925137526903,
      "loss": 0.0638,
      "step": 2475
    },
    {
      "epoch": 0.34702172389628594,
      "grad_norm": 0.33546149730682373,
      "learning_rate": 0.00027171490074144936,
      "loss": 0.0822,
      "step": 2476
    },
    {
      "epoch": 0.34716187806587245,
      "grad_norm": 0.9183886051177979,
      "learning_rate": 0.00027170055010762974,
      "loss": 0.1159,
      "step": 2477
    },
    {
      "epoch": 0.347302032235459,
      "grad_norm": 0.5276036262512207,
      "learning_rate": 0.00027168619947381007,
      "loss": 0.1684,
      "step": 2478
    },
    {
      "epoch": 0.34744218640504554,
      "grad_norm": 0.33101022243499756,
      "learning_rate": 0.0002716718488399904,
      "loss": 0.0965,
      "step": 2479
    },
    {
      "epoch": 0.3475823405746321,
      "grad_norm": 0.34179821610450745,
      "learning_rate": 0.0002716574982061708,
      "loss": 0.111,
      "step": 2480
    },
    {
      "epoch": 0.3477224947442186,
      "grad_norm": 0.3510150909423828,
      "learning_rate": 0.0002716431475723511,
      "loss": 0.0954,
      "step": 2481
    },
    {
      "epoch": 0.3478626489138052,
      "grad_norm": 0.39792853593826294,
      "learning_rate": 0.00027162879693853144,
      "loss": 0.0898,
      "step": 2482
    },
    {
      "epoch": 0.3480028030833917,
      "grad_norm": 0.3537274897098541,
      "learning_rate": 0.00027161444630471177,
      "loss": 0.1344,
      "step": 2483
    },
    {
      "epoch": 0.3481429572529783,
      "grad_norm": 0.3773926794528961,
      "learning_rate": 0.0002716000956708921,
      "loss": 0.0793,
      "step": 2484
    },
    {
      "epoch": 0.3482831114225648,
      "grad_norm": 0.5031774044036865,
      "learning_rate": 0.0002715857450370724,
      "loss": 0.1004,
      "step": 2485
    },
    {
      "epoch": 0.34842326559215137,
      "grad_norm": 0.2686522901058197,
      "learning_rate": 0.00027157139440325275,
      "loss": 0.066,
      "step": 2486
    },
    {
      "epoch": 0.34856341976173794,
      "grad_norm": 0.4645906984806061,
      "learning_rate": 0.00027155704376943313,
      "loss": 0.1754,
      "step": 2487
    },
    {
      "epoch": 0.34870357393132445,
      "grad_norm": 0.34407687187194824,
      "learning_rate": 0.00027154269313561346,
      "loss": 0.0742,
      "step": 2488
    },
    {
      "epoch": 0.348843728100911,
      "grad_norm": 0.34344732761383057,
      "learning_rate": 0.0002715283425017938,
      "loss": 0.1301,
      "step": 2489
    },
    {
      "epoch": 0.34898388227049754,
      "grad_norm": 0.514249324798584,
      "learning_rate": 0.0002715139918679742,
      "loss": 0.1335,
      "step": 2490
    },
    {
      "epoch": 0.3491240364400841,
      "grad_norm": 0.40635946393013,
      "learning_rate": 0.0002714996412341545,
      "loss": 0.0584,
      "step": 2491
    },
    {
      "epoch": 0.3492641906096706,
      "grad_norm": 0.8239240646362305,
      "learning_rate": 0.00027148529060033483,
      "loss": 0.3027,
      "step": 2492
    },
    {
      "epoch": 0.3494043447792572,
      "grad_norm": 0.2898925244808197,
      "learning_rate": 0.00027147093996651516,
      "loss": 0.0396,
      "step": 2493
    },
    {
      "epoch": 0.3495444989488437,
      "grad_norm": 0.6368575692176819,
      "learning_rate": 0.0002714565893326955,
      "loss": 0.1457,
      "step": 2494
    },
    {
      "epoch": 0.3496846531184303,
      "grad_norm": 0.664352297782898,
      "learning_rate": 0.0002714422386988758,
      "loss": 0.1002,
      "step": 2495
    },
    {
      "epoch": 0.3498248072880168,
      "grad_norm": 0.7303014397621155,
      "learning_rate": 0.0002714278880650562,
      "loss": 0.1534,
      "step": 2496
    },
    {
      "epoch": 0.34996496145760336,
      "grad_norm": 0.4517455995082855,
      "learning_rate": 0.00027141353743123653,
      "loss": 0.0741,
      "step": 2497
    },
    {
      "epoch": 0.35010511562718993,
      "grad_norm": 1.1914290189743042,
      "learning_rate": 0.00027139918679741686,
      "loss": 0.1185,
      "step": 2498
    },
    {
      "epoch": 0.35024526979677645,
      "grad_norm": 2.456247568130493,
      "learning_rate": 0.0002713848361635972,
      "loss": 0.1924,
      "step": 2499
    },
    {
      "epoch": 0.350385423966363,
      "grad_norm": 2.8768482208251953,
      "learning_rate": 0.00027137048552977757,
      "loss": 0.3982,
      "step": 2500
    },
    {
      "epoch": 0.35052557813594953,
      "grad_norm": 0.44898679852485657,
      "learning_rate": 0.0002713561348959579,
      "loss": 0.0891,
      "step": 2501
    },
    {
      "epoch": 0.3506657323055361,
      "grad_norm": 0.4719737768173218,
      "learning_rate": 0.0002713417842621382,
      "loss": 0.1062,
      "step": 2502
    },
    {
      "epoch": 0.3508058864751226,
      "grad_norm": 0.5575537085533142,
      "learning_rate": 0.00027132743362831855,
      "loss": 0.1219,
      "step": 2503
    },
    {
      "epoch": 0.3509460406447092,
      "grad_norm": 0.5102971792221069,
      "learning_rate": 0.0002713130829944989,
      "loss": 0.1151,
      "step": 2504
    },
    {
      "epoch": 0.3510861948142957,
      "grad_norm": 0.8450426459312439,
      "learning_rate": 0.0002712987323606792,
      "loss": 0.1439,
      "step": 2505
    },
    {
      "epoch": 0.3512263489838823,
      "grad_norm": 0.608574390411377,
      "learning_rate": 0.0002712843817268596,
      "loss": 0.0638,
      "step": 2506
    },
    {
      "epoch": 0.3513665031534688,
      "grad_norm": 0.38713815808296204,
      "learning_rate": 0.0002712700310930399,
      "loss": 0.1224,
      "step": 2507
    },
    {
      "epoch": 0.35150665732305536,
      "grad_norm": 0.5570006370544434,
      "learning_rate": 0.00027125568045922025,
      "loss": 0.0665,
      "step": 2508
    },
    {
      "epoch": 0.35164681149264193,
      "grad_norm": 0.47272542119026184,
      "learning_rate": 0.00027124132982540063,
      "loss": 0.1593,
      "step": 2509
    },
    {
      "epoch": 0.35178696566222845,
      "grad_norm": 0.32512497901916504,
      "learning_rate": 0.00027122697919158096,
      "loss": 0.097,
      "step": 2510
    },
    {
      "epoch": 0.351927119831815,
      "grad_norm": 0.3275410830974579,
      "learning_rate": 0.0002712126285577613,
      "loss": 0.0978,
      "step": 2511
    },
    {
      "epoch": 0.35206727400140153,
      "grad_norm": 0.9663432240486145,
      "learning_rate": 0.0002711982779239416,
      "loss": 0.1423,
      "step": 2512
    },
    {
      "epoch": 0.3522074281709881,
      "grad_norm": 0.8780844807624817,
      "learning_rate": 0.00027118392729012195,
      "loss": 0.0753,
      "step": 2513
    },
    {
      "epoch": 0.3523475823405746,
      "grad_norm": 0.4171141982078552,
      "learning_rate": 0.0002711695766563023,
      "loss": 0.1024,
      "step": 2514
    },
    {
      "epoch": 0.3524877365101612,
      "grad_norm": 0.38195255398750305,
      "learning_rate": 0.00027115522602248266,
      "loss": 0.0956,
      "step": 2515
    },
    {
      "epoch": 0.3526278906797477,
      "grad_norm": 0.616711437702179,
      "learning_rate": 0.000271140875388663,
      "loss": 0.0632,
      "step": 2516
    },
    {
      "epoch": 0.3527680448493343,
      "grad_norm": 0.3601103127002716,
      "learning_rate": 0.0002711265247548433,
      "loss": 0.1041,
      "step": 2517
    },
    {
      "epoch": 0.3529081990189208,
      "grad_norm": 0.4452815055847168,
      "learning_rate": 0.00027111217412102365,
      "loss": 0.0983,
      "step": 2518
    },
    {
      "epoch": 0.35304835318850736,
      "grad_norm": 0.618134081363678,
      "learning_rate": 0.00027109782348720403,
      "loss": 0.1594,
      "step": 2519
    },
    {
      "epoch": 0.35318850735809393,
      "grad_norm": 0.31251460313796997,
      "learning_rate": 0.00027108347285338436,
      "loss": 0.0936,
      "step": 2520
    },
    {
      "epoch": 0.35332866152768044,
      "grad_norm": 0.37265312671661377,
      "learning_rate": 0.0002710691222195647,
      "loss": 0.0989,
      "step": 2521
    },
    {
      "epoch": 0.353468815697267,
      "grad_norm": 0.629951000213623,
      "learning_rate": 0.000271054771585745,
      "loss": 0.0924,
      "step": 2522
    },
    {
      "epoch": 0.35360896986685353,
      "grad_norm": 0.3488209843635559,
      "learning_rate": 0.00027104042095192534,
      "loss": 0.0513,
      "step": 2523
    },
    {
      "epoch": 0.3537491240364401,
      "grad_norm": 0.8103635907173157,
      "learning_rate": 0.00027102607031810567,
      "loss": 0.1135,
      "step": 2524
    },
    {
      "epoch": 0.3538892782060266,
      "grad_norm": 0.3749235272407532,
      "learning_rate": 0.00027101171968428605,
      "loss": 0.1301,
      "step": 2525
    },
    {
      "epoch": 0.3540294323756132,
      "grad_norm": 0.2901221215724945,
      "learning_rate": 0.0002709973690504664,
      "loss": 0.0809,
      "step": 2526
    },
    {
      "epoch": 0.3541695865451997,
      "grad_norm": 0.6542328596115112,
      "learning_rate": 0.0002709830184166467,
      "loss": 0.1007,
      "step": 2527
    },
    {
      "epoch": 0.35430974071478627,
      "grad_norm": 0.49765893816947937,
      "learning_rate": 0.00027096866778282704,
      "loss": 0.0822,
      "step": 2528
    },
    {
      "epoch": 0.3544498948843728,
      "grad_norm": 0.44325828552246094,
      "learning_rate": 0.00027095431714900737,
      "loss": 0.0677,
      "step": 2529
    },
    {
      "epoch": 0.35459004905395936,
      "grad_norm": 0.32594650983810425,
      "learning_rate": 0.0002709399665151877,
      "loss": 0.0702,
      "step": 2530
    },
    {
      "epoch": 0.3547302032235459,
      "grad_norm": 0.2752164900302887,
      "learning_rate": 0.0002709256158813681,
      "loss": 0.0291,
      "step": 2531
    },
    {
      "epoch": 0.35487035739313244,
      "grad_norm": 0.43771252036094666,
      "learning_rate": 0.0002709112652475484,
      "loss": 0.1288,
      "step": 2532
    },
    {
      "epoch": 0.355010511562719,
      "grad_norm": 0.27745795249938965,
      "learning_rate": 0.00027089691461372874,
      "loss": 0.0717,
      "step": 2533
    },
    {
      "epoch": 0.3551506657323055,
      "grad_norm": 0.3869256377220154,
      "learning_rate": 0.00027088256397990907,
      "loss": 0.1279,
      "step": 2534
    },
    {
      "epoch": 0.3552908199018921,
      "grad_norm": 0.37020087242126465,
      "learning_rate": 0.00027086821334608945,
      "loss": 0.0666,
      "step": 2535
    },
    {
      "epoch": 0.3554309740714786,
      "grad_norm": 1.041060447692871,
      "learning_rate": 0.0002708538627122698,
      "loss": 0.1882,
      "step": 2536
    },
    {
      "epoch": 0.3555711282410652,
      "grad_norm": 1.5964288711547852,
      "learning_rate": 0.0002708395120784501,
      "loss": 0.1383,
      "step": 2537
    },
    {
      "epoch": 0.3557112824106517,
      "grad_norm": 0.39310428500175476,
      "learning_rate": 0.00027082516144463043,
      "loss": 0.0718,
      "step": 2538
    },
    {
      "epoch": 0.35585143658023827,
      "grad_norm": 0.22166696190834045,
      "learning_rate": 0.00027081081081081076,
      "loss": 0.0422,
      "step": 2539
    },
    {
      "epoch": 0.3559915907498248,
      "grad_norm": 0.4267207980155945,
      "learning_rate": 0.0002707964601769911,
      "loss": 0.1146,
      "step": 2540
    },
    {
      "epoch": 0.35613174491941135,
      "grad_norm": 0.7645246982574463,
      "learning_rate": 0.0002707821095431715,
      "loss": 0.1091,
      "step": 2541
    },
    {
      "epoch": 0.3562718990889979,
      "grad_norm": 0.37269437313079834,
      "learning_rate": 0.0002707677589093518,
      "loss": 0.113,
      "step": 2542
    },
    {
      "epoch": 0.35641205325858444,
      "grad_norm": 0.5857687592506409,
      "learning_rate": 0.00027075340827553213,
      "loss": 0.0876,
      "step": 2543
    },
    {
      "epoch": 0.356552207428171,
      "grad_norm": 0.40357160568237305,
      "learning_rate": 0.0002707390576417125,
      "loss": 0.1032,
      "step": 2544
    },
    {
      "epoch": 0.3566923615977575,
      "grad_norm": 0.44467687606811523,
      "learning_rate": 0.00027072470700789284,
      "loss": 0.1375,
      "step": 2545
    },
    {
      "epoch": 0.3568325157673441,
      "grad_norm": 0.4880650043487549,
      "learning_rate": 0.00027071035637407317,
      "loss": 0.1046,
      "step": 2546
    },
    {
      "epoch": 0.3569726699369306,
      "grad_norm": 2.2533798217773438,
      "learning_rate": 0.0002706960057402535,
      "loss": 0.2249,
      "step": 2547
    },
    {
      "epoch": 0.3571128241065172,
      "grad_norm": 4.149808406829834,
      "learning_rate": 0.00027068165510643383,
      "loss": 0.1759,
      "step": 2548
    },
    {
      "epoch": 0.3572529782761037,
      "grad_norm": 0.460557222366333,
      "learning_rate": 0.00027066730447261416,
      "loss": 0.0493,
      "step": 2549
    },
    {
      "epoch": 0.35739313244569026,
      "grad_norm": 0.47430360317230225,
      "learning_rate": 0.00027065295383879454,
      "loss": 0.1308,
      "step": 2550
    },
    {
      "epoch": 0.3575332866152768,
      "grad_norm": 0.6032901406288147,
      "learning_rate": 0.00027063860320497487,
      "loss": 0.1162,
      "step": 2551
    },
    {
      "epoch": 0.35767344078486335,
      "grad_norm": 0.5174827575683594,
      "learning_rate": 0.0002706242525711552,
      "loss": 0.141,
      "step": 2552
    },
    {
      "epoch": 0.3578135949544499,
      "grad_norm": 0.4386367201805115,
      "learning_rate": 0.0002706099019373355,
      "loss": 0.1585,
      "step": 2553
    },
    {
      "epoch": 0.35795374912403644,
      "grad_norm": 0.3321300446987152,
      "learning_rate": 0.0002705955513035159,
      "loss": 0.0801,
      "step": 2554
    },
    {
      "epoch": 0.358093903293623,
      "grad_norm": 0.8595570921897888,
      "learning_rate": 0.00027058120066969624,
      "loss": 0.0832,
      "step": 2555
    },
    {
      "epoch": 0.3582340574632095,
      "grad_norm": 0.34902796149253845,
      "learning_rate": 0.00027056685003587656,
      "loss": 0.0685,
      "step": 2556
    },
    {
      "epoch": 0.3583742116327961,
      "grad_norm": 0.7868647575378418,
      "learning_rate": 0.0002705524994020569,
      "loss": 0.1004,
      "step": 2557
    },
    {
      "epoch": 0.3585143658023826,
      "grad_norm": 0.2486821860074997,
      "learning_rate": 0.0002705381487682372,
      "loss": 0.0649,
      "step": 2558
    },
    {
      "epoch": 0.3586545199719692,
      "grad_norm": 0.37497857213020325,
      "learning_rate": 0.00027052379813441755,
      "loss": 0.1346,
      "step": 2559
    },
    {
      "epoch": 0.3587946741415557,
      "grad_norm": 0.4706312119960785,
      "learning_rate": 0.00027050944750059793,
      "loss": 0.0804,
      "step": 2560
    },
    {
      "epoch": 0.35893482831114226,
      "grad_norm": 1.0178109407424927,
      "learning_rate": 0.00027049509686677826,
      "loss": 0.2915,
      "step": 2561
    },
    {
      "epoch": 0.3590749824807288,
      "grad_norm": 0.5455099940299988,
      "learning_rate": 0.0002704807462329586,
      "loss": 0.098,
      "step": 2562
    },
    {
      "epoch": 0.35921513665031535,
      "grad_norm": 0.5274816751480103,
      "learning_rate": 0.00027046639559913897,
      "loss": 0.0971,
      "step": 2563
    },
    {
      "epoch": 0.3593552908199019,
      "grad_norm": 0.6896162033081055,
      "learning_rate": 0.0002704520449653193,
      "loss": 0.2009,
      "step": 2564
    },
    {
      "epoch": 0.35949544498948843,
      "grad_norm": 0.36828720569610596,
      "learning_rate": 0.00027043769433149963,
      "loss": 0.0989,
      "step": 2565
    },
    {
      "epoch": 0.359635599159075,
      "grad_norm": 0.656211256980896,
      "learning_rate": 0.00027042334369767996,
      "loss": 0.1403,
      "step": 2566
    },
    {
      "epoch": 0.3597757533286615,
      "grad_norm": 0.29173436760902405,
      "learning_rate": 0.0002704089930638603,
      "loss": 0.089,
      "step": 2567
    },
    {
      "epoch": 0.3599159074982481,
      "grad_norm": 0.24642011523246765,
      "learning_rate": 0.0002703946424300406,
      "loss": 0.0272,
      "step": 2568
    },
    {
      "epoch": 0.3600560616678346,
      "grad_norm": 0.6337753534317017,
      "learning_rate": 0.00027038029179622094,
      "loss": 0.0785,
      "step": 2569
    },
    {
      "epoch": 0.3601962158374212,
      "grad_norm": 0.4922681450843811,
      "learning_rate": 0.0002703659411624013,
      "loss": 0.1003,
      "step": 2570
    },
    {
      "epoch": 0.3603363700070077,
      "grad_norm": 0.3286852538585663,
      "learning_rate": 0.00027035159052858166,
      "loss": 0.0703,
      "step": 2571
    },
    {
      "epoch": 0.36047652417659426,
      "grad_norm": 0.5443809032440186,
      "learning_rate": 0.000270337239894762,
      "loss": 0.1734,
      "step": 2572
    },
    {
      "epoch": 0.3606166783461808,
      "grad_norm": 0.3691226840019226,
      "learning_rate": 0.00027032288926094237,
      "loss": 0.0935,
      "step": 2573
    },
    {
      "epoch": 0.36075683251576735,
      "grad_norm": 0.6144779324531555,
      "learning_rate": 0.0002703085386271227,
      "loss": 0.0791,
      "step": 2574
    },
    {
      "epoch": 0.3608969866853539,
      "grad_norm": 0.3534158170223236,
      "learning_rate": 0.000270294187993303,
      "loss": 0.0963,
      "step": 2575
    },
    {
      "epoch": 0.36103714085494043,
      "grad_norm": 0.23296035826206207,
      "learning_rate": 0.00027027983735948335,
      "loss": 0.0369,
      "step": 2576
    },
    {
      "epoch": 0.361177295024527,
      "grad_norm": 0.3788452446460724,
      "learning_rate": 0.0002702654867256637,
      "loss": 0.0674,
      "step": 2577
    },
    {
      "epoch": 0.3613174491941135,
      "grad_norm": 0.6378711462020874,
      "learning_rate": 0.000270251136091844,
      "loss": 0.1474,
      "step": 2578
    },
    {
      "epoch": 0.3614576033637001,
      "grad_norm": 0.23953083157539368,
      "learning_rate": 0.0002702367854580244,
      "loss": 0.0452,
      "step": 2579
    },
    {
      "epoch": 0.3615977575332866,
      "grad_norm": 0.37547391653060913,
      "learning_rate": 0.0002702224348242047,
      "loss": 0.0911,
      "step": 2580
    },
    {
      "epoch": 0.36173791170287317,
      "grad_norm": 0.5833947062492371,
      "learning_rate": 0.00027020808419038505,
      "loss": 0.0793,
      "step": 2581
    },
    {
      "epoch": 0.3618780658724597,
      "grad_norm": 0.4846550226211548,
      "learning_rate": 0.00027019373355656543,
      "loss": 0.1074,
      "step": 2582
    },
    {
      "epoch": 0.36201822004204626,
      "grad_norm": 2.035660982131958,
      "learning_rate": 0.00027017938292274576,
      "loss": 0.2078,
      "step": 2583
    },
    {
      "epoch": 0.36215837421163277,
      "grad_norm": 0.30998244881629944,
      "learning_rate": 0.0002701650322889261,
      "loss": 0.0521,
      "step": 2584
    },
    {
      "epoch": 0.36229852838121934,
      "grad_norm": 0.8155659437179565,
      "learning_rate": 0.0002701506816551064,
      "loss": 0.0759,
      "step": 2585
    },
    {
      "epoch": 0.3624386825508059,
      "grad_norm": 0.32316064834594727,
      "learning_rate": 0.00027013633102128675,
      "loss": 0.1013,
      "step": 2586
    },
    {
      "epoch": 0.36257883672039243,
      "grad_norm": 0.19872768223285675,
      "learning_rate": 0.0002701219803874671,
      "loss": 0.0499,
      "step": 2587
    },
    {
      "epoch": 0.362718990889979,
      "grad_norm": 0.6700904369354248,
      "learning_rate": 0.0002701076297536474,
      "loss": 0.125,
      "step": 2588
    },
    {
      "epoch": 0.3628591450595655,
      "grad_norm": 0.41210803389549255,
      "learning_rate": 0.0002700932791198278,
      "loss": 0.1046,
      "step": 2589
    },
    {
      "epoch": 0.3629992992291521,
      "grad_norm": 0.6869544982910156,
      "learning_rate": 0.0002700789284860081,
      "loss": 0.0856,
      "step": 2590
    },
    {
      "epoch": 0.3631394533987386,
      "grad_norm": 1.2647379636764526,
      "learning_rate": 0.00027006457785218844,
      "loss": 0.2693,
      "step": 2591
    },
    {
      "epoch": 0.36327960756832517,
      "grad_norm": 0.34526896476745605,
      "learning_rate": 0.00027005022721836877,
      "loss": 0.0904,
      "step": 2592
    },
    {
      "epoch": 0.3634197617379117,
      "grad_norm": 0.43056705594062805,
      "learning_rate": 0.0002700358765845491,
      "loss": 0.1129,
      "step": 2593
    },
    {
      "epoch": 0.36355991590749825,
      "grad_norm": 1.3661410808563232,
      "learning_rate": 0.00027002152595072943,
      "loss": 0.1855,
      "step": 2594
    },
    {
      "epoch": 0.36370007007708477,
      "grad_norm": 0.49362099170684814,
      "learning_rate": 0.0002700071753169098,
      "loss": 0.091,
      "step": 2595
    },
    {
      "epoch": 0.36384022424667134,
      "grad_norm": 0.9674927592277527,
      "learning_rate": 0.00026999282468309014,
      "loss": 0.1332,
      "step": 2596
    },
    {
      "epoch": 0.3639803784162579,
      "grad_norm": 0.7007322311401367,
      "learning_rate": 0.00026997847404927047,
      "loss": 0.2655,
      "step": 2597
    },
    {
      "epoch": 0.3641205325858444,
      "grad_norm": 1.149762511253357,
      "learning_rate": 0.00026996412341545085,
      "loss": 0.067,
      "step": 2598
    },
    {
      "epoch": 0.364260686755431,
      "grad_norm": 2.4529738426208496,
      "learning_rate": 0.0002699497727816312,
      "loss": 0.1923,
      "step": 2599
    },
    {
      "epoch": 0.3644008409250175,
      "grad_norm": 2.5562126636505127,
      "learning_rate": 0.0002699354221478115,
      "loss": 0.3356,
      "step": 2600
    },
    {
      "epoch": 0.3645409950946041,
      "grad_norm": 0.4913303554058075,
      "learning_rate": 0.00026992107151399184,
      "loss": 0.106,
      "step": 2601
    },
    {
      "epoch": 0.3646811492641906,
      "grad_norm": 0.35753095149993896,
      "learning_rate": 0.00026990672088017217,
      "loss": 0.0644,
      "step": 2602
    },
    {
      "epoch": 0.36482130343377717,
      "grad_norm": 0.27814993262290955,
      "learning_rate": 0.0002698923702463525,
      "loss": 0.0594,
      "step": 2603
    },
    {
      "epoch": 0.3649614576033637,
      "grad_norm": 0.5234560966491699,
      "learning_rate": 0.0002698780196125328,
      "loss": 0.2094,
      "step": 2604
    },
    {
      "epoch": 0.36510161177295025,
      "grad_norm": 0.45245644450187683,
      "learning_rate": 0.0002698636689787132,
      "loss": 0.1318,
      "step": 2605
    },
    {
      "epoch": 0.36524176594253677,
      "grad_norm": 0.4091087281703949,
      "learning_rate": 0.00026984931834489353,
      "loss": 0.0882,
      "step": 2606
    },
    {
      "epoch": 0.36538192011212334,
      "grad_norm": 0.45572516322135925,
      "learning_rate": 0.00026983496771107386,
      "loss": 0.1447,
      "step": 2607
    },
    {
      "epoch": 0.3655220742817099,
      "grad_norm": 0.2645706236362457,
      "learning_rate": 0.00026982061707725425,
      "loss": 0.0735,
      "step": 2608
    },
    {
      "epoch": 0.3656622284512964,
      "grad_norm": 0.2757112681865692,
      "learning_rate": 0.0002698062664434346,
      "loss": 0.0578,
      "step": 2609
    },
    {
      "epoch": 0.365802382620883,
      "grad_norm": 0.4935554265975952,
      "learning_rate": 0.0002697919158096149,
      "loss": 0.0988,
      "step": 2610
    },
    {
      "epoch": 0.3659425367904695,
      "grad_norm": 0.5613978505134583,
      "learning_rate": 0.00026977756517579523,
      "loss": 0.0946,
      "step": 2611
    },
    {
      "epoch": 0.3660826909600561,
      "grad_norm": 0.7950171828269958,
      "learning_rate": 0.00026976321454197556,
      "loss": 0.1646,
      "step": 2612
    },
    {
      "epoch": 0.3662228451296426,
      "grad_norm": 0.4371897280216217,
      "learning_rate": 0.0002697488639081559,
      "loss": 0.0887,
      "step": 2613
    },
    {
      "epoch": 0.36636299929922916,
      "grad_norm": 0.36467158794403076,
      "learning_rate": 0.00026973451327433627,
      "loss": 0.0458,
      "step": 2614
    },
    {
      "epoch": 0.3665031534688157,
      "grad_norm": 0.3143620193004608,
      "learning_rate": 0.0002697201626405166,
      "loss": 0.0531,
      "step": 2615
    },
    {
      "epoch": 0.36664330763840225,
      "grad_norm": 0.379914790391922,
      "learning_rate": 0.00026970581200669693,
      "loss": 0.1284,
      "step": 2616
    },
    {
      "epoch": 0.36678346180798876,
      "grad_norm": 2.0873301029205322,
      "learning_rate": 0.0002696914613728773,
      "loss": 0.0807,
      "step": 2617
    },
    {
      "epoch": 0.36692361597757533,
      "grad_norm": 0.43054527044296265,
      "learning_rate": 0.00026967711073905764,
      "loss": 0.0463,
      "step": 2618
    },
    {
      "epoch": 0.3670637701471619,
      "grad_norm": 0.4760350286960602,
      "learning_rate": 0.00026966276010523797,
      "loss": 0.0826,
      "step": 2619
    },
    {
      "epoch": 0.3672039243167484,
      "grad_norm": 0.9995202422142029,
      "learning_rate": 0.0002696484094714183,
      "loss": 0.1378,
      "step": 2620
    },
    {
      "epoch": 0.367344078486335,
      "grad_norm": 0.7339271903038025,
      "learning_rate": 0.0002696340588375986,
      "loss": 0.1573,
      "step": 2621
    },
    {
      "epoch": 0.3674842326559215,
      "grad_norm": 0.6659233570098877,
      "learning_rate": 0.00026961970820377895,
      "loss": 0.0768,
      "step": 2622
    },
    {
      "epoch": 0.3676243868255081,
      "grad_norm": 0.38232073187828064,
      "learning_rate": 0.0002696053575699593,
      "loss": 0.0535,
      "step": 2623
    },
    {
      "epoch": 0.3677645409950946,
      "grad_norm": 0.6639416217803955,
      "learning_rate": 0.00026959100693613967,
      "loss": 0.1746,
      "step": 2624
    },
    {
      "epoch": 0.36790469516468116,
      "grad_norm": 0.614084005355835,
      "learning_rate": 0.00026957665630232,
      "loss": 0.1573,
      "step": 2625
    },
    {
      "epoch": 0.3680448493342677,
      "grad_norm": 0.4498412311077118,
      "learning_rate": 0.0002695623056685003,
      "loss": 0.1079,
      "step": 2626
    },
    {
      "epoch": 0.36818500350385425,
      "grad_norm": 0.6188480854034424,
      "learning_rate": 0.0002695479550346807,
      "loss": 0.065,
      "step": 2627
    },
    {
      "epoch": 0.36832515767344076,
      "grad_norm": 0.4791368544101715,
      "learning_rate": 0.00026953360440086103,
      "loss": 0.0572,
      "step": 2628
    },
    {
      "epoch": 0.36846531184302733,
      "grad_norm": 0.6614050269126892,
      "learning_rate": 0.00026951925376704136,
      "loss": 0.1763,
      "step": 2629
    },
    {
      "epoch": 0.3686054660126139,
      "grad_norm": 0.4944930076599121,
      "learning_rate": 0.0002695049031332217,
      "loss": 0.0964,
      "step": 2630
    },
    {
      "epoch": 0.3687456201822004,
      "grad_norm": 0.38349610567092896,
      "learning_rate": 0.000269490552499402,
      "loss": 0.0554,
      "step": 2631
    },
    {
      "epoch": 0.368885774351787,
      "grad_norm": 0.3273887038230896,
      "learning_rate": 0.00026947620186558235,
      "loss": 0.0505,
      "step": 2632
    },
    {
      "epoch": 0.3690259285213735,
      "grad_norm": 0.5440751314163208,
      "learning_rate": 0.00026946185123176273,
      "loss": 0.1368,
      "step": 2633
    },
    {
      "epoch": 0.3691660826909601,
      "grad_norm": 0.2655757963657379,
      "learning_rate": 0.00026944750059794306,
      "loss": 0.061,
      "step": 2634
    },
    {
      "epoch": 0.3693062368605466,
      "grad_norm": 0.4824255406856537,
      "learning_rate": 0.0002694331499641234,
      "loss": 0.1301,
      "step": 2635
    },
    {
      "epoch": 0.36944639103013316,
      "grad_norm": 0.4690677225589752,
      "learning_rate": 0.00026941879933030377,
      "loss": 0.1251,
      "step": 2636
    },
    {
      "epoch": 0.3695865451997197,
      "grad_norm": 0.40310895442962646,
      "learning_rate": 0.0002694044486964841,
      "loss": 0.1116,
      "step": 2637
    },
    {
      "epoch": 0.36972669936930624,
      "grad_norm": 0.3774832785129547,
      "learning_rate": 0.00026939009806266443,
      "loss": 0.0761,
      "step": 2638
    },
    {
      "epoch": 0.36986685353889276,
      "grad_norm": 0.41188836097717285,
      "learning_rate": 0.00026937574742884476,
      "loss": 0.0666,
      "step": 2639
    },
    {
      "epoch": 0.37000700770847933,
      "grad_norm": 0.6129133701324463,
      "learning_rate": 0.0002693613967950251,
      "loss": 0.1096,
      "step": 2640
    },
    {
      "epoch": 0.3701471618780659,
      "grad_norm": 0.35266202688217163,
      "learning_rate": 0.0002693470461612054,
      "loss": 0.0914,
      "step": 2641
    },
    {
      "epoch": 0.3702873160476524,
      "grad_norm": 0.6404738426208496,
      "learning_rate": 0.00026933269552738574,
      "loss": 0.104,
      "step": 2642
    },
    {
      "epoch": 0.370427470217239,
      "grad_norm": 1.3113828897476196,
      "learning_rate": 0.0002693183448935661,
      "loss": 0.1766,
      "step": 2643
    },
    {
      "epoch": 0.3705676243868255,
      "grad_norm": 1.2462184429168701,
      "learning_rate": 0.00026930399425974645,
      "loss": 0.142,
      "step": 2644
    },
    {
      "epoch": 0.37070777855641207,
      "grad_norm": 0.9791223406791687,
      "learning_rate": 0.0002692896436259268,
      "loss": 0.0822,
      "step": 2645
    },
    {
      "epoch": 0.3708479327259986,
      "grad_norm": 1.0689698457717896,
      "learning_rate": 0.00026927529299210716,
      "loss": 0.2384,
      "step": 2646
    },
    {
      "epoch": 0.37098808689558516,
      "grad_norm": 0.2582118511199951,
      "learning_rate": 0.0002692609423582875,
      "loss": 0.0345,
      "step": 2647
    },
    {
      "epoch": 0.37112824106517167,
      "grad_norm": 2.33042311668396,
      "learning_rate": 0.0002692465917244678,
      "loss": 0.4221,
      "step": 2648
    },
    {
      "epoch": 0.37126839523475824,
      "grad_norm": 9.557506561279297,
      "learning_rate": 0.00026923224109064815,
      "loss": 0.5129,
      "step": 2649
    },
    {
      "epoch": 0.37140854940434476,
      "grad_norm": 2.5115840435028076,
      "learning_rate": 0.0002692178904568285,
      "loss": 0.515,
      "step": 2650
    },
    {
      "epoch": 0.3715487035739313,
      "grad_norm": 0.2643502354621887,
      "learning_rate": 0.0002692035398230088,
      "loss": 0.1052,
      "step": 2651
    },
    {
      "epoch": 0.37168885774351784,
      "grad_norm": 0.5183128118515015,
      "learning_rate": 0.0002691891891891892,
      "loss": 0.13,
      "step": 2652
    },
    {
      "epoch": 0.3718290119131044,
      "grad_norm": 0.5094909071922302,
      "learning_rate": 0.0002691748385553695,
      "loss": 0.0855,
      "step": 2653
    },
    {
      "epoch": 0.371969166082691,
      "grad_norm": 0.37950798869132996,
      "learning_rate": 0.00026916048792154985,
      "loss": 0.0853,
      "step": 2654
    },
    {
      "epoch": 0.3721093202522775,
      "grad_norm": 0.3840545117855072,
      "learning_rate": 0.0002691461372877302,
      "loss": 0.0783,
      "step": 2655
    },
    {
      "epoch": 0.37224947442186407,
      "grad_norm": 0.41216596961021423,
      "learning_rate": 0.0002691317866539105,
      "loss": 0.1514,
      "step": 2656
    },
    {
      "epoch": 0.3723896285914506,
      "grad_norm": 0.39185765385627747,
      "learning_rate": 0.00026911743602009083,
      "loss": 0.1446,
      "step": 2657
    },
    {
      "epoch": 0.37252978276103715,
      "grad_norm": 0.4257045388221741,
      "learning_rate": 0.00026910308538627116,
      "loss": 0.0753,
      "step": 2658
    },
    {
      "epoch": 0.37266993693062367,
      "grad_norm": 0.24160675704479218,
      "learning_rate": 0.00026908873475245154,
      "loss": 0.0529,
      "step": 2659
    },
    {
      "epoch": 0.37281009110021024,
      "grad_norm": 0.38580718636512756,
      "learning_rate": 0.0002690743841186319,
      "loss": 0.1124,
      "step": 2660
    },
    {
      "epoch": 0.37295024526979675,
      "grad_norm": 0.5381124019622803,
      "learning_rate": 0.0002690600334848122,
      "loss": 0.1175,
      "step": 2661
    },
    {
      "epoch": 0.3730903994393833,
      "grad_norm": 0.6174787282943726,
      "learning_rate": 0.0002690456828509926,
      "loss": 0.1924,
      "step": 2662
    },
    {
      "epoch": 0.37323055360896984,
      "grad_norm": 0.4316098690032959,
      "learning_rate": 0.0002690313322171729,
      "loss": 0.0926,
      "step": 2663
    },
    {
      "epoch": 0.3733707077785564,
      "grad_norm": 0.23370975255966187,
      "learning_rate": 0.00026901698158335324,
      "loss": 0.063,
      "step": 2664
    },
    {
      "epoch": 0.373510861948143,
      "grad_norm": 0.6281079053878784,
      "learning_rate": 0.00026900263094953357,
      "loss": 0.1527,
      "step": 2665
    },
    {
      "epoch": 0.3736510161177295,
      "grad_norm": 0.6845026016235352,
      "learning_rate": 0.0002689882803157139,
      "loss": 0.1397,
      "step": 2666
    },
    {
      "epoch": 0.37379117028731607,
      "grad_norm": 0.43456852436065674,
      "learning_rate": 0.00026897392968189423,
      "loss": 0.1373,
      "step": 2667
    },
    {
      "epoch": 0.3739313244569026,
      "grad_norm": 0.46237489581108093,
      "learning_rate": 0.0002689595790480746,
      "loss": 0.2164,
      "step": 2668
    },
    {
      "epoch": 0.37407147862648915,
      "grad_norm": 0.2265549749135971,
      "learning_rate": 0.00026894522841425494,
      "loss": 0.0514,
      "step": 2669
    },
    {
      "epoch": 0.37421163279607567,
      "grad_norm": 0.6893908381462097,
      "learning_rate": 0.00026893087778043527,
      "loss": 0.1429,
      "step": 2670
    },
    {
      "epoch": 0.37435178696566224,
      "grad_norm": 0.46016186475753784,
      "learning_rate": 0.00026891652714661565,
      "loss": 0.1524,
      "step": 2671
    },
    {
      "epoch": 0.37449194113524875,
      "grad_norm": 0.7288147807121277,
      "learning_rate": 0.000268902176512796,
      "loss": 0.0862,
      "step": 2672
    },
    {
      "epoch": 0.3746320953048353,
      "grad_norm": 0.5569015145301819,
      "learning_rate": 0.0002688878258789763,
      "loss": 0.0597,
      "step": 2673
    },
    {
      "epoch": 0.37477224947442184,
      "grad_norm": 0.6619922518730164,
      "learning_rate": 0.00026887347524515664,
      "loss": 0.1766,
      "step": 2674
    },
    {
      "epoch": 0.3749124036440084,
      "grad_norm": 0.43870124220848083,
      "learning_rate": 0.00026885912461133696,
      "loss": 0.0611,
      "step": 2675
    },
    {
      "epoch": 0.375052557813595,
      "grad_norm": 0.5007179975509644,
      "learning_rate": 0.0002688447739775173,
      "loss": 0.0612,
      "step": 2676
    },
    {
      "epoch": 0.3751927119831815,
      "grad_norm": 0.668543815612793,
      "learning_rate": 0.0002688304233436976,
      "loss": 0.1205,
      "step": 2677
    },
    {
      "epoch": 0.37533286615276806,
      "grad_norm": 0.635049045085907,
      "learning_rate": 0.000268816072709878,
      "loss": 0.1249,
      "step": 2678
    },
    {
      "epoch": 0.3754730203223546,
      "grad_norm": 0.4230019152164459,
      "learning_rate": 0.00026880172207605833,
      "loss": 0.0506,
      "step": 2679
    },
    {
      "epoch": 0.37561317449194115,
      "grad_norm": 0.522056519985199,
      "learning_rate": 0.00026878737144223866,
      "loss": 0.0971,
      "step": 2680
    },
    {
      "epoch": 0.37575332866152766,
      "grad_norm": 0.4881804287433624,
      "learning_rate": 0.00026877302080841904,
      "loss": 0.0421,
      "step": 2681
    },
    {
      "epoch": 0.37589348283111423,
      "grad_norm": 0.4913877248764038,
      "learning_rate": 0.00026875867017459937,
      "loss": 0.0579,
      "step": 2682
    },
    {
      "epoch": 0.37603363700070075,
      "grad_norm": 0.39831966161727905,
      "learning_rate": 0.0002687443195407797,
      "loss": 0.0855,
      "step": 2683
    },
    {
      "epoch": 0.3761737911702873,
      "grad_norm": 0.7055082321166992,
      "learning_rate": 0.00026872996890696003,
      "loss": 0.1354,
      "step": 2684
    },
    {
      "epoch": 0.37631394533987383,
      "grad_norm": 0.6139459013938904,
      "learning_rate": 0.00026871561827314036,
      "loss": 0.1284,
      "step": 2685
    },
    {
      "epoch": 0.3764540995094604,
      "grad_norm": 0.4901534616947174,
      "learning_rate": 0.0002687012676393207,
      "loss": 0.0991,
      "step": 2686
    },
    {
      "epoch": 0.376594253679047,
      "grad_norm": 0.47301438450813293,
      "learning_rate": 0.00026868691700550107,
      "loss": 0.0584,
      "step": 2687
    },
    {
      "epoch": 0.3767344078486335,
      "grad_norm": 0.28055500984191895,
      "learning_rate": 0.0002686725663716814,
      "loss": 0.0614,
      "step": 2688
    },
    {
      "epoch": 0.37687456201822006,
      "grad_norm": 0.27549150586128235,
      "learning_rate": 0.0002686582157378617,
      "loss": 0.0478,
      "step": 2689
    },
    {
      "epoch": 0.3770147161878066,
      "grad_norm": 0.4826743006706238,
      "learning_rate": 0.00026864386510404206,
      "loss": 0.1763,
      "step": 2690
    },
    {
      "epoch": 0.37715487035739315,
      "grad_norm": 0.6995766162872314,
      "learning_rate": 0.00026862951447022244,
      "loss": 0.1223,
      "step": 2691
    },
    {
      "epoch": 0.37729502452697966,
      "grad_norm": 0.5631874799728394,
      "learning_rate": 0.00026861516383640277,
      "loss": 0.1884,
      "step": 2692
    },
    {
      "epoch": 0.37743517869656623,
      "grad_norm": 1.017300009727478,
      "learning_rate": 0.0002686008132025831,
      "loss": 0.3021,
      "step": 2693
    },
    {
      "epoch": 0.37757533286615275,
      "grad_norm": 0.5196384787559509,
      "learning_rate": 0.0002685864625687634,
      "loss": 0.0587,
      "step": 2694
    },
    {
      "epoch": 0.3777154870357393,
      "grad_norm": 1.3323756456375122,
      "learning_rate": 0.00026857211193494375,
      "loss": 0.0605,
      "step": 2695
    },
    {
      "epoch": 0.37785564120532583,
      "grad_norm": 5.389350891113281,
      "learning_rate": 0.0002685577613011241,
      "loss": 0.4775,
      "step": 2696
    },
    {
      "epoch": 0.3779957953749124,
      "grad_norm": 0.7972546815872192,
      "learning_rate": 0.00026854341066730446,
      "loss": 0.2121,
      "step": 2697
    },
    {
      "epoch": 0.378135949544499,
      "grad_norm": 1.7250601053237915,
      "learning_rate": 0.0002685290600334848,
      "loss": 0.4052,
      "step": 2698
    },
    {
      "epoch": 0.3782761037140855,
      "grad_norm": 1.4964070320129395,
      "learning_rate": 0.0002685147093996651,
      "loss": 0.1881,
      "step": 2699
    },
    {
      "epoch": 0.37841625788367206,
      "grad_norm": 0.9878752827644348,
      "learning_rate": 0.0002685003587658455,
      "loss": 0.1696,
      "step": 2700
    },
    {
      "epoch": 0.3785564120532586,
      "grad_norm": 0.4222157597541809,
      "learning_rate": 0.00026848600813202583,
      "loss": 0.1127,
      "step": 2701
    },
    {
      "epoch": 0.37869656622284514,
      "grad_norm": 0.7448936104774475,
      "learning_rate": 0.00026847165749820616,
      "loss": 0.1394,
      "step": 2702
    },
    {
      "epoch": 0.37883672039243166,
      "grad_norm": 0.3859920799732208,
      "learning_rate": 0.0002684573068643865,
      "loss": 0.0808,
      "step": 2703
    },
    {
      "epoch": 0.37897687456201823,
      "grad_norm": 0.29533955454826355,
      "learning_rate": 0.0002684429562305668,
      "loss": 0.0698,
      "step": 2704
    },
    {
      "epoch": 0.37911702873160474,
      "grad_norm": 0.5440114736557007,
      "learning_rate": 0.00026842860559674715,
      "loss": 0.1567,
      "step": 2705
    },
    {
      "epoch": 0.3792571829011913,
      "grad_norm": 0.2795041799545288,
      "learning_rate": 0.00026841425496292753,
      "loss": 0.0693,
      "step": 2706
    },
    {
      "epoch": 0.37939733707077783,
      "grad_norm": 0.21150372922420502,
      "learning_rate": 0.00026839990432910786,
      "loss": 0.0974,
      "step": 2707
    },
    {
      "epoch": 0.3795374912403644,
      "grad_norm": 0.4844374358654022,
      "learning_rate": 0.0002683855536952882,
      "loss": 0.1669,
      "step": 2708
    },
    {
      "epoch": 0.37967764540995097,
      "grad_norm": 0.3226596713066101,
      "learning_rate": 0.0002683712030614685,
      "loss": 0.0809,
      "step": 2709
    },
    {
      "epoch": 0.3798177995795375,
      "grad_norm": 0.6389403939247131,
      "learning_rate": 0.0002683568524276489,
      "loss": 0.0491,
      "step": 2710
    },
    {
      "epoch": 0.37995795374912406,
      "grad_norm": 0.33429408073425293,
      "learning_rate": 0.0002683425017938292,
      "loss": 0.1099,
      "step": 2711
    },
    {
      "epoch": 0.38009810791871057,
      "grad_norm": 0.7514497637748718,
      "learning_rate": 0.00026832815116000955,
      "loss": 0.1274,
      "step": 2712
    },
    {
      "epoch": 0.38023826208829714,
      "grad_norm": 0.2460016906261444,
      "learning_rate": 0.0002683138005261899,
      "loss": 0.0586,
      "step": 2713
    },
    {
      "epoch": 0.38037841625788366,
      "grad_norm": 0.424694687128067,
      "learning_rate": 0.0002682994498923702,
      "loss": 0.0435,
      "step": 2714
    },
    {
      "epoch": 0.3805185704274702,
      "grad_norm": 0.24495062232017517,
      "learning_rate": 0.00026828509925855054,
      "loss": 0.0467,
      "step": 2715
    },
    {
      "epoch": 0.38065872459705674,
      "grad_norm": 0.631500244140625,
      "learning_rate": 0.0002682707486247309,
      "loss": 0.1284,
      "step": 2716
    },
    {
      "epoch": 0.3807988787666433,
      "grad_norm": 0.5932847261428833,
      "learning_rate": 0.00026825639799091125,
      "loss": 0.125,
      "step": 2717
    },
    {
      "epoch": 0.3809390329362298,
      "grad_norm": 0.7991192936897278,
      "learning_rate": 0.0002682420473570916,
      "loss": 0.0797,
      "step": 2718
    },
    {
      "epoch": 0.3810791871058164,
      "grad_norm": 0.8800172209739685,
      "learning_rate": 0.0002682276967232719,
      "loss": 0.1538,
      "step": 2719
    },
    {
      "epoch": 0.38121934127540297,
      "grad_norm": 0.6050055623054504,
      "learning_rate": 0.00026821334608945224,
      "loss": 0.1026,
      "step": 2720
    },
    {
      "epoch": 0.3813594954449895,
      "grad_norm": 0.3597300350666046,
      "learning_rate": 0.00026819899545563257,
      "loss": 0.0788,
      "step": 2721
    },
    {
      "epoch": 0.38149964961457605,
      "grad_norm": 0.45745959877967834,
      "learning_rate": 0.00026818464482181295,
      "loss": 0.1535,
      "step": 2722
    },
    {
      "epoch": 0.38163980378416257,
      "grad_norm": 0.38664647936820984,
      "learning_rate": 0.0002681702941879933,
      "loss": 0.112,
      "step": 2723
    },
    {
      "epoch": 0.38177995795374914,
      "grad_norm": 0.2691517174243927,
      "learning_rate": 0.0002681559435541736,
      "loss": 0.0814,
      "step": 2724
    },
    {
      "epoch": 0.38192011212333565,
      "grad_norm": 1.1752463579177856,
      "learning_rate": 0.00026814159292035393,
      "loss": 0.2208,
      "step": 2725
    },
    {
      "epoch": 0.3820602662929222,
      "grad_norm": 0.2488446980714798,
      "learning_rate": 0.0002681272422865343,
      "loss": 0.0715,
      "step": 2726
    },
    {
      "epoch": 0.38220042046250874,
      "grad_norm": 0.3645718991756439,
      "learning_rate": 0.00026811289165271465,
      "loss": 0.0512,
      "step": 2727
    },
    {
      "epoch": 0.3823405746320953,
      "grad_norm": 0.581358015537262,
      "learning_rate": 0.000268098541018895,
      "loss": 0.1292,
      "step": 2728
    },
    {
      "epoch": 0.3824807288016818,
      "grad_norm": 0.26961782574653625,
      "learning_rate": 0.0002680841903850753,
      "loss": 0.0528,
      "step": 2729
    },
    {
      "epoch": 0.3826208829712684,
      "grad_norm": 0.6025241017341614,
      "learning_rate": 0.00026806983975125563,
      "loss": 0.1213,
      "step": 2730
    },
    {
      "epoch": 0.38276103714085496,
      "grad_norm": 0.39697104692459106,
      "learning_rate": 0.00026805548911743596,
      "loss": 0.1117,
      "step": 2731
    },
    {
      "epoch": 0.3829011913104415,
      "grad_norm": 0.5270704627037048,
      "learning_rate": 0.00026804113848361634,
      "loss": 0.0867,
      "step": 2732
    },
    {
      "epoch": 0.38304134548002805,
      "grad_norm": 2.1186790466308594,
      "learning_rate": 0.00026802678784979667,
      "loss": 0.2197,
      "step": 2733
    },
    {
      "epoch": 0.38318149964961457,
      "grad_norm": 0.2457168698310852,
      "learning_rate": 0.000268012437215977,
      "loss": 0.0489,
      "step": 2734
    },
    {
      "epoch": 0.38332165381920114,
      "grad_norm": 0.6924535036087036,
      "learning_rate": 0.0002679980865821574,
      "loss": 0.0848,
      "step": 2735
    },
    {
      "epoch": 0.38346180798878765,
      "grad_norm": 0.2335180789232254,
      "learning_rate": 0.0002679837359483377,
      "loss": 0.0556,
      "step": 2736
    },
    {
      "epoch": 0.3836019621583742,
      "grad_norm": 0.23670127987861633,
      "learning_rate": 0.00026796938531451804,
      "loss": 0.0452,
      "step": 2737
    },
    {
      "epoch": 0.38374211632796074,
      "grad_norm": 0.27039679884910583,
      "learning_rate": 0.00026795503468069837,
      "loss": 0.1049,
      "step": 2738
    },
    {
      "epoch": 0.3838822704975473,
      "grad_norm": 0.4019028842449188,
      "learning_rate": 0.0002679406840468787,
      "loss": 0.0846,
      "step": 2739
    },
    {
      "epoch": 0.3840224246671338,
      "grad_norm": 1.0159893035888672,
      "learning_rate": 0.000267926333413059,
      "loss": 0.1307,
      "step": 2740
    },
    {
      "epoch": 0.3841625788367204,
      "grad_norm": 0.2937857508659363,
      "learning_rate": 0.0002679119827792394,
      "loss": 0.0524,
      "step": 2741
    },
    {
      "epoch": 0.38430273300630696,
      "grad_norm": 1.8584258556365967,
      "learning_rate": 0.00026789763214541974,
      "loss": 0.3712,
      "step": 2742
    },
    {
      "epoch": 0.3844428871758935,
      "grad_norm": 0.5510191917419434,
      "learning_rate": 0.00026788328151160007,
      "loss": 0.0927,
      "step": 2743
    },
    {
      "epoch": 0.38458304134548005,
      "grad_norm": 1.2359387874603271,
      "learning_rate": 0.0002678689308777804,
      "loss": 0.2157,
      "step": 2744
    },
    {
      "epoch": 0.38472319551506656,
      "grad_norm": 0.6061068177223206,
      "learning_rate": 0.0002678545802439608,
      "loss": 0.1064,
      "step": 2745
    },
    {
      "epoch": 0.38486334968465313,
      "grad_norm": 0.4549236297607422,
      "learning_rate": 0.0002678402296101411,
      "loss": 0.0746,
      "step": 2746
    },
    {
      "epoch": 0.38500350385423965,
      "grad_norm": 0.34904977679252625,
      "learning_rate": 0.00026782587897632143,
      "loss": 0.0519,
      "step": 2747
    },
    {
      "epoch": 0.3851436580238262,
      "grad_norm": 0.4819194972515106,
      "learning_rate": 0.00026781152834250176,
      "loss": 0.2932,
      "step": 2748
    },
    {
      "epoch": 0.38528381219341273,
      "grad_norm": 2.390181541442871,
      "learning_rate": 0.0002677971777086821,
      "loss": 0.1389,
      "step": 2749
    },
    {
      "epoch": 0.3854239663629993,
      "grad_norm": 2.2200119495391846,
      "learning_rate": 0.0002677828270748624,
      "loss": 0.5359,
      "step": 2750
    },
    {
      "epoch": 0.3855641205325858,
      "grad_norm": 0.33571431040763855,
      "learning_rate": 0.0002677684764410428,
      "loss": 0.0552,
      "step": 2751
    },
    {
      "epoch": 0.3857042747021724,
      "grad_norm": 0.41551390290260315,
      "learning_rate": 0.00026775412580722313,
      "loss": 0.1274,
      "step": 2752
    },
    {
      "epoch": 0.38584442887175896,
      "grad_norm": 0.4421321451663971,
      "learning_rate": 0.00026773977517340346,
      "loss": 0.1504,
      "step": 2753
    },
    {
      "epoch": 0.3859845830413455,
      "grad_norm": 0.39728522300720215,
      "learning_rate": 0.00026772542453958384,
      "loss": 0.144,
      "step": 2754
    },
    {
      "epoch": 0.38612473721093205,
      "grad_norm": 0.4899279773235321,
      "learning_rate": 0.00026771107390576417,
      "loss": 0.1237,
      "step": 2755
    },
    {
      "epoch": 0.38626489138051856,
      "grad_norm": 0.6530525088310242,
      "learning_rate": 0.0002676967232719445,
      "loss": 0.124,
      "step": 2756
    },
    {
      "epoch": 0.38640504555010513,
      "grad_norm": 0.30461323261260986,
      "learning_rate": 0.00026768237263812483,
      "loss": 0.0963,
      "step": 2757
    },
    {
      "epoch": 0.38654519971969165,
      "grad_norm": 0.4590955376625061,
      "learning_rate": 0.00026766802200430516,
      "loss": 0.1259,
      "step": 2758
    },
    {
      "epoch": 0.3866853538892782,
      "grad_norm": 0.30805009603500366,
      "learning_rate": 0.0002676536713704855,
      "loss": 0.0824,
      "step": 2759
    },
    {
      "epoch": 0.38682550805886473,
      "grad_norm": 0.3916451930999756,
      "learning_rate": 0.0002676393207366658,
      "loss": 0.113,
      "step": 2760
    },
    {
      "epoch": 0.3869656622284513,
      "grad_norm": 0.33900728821754456,
      "learning_rate": 0.0002676249701028462,
      "loss": 0.0649,
      "step": 2761
    },
    {
      "epoch": 0.3871058163980378,
      "grad_norm": 0.37784919142723083,
      "learning_rate": 0.0002676106194690265,
      "loss": 0.053,
      "step": 2762
    },
    {
      "epoch": 0.3872459705676244,
      "grad_norm": 0.47026288509368896,
      "learning_rate": 0.00026759626883520685,
      "loss": 0.1056,
      "step": 2763
    },
    {
      "epoch": 0.38738612473721096,
      "grad_norm": 0.3922427296638489,
      "learning_rate": 0.00026758191820138724,
      "loss": 0.0546,
      "step": 2764
    },
    {
      "epoch": 0.3875262789067975,
      "grad_norm": 0.7709530591964722,
      "learning_rate": 0.00026756756756756756,
      "loss": 0.143,
      "step": 2765
    },
    {
      "epoch": 0.38766643307638404,
      "grad_norm": 0.4271753132343292,
      "learning_rate": 0.0002675532169337479,
      "loss": 0.0875,
      "step": 2766
    },
    {
      "epoch": 0.38780658724597056,
      "grad_norm": 0.262979656457901,
      "learning_rate": 0.0002675388662999282,
      "loss": 0.069,
      "step": 2767
    },
    {
      "epoch": 0.38794674141555713,
      "grad_norm": 0.5255424976348877,
      "learning_rate": 0.00026752451566610855,
      "loss": 0.0924,
      "step": 2768
    },
    {
      "epoch": 0.38808689558514364,
      "grad_norm": 0.30937355756759644,
      "learning_rate": 0.0002675101650322889,
      "loss": 0.1292,
      "step": 2769
    },
    {
      "epoch": 0.3882270497547302,
      "grad_norm": 0.37371179461479187,
      "learning_rate": 0.00026749581439846926,
      "loss": 0.0752,
      "step": 2770
    },
    {
      "epoch": 0.38836720392431673,
      "grad_norm": 0.5643699765205383,
      "learning_rate": 0.0002674814637646496,
      "loss": 0.2093,
      "step": 2771
    },
    {
      "epoch": 0.3885073580939033,
      "grad_norm": 0.38963592052459717,
      "learning_rate": 0.0002674671131308299,
      "loss": 0.1255,
      "step": 2772
    },
    {
      "epoch": 0.3886475122634898,
      "grad_norm": 0.38058674335479736,
      "learning_rate": 0.0002674527624970103,
      "loss": 0.1023,
      "step": 2773
    },
    {
      "epoch": 0.3887876664330764,
      "grad_norm": 0.8996592164039612,
      "learning_rate": 0.00026743841186319063,
      "loss": 0.0876,
      "step": 2774
    },
    {
      "epoch": 0.38892782060266295,
      "grad_norm": 0.6100387573242188,
      "learning_rate": 0.00026742406122937096,
      "loss": 0.1035,
      "step": 2775
    },
    {
      "epoch": 0.38906797477224947,
      "grad_norm": 0.662075400352478,
      "learning_rate": 0.0002674097105955513,
      "loss": 0.1833,
      "step": 2776
    },
    {
      "epoch": 0.38920812894183604,
      "grad_norm": 0.5425719022750854,
      "learning_rate": 0.0002673953599617316,
      "loss": 0.1492,
      "step": 2777
    },
    {
      "epoch": 0.38934828311142256,
      "grad_norm": 0.5210924744606018,
      "learning_rate": 0.00026738100932791194,
      "loss": 0.1409,
      "step": 2778
    },
    {
      "epoch": 0.3894884372810091,
      "grad_norm": 0.6877641677856445,
      "learning_rate": 0.0002673666586940923,
      "loss": 0.1507,
      "step": 2779
    },
    {
      "epoch": 0.38962859145059564,
      "grad_norm": 0.5524642467498779,
      "learning_rate": 0.00026735230806027266,
      "loss": 0.0747,
      "step": 2780
    },
    {
      "epoch": 0.3897687456201822,
      "grad_norm": 0.33044764399528503,
      "learning_rate": 0.000267337957426453,
      "loss": 0.0613,
      "step": 2781
    },
    {
      "epoch": 0.3899088997897687,
      "grad_norm": 0.7215332388877869,
      "learning_rate": 0.0002673236067926333,
      "loss": 0.1626,
      "step": 2782
    },
    {
      "epoch": 0.3900490539593553,
      "grad_norm": 0.5219545960426331,
      "learning_rate": 0.00026730925615881364,
      "loss": 0.0705,
      "step": 2783
    },
    {
      "epoch": 0.3901892081289418,
      "grad_norm": 0.5790599584579468,
      "learning_rate": 0.00026729490552499397,
      "loss": 0.0681,
      "step": 2784
    },
    {
      "epoch": 0.3903293622985284,
      "grad_norm": 0.4165177047252655,
      "learning_rate": 0.0002672805548911743,
      "loss": 0.0988,
      "step": 2785
    },
    {
      "epoch": 0.39046951646811495,
      "grad_norm": 0.4054563045501709,
      "learning_rate": 0.0002672662042573547,
      "loss": 0.1123,
      "step": 2786
    },
    {
      "epoch": 0.39060967063770147,
      "grad_norm": 0.2322995364665985,
      "learning_rate": 0.000267251853623535,
      "loss": 0.0698,
      "step": 2787
    },
    {
      "epoch": 0.39074982480728804,
      "grad_norm": 0.3329775035381317,
      "learning_rate": 0.00026723750298971534,
      "loss": 0.0646,
      "step": 2788
    },
    {
      "epoch": 0.39088997897687455,
      "grad_norm": 0.3626221716403961,
      "learning_rate": 0.0002672231523558957,
      "loss": 0.1145,
      "step": 2789
    },
    {
      "epoch": 0.3910301331464611,
      "grad_norm": 0.36956289410591125,
      "learning_rate": 0.00026720880172207605,
      "loss": 0.079,
      "step": 2790
    },
    {
      "epoch": 0.39117028731604764,
      "grad_norm": 1.1502017974853516,
      "learning_rate": 0.0002671944510882564,
      "loss": 0.1271,
      "step": 2791
    },
    {
      "epoch": 0.3913104414856342,
      "grad_norm": 0.8659423589706421,
      "learning_rate": 0.0002671801004544367,
      "loss": 0.1307,
      "step": 2792
    },
    {
      "epoch": 0.3914505956552207,
      "grad_norm": 0.8645521998405457,
      "learning_rate": 0.00026716574982061704,
      "loss": 0.0852,
      "step": 2793
    },
    {
      "epoch": 0.3915907498248073,
      "grad_norm": 0.6566555500030518,
      "learning_rate": 0.00026715139918679736,
      "loss": 0.1364,
      "step": 2794
    },
    {
      "epoch": 0.3917309039943938,
      "grad_norm": 0.8128021955490112,
      "learning_rate": 0.0002671370485529777,
      "loss": 0.1103,
      "step": 2795
    },
    {
      "epoch": 0.3918710581639804,
      "grad_norm": 1.0918937921524048,
      "learning_rate": 0.0002671226979191581,
      "loss": 0.2245,
      "step": 2796
    },
    {
      "epoch": 0.39201121233356695,
      "grad_norm": 0.741801381111145,
      "learning_rate": 0.0002671083472853384,
      "loss": 0.2202,
      "step": 2797
    },
    {
      "epoch": 0.39215136650315346,
      "grad_norm": 0.4222448170185089,
      "learning_rate": 0.00026709399665151873,
      "loss": 0.0374,
      "step": 2798
    },
    {
      "epoch": 0.39229152067274004,
      "grad_norm": 0.5807374119758606,
      "learning_rate": 0.0002670796460176991,
      "loss": 0.1535,
      "step": 2799
    },
    {
      "epoch": 0.39243167484232655,
      "grad_norm": 1.4028120040893555,
      "learning_rate": 0.00026706529538387944,
      "loss": 0.1488,
      "step": 2800
    },
    {
      "epoch": 0.3925718290119131,
      "grad_norm": 0.41136837005615234,
      "learning_rate": 0.00026705094475005977,
      "loss": 0.1106,
      "step": 2801
    },
    {
      "epoch": 0.39271198318149964,
      "grad_norm": 0.35249119997024536,
      "learning_rate": 0.0002670365941162401,
      "loss": 0.0743,
      "step": 2802
    },
    {
      "epoch": 0.3928521373510862,
      "grad_norm": 0.5796114206314087,
      "learning_rate": 0.00026702224348242043,
      "loss": 0.15,
      "step": 2803
    },
    {
      "epoch": 0.3929922915206727,
      "grad_norm": 0.28602728247642517,
      "learning_rate": 0.00026700789284860076,
      "loss": 0.0789,
      "step": 2804
    },
    {
      "epoch": 0.3931324456902593,
      "grad_norm": 0.39246585965156555,
      "learning_rate": 0.00026699354221478114,
      "loss": 0.0915,
      "step": 2805
    },
    {
      "epoch": 0.3932725998598458,
      "grad_norm": 0.7621869444847107,
      "learning_rate": 0.00026697919158096147,
      "loss": 0.1138,
      "step": 2806
    },
    {
      "epoch": 0.3934127540294324,
      "grad_norm": 0.52781081199646,
      "learning_rate": 0.0002669648409471418,
      "loss": 0.1115,
      "step": 2807
    },
    {
      "epoch": 0.39355290819901895,
      "grad_norm": 0.5050719976425171,
      "learning_rate": 0.0002669504903133222,
      "loss": 0.1652,
      "step": 2808
    },
    {
      "epoch": 0.39369306236860546,
      "grad_norm": 0.8677018880844116,
      "learning_rate": 0.0002669361396795025,
      "loss": 0.2224,
      "step": 2809
    },
    {
      "epoch": 0.39383321653819203,
      "grad_norm": 0.6989110708236694,
      "learning_rate": 0.00026692178904568284,
      "loss": 0.1402,
      "step": 2810
    },
    {
      "epoch": 0.39397337070777855,
      "grad_norm": 0.4217401146888733,
      "learning_rate": 0.00026690743841186317,
      "loss": 0.0703,
      "step": 2811
    },
    {
      "epoch": 0.3941135248773651,
      "grad_norm": 0.6364090442657471,
      "learning_rate": 0.0002668930877780435,
      "loss": 0.1797,
      "step": 2812
    },
    {
      "epoch": 0.39425367904695163,
      "grad_norm": 0.7878068685531616,
      "learning_rate": 0.0002668787371442238,
      "loss": 0.1714,
      "step": 2813
    },
    {
      "epoch": 0.3943938332165382,
      "grad_norm": 0.5863538384437561,
      "learning_rate": 0.00026686438651040415,
      "loss": 0.1672,
      "step": 2814
    },
    {
      "epoch": 0.3945339873861247,
      "grad_norm": 0.7067665457725525,
      "learning_rate": 0.00026685003587658453,
      "loss": 0.1089,
      "step": 2815
    },
    {
      "epoch": 0.3946741415557113,
      "grad_norm": 0.903645932674408,
      "learning_rate": 0.00026683568524276486,
      "loss": 0.1005,
      "step": 2816
    },
    {
      "epoch": 0.3948142957252978,
      "grad_norm": 0.6870549917221069,
      "learning_rate": 0.0002668213346089452,
      "loss": 0.1138,
      "step": 2817
    },
    {
      "epoch": 0.3949544498948844,
      "grad_norm": 0.5287951827049255,
      "learning_rate": 0.0002668069839751256,
      "loss": 0.1159,
      "step": 2818
    },
    {
      "epoch": 0.39509460406447094,
      "grad_norm": 0.5331532955169678,
      "learning_rate": 0.0002667926333413059,
      "loss": 0.0755,
      "step": 2819
    },
    {
      "epoch": 0.39523475823405746,
      "grad_norm": 0.5666272044181824,
      "learning_rate": 0.00026677828270748623,
      "loss": 0.1031,
      "step": 2820
    },
    {
      "epoch": 0.39537491240364403,
      "grad_norm": 0.36146727204322815,
      "learning_rate": 0.00026676393207366656,
      "loss": 0.1319,
      "step": 2821
    },
    {
      "epoch": 0.39551506657323054,
      "grad_norm": 0.2870951294898987,
      "learning_rate": 0.0002667495814398469,
      "loss": 0.0896,
      "step": 2822
    },
    {
      "epoch": 0.3956552207428171,
      "grad_norm": 0.274265855550766,
      "learning_rate": 0.0002667352308060272,
      "loss": 0.0828,
      "step": 2823
    },
    {
      "epoch": 0.39579537491240363,
      "grad_norm": 0.33852601051330566,
      "learning_rate": 0.0002667208801722076,
      "loss": 0.0799,
      "step": 2824
    },
    {
      "epoch": 0.3959355290819902,
      "grad_norm": 0.6449515223503113,
      "learning_rate": 0.00026670652953838793,
      "loss": 0.1358,
      "step": 2825
    },
    {
      "epoch": 0.3960756832515767,
      "grad_norm": 0.3763137459754944,
      "learning_rate": 0.00026669217890456826,
      "loss": 0.0643,
      "step": 2826
    },
    {
      "epoch": 0.3962158374211633,
      "grad_norm": 0.41076070070266724,
      "learning_rate": 0.00026667782827074864,
      "loss": 0.063,
      "step": 2827
    },
    {
      "epoch": 0.3963559915907498,
      "grad_norm": 0.5818364024162292,
      "learning_rate": 0.00026666347763692897,
      "loss": 0.124,
      "step": 2828
    },
    {
      "epoch": 0.39649614576033637,
      "grad_norm": 0.6228339672088623,
      "learning_rate": 0.0002666491270031093,
      "loss": 0.1254,
      "step": 2829
    },
    {
      "epoch": 0.39663629992992294,
      "grad_norm": 0.47882217168807983,
      "learning_rate": 0.0002666347763692896,
      "loss": 0.0891,
      "step": 2830
    },
    {
      "epoch": 0.39677645409950946,
      "grad_norm": 0.35936740040779114,
      "learning_rate": 0.00026662042573546995,
      "loss": 0.1275,
      "step": 2831
    },
    {
      "epoch": 0.396916608269096,
      "grad_norm": 0.5200803279876709,
      "learning_rate": 0.0002666060751016503,
      "loss": 0.1202,
      "step": 2832
    },
    {
      "epoch": 0.39705676243868254,
      "grad_norm": 0.3239125907421112,
      "learning_rate": 0.0002665917244678306,
      "loss": 0.0439,
      "step": 2833
    },
    {
      "epoch": 0.3971969166082691,
      "grad_norm": 0.44537585973739624,
      "learning_rate": 0.000266577373834011,
      "loss": 0.1007,
      "step": 2834
    },
    {
      "epoch": 0.3973370707778556,
      "grad_norm": 0.3647289276123047,
      "learning_rate": 0.0002665630232001913,
      "loss": 0.0237,
      "step": 2835
    },
    {
      "epoch": 0.3974772249474422,
      "grad_norm": 0.46190890669822693,
      "learning_rate": 0.00026654867256637165,
      "loss": 0.1221,
      "step": 2836
    },
    {
      "epoch": 0.3976173791170287,
      "grad_norm": 0.4426904618740082,
      "learning_rate": 0.00026653432193255203,
      "loss": 0.14,
      "step": 2837
    },
    {
      "epoch": 0.3977575332866153,
      "grad_norm": 1.7598018646240234,
      "learning_rate": 0.00026651997129873236,
      "loss": 0.1223,
      "step": 2838
    },
    {
      "epoch": 0.3978976874562018,
      "grad_norm": 1.075234055519104,
      "learning_rate": 0.0002665056206649127,
      "loss": 0.1454,
      "step": 2839
    },
    {
      "epoch": 0.39803784162578837,
      "grad_norm": 1.100995421409607,
      "learning_rate": 0.000266491270031093,
      "loss": 0.1231,
      "step": 2840
    },
    {
      "epoch": 0.39817799579537494,
      "grad_norm": 0.5203006267547607,
      "learning_rate": 0.00026647691939727335,
      "loss": 0.1033,
      "step": 2841
    },
    {
      "epoch": 0.39831814996496145,
      "grad_norm": 0.49494871497154236,
      "learning_rate": 0.0002664625687634537,
      "loss": 0.1666,
      "step": 2842
    },
    {
      "epoch": 0.398458304134548,
      "grad_norm": 0.6560659408569336,
      "learning_rate": 0.00026644821812963406,
      "loss": 0.1705,
      "step": 2843
    },
    {
      "epoch": 0.39859845830413454,
      "grad_norm": 0.702608585357666,
      "learning_rate": 0.0002664338674958144,
      "loss": 0.1548,
      "step": 2844
    },
    {
      "epoch": 0.3987386124737211,
      "grad_norm": 1.7253023386001587,
      "learning_rate": 0.0002664195168619947,
      "loss": 0.1621,
      "step": 2845
    },
    {
      "epoch": 0.3988787666433076,
      "grad_norm": 0.5822128057479858,
      "learning_rate": 0.00026640516622817505,
      "loss": 0.1396,
      "step": 2846
    },
    {
      "epoch": 0.3990189208128942,
      "grad_norm": 1.302255392074585,
      "learning_rate": 0.0002663908155943554,
      "loss": 0.2037,
      "step": 2847
    },
    {
      "epoch": 0.3991590749824807,
      "grad_norm": 0.3844779133796692,
      "learning_rate": 0.0002663764649605357,
      "loss": 0.0452,
      "step": 2848
    },
    {
      "epoch": 0.3992992291520673,
      "grad_norm": 1.5568729639053345,
      "learning_rate": 0.0002663621143267161,
      "loss": 0.2632,
      "step": 2849
    },
    {
      "epoch": 0.3994393833216538,
      "grad_norm": 1.6648600101470947,
      "learning_rate": 0.0002663477636928964,
      "loss": 0.2718,
      "step": 2850
    },
    {
      "epoch": 0.39957953749124037,
      "grad_norm": 0.37880682945251465,
      "learning_rate": 0.00026633341305907674,
      "loss": 0.1197,
      "step": 2851
    },
    {
      "epoch": 0.39971969166082694,
      "grad_norm": 0.6926692724227905,
      "learning_rate": 0.00026631906242525707,
      "loss": 0.1162,
      "step": 2852
    },
    {
      "epoch": 0.39985984583041345,
      "grad_norm": 0.34722068905830383,
      "learning_rate": 0.00026630471179143745,
      "loss": 0.1213,
      "step": 2853
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.29237687587738037,
      "learning_rate": 0.0002662903611576178,
      "loss": 0.043,
      "step": 2854
    },
    {
      "epoch": 0.40014015416958654,
      "grad_norm": 0.3461633622646332,
      "learning_rate": 0.0002662760105237981,
      "loss": 0.1524,
      "step": 2855
    },
    {
      "epoch": 0.4002803083391731,
      "grad_norm": 0.3609064519405365,
      "learning_rate": 0.00026626165988997844,
      "loss": 0.1184,
      "step": 2856
    },
    {
      "epoch": 0.4004204625087596,
      "grad_norm": 0.3734062910079956,
      "learning_rate": 0.00026624730925615877,
      "loss": 0.1052,
      "step": 2857
    },
    {
      "epoch": 0.4005606166783462,
      "grad_norm": 0.1862742304801941,
      "learning_rate": 0.0002662329586223391,
      "loss": 0.053,
      "step": 2858
    },
    {
      "epoch": 0.4007007708479327,
      "grad_norm": 0.3383943736553192,
      "learning_rate": 0.0002662186079885195,
      "loss": 0.1108,
      "step": 2859
    },
    {
      "epoch": 0.4008409250175193,
      "grad_norm": 0.6033865213394165,
      "learning_rate": 0.0002662042573546998,
      "loss": 0.12,
      "step": 2860
    },
    {
      "epoch": 0.4009810791871058,
      "grad_norm": 0.3873053193092346,
      "learning_rate": 0.00026618990672088014,
      "loss": 0.1069,
      "step": 2861
    },
    {
      "epoch": 0.40112123335669236,
      "grad_norm": 0.491430401802063,
      "learning_rate": 0.0002661755560870605,
      "loss": 0.1415,
      "step": 2862
    },
    {
      "epoch": 0.40126138752627893,
      "grad_norm": 0.44992104172706604,
      "learning_rate": 0.00026616120545324085,
      "loss": 0.0965,
      "step": 2863
    },
    {
      "epoch": 0.40140154169586545,
      "grad_norm": 0.5101977586746216,
      "learning_rate": 0.0002661468548194212,
      "loss": 0.1512,
      "step": 2864
    },
    {
      "epoch": 0.401541695865452,
      "grad_norm": 0.2565525472164154,
      "learning_rate": 0.0002661325041856015,
      "loss": 0.0809,
      "step": 2865
    },
    {
      "epoch": 0.40168185003503853,
      "grad_norm": 0.4568197429180145,
      "learning_rate": 0.00026611815355178183,
      "loss": 0.1424,
      "step": 2866
    },
    {
      "epoch": 0.4018220042046251,
      "grad_norm": 0.3124846816062927,
      "learning_rate": 0.00026610380291796216,
      "loss": 0.0816,
      "step": 2867
    },
    {
      "epoch": 0.4019621583742116,
      "grad_norm": 0.4760637581348419,
      "learning_rate": 0.0002660894522841425,
      "loss": 0.0587,
      "step": 2868
    },
    {
      "epoch": 0.4021023125437982,
      "grad_norm": 0.40039730072021484,
      "learning_rate": 0.0002660751016503229,
      "loss": 0.1076,
      "step": 2869
    },
    {
      "epoch": 0.4022424667133847,
      "grad_norm": 0.37872010469436646,
      "learning_rate": 0.0002660607510165032,
      "loss": 0.0934,
      "step": 2870
    },
    {
      "epoch": 0.4023826208829713,
      "grad_norm": 0.657885730266571,
      "learning_rate": 0.00026604640038268353,
      "loss": 0.1092,
      "step": 2871
    },
    {
      "epoch": 0.4025227750525578,
      "grad_norm": 0.3977409601211548,
      "learning_rate": 0.0002660320497488639,
      "loss": 0.1252,
      "step": 2872
    },
    {
      "epoch": 0.40266292922214436,
      "grad_norm": 0.4142765700817108,
      "learning_rate": 0.00026601769911504424,
      "loss": 0.1236,
      "step": 2873
    },
    {
      "epoch": 0.40280308339173093,
      "grad_norm": 0.4369869828224182,
      "learning_rate": 0.00026600334848122457,
      "loss": 0.0793,
      "step": 2874
    },
    {
      "epoch": 0.40294323756131745,
      "grad_norm": 0.6777017712593079,
      "learning_rate": 0.0002659889978474049,
      "loss": 0.1849,
      "step": 2875
    },
    {
      "epoch": 0.403083391730904,
      "grad_norm": 0.5213425755500793,
      "learning_rate": 0.00026597464721358523,
      "loss": 0.0817,
      "step": 2876
    },
    {
      "epoch": 0.40322354590049053,
      "grad_norm": 0.2951483130455017,
      "learning_rate": 0.00026596029657976556,
      "loss": 0.0941,
      "step": 2877
    },
    {
      "epoch": 0.4033637000700771,
      "grad_norm": 0.41839635372161865,
      "learning_rate": 0.00026594594594594594,
      "loss": 0.1444,
      "step": 2878
    },
    {
      "epoch": 0.4035038542396636,
      "grad_norm": 0.7733237743377686,
      "learning_rate": 0.00026593159531212627,
      "loss": 0.1435,
      "step": 2879
    },
    {
      "epoch": 0.4036440084092502,
      "grad_norm": 0.43835964798927307,
      "learning_rate": 0.0002659172446783066,
      "loss": 0.0766,
      "step": 2880
    },
    {
      "epoch": 0.4037841625788367,
      "grad_norm": 0.6225267648696899,
      "learning_rate": 0.0002659028940444869,
      "loss": 0.1538,
      "step": 2881
    },
    {
      "epoch": 0.4039243167484233,
      "grad_norm": 0.5458643436431885,
      "learning_rate": 0.0002658885434106673,
      "loss": 0.1374,
      "step": 2882
    },
    {
      "epoch": 0.4040644709180098,
      "grad_norm": 0.39032381772994995,
      "learning_rate": 0.00026587419277684764,
      "loss": 0.0909,
      "step": 2883
    },
    {
      "epoch": 0.40420462508759636,
      "grad_norm": 0.14219030737876892,
      "learning_rate": 0.00026585984214302796,
      "loss": 0.045,
      "step": 2884
    },
    {
      "epoch": 0.4043447792571829,
      "grad_norm": 0.5623557567596436,
      "learning_rate": 0.0002658454915092083,
      "loss": 0.1137,
      "step": 2885
    },
    {
      "epoch": 0.40448493342676944,
      "grad_norm": 0.05237596482038498,
      "learning_rate": 0.0002658311408753886,
      "loss": 0.0092,
      "step": 2886
    },
    {
      "epoch": 0.404625087596356,
      "grad_norm": 0.5156688690185547,
      "learning_rate": 0.00026581679024156895,
      "loss": 0.1508,
      "step": 2887
    },
    {
      "epoch": 0.40476524176594253,
      "grad_norm": 0.4336191713809967,
      "learning_rate": 0.00026580243960774933,
      "loss": 0.1281,
      "step": 2888
    },
    {
      "epoch": 0.4049053959355291,
      "grad_norm": 0.5453097224235535,
      "learning_rate": 0.00026578808897392966,
      "loss": 0.0864,
      "step": 2889
    },
    {
      "epoch": 0.4050455501051156,
      "grad_norm": 0.49457165598869324,
      "learning_rate": 0.00026577373834011,
      "loss": 0.0727,
      "step": 2890
    },
    {
      "epoch": 0.4051857042747022,
      "grad_norm": 0.42322638630867004,
      "learning_rate": 0.00026575938770629037,
      "loss": 0.0763,
      "step": 2891
    },
    {
      "epoch": 0.4053258584442887,
      "grad_norm": 0.2601456344127655,
      "learning_rate": 0.0002657450370724707,
      "loss": 0.049,
      "step": 2892
    },
    {
      "epoch": 0.40546601261387527,
      "grad_norm": 0.38877516984939575,
      "learning_rate": 0.00026573068643865103,
      "loss": 0.0971,
      "step": 2893
    },
    {
      "epoch": 0.4056061667834618,
      "grad_norm": 0.5565824508666992,
      "learning_rate": 0.00026571633580483136,
      "loss": 0.072,
      "step": 2894
    },
    {
      "epoch": 0.40574632095304836,
      "grad_norm": 0.3686055541038513,
      "learning_rate": 0.0002657019851710117,
      "loss": 0.1028,
      "step": 2895
    },
    {
      "epoch": 0.40588647512263487,
      "grad_norm": 0.8749672770500183,
      "learning_rate": 0.000265687634537192,
      "loss": 0.1768,
      "step": 2896
    },
    {
      "epoch": 0.40602662929222144,
      "grad_norm": 3.2944326400756836,
      "learning_rate": 0.0002656732839033724,
      "loss": 0.1289,
      "step": 2897
    },
    {
      "epoch": 0.406166783461808,
      "grad_norm": 2.763009548187256,
      "learning_rate": 0.0002656589332695527,
      "loss": 0.0903,
      "step": 2898
    },
    {
      "epoch": 0.4063069376313945,
      "grad_norm": 0.6687091588973999,
      "learning_rate": 0.00026564458263573306,
      "loss": 0.0438,
      "step": 2899
    },
    {
      "epoch": 0.4064470918009811,
      "grad_norm": 1.7816590070724487,
      "learning_rate": 0.0002656302320019134,
      "loss": 0.2906,
      "step": 2900
    },
    {
      "epoch": 0.4065872459705676,
      "grad_norm": 0.31545910239219666,
      "learning_rate": 0.00026561588136809377,
      "loss": 0.0616,
      "step": 2901
    },
    {
      "epoch": 0.4067274001401542,
      "grad_norm": 0.8465245366096497,
      "learning_rate": 0.0002656015307342741,
      "loss": 0.0696,
      "step": 2902
    },
    {
      "epoch": 0.4068675543097407,
      "grad_norm": 0.5099138021469116,
      "learning_rate": 0.0002655871801004544,
      "loss": 0.1222,
      "step": 2903
    },
    {
      "epoch": 0.40700770847932727,
      "grad_norm": 0.30457115173339844,
      "learning_rate": 0.00026557282946663475,
      "loss": 0.0744,
      "step": 2904
    },
    {
      "epoch": 0.4071478626489138,
      "grad_norm": 0.37230920791625977,
      "learning_rate": 0.0002655584788328151,
      "loss": 0.0819,
      "step": 2905
    },
    {
      "epoch": 0.40728801681850035,
      "grad_norm": 0.3571549654006958,
      "learning_rate": 0.0002655441281989954,
      "loss": 0.1272,
      "step": 2906
    },
    {
      "epoch": 0.40742817098808687,
      "grad_norm": 0.7391756772994995,
      "learning_rate": 0.0002655297775651758,
      "loss": 0.1658,
      "step": 2907
    },
    {
      "epoch": 0.40756832515767344,
      "grad_norm": 0.27977412939071655,
      "learning_rate": 0.0002655154269313561,
      "loss": 0.1013,
      "step": 2908
    },
    {
      "epoch": 0.40770847932726,
      "grad_norm": 0.5675486326217651,
      "learning_rate": 0.00026550107629753645,
      "loss": 0.093,
      "step": 2909
    },
    {
      "epoch": 0.4078486334968465,
      "grad_norm": 0.4636172950267792,
      "learning_rate": 0.0002654867256637168,
      "loss": 0.1128,
      "step": 2910
    },
    {
      "epoch": 0.4079887876664331,
      "grad_norm": 0.4461061954498291,
      "learning_rate": 0.0002654723750298971,
      "loss": 0.0874,
      "step": 2911
    },
    {
      "epoch": 0.4081289418360196,
      "grad_norm": 0.8013656139373779,
      "learning_rate": 0.0002654580243960775,
      "loss": 0.2077,
      "step": 2912
    },
    {
      "epoch": 0.4082690960056062,
      "grad_norm": 0.41661831736564636,
      "learning_rate": 0.0002654436737622578,
      "loss": 0.1508,
      "step": 2913
    },
    {
      "epoch": 0.4084092501751927,
      "grad_norm": 0.6267321109771729,
      "learning_rate": 0.00026542932312843815,
      "loss": 0.1069,
      "step": 2914
    },
    {
      "epoch": 0.40854940434477927,
      "grad_norm": 0.41701653599739075,
      "learning_rate": 0.0002654149724946185,
      "loss": 0.0746,
      "step": 2915
    },
    {
      "epoch": 0.4086895585143658,
      "grad_norm": 0.5108296871185303,
      "learning_rate": 0.0002654006218607988,
      "loss": 0.1085,
      "step": 2916
    },
    {
      "epoch": 0.40882971268395235,
      "grad_norm": 0.45943978428840637,
      "learning_rate": 0.0002653862712269792,
      "loss": 0.075,
      "step": 2917
    },
    {
      "epoch": 0.40896986685353887,
      "grad_norm": 0.1671084612607956,
      "learning_rate": 0.0002653719205931595,
      "loss": 0.0355,
      "step": 2918
    },
    {
      "epoch": 0.40911002102312544,
      "grad_norm": 0.684384822845459,
      "learning_rate": 0.00026535756995933984,
      "loss": 0.0888,
      "step": 2919
    },
    {
      "epoch": 0.409250175192712,
      "grad_norm": 0.6261886358261108,
      "learning_rate": 0.00026534321932552017,
      "loss": 0.1541,
      "step": 2920
    },
    {
      "epoch": 0.4093903293622985,
      "grad_norm": 0.816017746925354,
      "learning_rate": 0.0002653288686917005,
      "loss": 0.1254,
      "step": 2921
    },
    {
      "epoch": 0.4095304835318851,
      "grad_norm": 0.5166419148445129,
      "learning_rate": 0.00026531451805788083,
      "loss": 0.1393,
      "step": 2922
    },
    {
      "epoch": 0.4096706377014716,
      "grad_norm": 0.7215385437011719,
      "learning_rate": 0.0002653001674240612,
      "loss": 0.2085,
      "step": 2923
    },
    {
      "epoch": 0.4098107918710582,
      "grad_norm": 0.4829675853252411,
      "learning_rate": 0.00026528581679024154,
      "loss": 0.1129,
      "step": 2924
    },
    {
      "epoch": 0.4099509460406447,
      "grad_norm": 0.4476076364517212,
      "learning_rate": 0.00026527146615642187,
      "loss": 0.0855,
      "step": 2925
    },
    {
      "epoch": 0.41009110021023126,
      "grad_norm": 0.42605283856391907,
      "learning_rate": 0.00026525711552260225,
      "loss": 0.0879,
      "step": 2926
    },
    {
      "epoch": 0.4102312543798178,
      "grad_norm": 0.5707522034645081,
      "learning_rate": 0.0002652427648887826,
      "loss": 0.1281,
      "step": 2927
    },
    {
      "epoch": 0.41037140854940435,
      "grad_norm": 0.31168338656425476,
      "learning_rate": 0.0002652284142549629,
      "loss": 0.0618,
      "step": 2928
    },
    {
      "epoch": 0.41051156271899086,
      "grad_norm": 0.7115208506584167,
      "learning_rate": 0.00026521406362114324,
      "loss": 0.1368,
      "step": 2929
    },
    {
      "epoch": 0.41065171688857743,
      "grad_norm": 0.38857266306877136,
      "learning_rate": 0.00026519971298732357,
      "loss": 0.0938,
      "step": 2930
    },
    {
      "epoch": 0.410791871058164,
      "grad_norm": 0.9774143695831299,
      "learning_rate": 0.0002651853623535039,
      "loss": 0.1251,
      "step": 2931
    },
    {
      "epoch": 0.4109320252277505,
      "grad_norm": 0.4362261891365051,
      "learning_rate": 0.0002651710117196843,
      "loss": 0.0557,
      "step": 2932
    },
    {
      "epoch": 0.4110721793973371,
      "grad_norm": 0.33460697531700134,
      "learning_rate": 0.0002651566610858646,
      "loss": 0.0686,
      "step": 2933
    },
    {
      "epoch": 0.4112123335669236,
      "grad_norm": 0.2996613681316376,
      "learning_rate": 0.00026514231045204493,
      "loss": 0.0786,
      "step": 2934
    },
    {
      "epoch": 0.4113524877365102,
      "grad_norm": 0.5538722276687622,
      "learning_rate": 0.00026512795981822526,
      "loss": 0.0586,
      "step": 2935
    },
    {
      "epoch": 0.4114926419060967,
      "grad_norm": 0.9747188091278076,
      "learning_rate": 0.00026511360918440565,
      "loss": 0.2031,
      "step": 2936
    },
    {
      "epoch": 0.41163279607568326,
      "grad_norm": 0.3425869047641754,
      "learning_rate": 0.000265099258550586,
      "loss": 0.0454,
      "step": 2937
    },
    {
      "epoch": 0.4117729502452698,
      "grad_norm": 0.820833683013916,
      "learning_rate": 0.0002650849079167663,
      "loss": 0.0933,
      "step": 2938
    },
    {
      "epoch": 0.41191310441485635,
      "grad_norm": 0.562213659286499,
      "learning_rate": 0.00026507055728294663,
      "loss": 0.1649,
      "step": 2939
    },
    {
      "epoch": 0.41205325858444286,
      "grad_norm": 0.28278765082359314,
      "learning_rate": 0.00026505620664912696,
      "loss": 0.0369,
      "step": 2940
    },
    {
      "epoch": 0.41219341275402943,
      "grad_norm": 0.8125547766685486,
      "learning_rate": 0.0002650418560153073,
      "loss": 0.1452,
      "step": 2941
    },
    {
      "epoch": 0.412333566923616,
      "grad_norm": 0.45113614201545715,
      "learning_rate": 0.00026502750538148767,
      "loss": 0.1266,
      "step": 2942
    },
    {
      "epoch": 0.4124737210932025,
      "grad_norm": 0.933447539806366,
      "learning_rate": 0.000265013154747668,
      "loss": 0.0754,
      "step": 2943
    },
    {
      "epoch": 0.4126138752627891,
      "grad_norm": 0.5742002725601196,
      "learning_rate": 0.00026499880411384833,
      "loss": 0.0769,
      "step": 2944
    },
    {
      "epoch": 0.4127540294323756,
      "grad_norm": 0.9658991098403931,
      "learning_rate": 0.0002649844534800287,
      "loss": 0.0815,
      "step": 2945
    },
    {
      "epoch": 0.4128941836019622,
      "grad_norm": 1.565278172492981,
      "learning_rate": 0.00026497010284620904,
      "loss": 0.2244,
      "step": 2946
    },
    {
      "epoch": 0.4130343377715487,
      "grad_norm": 0.43459033966064453,
      "learning_rate": 0.00026495575221238937,
      "loss": 0.0706,
      "step": 2947
    },
    {
      "epoch": 0.41317449194113526,
      "grad_norm": 0.7879146337509155,
      "learning_rate": 0.0002649414015785697,
      "loss": 0.1087,
      "step": 2948
    },
    {
      "epoch": 0.4133146461107218,
      "grad_norm": 1.9656546115875244,
      "learning_rate": 0.00026492705094475,
      "loss": 0.2337,
      "step": 2949
    },
    {
      "epoch": 0.41345480028030834,
      "grad_norm": 4.229058742523193,
      "learning_rate": 0.00026491270031093035,
      "loss": 0.2678,
      "step": 2950
    },
    {
      "epoch": 0.41359495444989486,
      "grad_norm": 0.1756371259689331,
      "learning_rate": 0.0002648983496771107,
      "loss": 0.0332,
      "step": 2951
    },
    {
      "epoch": 0.41373510861948143,
      "grad_norm": 0.4779178202152252,
      "learning_rate": 0.00026488399904329107,
      "loss": 0.1091,
      "step": 2952
    },
    {
      "epoch": 0.413875262789068,
      "grad_norm": 0.254630446434021,
      "learning_rate": 0.0002648696484094714,
      "loss": 0.1228,
      "step": 2953
    },
    {
      "epoch": 0.4140154169586545,
      "grad_norm": 0.7055188417434692,
      "learning_rate": 0.0002648552977756517,
      "loss": 0.1322,
      "step": 2954
    },
    {
      "epoch": 0.4141555711282411,
      "grad_norm": 0.39890941977500916,
      "learning_rate": 0.0002648409471418321,
      "loss": 0.0892,
      "step": 2955
    },
    {
      "epoch": 0.4142957252978276,
      "grad_norm": 0.37668585777282715,
      "learning_rate": 0.00026482659650801243,
      "loss": 0.1193,
      "step": 2956
    },
    {
      "epoch": 0.41443587946741417,
      "grad_norm": 0.38553062081336975,
      "learning_rate": 0.00026481224587419276,
      "loss": 0.1208,
      "step": 2957
    },
    {
      "epoch": 0.4145760336370007,
      "grad_norm": 0.4968641400337219,
      "learning_rate": 0.0002647978952403731,
      "loss": 0.1963,
      "step": 2958
    },
    {
      "epoch": 0.41471618780658726,
      "grad_norm": 0.25200724601745605,
      "learning_rate": 0.0002647835446065534,
      "loss": 0.0918,
      "step": 2959
    },
    {
      "epoch": 0.41485634197617377,
      "grad_norm": 0.49894216656684875,
      "learning_rate": 0.00026476919397273375,
      "loss": 0.1804,
      "step": 2960
    },
    {
      "epoch": 0.41499649614576034,
      "grad_norm": 0.3575514554977417,
      "learning_rate": 0.00026475484333891413,
      "loss": 0.1462,
      "step": 2961
    },
    {
      "epoch": 0.41513665031534686,
      "grad_norm": 0.27937018871307373,
      "learning_rate": 0.00026474049270509446,
      "loss": 0.1234,
      "step": 2962
    },
    {
      "epoch": 0.4152768044849334,
      "grad_norm": 0.20667894184589386,
      "learning_rate": 0.0002647261420712748,
      "loss": 0.0405,
      "step": 2963
    },
    {
      "epoch": 0.41541695865452,
      "grad_norm": 0.39918091893196106,
      "learning_rate": 0.00026471179143745517,
      "loss": 0.1539,
      "step": 2964
    },
    {
      "epoch": 0.4155571128241065,
      "grad_norm": 0.4819334149360657,
      "learning_rate": 0.0002646974408036355,
      "loss": 0.0692,
      "step": 2965
    },
    {
      "epoch": 0.4156972669936931,
      "grad_norm": 0.32738980650901794,
      "learning_rate": 0.00026468309016981583,
      "loss": 0.1191,
      "step": 2966
    },
    {
      "epoch": 0.4158374211632796,
      "grad_norm": 0.353799432516098,
      "learning_rate": 0.00026466873953599616,
      "loss": 0.1041,
      "step": 2967
    },
    {
      "epoch": 0.41597757533286617,
      "grad_norm": 0.32365113496780396,
      "learning_rate": 0.0002646543889021765,
      "loss": 0.1183,
      "step": 2968
    },
    {
      "epoch": 0.4161177295024527,
      "grad_norm": 0.3455530107021332,
      "learning_rate": 0.0002646400382683568,
      "loss": 0.0691,
      "step": 2969
    },
    {
      "epoch": 0.41625788367203925,
      "grad_norm": 0.28271186351776123,
      "learning_rate": 0.00026462568763453714,
      "loss": 0.0503,
      "step": 2970
    },
    {
      "epoch": 0.41639803784162577,
      "grad_norm": 0.26132267713546753,
      "learning_rate": 0.0002646113370007175,
      "loss": 0.1201,
      "step": 2971
    },
    {
      "epoch": 0.41653819201121234,
      "grad_norm": 0.2868199050426483,
      "learning_rate": 0.00026459698636689785,
      "loss": 0.129,
      "step": 2972
    },
    {
      "epoch": 0.41667834618079885,
      "grad_norm": 0.41328686475753784,
      "learning_rate": 0.0002645826357330782,
      "loss": 0.0912,
      "step": 2973
    },
    {
      "epoch": 0.4168185003503854,
      "grad_norm": 0.42311835289001465,
      "learning_rate": 0.0002645682850992585,
      "loss": 0.0916,
      "step": 2974
    },
    {
      "epoch": 0.416958654519972,
      "grad_norm": 0.23855365812778473,
      "learning_rate": 0.00026455393446543884,
      "loss": 0.0711,
      "step": 2975
    },
    {
      "epoch": 0.4170988086895585,
      "grad_norm": 0.6829186081886292,
      "learning_rate": 0.0002645395838316192,
      "loss": 0.1439,
      "step": 2976
    },
    {
      "epoch": 0.4172389628591451,
      "grad_norm": 0.35768362879753113,
      "learning_rate": 0.00026452523319779955,
      "loss": 0.1002,
      "step": 2977
    },
    {
      "epoch": 0.4173791170287316,
      "grad_norm": 0.9309812188148499,
      "learning_rate": 0.0002645108825639799,
      "loss": 0.1423,
      "step": 2978
    },
    {
      "epoch": 0.41751927119831816,
      "grad_norm": 0.3020429313182831,
      "learning_rate": 0.0002644965319301602,
      "loss": 0.0776,
      "step": 2979
    },
    {
      "epoch": 0.4176594253679047,
      "grad_norm": 0.8399023413658142,
      "learning_rate": 0.0002644821812963406,
      "loss": 0.1276,
      "step": 2980
    },
    {
      "epoch": 0.41779957953749125,
      "grad_norm": 0.41155946254730225,
      "learning_rate": 0.0002644678306625209,
      "loss": 0.0882,
      "step": 2981
    },
    {
      "epoch": 0.41793973370707777,
      "grad_norm": 1.7410348653793335,
      "learning_rate": 0.00026445348002870125,
      "loss": 0.0627,
      "step": 2982
    },
    {
      "epoch": 0.41807988787666434,
      "grad_norm": 0.4673553705215454,
      "learning_rate": 0.0002644391293948816,
      "loss": 0.1099,
      "step": 2983
    },
    {
      "epoch": 0.41822004204625085,
      "grad_norm": 0.652702271938324,
      "learning_rate": 0.0002644247787610619,
      "loss": 0.1006,
      "step": 2984
    },
    {
      "epoch": 0.4183601962158374,
      "grad_norm": 0.5195550918579102,
      "learning_rate": 0.00026441042812724223,
      "loss": 0.1085,
      "step": 2985
    },
    {
      "epoch": 0.418500350385424,
      "grad_norm": 0.535480260848999,
      "learning_rate": 0.00026439607749342256,
      "loss": 0.06,
      "step": 2986
    },
    {
      "epoch": 0.4186405045550105,
      "grad_norm": 0.47558096051216125,
      "learning_rate": 0.00026438172685960294,
      "loss": 0.0361,
      "step": 2987
    },
    {
      "epoch": 0.4187806587245971,
      "grad_norm": 0.34890055656433105,
      "learning_rate": 0.0002643673762257833,
      "loss": 0.0718,
      "step": 2988
    },
    {
      "epoch": 0.4189208128941836,
      "grad_norm": 0.4973667860031128,
      "learning_rate": 0.0002643530255919636,
      "loss": 0.1232,
      "step": 2989
    },
    {
      "epoch": 0.41906096706377016,
      "grad_norm": 0.36426812410354614,
      "learning_rate": 0.000264338674958144,
      "loss": 0.0783,
      "step": 2990
    },
    {
      "epoch": 0.4192011212333567,
      "grad_norm": 0.41723981499671936,
      "learning_rate": 0.0002643243243243243,
      "loss": 0.0778,
      "step": 2991
    },
    {
      "epoch": 0.41934127540294325,
      "grad_norm": 0.642321765422821,
      "learning_rate": 0.00026430997369050464,
      "loss": 0.1686,
      "step": 2992
    },
    {
      "epoch": 0.41948142957252976,
      "grad_norm": 0.562143862247467,
      "learning_rate": 0.00026429562305668497,
      "loss": 0.0699,
      "step": 2993
    },
    {
      "epoch": 0.41962158374211633,
      "grad_norm": 0.5614160895347595,
      "learning_rate": 0.0002642812724228653,
      "loss": 0.0604,
      "step": 2994
    },
    {
      "epoch": 0.41976173791170285,
      "grad_norm": 0.9486021399497986,
      "learning_rate": 0.00026426692178904563,
      "loss": 0.2225,
      "step": 2995
    },
    {
      "epoch": 0.4199018920812894,
      "grad_norm": 0.42924001812934875,
      "learning_rate": 0.000264252571155226,
      "loss": 0.0718,
      "step": 2996
    },
    {
      "epoch": 0.420042046250876,
      "grad_norm": 0.39702117443084717,
      "learning_rate": 0.00026423822052140634,
      "loss": 0.0527,
      "step": 2997
    },
    {
      "epoch": 0.4201822004204625,
      "grad_norm": 0.39967837929725647,
      "learning_rate": 0.00026422386988758667,
      "loss": 0.1176,
      "step": 2998
    },
    {
      "epoch": 0.4203223545900491,
      "grad_norm": 0.949603259563446,
      "learning_rate": 0.00026420951925376705,
      "loss": 0.2238,
      "step": 2999
    },
    {
      "epoch": 0.4204625087596356,
      "grad_norm": 0.5017886161804199,
      "learning_rate": 0.0002641951686199474,
      "loss": 0.0277,
      "step": 3000
    },
    {
      "epoch": 0.42060266292922216,
      "grad_norm": 0.7275369763374329,
      "learning_rate": 0.0002641808179861277,
      "loss": 0.1272,
      "step": 3001
    },
    {
      "epoch": 0.4207428170988087,
      "grad_norm": 0.4509636461734772,
      "learning_rate": 0.00026416646735230804,
      "loss": 0.1376,
      "step": 3002
    },
    {
      "epoch": 0.42088297126839525,
      "grad_norm": 0.5083564519882202,
      "learning_rate": 0.00026415211671848836,
      "loss": 0.1072,
      "step": 3003
    },
    {
      "epoch": 0.42102312543798176,
      "grad_norm": 0.5507363080978394,
      "learning_rate": 0.0002641377660846687,
      "loss": 0.1756,
      "step": 3004
    },
    {
      "epoch": 0.42116327960756833,
      "grad_norm": 0.6499862670898438,
      "learning_rate": 0.000264123415450849,
      "loss": 0.0708,
      "step": 3005
    },
    {
      "epoch": 0.42130343377715485,
      "grad_norm": 0.5804831981658936,
      "learning_rate": 0.0002641090648170294,
      "loss": 0.0764,
      "step": 3006
    },
    {
      "epoch": 0.4214435879467414,
      "grad_norm": 0.3736114501953125,
      "learning_rate": 0.00026409471418320973,
      "loss": 0.1008,
      "step": 3007
    },
    {
      "epoch": 0.421583742116328,
      "grad_norm": 0.35122519731521606,
      "learning_rate": 0.00026408036354939006,
      "loss": 0.1304,
      "step": 3008
    },
    {
      "epoch": 0.4217238962859145,
      "grad_norm": 0.5111154913902283,
      "learning_rate": 0.00026406601291557044,
      "loss": 0.1603,
      "step": 3009
    },
    {
      "epoch": 0.42186405045550107,
      "grad_norm": 0.638640284538269,
      "learning_rate": 0.00026405166228175077,
      "loss": 0.1403,
      "step": 3010
    },
    {
      "epoch": 0.4220042046250876,
      "grad_norm": 0.3681301772594452,
      "learning_rate": 0.0002640373116479311,
      "loss": 0.0563,
      "step": 3011
    },
    {
      "epoch": 0.42214435879467416,
      "grad_norm": 0.17159447073936462,
      "learning_rate": 0.00026402296101411143,
      "loss": 0.0441,
      "step": 3012
    },
    {
      "epoch": 0.42228451296426067,
      "grad_norm": 0.12306694686412811,
      "learning_rate": 0.00026400861038029176,
      "loss": 0.0191,
      "step": 3013
    },
    {
      "epoch": 0.42242466713384724,
      "grad_norm": 0.36573100090026855,
      "learning_rate": 0.0002639942597464721,
      "loss": 0.0903,
      "step": 3014
    },
    {
      "epoch": 0.42256482130343376,
      "grad_norm": 0.5017110109329224,
      "learning_rate": 0.00026397990911265247,
      "loss": 0.159,
      "step": 3015
    },
    {
      "epoch": 0.42270497547302033,
      "grad_norm": 0.4422838091850281,
      "learning_rate": 0.0002639655584788328,
      "loss": 0.0761,
      "step": 3016
    },
    {
      "epoch": 0.42284512964260684,
      "grad_norm": 0.34144967794418335,
      "learning_rate": 0.0002639512078450131,
      "loss": 0.1163,
      "step": 3017
    },
    {
      "epoch": 0.4229852838121934,
      "grad_norm": 0.5035829544067383,
      "learning_rate": 0.0002639368572111935,
      "loss": 0.2362,
      "step": 3018
    },
    {
      "epoch": 0.42312543798178,
      "grad_norm": 0.5085100531578064,
      "learning_rate": 0.00026392250657737384,
      "loss": 0.1074,
      "step": 3019
    },
    {
      "epoch": 0.4232655921513665,
      "grad_norm": 0.6563408374786377,
      "learning_rate": 0.00026390815594355417,
      "loss": 0.1482,
      "step": 3020
    },
    {
      "epoch": 0.42340574632095307,
      "grad_norm": 0.4050714373588562,
      "learning_rate": 0.0002638938053097345,
      "loss": 0.1033,
      "step": 3021
    },
    {
      "epoch": 0.4235459004905396,
      "grad_norm": 0.6079158186912537,
      "learning_rate": 0.0002638794546759148,
      "loss": 0.1027,
      "step": 3022
    },
    {
      "epoch": 0.42368605466012615,
      "grad_norm": 0.36508792638778687,
      "learning_rate": 0.00026386510404209515,
      "loss": 0.0756,
      "step": 3023
    },
    {
      "epoch": 0.42382620882971267,
      "grad_norm": 0.5695202946662903,
      "learning_rate": 0.0002638507534082755,
      "loss": 0.0869,
      "step": 3024
    },
    {
      "epoch": 0.42396636299929924,
      "grad_norm": 0.2257179319858551,
      "learning_rate": 0.00026383640277445586,
      "loss": 0.053,
      "step": 3025
    },
    {
      "epoch": 0.42410651716888575,
      "grad_norm": 0.3668532073497772,
      "learning_rate": 0.0002638220521406362,
      "loss": 0.0697,
      "step": 3026
    },
    {
      "epoch": 0.4242466713384723,
      "grad_norm": 0.33470430970191956,
      "learning_rate": 0.0002638077015068165,
      "loss": 0.1208,
      "step": 3027
    },
    {
      "epoch": 0.42438682550805884,
      "grad_norm": 0.24994029104709625,
      "learning_rate": 0.0002637933508729969,
      "loss": 0.0927,
      "step": 3028
    },
    {
      "epoch": 0.4245269796776454,
      "grad_norm": 0.5565463900566101,
      "learning_rate": 0.00026377900023917723,
      "loss": 0.0777,
      "step": 3029
    },
    {
      "epoch": 0.424667133847232,
      "grad_norm": 0.4051344692707062,
      "learning_rate": 0.00026376464960535756,
      "loss": 0.1001,
      "step": 3030
    },
    {
      "epoch": 0.4248072880168185,
      "grad_norm": 0.34019601345062256,
      "learning_rate": 0.0002637502989715379,
      "loss": 0.1183,
      "step": 3031
    },
    {
      "epoch": 0.42494744218640507,
      "grad_norm": 0.42420047521591187,
      "learning_rate": 0.0002637359483377182,
      "loss": 0.0537,
      "step": 3032
    },
    {
      "epoch": 0.4250875963559916,
      "grad_norm": 0.25984880328178406,
      "learning_rate": 0.00026372159770389855,
      "loss": 0.1038,
      "step": 3033
    },
    {
      "epoch": 0.42522775052557815,
      "grad_norm": 1.339253306388855,
      "learning_rate": 0.00026370724707007893,
      "loss": 0.1437,
      "step": 3034
    },
    {
      "epoch": 0.42536790469516467,
      "grad_norm": 0.732659637928009,
      "learning_rate": 0.00026369289643625926,
      "loss": 0.127,
      "step": 3035
    },
    {
      "epoch": 0.42550805886475124,
      "grad_norm": 0.4371141493320465,
      "learning_rate": 0.0002636785458024396,
      "loss": 0.1288,
      "step": 3036
    },
    {
      "epoch": 0.42564821303433775,
      "grad_norm": 0.6234946250915527,
      "learning_rate": 0.0002636641951686199,
      "loss": 0.1709,
      "step": 3037
    },
    {
      "epoch": 0.4257883672039243,
      "grad_norm": 0.3678855001926422,
      "learning_rate": 0.00026364984453480024,
      "loss": 0.1082,
      "step": 3038
    },
    {
      "epoch": 0.42592852137351084,
      "grad_norm": 0.5632939338684082,
      "learning_rate": 0.0002636354939009806,
      "loss": 0.1043,
      "step": 3039
    },
    {
      "epoch": 0.4260686755430974,
      "grad_norm": 0.761344850063324,
      "learning_rate": 0.00026362114326716095,
      "loss": 0.2318,
      "step": 3040
    },
    {
      "epoch": 0.426208829712684,
      "grad_norm": 0.28364187479019165,
      "learning_rate": 0.0002636067926333413,
      "loss": 0.066,
      "step": 3041
    },
    {
      "epoch": 0.4263489838822705,
      "grad_norm": 0.20622359216213226,
      "learning_rate": 0.0002635924419995216,
      "loss": 0.0352,
      "step": 3042
    },
    {
      "epoch": 0.42648913805185706,
      "grad_norm": 0.8606165647506714,
      "learning_rate": 0.00026357809136570194,
      "loss": 0.202,
      "step": 3043
    },
    {
      "epoch": 0.4266292922214436,
      "grad_norm": 0.6587481498718262,
      "learning_rate": 0.0002635637407318823,
      "loss": 0.2369,
      "step": 3044
    },
    {
      "epoch": 0.42676944639103015,
      "grad_norm": 0.7817203402519226,
      "learning_rate": 0.00026354939009806265,
      "loss": 0.0755,
      "step": 3045
    },
    {
      "epoch": 0.42690960056061666,
      "grad_norm": 0.9240639209747314,
      "learning_rate": 0.000263535039464243,
      "loss": 0.0997,
      "step": 3046
    },
    {
      "epoch": 0.42704975473020323,
      "grad_norm": 0.5465213656425476,
      "learning_rate": 0.0002635206888304233,
      "loss": 0.0573,
      "step": 3047
    },
    {
      "epoch": 0.42718990889978975,
      "grad_norm": 1.686435341835022,
      "learning_rate": 0.00026350633819660364,
      "loss": 0.2592,
      "step": 3048
    },
    {
      "epoch": 0.4273300630693763,
      "grad_norm": 1.7949013710021973,
      "learning_rate": 0.00026349198756278397,
      "loss": 0.3234,
      "step": 3049
    },
    {
      "epoch": 0.42747021723896284,
      "grad_norm": 1.3989791870117188,
      "learning_rate": 0.00026347763692896435,
      "loss": 0.3474,
      "step": 3050
    },
    {
      "epoch": 0.4276103714085494,
      "grad_norm": 0.4435987174510956,
      "learning_rate": 0.0002634632862951447,
      "loss": 0.1072,
      "step": 3051
    },
    {
      "epoch": 0.427750525578136,
      "grad_norm": 0.3666754961013794,
      "learning_rate": 0.000263448935661325,
      "loss": 0.1159,
      "step": 3052
    },
    {
      "epoch": 0.4278906797477225,
      "grad_norm": 0.37432336807250977,
      "learning_rate": 0.0002634345850275054,
      "loss": 0.0645,
      "step": 3053
    },
    {
      "epoch": 0.42803083391730906,
      "grad_norm": 0.26446565985679626,
      "learning_rate": 0.0002634202343936857,
      "loss": 0.0672,
      "step": 3054
    },
    {
      "epoch": 0.4281709880868956,
      "grad_norm": 0.4468914270401001,
      "learning_rate": 0.00026340588375986605,
      "loss": 0.1082,
      "step": 3055
    },
    {
      "epoch": 0.42831114225648215,
      "grad_norm": 0.3955155611038208,
      "learning_rate": 0.0002633915331260464,
      "loss": 0.0851,
      "step": 3056
    },
    {
      "epoch": 0.42845129642606866,
      "grad_norm": 0.42428505420684814,
      "learning_rate": 0.0002633771824922267,
      "loss": 0.1129,
      "step": 3057
    },
    {
      "epoch": 0.42859145059565523,
      "grad_norm": 0.3192806839942932,
      "learning_rate": 0.00026336283185840703,
      "loss": 0.0899,
      "step": 3058
    },
    {
      "epoch": 0.42873160476524175,
      "grad_norm": 0.1797448843717575,
      "learning_rate": 0.00026334848122458736,
      "loss": 0.0711,
      "step": 3059
    },
    {
      "epoch": 0.4288717589348283,
      "grad_norm": 0.5377113223075867,
      "learning_rate": 0.00026333413059076774,
      "loss": 0.1953,
      "step": 3060
    },
    {
      "epoch": 0.42901191310441483,
      "grad_norm": 0.32639339566230774,
      "learning_rate": 0.00026331977995694807,
      "loss": 0.1044,
      "step": 3061
    },
    {
      "epoch": 0.4291520672740014,
      "grad_norm": 0.292711079120636,
      "learning_rate": 0.0002633054293231284,
      "loss": 0.0798,
      "step": 3062
    },
    {
      "epoch": 0.429292221443588,
      "grad_norm": 0.3188745975494385,
      "learning_rate": 0.0002632910786893088,
      "loss": 0.0716,
      "step": 3063
    },
    {
      "epoch": 0.4294323756131745,
      "grad_norm": 0.35329943895339966,
      "learning_rate": 0.0002632767280554891,
      "loss": 0.1174,
      "step": 3064
    },
    {
      "epoch": 0.42957252978276106,
      "grad_norm": 0.254645437002182,
      "learning_rate": 0.00026326237742166944,
      "loss": 0.063,
      "step": 3065
    },
    {
      "epoch": 0.4297126839523476,
      "grad_norm": 0.2007201761007309,
      "learning_rate": 0.00026324802678784977,
      "loss": 0.0603,
      "step": 3066
    },
    {
      "epoch": 0.42985283812193414,
      "grad_norm": 0.18286974728107452,
      "learning_rate": 0.0002632336761540301,
      "loss": 0.0271,
      "step": 3067
    },
    {
      "epoch": 0.42999299229152066,
      "grad_norm": 0.2654685080051422,
      "learning_rate": 0.0002632193255202104,
      "loss": 0.0739,
      "step": 3068
    },
    {
      "epoch": 0.43013314646110723,
      "grad_norm": 0.4285154342651367,
      "learning_rate": 0.0002632049748863908,
      "loss": 0.1153,
      "step": 3069
    },
    {
      "epoch": 0.43027330063069374,
      "grad_norm": 0.6475933790206909,
      "learning_rate": 0.00026319062425257114,
      "loss": 0.1119,
      "step": 3070
    },
    {
      "epoch": 0.4304134548002803,
      "grad_norm": 0.40413999557495117,
      "learning_rate": 0.00026317627361875147,
      "loss": 0.083,
      "step": 3071
    },
    {
      "epoch": 0.43055360896986683,
      "grad_norm": 1.027583360671997,
      "learning_rate": 0.00026316192298493185,
      "loss": 0.1676,
      "step": 3072
    },
    {
      "epoch": 0.4306937631394534,
      "grad_norm": 0.8123639225959778,
      "learning_rate": 0.0002631475723511122,
      "loss": 0.0651,
      "step": 3073
    },
    {
      "epoch": 0.43083391730903997,
      "grad_norm": 0.3901086151599884,
      "learning_rate": 0.0002631332217172925,
      "loss": 0.0763,
      "step": 3074
    },
    {
      "epoch": 0.4309740714786265,
      "grad_norm": 0.15434011816978455,
      "learning_rate": 0.00026311887108347283,
      "loss": 0.0501,
      "step": 3075
    },
    {
      "epoch": 0.43111422564821306,
      "grad_norm": 0.6387568116188049,
      "learning_rate": 0.00026310452044965316,
      "loss": 0.0862,
      "step": 3076
    },
    {
      "epoch": 0.43125437981779957,
      "grad_norm": 0.44356054067611694,
      "learning_rate": 0.0002630901698158335,
      "loss": 0.0906,
      "step": 3077
    },
    {
      "epoch": 0.43139453398738614,
      "grad_norm": 1.3771541118621826,
      "learning_rate": 0.0002630758191820138,
      "loss": 0.1049,
      "step": 3078
    },
    {
      "epoch": 0.43153468815697266,
      "grad_norm": 0.4895625412464142,
      "learning_rate": 0.0002630614685481942,
      "loss": 0.0343,
      "step": 3079
    },
    {
      "epoch": 0.4316748423265592,
      "grad_norm": 0.5825363397598267,
      "learning_rate": 0.00026304711791437453,
      "loss": 0.0785,
      "step": 3080
    },
    {
      "epoch": 0.43181499649614574,
      "grad_norm": 0.4963255524635315,
      "learning_rate": 0.00026303276728055486,
      "loss": 0.0911,
      "step": 3081
    },
    {
      "epoch": 0.4319551506657323,
      "grad_norm": 0.3956547975540161,
      "learning_rate": 0.00026301841664673524,
      "loss": 0.0966,
      "step": 3082
    },
    {
      "epoch": 0.4320953048353188,
      "grad_norm": 0.18339678645133972,
      "learning_rate": 0.00026300406601291557,
      "loss": 0.0325,
      "step": 3083
    },
    {
      "epoch": 0.4322354590049054,
      "grad_norm": 0.2436743974685669,
      "learning_rate": 0.0002629897153790959,
      "loss": 0.045,
      "step": 3084
    },
    {
      "epoch": 0.43237561317449197,
      "grad_norm": 0.24353475868701935,
      "learning_rate": 0.00026297536474527623,
      "loss": 0.1093,
      "step": 3085
    },
    {
      "epoch": 0.4325157673440785,
      "grad_norm": 0.5746607184410095,
      "learning_rate": 0.00026296101411145656,
      "loss": 0.1071,
      "step": 3086
    },
    {
      "epoch": 0.43265592151366505,
      "grad_norm": 0.3748045861721039,
      "learning_rate": 0.0002629466634776369,
      "loss": 0.086,
      "step": 3087
    },
    {
      "epoch": 0.43279607568325157,
      "grad_norm": 0.2613745331764221,
      "learning_rate": 0.00026293231284381727,
      "loss": 0.0349,
      "step": 3088
    },
    {
      "epoch": 0.43293622985283814,
      "grad_norm": 0.3729539215564728,
      "learning_rate": 0.0002629179622099976,
      "loss": 0.0576,
      "step": 3089
    },
    {
      "epoch": 0.43307638402242465,
      "grad_norm": 0.5155344009399414,
      "learning_rate": 0.0002629036115761779,
      "loss": 0.1923,
      "step": 3090
    },
    {
      "epoch": 0.4332165381920112,
      "grad_norm": 0.1734478771686554,
      "learning_rate": 0.00026288926094235825,
      "loss": 0.0357,
      "step": 3091
    },
    {
      "epoch": 0.43335669236159774,
      "grad_norm": 0.5360126495361328,
      "learning_rate": 0.00026287491030853864,
      "loss": 0.1009,
      "step": 3092
    },
    {
      "epoch": 0.4334968465311843,
      "grad_norm": 0.43895745277404785,
      "learning_rate": 0.00026286055967471896,
      "loss": 0.0717,
      "step": 3093
    },
    {
      "epoch": 0.4336370007007708,
      "grad_norm": 0.3264348804950714,
      "learning_rate": 0.0002628462090408993,
      "loss": 0.041,
      "step": 3094
    },
    {
      "epoch": 0.4337771548703574,
      "grad_norm": 0.28762224316596985,
      "learning_rate": 0.0002628318584070796,
      "loss": 0.0367,
      "step": 3095
    },
    {
      "epoch": 0.43391730903994397,
      "grad_norm": 0.3108057975769043,
      "learning_rate": 0.00026281750777325995,
      "loss": 0.0428,
      "step": 3096
    },
    {
      "epoch": 0.4340574632095305,
      "grad_norm": 0.7983070015907288,
      "learning_rate": 0.0002628031571394403,
      "loss": 0.1611,
      "step": 3097
    },
    {
      "epoch": 0.43419761737911705,
      "grad_norm": 2.7643096446990967,
      "learning_rate": 0.00026278880650562066,
      "loss": 0.1842,
      "step": 3098
    },
    {
      "epoch": 0.43433777154870357,
      "grad_norm": 1.5378133058547974,
      "learning_rate": 0.000262774455871801,
      "loss": 0.1501,
      "step": 3099
    },
    {
      "epoch": 0.43447792571829014,
      "grad_norm": 1.134103536605835,
      "learning_rate": 0.0002627601052379813,
      "loss": 0.0855,
      "step": 3100
    },
    {
      "epoch": 0.43461807988787665,
      "grad_norm": 0.6849503517150879,
      "learning_rate": 0.00026274575460416165,
      "loss": 0.1099,
      "step": 3101
    },
    {
      "epoch": 0.4347582340574632,
      "grad_norm": 0.5533718466758728,
      "learning_rate": 0.000262731403970342,
      "loss": 0.128,
      "step": 3102
    },
    {
      "epoch": 0.43489838822704974,
      "grad_norm": 0.428962767124176,
      "learning_rate": 0.00026271705333652236,
      "loss": 0.0804,
      "step": 3103
    },
    {
      "epoch": 0.4350385423966363,
      "grad_norm": 0.6989350318908691,
      "learning_rate": 0.0002627027027027027,
      "loss": 0.1008,
      "step": 3104
    },
    {
      "epoch": 0.4351786965662228,
      "grad_norm": 0.7078231573104858,
      "learning_rate": 0.000262688352068883,
      "loss": 0.1058,
      "step": 3105
    },
    {
      "epoch": 0.4353188507358094,
      "grad_norm": 0.3282186985015869,
      "learning_rate": 0.00026267400143506334,
      "loss": 0.0401,
      "step": 3106
    },
    {
      "epoch": 0.43545900490539596,
      "grad_norm": 0.38179269433021545,
      "learning_rate": 0.0002626596508012437,
      "loss": 0.1186,
      "step": 3107
    },
    {
      "epoch": 0.4355991590749825,
      "grad_norm": 0.40050217509269714,
      "learning_rate": 0.00026264530016742406,
      "loss": 0.0942,
      "step": 3108
    },
    {
      "epoch": 0.43573931324456905,
      "grad_norm": 0.582998514175415,
      "learning_rate": 0.0002626309495336044,
      "loss": 0.1446,
      "step": 3109
    },
    {
      "epoch": 0.43587946741415556,
      "grad_norm": 0.479499876499176,
      "learning_rate": 0.0002626165988997847,
      "loss": 0.1304,
      "step": 3110
    },
    {
      "epoch": 0.43601962158374213,
      "grad_norm": 0.43234512209892273,
      "learning_rate": 0.00026260224826596504,
      "loss": 0.0875,
      "step": 3111
    },
    {
      "epoch": 0.43615977575332865,
      "grad_norm": 0.5029392242431641,
      "learning_rate": 0.00026258789763214537,
      "loss": 0.0653,
      "step": 3112
    },
    {
      "epoch": 0.4362999299229152,
      "grad_norm": 0.4136073887348175,
      "learning_rate": 0.0002625735469983257,
      "loss": 0.0685,
      "step": 3113
    },
    {
      "epoch": 0.43644008409250173,
      "grad_norm": 0.2521909773349762,
      "learning_rate": 0.0002625591963645061,
      "loss": 0.0494,
      "step": 3114
    },
    {
      "epoch": 0.4365802382620883,
      "grad_norm": 0.32898223400115967,
      "learning_rate": 0.0002625448457306864,
      "loss": 0.0833,
      "step": 3115
    },
    {
      "epoch": 0.4367203924316748,
      "grad_norm": 0.354140967130661,
      "learning_rate": 0.00026253049509686674,
      "loss": 0.057,
      "step": 3116
    },
    {
      "epoch": 0.4368605466012614,
      "grad_norm": 0.5951328277587891,
      "learning_rate": 0.0002625161444630471,
      "loss": 0.1146,
      "step": 3117
    },
    {
      "epoch": 0.4370007007708479,
      "grad_norm": 0.4011349380016327,
      "learning_rate": 0.00026250179382922745,
      "loss": 0.1438,
      "step": 3118
    },
    {
      "epoch": 0.4371408549404345,
      "grad_norm": 0.28471773862838745,
      "learning_rate": 0.0002624874431954078,
      "loss": 0.0847,
      "step": 3119
    },
    {
      "epoch": 0.43728100911002105,
      "grad_norm": 0.269428551197052,
      "learning_rate": 0.0002624730925615881,
      "loss": 0.0586,
      "step": 3120
    },
    {
      "epoch": 0.43742116327960756,
      "grad_norm": 0.6035513281822205,
      "learning_rate": 0.00026245874192776844,
      "loss": 0.0608,
      "step": 3121
    },
    {
      "epoch": 0.43756131744919413,
      "grad_norm": 0.43457916378974915,
      "learning_rate": 0.00026244439129394876,
      "loss": 0.0613,
      "step": 3122
    },
    {
      "epoch": 0.43770147161878065,
      "grad_norm": 0.36663711071014404,
      "learning_rate": 0.00026243004066012915,
      "loss": 0.0665,
      "step": 3123
    },
    {
      "epoch": 0.4378416257883672,
      "grad_norm": 0.26888251304626465,
      "learning_rate": 0.0002624156900263095,
      "loss": 0.0907,
      "step": 3124
    },
    {
      "epoch": 0.43798177995795373,
      "grad_norm": 0.2705402970314026,
      "learning_rate": 0.0002624013393924898,
      "loss": 0.0944,
      "step": 3125
    },
    {
      "epoch": 0.4381219341275403,
      "grad_norm": 0.5814818143844604,
      "learning_rate": 0.00026238698875867013,
      "loss": 0.1185,
      "step": 3126
    },
    {
      "epoch": 0.4382620882971268,
      "grad_norm": 1.2210744619369507,
      "learning_rate": 0.0002623726381248505,
      "loss": 0.1448,
      "step": 3127
    },
    {
      "epoch": 0.4384022424667134,
      "grad_norm": 0.1287975311279297,
      "learning_rate": 0.00026235828749103084,
      "loss": 0.0172,
      "step": 3128
    },
    {
      "epoch": 0.4385423966362999,
      "grad_norm": 0.40411970019340515,
      "learning_rate": 0.00026234393685721117,
      "loss": 0.0874,
      "step": 3129
    },
    {
      "epoch": 0.4386825508058865,
      "grad_norm": 0.43082526326179504,
      "learning_rate": 0.0002623295862233915,
      "loss": 0.1773,
      "step": 3130
    },
    {
      "epoch": 0.43882270497547304,
      "grad_norm": 0.19812557101249695,
      "learning_rate": 0.00026231523558957183,
      "loss": 0.0472,
      "step": 3131
    },
    {
      "epoch": 0.43896285914505956,
      "grad_norm": 0.9664269685745239,
      "learning_rate": 0.00026230088495575216,
      "loss": 0.1734,
      "step": 3132
    },
    {
      "epoch": 0.43910301331464613,
      "grad_norm": 0.332105427980423,
      "learning_rate": 0.00026228653432193254,
      "loss": 0.0914,
      "step": 3133
    },
    {
      "epoch": 0.43924316748423264,
      "grad_norm": 0.2724143862724304,
      "learning_rate": 0.00026227218368811287,
      "loss": 0.0423,
      "step": 3134
    },
    {
      "epoch": 0.4393833216538192,
      "grad_norm": 0.4498167634010315,
      "learning_rate": 0.0002622578330542932,
      "loss": 0.0532,
      "step": 3135
    },
    {
      "epoch": 0.43952347582340573,
      "grad_norm": 0.52696293592453,
      "learning_rate": 0.0002622434824204736,
      "loss": 0.0794,
      "step": 3136
    },
    {
      "epoch": 0.4396636299929923,
      "grad_norm": 0.719615638256073,
      "learning_rate": 0.0002622291317866539,
      "loss": 0.0933,
      "step": 3137
    },
    {
      "epoch": 0.4398037841625788,
      "grad_norm": 0.8044444918632507,
      "learning_rate": 0.00026221478115283424,
      "loss": 0.1427,
      "step": 3138
    },
    {
      "epoch": 0.4399439383321654,
      "grad_norm": 0.3632006347179413,
      "learning_rate": 0.00026220043051901457,
      "loss": 0.0681,
      "step": 3139
    },
    {
      "epoch": 0.4400840925017519,
      "grad_norm": 0.22276180982589722,
      "learning_rate": 0.0002621860798851949,
      "loss": 0.0323,
      "step": 3140
    },
    {
      "epoch": 0.44022424667133847,
      "grad_norm": 0.45727425813674927,
      "learning_rate": 0.0002621717292513752,
      "loss": 0.0811,
      "step": 3141
    },
    {
      "epoch": 0.44036440084092504,
      "grad_norm": 0.8930152058601379,
      "learning_rate": 0.0002621573786175556,
      "loss": 0.2072,
      "step": 3142
    },
    {
      "epoch": 0.44050455501051156,
      "grad_norm": 1.0207765102386475,
      "learning_rate": 0.00026214302798373593,
      "loss": 0.1582,
      "step": 3143
    },
    {
      "epoch": 0.4406447091800981,
      "grad_norm": 0.5111526250839233,
      "learning_rate": 0.00026212867734991626,
      "loss": 0.0848,
      "step": 3144
    },
    {
      "epoch": 0.44078486334968464,
      "grad_norm": 0.6272589564323425,
      "learning_rate": 0.0002621143267160966,
      "loss": 0.1635,
      "step": 3145
    },
    {
      "epoch": 0.4409250175192712,
      "grad_norm": 0.732872724533081,
      "learning_rate": 0.000262099976082277,
      "loss": 0.1136,
      "step": 3146
    },
    {
      "epoch": 0.4410651716888577,
      "grad_norm": 0.5442157983779907,
      "learning_rate": 0.0002620856254484573,
      "loss": 0.1094,
      "step": 3147
    },
    {
      "epoch": 0.4412053258584443,
      "grad_norm": 0.8131890892982483,
      "learning_rate": 0.00026207127481463763,
      "loss": 0.0801,
      "step": 3148
    },
    {
      "epoch": 0.4413454800280308,
      "grad_norm": 2.3645200729370117,
      "learning_rate": 0.00026205692418081796,
      "loss": 0.3437,
      "step": 3149
    },
    {
      "epoch": 0.4414856341976174,
      "grad_norm": 0.8918802738189697,
      "learning_rate": 0.0002620425735469983,
      "loss": 0.1558,
      "step": 3150
    },
    {
      "epoch": 0.4416257883672039,
      "grad_norm": 0.7216973900794983,
      "learning_rate": 0.0002620282229131786,
      "loss": 0.1064,
      "step": 3151
    },
    {
      "epoch": 0.44176594253679047,
      "grad_norm": 0.4748555123806,
      "learning_rate": 0.000262013872279359,
      "loss": 0.1197,
      "step": 3152
    },
    {
      "epoch": 0.44190609670637704,
      "grad_norm": 0.37840035557746887,
      "learning_rate": 0.00026199952164553933,
      "loss": 0.1133,
      "step": 3153
    },
    {
      "epoch": 0.44204625087596355,
      "grad_norm": 0.4857287108898163,
      "learning_rate": 0.00026198517101171966,
      "loss": 0.0831,
      "step": 3154
    },
    {
      "epoch": 0.4421864050455501,
      "grad_norm": 0.24071750044822693,
      "learning_rate": 0.00026197082037790004,
      "loss": 0.0364,
      "step": 3155
    },
    {
      "epoch": 0.44232655921513664,
      "grad_norm": 1.2475543022155762,
      "learning_rate": 0.00026195646974408037,
      "loss": 0.1407,
      "step": 3156
    },
    {
      "epoch": 0.4424667133847232,
      "grad_norm": 1.206235647201538,
      "learning_rate": 0.0002619421191102607,
      "loss": 0.148,
      "step": 3157
    },
    {
      "epoch": 0.4426068675543097,
      "grad_norm": 0.5146431922912598,
      "learning_rate": 0.000261927768476441,
      "loss": 0.0851,
      "step": 3158
    },
    {
      "epoch": 0.4427470217238963,
      "grad_norm": 0.21378140151500702,
      "learning_rate": 0.00026191341784262135,
      "loss": 0.0452,
      "step": 3159
    },
    {
      "epoch": 0.4428871758934828,
      "grad_norm": 0.690898597240448,
      "learning_rate": 0.0002618990672088017,
      "loss": 0.1493,
      "step": 3160
    },
    {
      "epoch": 0.4430273300630694,
      "grad_norm": 0.20644928514957428,
      "learning_rate": 0.000261884716574982,
      "loss": 0.0237,
      "step": 3161
    },
    {
      "epoch": 0.4431674842326559,
      "grad_norm": 0.6293278932571411,
      "learning_rate": 0.0002618703659411624,
      "loss": 0.138,
      "step": 3162
    },
    {
      "epoch": 0.44330763840224247,
      "grad_norm": 0.40677598118782043,
      "learning_rate": 0.0002618560153073427,
      "loss": 0.0668,
      "step": 3163
    },
    {
      "epoch": 0.44344779257182904,
      "grad_norm": 0.5735763311386108,
      "learning_rate": 0.00026184166467352305,
      "loss": 0.1093,
      "step": 3164
    },
    {
      "epoch": 0.44358794674141555,
      "grad_norm": 0.3656761646270752,
      "learning_rate": 0.0002618273140397034,
      "loss": 0.0927,
      "step": 3165
    },
    {
      "epoch": 0.4437281009110021,
      "grad_norm": 0.6591564416885376,
      "learning_rate": 0.00026181296340588376,
      "loss": 0.0832,
      "step": 3166
    },
    {
      "epoch": 0.44386825508058864,
      "grad_norm": 0.2716527581214905,
      "learning_rate": 0.0002617986127720641,
      "loss": 0.0705,
      "step": 3167
    },
    {
      "epoch": 0.4440084092501752,
      "grad_norm": 0.5181809067726135,
      "learning_rate": 0.0002617842621382444,
      "loss": 0.1406,
      "step": 3168
    },
    {
      "epoch": 0.4441485634197617,
      "grad_norm": 0.6783599257469177,
      "learning_rate": 0.00026176991150442475,
      "loss": 0.1937,
      "step": 3169
    },
    {
      "epoch": 0.4442887175893483,
      "grad_norm": 0.434597373008728,
      "learning_rate": 0.0002617555608706051,
      "loss": 0.0832,
      "step": 3170
    },
    {
      "epoch": 0.4444288717589348,
      "grad_norm": 0.20813724398612976,
      "learning_rate": 0.00026174121023678546,
      "loss": 0.0746,
      "step": 3171
    },
    {
      "epoch": 0.4445690259285214,
      "grad_norm": 0.5710974931716919,
      "learning_rate": 0.0002617268596029658,
      "loss": 0.1291,
      "step": 3172
    },
    {
      "epoch": 0.4447091800981079,
      "grad_norm": 0.4765787720680237,
      "learning_rate": 0.0002617125089691461,
      "loss": 0.1318,
      "step": 3173
    },
    {
      "epoch": 0.44484933426769446,
      "grad_norm": 0.15755750238895416,
      "learning_rate": 0.00026169815833532645,
      "loss": 0.0489,
      "step": 3174
    },
    {
      "epoch": 0.44498948843728103,
      "grad_norm": 0.4330451786518097,
      "learning_rate": 0.0002616838077015068,
      "loss": 0.1133,
      "step": 3175
    },
    {
      "epoch": 0.44512964260686755,
      "grad_norm": 0.4515622854232788,
      "learning_rate": 0.0002616694570676871,
      "loss": 0.0789,
      "step": 3176
    },
    {
      "epoch": 0.4452697967764541,
      "grad_norm": 0.29335352778434753,
      "learning_rate": 0.00026165510643386743,
      "loss": 0.0448,
      "step": 3177
    },
    {
      "epoch": 0.44540995094604063,
      "grad_norm": 0.18906913697719574,
      "learning_rate": 0.0002616407558000478,
      "loss": 0.0611,
      "step": 3178
    },
    {
      "epoch": 0.4455501051156272,
      "grad_norm": 0.7547434568405151,
      "learning_rate": 0.00026162640516622814,
      "loss": 0.1624,
      "step": 3179
    },
    {
      "epoch": 0.4456902592852137,
      "grad_norm": 0.283898264169693,
      "learning_rate": 0.00026161205453240847,
      "loss": 0.0698,
      "step": 3180
    },
    {
      "epoch": 0.4458304134548003,
      "grad_norm": 0.4482957124710083,
      "learning_rate": 0.00026159770389858885,
      "loss": 0.0663,
      "step": 3181
    },
    {
      "epoch": 0.4459705676243868,
      "grad_norm": 0.4283231198787689,
      "learning_rate": 0.0002615833532647692,
      "loss": 0.1118,
      "step": 3182
    },
    {
      "epoch": 0.4461107217939734,
      "grad_norm": 0.7021906971931458,
      "learning_rate": 0.0002615690026309495,
      "loss": 0.1235,
      "step": 3183
    },
    {
      "epoch": 0.4462508759635599,
      "grad_norm": 0.42637211084365845,
      "learning_rate": 0.00026155465199712984,
      "loss": 0.2485,
      "step": 3184
    },
    {
      "epoch": 0.44639103013314646,
      "grad_norm": 0.6978999972343445,
      "learning_rate": 0.00026154030136331017,
      "loss": 0.0936,
      "step": 3185
    },
    {
      "epoch": 0.44653118430273303,
      "grad_norm": 1.0158802270889282,
      "learning_rate": 0.0002615259507294905,
      "loss": 0.1193,
      "step": 3186
    },
    {
      "epoch": 0.44667133847231955,
      "grad_norm": 0.4636441767215729,
      "learning_rate": 0.0002615116000956709,
      "loss": 0.1001,
      "step": 3187
    },
    {
      "epoch": 0.4468114926419061,
      "grad_norm": 0.5528663396835327,
      "learning_rate": 0.0002614972494618512,
      "loss": 0.1333,
      "step": 3188
    },
    {
      "epoch": 0.44695164681149263,
      "grad_norm": 0.4500793516635895,
      "learning_rate": 0.00026148289882803154,
      "loss": 0.0912,
      "step": 3189
    },
    {
      "epoch": 0.4470918009810792,
      "grad_norm": 0.2766513526439667,
      "learning_rate": 0.0002614685481942119,
      "loss": 0.074,
      "step": 3190
    },
    {
      "epoch": 0.4472319551506657,
      "grad_norm": 0.5607905983924866,
      "learning_rate": 0.00026145419756039225,
      "loss": 0.1008,
      "step": 3191
    },
    {
      "epoch": 0.4473721093202523,
      "grad_norm": 0.8401314616203308,
      "learning_rate": 0.0002614398469265726,
      "loss": 0.1189,
      "step": 3192
    },
    {
      "epoch": 0.4475122634898388,
      "grad_norm": 1.1839956045150757,
      "learning_rate": 0.0002614254962927529,
      "loss": 0.157,
      "step": 3193
    },
    {
      "epoch": 0.4476524176594254,
      "grad_norm": 0.49879908561706543,
      "learning_rate": 0.00026141114565893323,
      "loss": 0.0865,
      "step": 3194
    },
    {
      "epoch": 0.4477925718290119,
      "grad_norm": 0.6396143436431885,
      "learning_rate": 0.00026139679502511356,
      "loss": 0.0882,
      "step": 3195
    },
    {
      "epoch": 0.44793272599859846,
      "grad_norm": 0.859186053276062,
      "learning_rate": 0.0002613824443912939,
      "loss": 0.0736,
      "step": 3196
    },
    {
      "epoch": 0.44807288016818503,
      "grad_norm": 2.0409865379333496,
      "learning_rate": 0.0002613680937574743,
      "loss": 0.2236,
      "step": 3197
    },
    {
      "epoch": 0.44821303433777154,
      "grad_norm": 5.258183002471924,
      "learning_rate": 0.0002613537431236546,
      "loss": 0.3042,
      "step": 3198
    },
    {
      "epoch": 0.4483531885073581,
      "grad_norm": 0.9645035862922668,
      "learning_rate": 0.00026133939248983493,
      "loss": 0.0763,
      "step": 3199
    },
    {
      "epoch": 0.44849334267694463,
      "grad_norm": 2.257209539413452,
      "learning_rate": 0.0002613250418560153,
      "loss": 0.1904,
      "step": 3200
    },
    {
      "epoch": 0.4486334968465312,
      "grad_norm": 0.4620717465877533,
      "learning_rate": 0.00026131069122219564,
      "loss": 0.1101,
      "step": 3201
    },
    {
      "epoch": 0.4487736510161177,
      "grad_norm": 0.32010000944137573,
      "learning_rate": 0.00026129634058837597,
      "loss": 0.0874,
      "step": 3202
    },
    {
      "epoch": 0.4489138051857043,
      "grad_norm": 0.6990892291069031,
      "learning_rate": 0.0002612819899545563,
      "loss": 0.0711,
      "step": 3203
    },
    {
      "epoch": 0.4490539593552908,
      "grad_norm": 0.4191516637802124,
      "learning_rate": 0.00026126763932073663,
      "loss": 0.073,
      "step": 3204
    },
    {
      "epoch": 0.44919411352487737,
      "grad_norm": 0.3744420111179352,
      "learning_rate": 0.00026125328868691696,
      "loss": 0.0502,
      "step": 3205
    },
    {
      "epoch": 0.4493342676944639,
      "grad_norm": 1.075827717781067,
      "learning_rate": 0.00026123893805309734,
      "loss": 0.1288,
      "step": 3206
    },
    {
      "epoch": 0.44947442186405046,
      "grad_norm": 0.42123499512672424,
      "learning_rate": 0.00026122458741927767,
      "loss": 0.1185,
      "step": 3207
    },
    {
      "epoch": 0.449614576033637,
      "grad_norm": 1.8295400142669678,
      "learning_rate": 0.000261210236785458,
      "loss": 0.1123,
      "step": 3208
    },
    {
      "epoch": 0.44975473020322354,
      "grad_norm": 0.6613677144050598,
      "learning_rate": 0.0002611958861516384,
      "loss": 0.0904,
      "step": 3209
    },
    {
      "epoch": 0.4498948843728101,
      "grad_norm": 0.31467679142951965,
      "learning_rate": 0.0002611815355178187,
      "loss": 0.0749,
      "step": 3210
    },
    {
      "epoch": 0.4500350385423966,
      "grad_norm": 0.46356335282325745,
      "learning_rate": 0.00026116718488399904,
      "loss": 0.1156,
      "step": 3211
    },
    {
      "epoch": 0.4501751927119832,
      "grad_norm": 0.6180557608604431,
      "learning_rate": 0.00026115283425017936,
      "loss": 0.085,
      "step": 3212
    },
    {
      "epoch": 0.4503153468815697,
      "grad_norm": 0.37459442019462585,
      "learning_rate": 0.0002611384836163597,
      "loss": 0.0633,
      "step": 3213
    },
    {
      "epoch": 0.4504555010511563,
      "grad_norm": 1.0285698175430298,
      "learning_rate": 0.00026112413298254,
      "loss": 0.3154,
      "step": 3214
    },
    {
      "epoch": 0.4505956552207428,
      "grad_norm": 1.2040151357650757,
      "learning_rate": 0.00026110978234872035,
      "loss": 0.1674,
      "step": 3215
    },
    {
      "epoch": 0.45073580939032937,
      "grad_norm": 0.5814149975776672,
      "learning_rate": 0.00026109543171490073,
      "loss": 0.1476,
      "step": 3216
    },
    {
      "epoch": 0.4508759635599159,
      "grad_norm": 0.4208332300186157,
      "learning_rate": 0.00026108108108108106,
      "loss": 0.1416,
      "step": 3217
    },
    {
      "epoch": 0.45101611772950245,
      "grad_norm": 0.798169732093811,
      "learning_rate": 0.0002610667304472614,
      "loss": 0.102,
      "step": 3218
    },
    {
      "epoch": 0.451156271899089,
      "grad_norm": 0.6296437978744507,
      "learning_rate": 0.00026105237981344177,
      "loss": 0.0909,
      "step": 3219
    },
    {
      "epoch": 0.45129642606867554,
      "grad_norm": 0.5009991526603699,
      "learning_rate": 0.0002610380291796221,
      "loss": 0.1791,
      "step": 3220
    },
    {
      "epoch": 0.4514365802382621,
      "grad_norm": 0.3721868693828583,
      "learning_rate": 0.00026102367854580243,
      "loss": 0.103,
      "step": 3221
    },
    {
      "epoch": 0.4515767344078486,
      "grad_norm": 0.5548828840255737,
      "learning_rate": 0.00026100932791198276,
      "loss": 0.1465,
      "step": 3222
    },
    {
      "epoch": 0.4517168885774352,
      "grad_norm": 0.44447898864746094,
      "learning_rate": 0.0002609949772781631,
      "loss": 0.0832,
      "step": 3223
    },
    {
      "epoch": 0.4518570427470217,
      "grad_norm": 0.3573371171951294,
      "learning_rate": 0.0002609806266443434,
      "loss": 0.0988,
      "step": 3224
    },
    {
      "epoch": 0.4519971969166083,
      "grad_norm": 0.45428574085235596,
      "learning_rate": 0.0002609662760105238,
      "loss": 0.161,
      "step": 3225
    },
    {
      "epoch": 0.4521373510861948,
      "grad_norm": 0.23204457759857178,
      "learning_rate": 0.0002609519253767041,
      "loss": 0.0699,
      "step": 3226
    },
    {
      "epoch": 0.45227750525578136,
      "grad_norm": 0.19380664825439453,
      "learning_rate": 0.00026093757474288446,
      "loss": 0.0558,
      "step": 3227
    },
    {
      "epoch": 0.4524176594253679,
      "grad_norm": 0.4129902720451355,
      "learning_rate": 0.0002609232241090648,
      "loss": 0.1388,
      "step": 3228
    },
    {
      "epoch": 0.45255781359495445,
      "grad_norm": 0.27421775460243225,
      "learning_rate": 0.0002609088734752451,
      "loss": 0.0512,
      "step": 3229
    },
    {
      "epoch": 0.452697967764541,
      "grad_norm": 0.28085988759994507,
      "learning_rate": 0.0002608945228414255,
      "loss": 0.0969,
      "step": 3230
    },
    {
      "epoch": 0.45283812193412754,
      "grad_norm": 0.3167339861392975,
      "learning_rate": 0.0002608801722076058,
      "loss": 0.077,
      "step": 3231
    },
    {
      "epoch": 0.4529782761037141,
      "grad_norm": 0.8371204137802124,
      "learning_rate": 0.00026086582157378615,
      "loss": 0.1373,
      "step": 3232
    },
    {
      "epoch": 0.4531184302733006,
      "grad_norm": 0.3662284016609192,
      "learning_rate": 0.0002608514709399665,
      "loss": 0.0511,
      "step": 3233
    },
    {
      "epoch": 0.4532585844428872,
      "grad_norm": 0.44114458560943604,
      "learning_rate": 0.0002608371203061468,
      "loss": 0.1005,
      "step": 3234
    },
    {
      "epoch": 0.4533987386124737,
      "grad_norm": 0.2829136252403259,
      "learning_rate": 0.0002608227696723272,
      "loss": 0.0616,
      "step": 3235
    },
    {
      "epoch": 0.4535388927820603,
      "grad_norm": 0.27263015508651733,
      "learning_rate": 0.0002608084190385075,
      "loss": 0.1114,
      "step": 3236
    },
    {
      "epoch": 0.4536790469516468,
      "grad_norm": 0.4303293526172638,
      "learning_rate": 0.00026079406840468785,
      "loss": 0.052,
      "step": 3237
    },
    {
      "epoch": 0.45381920112123336,
      "grad_norm": 0.4607178568840027,
      "learning_rate": 0.0002607797177708682,
      "loss": 0.2341,
      "step": 3238
    },
    {
      "epoch": 0.4539593552908199,
      "grad_norm": 0.7156077027320862,
      "learning_rate": 0.0002607653671370485,
      "loss": 0.0473,
      "step": 3239
    },
    {
      "epoch": 0.45409950946040645,
      "grad_norm": 0.5410364866256714,
      "learning_rate": 0.00026075101650322883,
      "loss": 0.1205,
      "step": 3240
    },
    {
      "epoch": 0.454239663629993,
      "grad_norm": 0.42635980248451233,
      "learning_rate": 0.0002607366658694092,
      "loss": 0.0563,
      "step": 3241
    },
    {
      "epoch": 0.45437981779957953,
      "grad_norm": 0.8016296625137329,
      "learning_rate": 0.00026072231523558955,
      "loss": 0.1667,
      "step": 3242
    },
    {
      "epoch": 0.4545199719691661,
      "grad_norm": 0.5985812544822693,
      "learning_rate": 0.0002607079646017699,
      "loss": 0.1038,
      "step": 3243
    },
    {
      "epoch": 0.4546601261387526,
      "grad_norm": 0.49753767251968384,
      "learning_rate": 0.00026069361396795026,
      "loss": 0.0713,
      "step": 3244
    },
    {
      "epoch": 0.4548002803083392,
      "grad_norm": 0.711574375629425,
      "learning_rate": 0.0002606792633341306,
      "loss": 0.0612,
      "step": 3245
    },
    {
      "epoch": 0.4549404344779257,
      "grad_norm": 0.42461737990379333,
      "learning_rate": 0.0002606649127003109,
      "loss": 0.057,
      "step": 3246
    },
    {
      "epoch": 0.4550805886475123,
      "grad_norm": 1.2752445936203003,
      "learning_rate": 0.00026065056206649124,
      "loss": 0.2273,
      "step": 3247
    },
    {
      "epoch": 0.4552207428170988,
      "grad_norm": 1.6907846927642822,
      "learning_rate": 0.00026063621143267157,
      "loss": 0.1058,
      "step": 3248
    },
    {
      "epoch": 0.45536089698668536,
      "grad_norm": 0.2871204912662506,
      "learning_rate": 0.0002606218607988519,
      "loss": 0.0783,
      "step": 3249
    },
    {
      "epoch": 0.4555010511562719,
      "grad_norm": 5.855483531951904,
      "learning_rate": 0.00026060751016503223,
      "loss": 0.3161,
      "step": 3250
    },
    {
      "epoch": 0.45564120532585844,
      "grad_norm": 0.41367432475090027,
      "learning_rate": 0.0002605931595312126,
      "loss": 0.076,
      "step": 3251
    },
    {
      "epoch": 0.455781359495445,
      "grad_norm": 0.3567110002040863,
      "learning_rate": 0.00026057880889739294,
      "loss": 0.0884,
      "step": 3252
    },
    {
      "epoch": 0.45592151366503153,
      "grad_norm": 0.8167653679847717,
      "learning_rate": 0.00026056445826357327,
      "loss": 0.1222,
      "step": 3253
    },
    {
      "epoch": 0.4560616678346181,
      "grad_norm": 0.29383742809295654,
      "learning_rate": 0.00026055010762975365,
      "loss": 0.0604,
      "step": 3254
    },
    {
      "epoch": 0.4562018220042046,
      "grad_norm": 0.36296477913856506,
      "learning_rate": 0.000260535756995934,
      "loss": 0.0421,
      "step": 3255
    },
    {
      "epoch": 0.4563419761737912,
      "grad_norm": 0.886231005191803,
      "learning_rate": 0.0002605214063621143,
      "loss": 0.1239,
      "step": 3256
    },
    {
      "epoch": 0.4564821303433777,
      "grad_norm": 0.69781494140625,
      "learning_rate": 0.00026050705572829464,
      "loss": 0.114,
      "step": 3257
    },
    {
      "epoch": 0.45662228451296427,
      "grad_norm": 0.3201445937156677,
      "learning_rate": 0.00026049270509447497,
      "loss": 0.0915,
      "step": 3258
    },
    {
      "epoch": 0.4567624386825508,
      "grad_norm": 0.5376164317131042,
      "learning_rate": 0.0002604783544606553,
      "loss": 0.0868,
      "step": 3259
    },
    {
      "epoch": 0.45690259285213736,
      "grad_norm": 0.3929520547389984,
      "learning_rate": 0.0002604640038268357,
      "loss": 0.103,
      "step": 3260
    },
    {
      "epoch": 0.45704274702172387,
      "grad_norm": 0.5999123454093933,
      "learning_rate": 0.000260449653193016,
      "loss": 0.0957,
      "step": 3261
    },
    {
      "epoch": 0.45718290119131044,
      "grad_norm": 0.983153760433197,
      "learning_rate": 0.00026043530255919633,
      "loss": 0.181,
      "step": 3262
    },
    {
      "epoch": 0.457323055360897,
      "grad_norm": 0.48948073387145996,
      "learning_rate": 0.0002604209519253767,
      "loss": 0.1591,
      "step": 3263
    },
    {
      "epoch": 0.4574632095304835,
      "grad_norm": 0.7965589761734009,
      "learning_rate": 0.00026040660129155705,
      "loss": 0.0592,
      "step": 3264
    },
    {
      "epoch": 0.4576033637000701,
      "grad_norm": 0.7127774357795715,
      "learning_rate": 0.0002603922506577374,
      "loss": 0.085,
      "step": 3265
    },
    {
      "epoch": 0.4577435178696566,
      "grad_norm": 0.41104570031166077,
      "learning_rate": 0.0002603779000239177,
      "loss": 0.0611,
      "step": 3266
    },
    {
      "epoch": 0.4578836720392432,
      "grad_norm": 1.4193328619003296,
      "learning_rate": 0.00026036354939009803,
      "loss": 0.0796,
      "step": 3267
    },
    {
      "epoch": 0.4580238262088297,
      "grad_norm": 0.2687203884124756,
      "learning_rate": 0.00026034919875627836,
      "loss": 0.056,
      "step": 3268
    },
    {
      "epoch": 0.45816398037841627,
      "grad_norm": 0.4520364999771118,
      "learning_rate": 0.0002603348481224587,
      "loss": 0.0476,
      "step": 3269
    },
    {
      "epoch": 0.4583041345480028,
      "grad_norm": 0.4314827620983124,
      "learning_rate": 0.00026032049748863907,
      "loss": 0.1113,
      "step": 3270
    },
    {
      "epoch": 0.45844428871758935,
      "grad_norm": 0.26888686418533325,
      "learning_rate": 0.0002603061468548194,
      "loss": 0.0425,
      "step": 3271
    },
    {
      "epoch": 0.45858444288717587,
      "grad_norm": 0.4072524607181549,
      "learning_rate": 0.00026029179622099973,
      "loss": 0.07,
      "step": 3272
    },
    {
      "epoch": 0.45872459705676244,
      "grad_norm": 0.29469847679138184,
      "learning_rate": 0.0002602774455871801,
      "loss": 0.1197,
      "step": 3273
    },
    {
      "epoch": 0.458864751226349,
      "grad_norm": 0.5500627756118774,
      "learning_rate": 0.00026026309495336044,
      "loss": 0.1291,
      "step": 3274
    },
    {
      "epoch": 0.4590049053959355,
      "grad_norm": 0.3043517470359802,
      "learning_rate": 0.00026024874431954077,
      "loss": 0.048,
      "step": 3275
    },
    {
      "epoch": 0.4591450595655221,
      "grad_norm": 0.977124810218811,
      "learning_rate": 0.0002602343936857211,
      "loss": 0.0884,
      "step": 3276
    },
    {
      "epoch": 0.4592852137351086,
      "grad_norm": 0.5323735475540161,
      "learning_rate": 0.0002602200430519014,
      "loss": 0.077,
      "step": 3277
    },
    {
      "epoch": 0.4594253679046952,
      "grad_norm": 0.7924557328224182,
      "learning_rate": 0.00026020569241808175,
      "loss": 0.0493,
      "step": 3278
    },
    {
      "epoch": 0.4595655220742817,
      "grad_norm": 0.3552142381668091,
      "learning_rate": 0.00026019134178426214,
      "loss": 0.1585,
      "step": 3279
    },
    {
      "epoch": 0.45970567624386827,
      "grad_norm": 0.5548868179321289,
      "learning_rate": 0.00026017699115044247,
      "loss": 0.1511,
      "step": 3280
    },
    {
      "epoch": 0.4598458304134548,
      "grad_norm": 0.25325852632522583,
      "learning_rate": 0.0002601626405166228,
      "loss": 0.0745,
      "step": 3281
    },
    {
      "epoch": 0.45998598458304135,
      "grad_norm": 0.14024783670902252,
      "learning_rate": 0.0002601482898828031,
      "loss": 0.0118,
      "step": 3282
    },
    {
      "epoch": 0.46012613875262787,
      "grad_norm": 0.530278205871582,
      "learning_rate": 0.0002601339392489835,
      "loss": 0.0631,
      "step": 3283
    },
    {
      "epoch": 0.46026629292221444,
      "grad_norm": 0.4354083836078644,
      "learning_rate": 0.00026011958861516383,
      "loss": 0.1464,
      "step": 3284
    },
    {
      "epoch": 0.460406447091801,
      "grad_norm": 0.44976821541786194,
      "learning_rate": 0.00026010523798134416,
      "loss": 0.1149,
      "step": 3285
    },
    {
      "epoch": 0.4605466012613875,
      "grad_norm": 0.9277240037918091,
      "learning_rate": 0.0002600908873475245,
      "loss": 0.1709,
      "step": 3286
    },
    {
      "epoch": 0.4606867554309741,
      "grad_norm": 0.2960893213748932,
      "learning_rate": 0.0002600765367137048,
      "loss": 0.0679,
      "step": 3287
    },
    {
      "epoch": 0.4608269096005606,
      "grad_norm": 0.5488150119781494,
      "learning_rate": 0.00026006218607988515,
      "loss": 0.0858,
      "step": 3288
    },
    {
      "epoch": 0.4609670637701472,
      "grad_norm": 0.5408404469490051,
      "learning_rate": 0.00026004783544606553,
      "loss": 0.0467,
      "step": 3289
    },
    {
      "epoch": 0.4611072179397337,
      "grad_norm": 0.4792618155479431,
      "learning_rate": 0.00026003348481224586,
      "loss": 0.1906,
      "step": 3290
    },
    {
      "epoch": 0.46124737210932026,
      "grad_norm": 0.2402646541595459,
      "learning_rate": 0.0002600191341784262,
      "loss": 0.0609,
      "step": 3291
    },
    {
      "epoch": 0.4613875262789068,
      "grad_norm": 0.4909380078315735,
      "learning_rate": 0.0002600047835446065,
      "loss": 0.0445,
      "step": 3292
    },
    {
      "epoch": 0.46152768044849335,
      "grad_norm": 0.6875008940696716,
      "learning_rate": 0.0002599904329107869,
      "loss": 0.1174,
      "step": 3293
    },
    {
      "epoch": 0.46166783461807986,
      "grad_norm": 1.5048495531082153,
      "learning_rate": 0.00025997608227696723,
      "loss": 0.0833,
      "step": 3294
    },
    {
      "epoch": 0.46180798878766643,
      "grad_norm": 0.39396318793296814,
      "learning_rate": 0.00025996173164314756,
      "loss": 0.0897,
      "step": 3295
    },
    {
      "epoch": 0.461948142957253,
      "grad_norm": 0.6658831238746643,
      "learning_rate": 0.0002599473810093279,
      "loss": 0.2576,
      "step": 3296
    },
    {
      "epoch": 0.4620882971268395,
      "grad_norm": 0.3684833347797394,
      "learning_rate": 0.0002599330303755082,
      "loss": 0.0584,
      "step": 3297
    },
    {
      "epoch": 0.4622284512964261,
      "grad_norm": 1.6234478950500488,
      "learning_rate": 0.0002599186797416886,
      "loss": 0.2005,
      "step": 3298
    },
    {
      "epoch": 0.4623686054660126,
      "grad_norm": 0.48739907145500183,
      "learning_rate": 0.0002599043291078689,
      "loss": 0.0398,
      "step": 3299
    },
    {
      "epoch": 0.4625087596355992,
      "grad_norm": 5.181631565093994,
      "learning_rate": 0.00025988997847404925,
      "loss": 0.5623,
      "step": 3300
    },
    {
      "epoch": 0.4626489138051857,
      "grad_norm": 0.41347596049308777,
      "learning_rate": 0.0002598756278402296,
      "loss": 0.1104,
      "step": 3301
    },
    {
      "epoch": 0.46278906797477226,
      "grad_norm": 0.6285414099693298,
      "learning_rate": 0.0002598612772064099,
      "loss": 0.1059,
      "step": 3302
    },
    {
      "epoch": 0.4629292221443588,
      "grad_norm": 0.146291583776474,
      "learning_rate": 0.00025984692657259024,
      "loss": 0.0398,
      "step": 3303
    },
    {
      "epoch": 0.46306937631394535,
      "grad_norm": 0.8626413941383362,
      "learning_rate": 0.00025983257593877057,
      "loss": 0.1472,
      "step": 3304
    },
    {
      "epoch": 0.46320953048353186,
      "grad_norm": 0.34750375151634216,
      "learning_rate": 0.00025981822530495095,
      "loss": 0.1122,
      "step": 3305
    },
    {
      "epoch": 0.46334968465311843,
      "grad_norm": 0.3703271746635437,
      "learning_rate": 0.0002598038746711313,
      "loss": 0.1117,
      "step": 3306
    },
    {
      "epoch": 0.463489838822705,
      "grad_norm": 0.42451179027557373,
      "learning_rate": 0.0002597895240373116,
      "loss": 0.1377,
      "step": 3307
    },
    {
      "epoch": 0.4636299929922915,
      "grad_norm": 0.6143540143966675,
      "learning_rate": 0.000259775173403492,
      "loss": 0.1384,
      "step": 3308
    },
    {
      "epoch": 0.4637701471618781,
      "grad_norm": 0.7038992643356323,
      "learning_rate": 0.0002597608227696723,
      "loss": 0.1523,
      "step": 3309
    },
    {
      "epoch": 0.4639103013314646,
      "grad_norm": 0.23753102123737335,
      "learning_rate": 0.00025974647213585265,
      "loss": 0.0499,
      "step": 3310
    },
    {
      "epoch": 0.4640504555010512,
      "grad_norm": 0.19012613594532013,
      "learning_rate": 0.000259732121502033,
      "loss": 0.0386,
      "step": 3311
    },
    {
      "epoch": 0.4641906096706377,
      "grad_norm": 0.2870272696018219,
      "learning_rate": 0.0002597177708682133,
      "loss": 0.0762,
      "step": 3312
    },
    {
      "epoch": 0.46433076384022426,
      "grad_norm": 0.35737189650535583,
      "learning_rate": 0.00025970342023439363,
      "loss": 0.0603,
      "step": 3313
    },
    {
      "epoch": 0.4644709180098108,
      "grad_norm": 0.36687299609184265,
      "learning_rate": 0.000259689069600574,
      "loss": 0.0919,
      "step": 3314
    },
    {
      "epoch": 0.46461107217939734,
      "grad_norm": 0.5802489519119263,
      "learning_rate": 0.00025967471896675434,
      "loss": 0.1383,
      "step": 3315
    },
    {
      "epoch": 0.46475122634898386,
      "grad_norm": 0.6128777265548706,
      "learning_rate": 0.00025966036833293467,
      "loss": 0.1645,
      "step": 3316
    },
    {
      "epoch": 0.46489138051857043,
      "grad_norm": 0.3619915544986725,
      "learning_rate": 0.000259646017699115,
      "loss": 0.0884,
      "step": 3317
    },
    {
      "epoch": 0.465031534688157,
      "grad_norm": 0.5430887937545776,
      "learning_rate": 0.0002596316670652954,
      "loss": 0.1596,
      "step": 3318
    },
    {
      "epoch": 0.4651716888577435,
      "grad_norm": 0.4762561023235321,
      "learning_rate": 0.0002596173164314757,
      "loss": 0.1249,
      "step": 3319
    },
    {
      "epoch": 0.4653118430273301,
      "grad_norm": 0.39257121086120605,
      "learning_rate": 0.00025960296579765604,
      "loss": 0.0724,
      "step": 3320
    },
    {
      "epoch": 0.4654519971969166,
      "grad_norm": 0.5943988561630249,
      "learning_rate": 0.00025958861516383637,
      "loss": 0.0888,
      "step": 3321
    },
    {
      "epoch": 0.46559215136650317,
      "grad_norm": 0.34692972898483276,
      "learning_rate": 0.0002595742645300167,
      "loss": 0.1072,
      "step": 3322
    },
    {
      "epoch": 0.4657323055360897,
      "grad_norm": 0.29024553298950195,
      "learning_rate": 0.000259559913896197,
      "loss": 0.0634,
      "step": 3323
    },
    {
      "epoch": 0.46587245970567626,
      "grad_norm": 0.36643385887145996,
      "learning_rate": 0.0002595455632623774,
      "loss": 0.0785,
      "step": 3324
    },
    {
      "epoch": 0.46601261387526277,
      "grad_norm": 0.3790509104728699,
      "learning_rate": 0.00025953121262855774,
      "loss": 0.0872,
      "step": 3325
    },
    {
      "epoch": 0.46615276804484934,
      "grad_norm": 0.2536316514015198,
      "learning_rate": 0.00025951686199473807,
      "loss": 0.1019,
      "step": 3326
    },
    {
      "epoch": 0.46629292221443586,
      "grad_norm": 0.3769552409648895,
      "learning_rate": 0.00025950251136091845,
      "loss": 0.0718,
      "step": 3327
    },
    {
      "epoch": 0.4664330763840224,
      "grad_norm": 0.1781432032585144,
      "learning_rate": 0.0002594881607270988,
      "loss": 0.0298,
      "step": 3328
    },
    {
      "epoch": 0.466573230553609,
      "grad_norm": 0.35734546184539795,
      "learning_rate": 0.0002594738100932791,
      "loss": 0.0477,
      "step": 3329
    },
    {
      "epoch": 0.4667133847231955,
      "grad_norm": 1.0124000310897827,
      "learning_rate": 0.00025945945945945944,
      "loss": 0.1497,
      "step": 3330
    },
    {
      "epoch": 0.4668535388927821,
      "grad_norm": 0.1277385950088501,
      "learning_rate": 0.00025944510882563976,
      "loss": 0.0372,
      "step": 3331
    },
    {
      "epoch": 0.4669936930623686,
      "grad_norm": 0.6442874073982239,
      "learning_rate": 0.0002594307581918201,
      "loss": 0.1137,
      "step": 3332
    },
    {
      "epoch": 0.46713384723195517,
      "grad_norm": 0.3189310133457184,
      "learning_rate": 0.0002594164075580005,
      "loss": 0.0777,
      "step": 3333
    },
    {
      "epoch": 0.4672740014015417,
      "grad_norm": 0.5977842807769775,
      "learning_rate": 0.0002594020569241808,
      "loss": 0.1426,
      "step": 3334
    },
    {
      "epoch": 0.46741415557112825,
      "grad_norm": 0.3326079547405243,
      "learning_rate": 0.00025938770629036113,
      "loss": 0.0406,
      "step": 3335
    },
    {
      "epoch": 0.46755430974071477,
      "grad_norm": 0.7526765465736389,
      "learning_rate": 0.00025937335565654146,
      "loss": 0.1369,
      "step": 3336
    },
    {
      "epoch": 0.46769446391030134,
      "grad_norm": 0.8409889340400696,
      "learning_rate": 0.00025935900502272184,
      "loss": 0.1491,
      "step": 3337
    },
    {
      "epoch": 0.46783461807988785,
      "grad_norm": 0.2698783576488495,
      "learning_rate": 0.00025934465438890217,
      "loss": 0.073,
      "step": 3338
    },
    {
      "epoch": 0.4679747722494744,
      "grad_norm": 1.8809798955917358,
      "learning_rate": 0.0002593303037550825,
      "loss": 0.166,
      "step": 3339
    },
    {
      "epoch": 0.46811492641906094,
      "grad_norm": 0.35007593035697937,
      "learning_rate": 0.00025931595312126283,
      "loss": 0.0447,
      "step": 3340
    },
    {
      "epoch": 0.4682550805886475,
      "grad_norm": 0.7924603223800659,
      "learning_rate": 0.00025930160248744316,
      "loss": 0.0846,
      "step": 3341
    },
    {
      "epoch": 0.4683952347582341,
      "grad_norm": 0.6533512473106384,
      "learning_rate": 0.0002592872518536235,
      "loss": 0.1827,
      "step": 3342
    },
    {
      "epoch": 0.4685353889278206,
      "grad_norm": 0.5200726985931396,
      "learning_rate": 0.00025927290121980387,
      "loss": 0.1278,
      "step": 3343
    },
    {
      "epoch": 0.46867554309740717,
      "grad_norm": 1.2893496751785278,
      "learning_rate": 0.0002592585505859842,
      "loss": 0.2467,
      "step": 3344
    },
    {
      "epoch": 0.4688156972669937,
      "grad_norm": 1.8457653522491455,
      "learning_rate": 0.0002592441999521645,
      "loss": 0.3387,
      "step": 3345
    },
    {
      "epoch": 0.46895585143658025,
      "grad_norm": 0.6498123407363892,
      "learning_rate": 0.0002592298493183449,
      "loss": 0.1305,
      "step": 3346
    },
    {
      "epoch": 0.46909600560616677,
      "grad_norm": 1.3193391561508179,
      "learning_rate": 0.00025921549868452524,
      "loss": 0.2333,
      "step": 3347
    },
    {
      "epoch": 0.46923615977575334,
      "grad_norm": 0.5757919549942017,
      "learning_rate": 0.00025920114805070557,
      "loss": 0.091,
      "step": 3348
    },
    {
      "epoch": 0.46937631394533985,
      "grad_norm": 1.8101173639297485,
      "learning_rate": 0.0002591867974168859,
      "loss": 0.2772,
      "step": 3349
    },
    {
      "epoch": 0.4695164681149264,
      "grad_norm": 1.8984631299972534,
      "learning_rate": 0.0002591724467830662,
      "loss": 0.1928,
      "step": 3350
    },
    {
      "epoch": 0.46965662228451294,
      "grad_norm": 0.33556658029556274,
      "learning_rate": 0.00025915809614924655,
      "loss": 0.1031,
      "step": 3351
    },
    {
      "epoch": 0.4697967764540995,
      "grad_norm": 0.4323277473449707,
      "learning_rate": 0.0002591437455154269,
      "loss": 0.0994,
      "step": 3352
    },
    {
      "epoch": 0.4699369306236861,
      "grad_norm": 0.2132180631160736,
      "learning_rate": 0.00025912939488160726,
      "loss": 0.1123,
      "step": 3353
    },
    {
      "epoch": 0.4700770847932726,
      "grad_norm": 0.24832658469676971,
      "learning_rate": 0.0002591150442477876,
      "loss": 0.0786,
      "step": 3354
    },
    {
      "epoch": 0.47021723896285916,
      "grad_norm": 0.28085970878601074,
      "learning_rate": 0.0002591006936139679,
      "loss": 0.0774,
      "step": 3355
    },
    {
      "epoch": 0.4703573931324457,
      "grad_norm": 0.3220592737197876,
      "learning_rate": 0.00025908634298014825,
      "loss": 0.1013,
      "step": 3356
    },
    {
      "epoch": 0.47049754730203225,
      "grad_norm": 0.24233441054821014,
      "learning_rate": 0.00025907199234632863,
      "loss": 0.0624,
      "step": 3357
    },
    {
      "epoch": 0.47063770147161876,
      "grad_norm": 1.9785246849060059,
      "learning_rate": 0.00025905764171250896,
      "loss": 0.1313,
      "step": 3358
    },
    {
      "epoch": 0.47077785564120533,
      "grad_norm": 0.7743126153945923,
      "learning_rate": 0.0002590432910786893,
      "loss": 0.1037,
      "step": 3359
    },
    {
      "epoch": 0.47091800981079185,
      "grad_norm": 0.4001225233078003,
      "learning_rate": 0.0002590289404448696,
      "loss": 0.0623,
      "step": 3360
    },
    {
      "epoch": 0.4710581639803784,
      "grad_norm": 0.5371206998825073,
      "learning_rate": 0.00025901458981104995,
      "loss": 0.158,
      "step": 3361
    },
    {
      "epoch": 0.47119831814996493,
      "grad_norm": 0.26361533999443054,
      "learning_rate": 0.00025900023917723033,
      "loss": 0.0843,
      "step": 3362
    },
    {
      "epoch": 0.4713384723195515,
      "grad_norm": 0.4106167256832123,
      "learning_rate": 0.00025898588854341066,
      "loss": 0.1343,
      "step": 3363
    },
    {
      "epoch": 0.4714786264891381,
      "grad_norm": 0.2720377445220947,
      "learning_rate": 0.000258971537909591,
      "loss": 0.0826,
      "step": 3364
    },
    {
      "epoch": 0.4716187806587246,
      "grad_norm": 0.7385671138763428,
      "learning_rate": 0.0002589571872757713,
      "loss": 0.1856,
      "step": 3365
    },
    {
      "epoch": 0.47175893482831116,
      "grad_norm": 0.19625452160835266,
      "learning_rate": 0.00025894283664195164,
      "loss": 0.04,
      "step": 3366
    },
    {
      "epoch": 0.4718990889978977,
      "grad_norm": 0.3260664939880371,
      "learning_rate": 0.00025892848600813197,
      "loss": 0.0909,
      "step": 3367
    },
    {
      "epoch": 0.47203924316748425,
      "grad_norm": 0.9174659848213196,
      "learning_rate": 0.00025891413537431235,
      "loss": 0.0885,
      "step": 3368
    },
    {
      "epoch": 0.47217939733707076,
      "grad_norm": 0.3643873631954193,
      "learning_rate": 0.0002588997847404927,
      "loss": 0.1192,
      "step": 3369
    },
    {
      "epoch": 0.47231955150665733,
      "grad_norm": 0.29093697667121887,
      "learning_rate": 0.000258885434106673,
      "loss": 0.0406,
      "step": 3370
    },
    {
      "epoch": 0.47245970567624385,
      "grad_norm": 0.31742843985557556,
      "learning_rate": 0.00025887108347285334,
      "loss": 0.0872,
      "step": 3371
    },
    {
      "epoch": 0.4725998598458304,
      "grad_norm": 0.3487047255039215,
      "learning_rate": 0.0002588567328390337,
      "loss": 0.0323,
      "step": 3372
    },
    {
      "epoch": 0.47274001401541693,
      "grad_norm": 0.5013546943664551,
      "learning_rate": 0.00025884238220521405,
      "loss": 0.1103,
      "step": 3373
    },
    {
      "epoch": 0.4728801681850035,
      "grad_norm": 1.6304062604904175,
      "learning_rate": 0.0002588280315713944,
      "loss": 0.1782,
      "step": 3374
    },
    {
      "epoch": 0.4730203223545901,
      "grad_norm": 0.18943342566490173,
      "learning_rate": 0.0002588136809375747,
      "loss": 0.0426,
      "step": 3375
    },
    {
      "epoch": 0.4731604765241766,
      "grad_norm": 0.8060956597328186,
      "learning_rate": 0.00025879933030375504,
      "loss": 0.1552,
      "step": 3376
    },
    {
      "epoch": 0.47330063069376316,
      "grad_norm": 0.3587362468242645,
      "learning_rate": 0.00025878497966993537,
      "loss": 0.0587,
      "step": 3377
    },
    {
      "epoch": 0.4734407848633497,
      "grad_norm": 0.7555720806121826,
      "learning_rate": 0.00025877062903611575,
      "loss": 0.1788,
      "step": 3378
    },
    {
      "epoch": 0.47358093903293624,
      "grad_norm": 0.1957497000694275,
      "learning_rate": 0.0002587562784022961,
      "loss": 0.0349,
      "step": 3379
    },
    {
      "epoch": 0.47372109320252276,
      "grad_norm": 0.5665290355682373,
      "learning_rate": 0.0002587419277684764,
      "loss": 0.1349,
      "step": 3380
    },
    {
      "epoch": 0.47386124737210933,
      "grad_norm": 0.6021792888641357,
      "learning_rate": 0.0002587275771346568,
      "loss": 0.1201,
      "step": 3381
    },
    {
      "epoch": 0.47400140154169584,
      "grad_norm": 0.7500127553939819,
      "learning_rate": 0.0002587132265008371,
      "loss": 0.0763,
      "step": 3382
    },
    {
      "epoch": 0.4741415557112824,
      "grad_norm": 0.3456926643848419,
      "learning_rate": 0.00025869887586701745,
      "loss": 0.1127,
      "step": 3383
    },
    {
      "epoch": 0.47428170988086893,
      "grad_norm": 0.1884390115737915,
      "learning_rate": 0.0002586845252331978,
      "loss": 0.042,
      "step": 3384
    },
    {
      "epoch": 0.4744218640504555,
      "grad_norm": 0.582822859287262,
      "learning_rate": 0.0002586701745993781,
      "loss": 0.0935,
      "step": 3385
    },
    {
      "epoch": 0.47456201822004207,
      "grad_norm": 0.46902984380722046,
      "learning_rate": 0.00025865582396555843,
      "loss": 0.1416,
      "step": 3386
    },
    {
      "epoch": 0.4747021723896286,
      "grad_norm": 0.9830765724182129,
      "learning_rate": 0.00025864147333173876,
      "loss": 0.2058,
      "step": 3387
    },
    {
      "epoch": 0.47484232655921516,
      "grad_norm": 0.7563696503639221,
      "learning_rate": 0.00025862712269791914,
      "loss": 0.1169,
      "step": 3388
    },
    {
      "epoch": 0.47498248072880167,
      "grad_norm": 0.6189610362052917,
      "learning_rate": 0.00025861277206409947,
      "loss": 0.1398,
      "step": 3389
    },
    {
      "epoch": 0.47512263489838824,
      "grad_norm": 0.36645278334617615,
      "learning_rate": 0.0002585984214302798,
      "loss": 0.0562,
      "step": 3390
    },
    {
      "epoch": 0.47526278906797476,
      "grad_norm": 0.5037279725074768,
      "learning_rate": 0.0002585840707964602,
      "loss": 0.145,
      "step": 3391
    },
    {
      "epoch": 0.4754029432375613,
      "grad_norm": 0.7555498480796814,
      "learning_rate": 0.0002585697201626405,
      "loss": 0.1208,
      "step": 3392
    },
    {
      "epoch": 0.47554309740714784,
      "grad_norm": 0.7340646982192993,
      "learning_rate": 0.00025855536952882084,
      "loss": 0.3138,
      "step": 3393
    },
    {
      "epoch": 0.4756832515767344,
      "grad_norm": 0.6181073188781738,
      "learning_rate": 0.00025854101889500117,
      "loss": 0.3114,
      "step": 3394
    },
    {
      "epoch": 0.4758234057463209,
      "grad_norm": 0.4733201563358307,
      "learning_rate": 0.0002585266682611815,
      "loss": 0.0942,
      "step": 3395
    },
    {
      "epoch": 0.4759635599159075,
      "grad_norm": 0.7280821800231934,
      "learning_rate": 0.0002585123176273618,
      "loss": 0.1755,
      "step": 3396
    },
    {
      "epoch": 0.47610371408549407,
      "grad_norm": 1.449250340461731,
      "learning_rate": 0.0002584979669935422,
      "loss": 0.3941,
      "step": 3397
    },
    {
      "epoch": 0.4762438682550806,
      "grad_norm": 0.5744331479072571,
      "learning_rate": 0.00025848361635972254,
      "loss": 0.1021,
      "step": 3398
    },
    {
      "epoch": 0.47638402242466715,
      "grad_norm": 1.5846236944198608,
      "learning_rate": 0.00025846926572590286,
      "loss": 0.1858,
      "step": 3399
    },
    {
      "epoch": 0.47652417659425367,
      "grad_norm": 0.42692530155181885,
      "learning_rate": 0.00025845491509208325,
      "loss": 0.0537,
      "step": 3400
    },
    {
      "epoch": 0.47666433076384024,
      "grad_norm": 0.32353678345680237,
      "learning_rate": 0.0002584405644582636,
      "loss": 0.0778,
      "step": 3401
    },
    {
      "epoch": 0.47680448493342675,
      "grad_norm": 0.6987240314483643,
      "learning_rate": 0.0002584262138244439,
      "loss": 0.1612,
      "step": 3402
    },
    {
      "epoch": 0.4769446391030133,
      "grad_norm": 0.34401682019233704,
      "learning_rate": 0.00025841186319062423,
      "loss": 0.1156,
      "step": 3403
    },
    {
      "epoch": 0.47708479327259984,
      "grad_norm": 0.37349554896354675,
      "learning_rate": 0.00025839751255680456,
      "loss": 0.1284,
      "step": 3404
    },
    {
      "epoch": 0.4772249474421864,
      "grad_norm": 0.6728724837303162,
      "learning_rate": 0.0002583831619229849,
      "loss": 0.1626,
      "step": 3405
    },
    {
      "epoch": 0.4773651016117729,
      "grad_norm": 0.4453674852848053,
      "learning_rate": 0.0002583688112891652,
      "loss": 0.156,
      "step": 3406
    },
    {
      "epoch": 0.4775052557813595,
      "grad_norm": 1.1556479930877686,
      "learning_rate": 0.0002583544606553456,
      "loss": 0.1456,
      "step": 3407
    },
    {
      "epoch": 0.47764540995094606,
      "grad_norm": 0.8492551445960999,
      "learning_rate": 0.00025834011002152593,
      "loss": 0.17,
      "step": 3408
    },
    {
      "epoch": 0.4777855641205326,
      "grad_norm": 0.416910856962204,
      "learning_rate": 0.00025832575938770626,
      "loss": 0.1274,
      "step": 3409
    },
    {
      "epoch": 0.47792571829011915,
      "grad_norm": 0.5362917184829712,
      "learning_rate": 0.00025831140875388664,
      "loss": 0.1398,
      "step": 3410
    },
    {
      "epoch": 0.47806587245970567,
      "grad_norm": 0.23010285198688507,
      "learning_rate": 0.00025829705812006697,
      "loss": 0.0932,
      "step": 3411
    },
    {
      "epoch": 0.47820602662929224,
      "grad_norm": 0.42692089080810547,
      "learning_rate": 0.0002582827074862473,
      "loss": 0.0968,
      "step": 3412
    },
    {
      "epoch": 0.47834618079887875,
      "grad_norm": 0.9318661689758301,
      "learning_rate": 0.00025826835685242763,
      "loss": 0.0882,
      "step": 3413
    },
    {
      "epoch": 0.4784863349684653,
      "grad_norm": 0.3168836534023285,
      "learning_rate": 0.00025825400621860796,
      "loss": 0.0579,
      "step": 3414
    },
    {
      "epoch": 0.47862648913805184,
      "grad_norm": 0.7715312838554382,
      "learning_rate": 0.0002582396555847883,
      "loss": 0.1876,
      "step": 3415
    },
    {
      "epoch": 0.4787666433076384,
      "grad_norm": 0.8471894860267639,
      "learning_rate": 0.00025822530495096867,
      "loss": 0.1608,
      "step": 3416
    },
    {
      "epoch": 0.4789067974772249,
      "grad_norm": 0.8708312511444092,
      "learning_rate": 0.000258210954317149,
      "loss": 0.1306,
      "step": 3417
    },
    {
      "epoch": 0.4790469516468115,
      "grad_norm": 0.41459688544273376,
      "learning_rate": 0.0002581966036833293,
      "loss": 0.1796,
      "step": 3418
    },
    {
      "epoch": 0.47918710581639806,
      "grad_norm": 0.3730902671813965,
      "learning_rate": 0.00025818225304950965,
      "loss": 0.1133,
      "step": 3419
    },
    {
      "epoch": 0.4793272599859846,
      "grad_norm": 0.288470596075058,
      "learning_rate": 0.00025816790241569004,
      "loss": 0.0992,
      "step": 3420
    },
    {
      "epoch": 0.47946741415557115,
      "grad_norm": 0.3027913570404053,
      "learning_rate": 0.00025815355178187036,
      "loss": 0.0575,
      "step": 3421
    },
    {
      "epoch": 0.47960756832515766,
      "grad_norm": 0.30857786536216736,
      "learning_rate": 0.0002581392011480507,
      "loss": 0.1214,
      "step": 3422
    },
    {
      "epoch": 0.47974772249474423,
      "grad_norm": 0.35822445154190063,
      "learning_rate": 0.000258124850514231,
      "loss": 0.0801,
      "step": 3423
    },
    {
      "epoch": 0.47988787666433075,
      "grad_norm": 0.3174699544906616,
      "learning_rate": 0.00025811049988041135,
      "loss": 0.0681,
      "step": 3424
    },
    {
      "epoch": 0.4800280308339173,
      "grad_norm": 0.3112591505050659,
      "learning_rate": 0.0002580961492465917,
      "loss": 0.1038,
      "step": 3425
    },
    {
      "epoch": 0.48016818500350383,
      "grad_norm": 0.7893008589744568,
      "learning_rate": 0.00025808179861277206,
      "loss": 0.1985,
      "step": 3426
    },
    {
      "epoch": 0.4803083391730904,
      "grad_norm": 0.49721333384513855,
      "learning_rate": 0.0002580674479789524,
      "loss": 0.1054,
      "step": 3427
    },
    {
      "epoch": 0.4804484933426769,
      "grad_norm": 0.6634990572929382,
      "learning_rate": 0.0002580530973451327,
      "loss": 0.1543,
      "step": 3428
    },
    {
      "epoch": 0.4805886475122635,
      "grad_norm": 0.7477396130561829,
      "learning_rate": 0.00025803874671131305,
      "loss": 0.1571,
      "step": 3429
    },
    {
      "epoch": 0.48072880168185006,
      "grad_norm": 0.397464394569397,
      "learning_rate": 0.0002580243960774934,
      "loss": 0.1483,
      "step": 3430
    },
    {
      "epoch": 0.4808689558514366,
      "grad_norm": 0.5868679881095886,
      "learning_rate": 0.0002580100454436737,
      "loss": 0.1224,
      "step": 3431
    },
    {
      "epoch": 0.48100911002102315,
      "grad_norm": 0.3429774045944214,
      "learning_rate": 0.0002579956948098541,
      "loss": 0.0623,
      "step": 3432
    },
    {
      "epoch": 0.48114926419060966,
      "grad_norm": 0.5995844602584839,
      "learning_rate": 0.0002579813441760344,
      "loss": 0.1672,
      "step": 3433
    },
    {
      "epoch": 0.48128941836019623,
      "grad_norm": 0.21627332270145416,
      "learning_rate": 0.00025796699354221474,
      "loss": 0.0277,
      "step": 3434
    },
    {
      "epoch": 0.48142957252978275,
      "grad_norm": 0.3642168939113617,
      "learning_rate": 0.0002579526429083951,
      "loss": 0.0414,
      "step": 3435
    },
    {
      "epoch": 0.4815697266993693,
      "grad_norm": 0.6434520483016968,
      "learning_rate": 0.00025793829227457546,
      "loss": 0.039,
      "step": 3436
    },
    {
      "epoch": 0.48170988086895583,
      "grad_norm": 0.1669931262731552,
      "learning_rate": 0.0002579239416407558,
      "loss": 0.0321,
      "step": 3437
    },
    {
      "epoch": 0.4818500350385424,
      "grad_norm": 0.3662303388118744,
      "learning_rate": 0.0002579095910069361,
      "loss": 0.1002,
      "step": 3438
    },
    {
      "epoch": 0.4819901892081289,
      "grad_norm": 0.9763455986976624,
      "learning_rate": 0.00025789524037311644,
      "loss": 0.1275,
      "step": 3439
    },
    {
      "epoch": 0.4821303433777155,
      "grad_norm": 0.7734778523445129,
      "learning_rate": 0.00025788088973929677,
      "loss": 0.2105,
      "step": 3440
    },
    {
      "epoch": 0.48227049754730206,
      "grad_norm": 0.6286949515342712,
      "learning_rate": 0.0002578665391054771,
      "loss": 0.0695,
      "step": 3441
    },
    {
      "epoch": 0.48241065171688857,
      "grad_norm": 0.19085323810577393,
      "learning_rate": 0.0002578521884716575,
      "loss": 0.0724,
      "step": 3442
    },
    {
      "epoch": 0.48255080588647514,
      "grad_norm": 0.78294837474823,
      "learning_rate": 0.0002578378378378378,
      "loss": 0.1144,
      "step": 3443
    },
    {
      "epoch": 0.48269096005606166,
      "grad_norm": 0.39815643429756165,
      "learning_rate": 0.00025782348720401814,
      "loss": 0.0887,
      "step": 3444
    },
    {
      "epoch": 0.48283111422564823,
      "grad_norm": 0.9066981077194214,
      "learning_rate": 0.0002578091365701985,
      "loss": 0.2532,
      "step": 3445
    },
    {
      "epoch": 0.48297126839523474,
      "grad_norm": 0.7563006281852722,
      "learning_rate": 0.00025779478593637885,
      "loss": 0.105,
      "step": 3446
    },
    {
      "epoch": 0.4831114225648213,
      "grad_norm": 0.5843519568443298,
      "learning_rate": 0.0002577804353025592,
      "loss": 0.2324,
      "step": 3447
    },
    {
      "epoch": 0.48325157673440783,
      "grad_norm": 1.0399240255355835,
      "learning_rate": 0.0002577660846687395,
      "loss": 0.1119,
      "step": 3448
    },
    {
      "epoch": 0.4833917309039944,
      "grad_norm": 1.2726222276687622,
      "learning_rate": 0.00025775173403491983,
      "loss": 0.1689,
      "step": 3449
    },
    {
      "epoch": 0.4835318850735809,
      "grad_norm": 1.1648828983306885,
      "learning_rate": 0.00025773738340110016,
      "loss": 0.0367,
      "step": 3450
    },
    {
      "epoch": 0.4836720392431675,
      "grad_norm": 0.4012549817562103,
      "learning_rate": 0.00025772303276728055,
      "loss": 0.1036,
      "step": 3451
    },
    {
      "epoch": 0.48381219341275405,
      "grad_norm": 0.4353487193584442,
      "learning_rate": 0.0002577086821334609,
      "loss": 0.1004,
      "step": 3452
    },
    {
      "epoch": 0.48395234758234057,
      "grad_norm": 0.7935703992843628,
      "learning_rate": 0.0002576943314996412,
      "loss": 0.0752,
      "step": 3453
    },
    {
      "epoch": 0.48409250175192714,
      "grad_norm": 0.5132361054420471,
      "learning_rate": 0.0002576799808658216,
      "loss": 0.092,
      "step": 3454
    },
    {
      "epoch": 0.48423265592151365,
      "grad_norm": 0.6994789838790894,
      "learning_rate": 0.0002576656302320019,
      "loss": 0.1422,
      "step": 3455
    },
    {
      "epoch": 0.4843728100911002,
      "grad_norm": 0.9248461127281189,
      "learning_rate": 0.00025765127959818224,
      "loss": 0.0677,
      "step": 3456
    },
    {
      "epoch": 0.48451296426068674,
      "grad_norm": 0.5581263303756714,
      "learning_rate": 0.00025763692896436257,
      "loss": 0.1042,
      "step": 3457
    },
    {
      "epoch": 0.4846531184302733,
      "grad_norm": 0.31875401735305786,
      "learning_rate": 0.0002576225783305429,
      "loss": 0.0621,
      "step": 3458
    },
    {
      "epoch": 0.4847932725998598,
      "grad_norm": 0.3572028577327728,
      "learning_rate": 0.00025760822769672323,
      "loss": 0.0955,
      "step": 3459
    },
    {
      "epoch": 0.4849334267694464,
      "grad_norm": 0.5527826547622681,
      "learning_rate": 0.00025759387706290356,
      "loss": 0.1427,
      "step": 3460
    },
    {
      "epoch": 0.4850735809390329,
      "grad_norm": 0.8580110669136047,
      "learning_rate": 0.00025757952642908394,
      "loss": 0.1618,
      "step": 3461
    },
    {
      "epoch": 0.4852137351086195,
      "grad_norm": 0.8033840656280518,
      "learning_rate": 0.00025756517579526427,
      "loss": 0.1871,
      "step": 3462
    },
    {
      "epoch": 0.48535388927820605,
      "grad_norm": 0.3018549680709839,
      "learning_rate": 0.0002575508251614446,
      "loss": 0.0674,
      "step": 3463
    },
    {
      "epoch": 0.48549404344779257,
      "grad_norm": 0.7677834033966064,
      "learning_rate": 0.000257536474527625,
      "loss": 0.0978,
      "step": 3464
    },
    {
      "epoch": 0.48563419761737914,
      "grad_norm": 0.13359415531158447,
      "learning_rate": 0.0002575221238938053,
      "loss": 0.0236,
      "step": 3465
    },
    {
      "epoch": 0.48577435178696565,
      "grad_norm": 0.4001922607421875,
      "learning_rate": 0.00025750777325998564,
      "loss": 0.1413,
      "step": 3466
    },
    {
      "epoch": 0.4859145059565522,
      "grad_norm": 0.4840030074119568,
      "learning_rate": 0.00025749342262616597,
      "loss": 0.1417,
      "step": 3467
    },
    {
      "epoch": 0.48605466012613874,
      "grad_norm": 0.5108070373535156,
      "learning_rate": 0.0002574790719923463,
      "loss": 0.1163,
      "step": 3468
    },
    {
      "epoch": 0.4861948142957253,
      "grad_norm": 0.5165778994560242,
      "learning_rate": 0.0002574647213585266,
      "loss": 0.1082,
      "step": 3469
    },
    {
      "epoch": 0.4863349684653118,
      "grad_norm": 0.37278759479522705,
      "learning_rate": 0.000257450370724707,
      "loss": 0.1251,
      "step": 3470
    },
    {
      "epoch": 0.4864751226348984,
      "grad_norm": 0.40070751309394836,
      "learning_rate": 0.00025743602009088733,
      "loss": 0.0892,
      "step": 3471
    },
    {
      "epoch": 0.4866152768044849,
      "grad_norm": 0.19404344260692596,
      "learning_rate": 0.00025742166945706766,
      "loss": 0.0361,
      "step": 3472
    },
    {
      "epoch": 0.4867554309740715,
      "grad_norm": 1.0165289640426636,
      "learning_rate": 0.000257407318823248,
      "loss": 0.1069,
      "step": 3473
    },
    {
      "epoch": 0.48689558514365805,
      "grad_norm": 0.36197417974472046,
      "learning_rate": 0.0002573929681894284,
      "loss": 0.1036,
      "step": 3474
    },
    {
      "epoch": 0.48703573931324456,
      "grad_norm": 0.550611674785614,
      "learning_rate": 0.0002573786175556087,
      "loss": 0.098,
      "step": 3475
    },
    {
      "epoch": 0.48717589348283113,
      "grad_norm": 0.6606674790382385,
      "learning_rate": 0.00025736426692178903,
      "loss": 0.093,
      "step": 3476
    },
    {
      "epoch": 0.48731604765241765,
      "grad_norm": 0.6661083102226257,
      "learning_rate": 0.00025734991628796936,
      "loss": 0.0694,
      "step": 3477
    },
    {
      "epoch": 0.4874562018220042,
      "grad_norm": 0.6594652533531189,
      "learning_rate": 0.0002573355656541497,
      "loss": 0.0879,
      "step": 3478
    },
    {
      "epoch": 0.48759635599159074,
      "grad_norm": 0.24664607644081116,
      "learning_rate": 0.00025732121502033,
      "loss": 0.0616,
      "step": 3479
    },
    {
      "epoch": 0.4877365101611773,
      "grad_norm": 0.4766852557659149,
      "learning_rate": 0.0002573068643865104,
      "loss": 0.0827,
      "step": 3480
    },
    {
      "epoch": 0.4878766643307638,
      "grad_norm": 0.6749598383903503,
      "learning_rate": 0.00025729251375269073,
      "loss": 0.141,
      "step": 3481
    },
    {
      "epoch": 0.4880168185003504,
      "grad_norm": 0.9280956983566284,
      "learning_rate": 0.00025727816311887106,
      "loss": 0.0943,
      "step": 3482
    },
    {
      "epoch": 0.4881569726699369,
      "grad_norm": 0.3066399097442627,
      "learning_rate": 0.00025726381248505144,
      "loss": 0.058,
      "step": 3483
    },
    {
      "epoch": 0.4882971268395235,
      "grad_norm": 0.3830578327178955,
      "learning_rate": 0.00025724946185123177,
      "loss": 0.093,
      "step": 3484
    },
    {
      "epoch": 0.48843728100911005,
      "grad_norm": 0.3635070323944092,
      "learning_rate": 0.0002572351112174121,
      "loss": 0.0965,
      "step": 3485
    },
    {
      "epoch": 0.48857743517869656,
      "grad_norm": 2.309699773788452,
      "learning_rate": 0.0002572207605835924,
      "loss": 0.1327,
      "step": 3486
    },
    {
      "epoch": 0.48871758934828313,
      "grad_norm": 0.6561049222946167,
      "learning_rate": 0.00025720640994977275,
      "loss": 0.1025,
      "step": 3487
    },
    {
      "epoch": 0.48885774351786965,
      "grad_norm": 0.3517068028450012,
      "learning_rate": 0.0002571920593159531,
      "loss": 0.0402,
      "step": 3488
    },
    {
      "epoch": 0.4889978976874562,
      "grad_norm": 0.3044321537017822,
      "learning_rate": 0.00025717770868213347,
      "loss": 0.0799,
      "step": 3489
    },
    {
      "epoch": 0.48913805185704273,
      "grad_norm": 0.36944735050201416,
      "learning_rate": 0.0002571633580483138,
      "loss": 0.1074,
      "step": 3490
    },
    {
      "epoch": 0.4892782060266293,
      "grad_norm": 1.395851731300354,
      "learning_rate": 0.0002571490074144941,
      "loss": 0.2275,
      "step": 3491
    },
    {
      "epoch": 0.4894183601962158,
      "grad_norm": 1.0201665163040161,
      "learning_rate": 0.00025713465678067445,
      "loss": 0.2468,
      "step": 3492
    },
    {
      "epoch": 0.4895585143658024,
      "grad_norm": 1.5163003206253052,
      "learning_rate": 0.0002571203061468548,
      "loss": 0.2178,
      "step": 3493
    },
    {
      "epoch": 0.4896986685353889,
      "grad_norm": 1.0206705331802368,
      "learning_rate": 0.0002571059555130351,
      "loss": 0.2149,
      "step": 3494
    },
    {
      "epoch": 0.4898388227049755,
      "grad_norm": 0.3390774130821228,
      "learning_rate": 0.00025709160487921544,
      "loss": 0.0213,
      "step": 3495
    },
    {
      "epoch": 0.48997897687456204,
      "grad_norm": 2.1452643871307373,
      "learning_rate": 0.0002570772542453958,
      "loss": 0.1764,
      "step": 3496
    },
    {
      "epoch": 0.49011913104414856,
      "grad_norm": 2.654993772506714,
      "learning_rate": 0.00025706290361157615,
      "loss": 0.4339,
      "step": 3497
    },
    {
      "epoch": 0.49025928521373513,
      "grad_norm": 0.6839194893836975,
      "learning_rate": 0.0002570485529777565,
      "loss": 0.1325,
      "step": 3498
    },
    {
      "epoch": 0.49039943938332164,
      "grad_norm": 6.174987316131592,
      "learning_rate": 0.00025703420234393686,
      "loss": 0.4508,
      "step": 3499
    },
    {
      "epoch": 0.4905395935529082,
      "grad_norm": 8.566628456115723,
      "learning_rate": 0.0002570198517101172,
      "loss": 0.325,
      "step": 3500
    },
    {
      "epoch": 0.49067974772249473,
      "grad_norm": 0.1667787879705429,
      "learning_rate": 0.0002570055010762975,
      "loss": 0.0534,
      "step": 3501
    },
    {
      "epoch": 0.4908199018920813,
      "grad_norm": 0.30611181259155273,
      "learning_rate": 0.00025699115044247784,
      "loss": 0.0828,
      "step": 3502
    },
    {
      "epoch": 0.4909600560616678,
      "grad_norm": 0.3280355632305145,
      "learning_rate": 0.0002569767998086582,
      "loss": 0.0964,
      "step": 3503
    },
    {
      "epoch": 0.4911002102312544,
      "grad_norm": 0.1987239569425583,
      "learning_rate": 0.0002569624491748385,
      "loss": 0.0428,
      "step": 3504
    },
    {
      "epoch": 0.4912403644008409,
      "grad_norm": 0.31350991129875183,
      "learning_rate": 0.0002569480985410189,
      "loss": 0.1097,
      "step": 3505
    },
    {
      "epoch": 0.49138051857042747,
      "grad_norm": 0.3353220224380493,
      "learning_rate": 0.0002569337479071992,
      "loss": 0.1157,
      "step": 3506
    },
    {
      "epoch": 0.49152067274001404,
      "grad_norm": 0.7625632286071777,
      "learning_rate": 0.00025691939727337954,
      "loss": 0.1397,
      "step": 3507
    },
    {
      "epoch": 0.49166082690960056,
      "grad_norm": 0.27706798911094666,
      "learning_rate": 0.00025690504663955987,
      "loss": 0.0687,
      "step": 3508
    },
    {
      "epoch": 0.4918009810791871,
      "grad_norm": 0.20815348625183105,
      "learning_rate": 0.00025689069600574025,
      "loss": 0.0612,
      "step": 3509
    },
    {
      "epoch": 0.49194113524877364,
      "grad_norm": 0.7141574621200562,
      "learning_rate": 0.0002568763453719206,
      "loss": 0.1223,
      "step": 3510
    },
    {
      "epoch": 0.4920812894183602,
      "grad_norm": 0.5768548250198364,
      "learning_rate": 0.0002568619947381009,
      "loss": 0.1378,
      "step": 3511
    },
    {
      "epoch": 0.4922214435879467,
      "grad_norm": 0.40852946043014526,
      "learning_rate": 0.00025684764410428124,
      "loss": 0.1122,
      "step": 3512
    },
    {
      "epoch": 0.4923615977575333,
      "grad_norm": 0.3819160759449005,
      "learning_rate": 0.00025683329347046157,
      "loss": 0.1262,
      "step": 3513
    },
    {
      "epoch": 0.4925017519271198,
      "grad_norm": 0.48255541920661926,
      "learning_rate": 0.0002568189428366419,
      "loss": 0.1082,
      "step": 3514
    },
    {
      "epoch": 0.4926419060967064,
      "grad_norm": 0.6783628463745117,
      "learning_rate": 0.0002568045922028223,
      "loss": 0.1658,
      "step": 3515
    },
    {
      "epoch": 0.4927820602662929,
      "grad_norm": 0.43569812178611755,
      "learning_rate": 0.0002567902415690026,
      "loss": 0.1546,
      "step": 3516
    },
    {
      "epoch": 0.49292221443587947,
      "grad_norm": 0.5011049509048462,
      "learning_rate": 0.00025677589093518294,
      "loss": 0.0901,
      "step": 3517
    },
    {
      "epoch": 0.49306236860546604,
      "grad_norm": 0.22206604480743408,
      "learning_rate": 0.0002567615403013633,
      "loss": 0.0677,
      "step": 3518
    },
    {
      "epoch": 0.49320252277505255,
      "grad_norm": 0.5132867693901062,
      "learning_rate": 0.00025674718966754365,
      "loss": 0.133,
      "step": 3519
    },
    {
      "epoch": 0.4933426769446391,
      "grad_norm": 0.7029857635498047,
      "learning_rate": 0.000256732839033724,
      "loss": 0.1019,
      "step": 3520
    },
    {
      "epoch": 0.49348283111422564,
      "grad_norm": 0.41205543279647827,
      "learning_rate": 0.0002567184883999043,
      "loss": 0.0791,
      "step": 3521
    },
    {
      "epoch": 0.4936229852838122,
      "grad_norm": 0.24129436910152435,
      "learning_rate": 0.00025670413776608463,
      "loss": 0.0907,
      "step": 3522
    },
    {
      "epoch": 0.4937631394533987,
      "grad_norm": 0.5285700559616089,
      "learning_rate": 0.00025668978713226496,
      "loss": 0.0735,
      "step": 3523
    },
    {
      "epoch": 0.4939032936229853,
      "grad_norm": 0.5766040086746216,
      "learning_rate": 0.00025667543649844534,
      "loss": 0.1658,
      "step": 3524
    },
    {
      "epoch": 0.4940434477925718,
      "grad_norm": 0.43381479382514954,
      "learning_rate": 0.00025666108586462567,
      "loss": 0.0587,
      "step": 3525
    },
    {
      "epoch": 0.4941836019621584,
      "grad_norm": 0.27597546577453613,
      "learning_rate": 0.000256646735230806,
      "loss": 0.1155,
      "step": 3526
    },
    {
      "epoch": 0.4943237561317449,
      "grad_norm": 0.5752700567245483,
      "learning_rate": 0.00025663238459698633,
      "loss": 0.1109,
      "step": 3527
    },
    {
      "epoch": 0.49446391030133147,
      "grad_norm": 0.3988467752933502,
      "learning_rate": 0.0002566180339631667,
      "loss": 0.0554,
      "step": 3528
    },
    {
      "epoch": 0.49460406447091804,
      "grad_norm": 0.7903053164482117,
      "learning_rate": 0.00025660368332934704,
      "loss": 0.1161,
      "step": 3529
    },
    {
      "epoch": 0.49474421864050455,
      "grad_norm": 0.5980937480926514,
      "learning_rate": 0.00025658933269552737,
      "loss": 0.1469,
      "step": 3530
    },
    {
      "epoch": 0.4948843728100911,
      "grad_norm": 0.5819936394691467,
      "learning_rate": 0.0002565749820617077,
      "loss": 0.1056,
      "step": 3531
    },
    {
      "epoch": 0.49502452697967764,
      "grad_norm": 0.8243651986122131,
      "learning_rate": 0.000256560631427888,
      "loss": 0.1388,
      "step": 3532
    },
    {
      "epoch": 0.4951646811492642,
      "grad_norm": 1.1008387804031372,
      "learning_rate": 0.00025654628079406836,
      "loss": 0.0849,
      "step": 3533
    },
    {
      "epoch": 0.4953048353188507,
      "grad_norm": 0.3780296742916107,
      "learning_rate": 0.00025653193016024874,
      "loss": 0.1,
      "step": 3534
    },
    {
      "epoch": 0.4954449894884373,
      "grad_norm": 0.351131796836853,
      "learning_rate": 0.00025651757952642907,
      "loss": 0.1352,
      "step": 3535
    },
    {
      "epoch": 0.4955851436580238,
      "grad_norm": 0.20896701514720917,
      "learning_rate": 0.0002565032288926094,
      "loss": 0.0608,
      "step": 3536
    },
    {
      "epoch": 0.4957252978276104,
      "grad_norm": 0.19262166321277618,
      "learning_rate": 0.0002564888782587898,
      "loss": 0.033,
      "step": 3537
    },
    {
      "epoch": 0.4958654519971969,
      "grad_norm": 0.6666895151138306,
      "learning_rate": 0.0002564745276249701,
      "loss": 0.1253,
      "step": 3538
    },
    {
      "epoch": 0.49600560616678346,
      "grad_norm": 0.5336970686912537,
      "learning_rate": 0.00025646017699115044,
      "loss": 0.2133,
      "step": 3539
    },
    {
      "epoch": 0.49614576033637003,
      "grad_norm": 0.5073505640029907,
      "learning_rate": 0.00025644582635733076,
      "loss": 0.0923,
      "step": 3540
    },
    {
      "epoch": 0.49628591450595655,
      "grad_norm": 0.29231294989585876,
      "learning_rate": 0.0002564314757235111,
      "loss": 0.0425,
      "step": 3541
    },
    {
      "epoch": 0.4964260686755431,
      "grad_norm": 0.30701613426208496,
      "learning_rate": 0.0002564171250896914,
      "loss": 0.1324,
      "step": 3542
    },
    {
      "epoch": 0.49656622284512963,
      "grad_norm": 0.6268811821937561,
      "learning_rate": 0.00025640277445587175,
      "loss": 0.1805,
      "step": 3543
    },
    {
      "epoch": 0.4967063770147162,
      "grad_norm": 0.4305824935436249,
      "learning_rate": 0.00025638842382205213,
      "loss": 0.047,
      "step": 3544
    },
    {
      "epoch": 0.4968465311843027,
      "grad_norm": 0.5213370323181152,
      "learning_rate": 0.00025637407318823246,
      "loss": 0.0772,
      "step": 3545
    },
    {
      "epoch": 0.4969866853538893,
      "grad_norm": 1.497339129447937,
      "learning_rate": 0.0002563597225544128,
      "loss": 0.2284,
      "step": 3546
    },
    {
      "epoch": 0.4971268395234758,
      "grad_norm": 1.0039198398590088,
      "learning_rate": 0.00025634537192059317,
      "loss": 0.1747,
      "step": 3547
    },
    {
      "epoch": 0.4972669936930624,
      "grad_norm": 2.4202349185943604,
      "learning_rate": 0.0002563310212867735,
      "loss": 0.2279,
      "step": 3548
    },
    {
      "epoch": 0.4974071478626489,
      "grad_norm": 1.3376708030700684,
      "learning_rate": 0.00025631667065295383,
      "loss": 0.2251,
      "step": 3549
    },
    {
      "epoch": 0.49754730203223546,
      "grad_norm": 0.34456321597099304,
      "learning_rate": 0.00025630232001913416,
      "loss": 0.0474,
      "step": 3550
    },
    {
      "epoch": 0.49768745620182203,
      "grad_norm": 0.4071559011936188,
      "learning_rate": 0.0002562879693853145,
      "loss": 0.063,
      "step": 3551
    },
    {
      "epoch": 0.49782761037140855,
      "grad_norm": 0.25602662563323975,
      "learning_rate": 0.0002562736187514948,
      "loss": 0.0572,
      "step": 3552
    },
    {
      "epoch": 0.4979677645409951,
      "grad_norm": 0.3327733278274536,
      "learning_rate": 0.0002562592681176752,
      "loss": 0.0515,
      "step": 3553
    },
    {
      "epoch": 0.49810791871058163,
      "grad_norm": 0.6401887536048889,
      "learning_rate": 0.0002562449174838555,
      "loss": 0.1129,
      "step": 3554
    },
    {
      "epoch": 0.4982480728801682,
      "grad_norm": 0.37373971939086914,
      "learning_rate": 0.00025623056685003585,
      "loss": 0.0688,
      "step": 3555
    },
    {
      "epoch": 0.4983882270497547,
      "grad_norm": 0.745818018913269,
      "learning_rate": 0.0002562162162162162,
      "loss": 0.0769,
      "step": 3556
    },
    {
      "epoch": 0.4985283812193413,
      "grad_norm": 0.5485659837722778,
      "learning_rate": 0.0002562018655823965,
      "loss": 0.095,
      "step": 3557
    },
    {
      "epoch": 0.4986685353889278,
      "grad_norm": 0.5274074077606201,
      "learning_rate": 0.00025618751494857684,
      "loss": 0.0817,
      "step": 3558
    },
    {
      "epoch": 0.4988086895585144,
      "grad_norm": 0.4357086420059204,
      "learning_rate": 0.0002561731643147572,
      "loss": 0.0887,
      "step": 3559
    },
    {
      "epoch": 0.4989488437281009,
      "grad_norm": 0.365132212638855,
      "learning_rate": 0.00025615881368093755,
      "loss": 0.0797,
      "step": 3560
    },
    {
      "epoch": 0.49908899789768746,
      "grad_norm": 0.305817186832428,
      "learning_rate": 0.0002561444630471179,
      "loss": 0.0603,
      "step": 3561
    },
    {
      "epoch": 0.49922915206727403,
      "grad_norm": 0.8316388130187988,
      "learning_rate": 0.0002561301124132982,
      "loss": 0.1481,
      "step": 3562
    },
    {
      "epoch": 0.49936930623686054,
      "grad_norm": 0.30767759680747986,
      "learning_rate": 0.0002561157617794786,
      "loss": 0.0433,
      "step": 3563
    },
    {
      "epoch": 0.4995094604064471,
      "grad_norm": 0.47065308690071106,
      "learning_rate": 0.0002561014111456589,
      "loss": 0.0997,
      "step": 3564
    },
    {
      "epoch": 0.49964961457603363,
      "grad_norm": 0.5954514741897583,
      "learning_rate": 0.00025608706051183925,
      "loss": 0.1261,
      "step": 3565
    },
    {
      "epoch": 0.4997897687456202,
      "grad_norm": 0.21949557960033417,
      "learning_rate": 0.0002560727098780196,
      "loss": 0.0429,
      "step": 3566
    },
    {
      "epoch": 0.4999299229152067,
      "grad_norm": 0.4462909400463104,
      "learning_rate": 0.0002560583592441999,
      "loss": 0.0542,
      "step": 3567
    },
    {
      "epoch": 0.5000700770847932,
      "grad_norm": 0.5338721871376038,
      "learning_rate": 0.00025604400861038023,
      "loss": 0.1435,
      "step": 3568
    },
    {
      "epoch": 0.5002102312543798,
      "grad_norm": 1.80253005027771,
      "learning_rate": 0.0002560296579765606,
      "loss": 0.1108,
      "step": 3569
    },
    {
      "epoch": 0.5003503854239664,
      "grad_norm": 0.5433024168014526,
      "learning_rate": 0.00025601530734274095,
      "loss": 0.1331,
      "step": 3570
    },
    {
      "epoch": 0.5004905395935529,
      "grad_norm": 0.5902112126350403,
      "learning_rate": 0.0002560009567089213,
      "loss": 0.0402,
      "step": 3571
    },
    {
      "epoch": 0.5006306937631394,
      "grad_norm": 0.4077446758747101,
      "learning_rate": 0.00025598660607510166,
      "loss": 0.0749,
      "step": 3572
    },
    {
      "epoch": 0.500770847932726,
      "grad_norm": 0.7763643264770508,
      "learning_rate": 0.000255972255441282,
      "loss": 0.0686,
      "step": 3573
    },
    {
      "epoch": 0.5009110021023125,
      "grad_norm": 0.5264174342155457,
      "learning_rate": 0.0002559579048074623,
      "loss": 0.1041,
      "step": 3574
    },
    {
      "epoch": 0.5010511562718991,
      "grad_norm": 0.5348429083824158,
      "learning_rate": 0.00025594355417364264,
      "loss": 0.1276,
      "step": 3575
    },
    {
      "epoch": 0.5011913104414857,
      "grad_norm": 0.15242506563663483,
      "learning_rate": 0.00025592920353982297,
      "loss": 0.0361,
      "step": 3576
    },
    {
      "epoch": 0.5013314646110721,
      "grad_norm": 0.3991434872150421,
      "learning_rate": 0.0002559148529060033,
      "loss": 0.075,
      "step": 3577
    },
    {
      "epoch": 0.5014716187806587,
      "grad_norm": 0.29129812121391296,
      "learning_rate": 0.00025590050227218363,
      "loss": 0.0662,
      "step": 3578
    },
    {
      "epoch": 0.5016117729502453,
      "grad_norm": 0.2888050079345703,
      "learning_rate": 0.000255886151638364,
      "loss": 0.0913,
      "step": 3579
    },
    {
      "epoch": 0.5017519271198319,
      "grad_norm": 0.5633611083030701,
      "learning_rate": 0.00025587180100454434,
      "loss": 0.064,
      "step": 3580
    },
    {
      "epoch": 0.5018920812894183,
      "grad_norm": 0.620185911655426,
      "learning_rate": 0.00025585745037072467,
      "loss": 0.1173,
      "step": 3581
    },
    {
      "epoch": 0.5020322354590049,
      "grad_norm": 0.3774522542953491,
      "learning_rate": 0.00025584309973690505,
      "loss": 0.1091,
      "step": 3582
    },
    {
      "epoch": 0.5021723896285915,
      "grad_norm": 0.610437273979187,
      "learning_rate": 0.0002558287491030854,
      "loss": 0.1707,
      "step": 3583
    },
    {
      "epoch": 0.502312543798178,
      "grad_norm": 0.7433859705924988,
      "learning_rate": 0.0002558143984692657,
      "loss": 0.168,
      "step": 3584
    },
    {
      "epoch": 0.5024526979677646,
      "grad_norm": 0.37567129731178284,
      "learning_rate": 0.00025580004783544604,
      "loss": 0.1214,
      "step": 3585
    },
    {
      "epoch": 0.502592852137351,
      "grad_norm": 0.504042387008667,
      "learning_rate": 0.00025578569720162637,
      "loss": 0.1119,
      "step": 3586
    },
    {
      "epoch": 0.5027330063069376,
      "grad_norm": 0.6103062629699707,
      "learning_rate": 0.0002557713465678067,
      "loss": 0.0884,
      "step": 3587
    },
    {
      "epoch": 0.5028731604765242,
      "grad_norm": 0.7486506104469299,
      "learning_rate": 0.0002557569959339871,
      "loss": 0.094,
      "step": 3588
    },
    {
      "epoch": 0.5030133146461108,
      "grad_norm": 0.6592079401016235,
      "learning_rate": 0.0002557426453001674,
      "loss": 0.0779,
      "step": 3589
    },
    {
      "epoch": 0.5031534688156972,
      "grad_norm": 0.3918934762477875,
      "learning_rate": 0.00025572829466634773,
      "loss": 0.0587,
      "step": 3590
    },
    {
      "epoch": 0.5032936229852838,
      "grad_norm": 0.38877201080322266,
      "learning_rate": 0.0002557139440325281,
      "loss": 0.0316,
      "step": 3591
    },
    {
      "epoch": 0.5034337771548704,
      "grad_norm": 0.1634371280670166,
      "learning_rate": 0.00025569959339870845,
      "loss": 0.014,
      "step": 3592
    },
    {
      "epoch": 0.5035739313244569,
      "grad_norm": 0.8864071369171143,
      "learning_rate": 0.0002556852427648888,
      "loss": 0.1071,
      "step": 3593
    },
    {
      "epoch": 0.5037140854940434,
      "grad_norm": 0.34695908427238464,
      "learning_rate": 0.0002556708921310691,
      "loss": 0.0266,
      "step": 3594
    },
    {
      "epoch": 0.50385423966363,
      "grad_norm": 0.4481075406074524,
      "learning_rate": 0.00025565654149724943,
      "loss": 0.0984,
      "step": 3595
    },
    {
      "epoch": 0.5039943938332165,
      "grad_norm": 1.075195550918579,
      "learning_rate": 0.00025564219086342976,
      "loss": 0.1276,
      "step": 3596
    },
    {
      "epoch": 0.5041345480028031,
      "grad_norm": 0.3739490211009979,
      "learning_rate": 0.0002556278402296101,
      "loss": 0.0578,
      "step": 3597
    },
    {
      "epoch": 0.5042747021723897,
      "grad_norm": 0.6815117001533508,
      "learning_rate": 0.00025561348959579047,
      "loss": 0.0491,
      "step": 3598
    },
    {
      "epoch": 0.5044148563419761,
      "grad_norm": 1.9127668142318726,
      "learning_rate": 0.0002555991389619708,
      "loss": 0.2415,
      "step": 3599
    },
    {
      "epoch": 0.5045550105115627,
      "grad_norm": 2.4393773078918457,
      "learning_rate": 0.00025558478832815113,
      "loss": 0.2747,
      "step": 3600
    },
    {
      "epoch": 0.5046951646811493,
      "grad_norm": 0.24543632566928864,
      "learning_rate": 0.0002555704376943315,
      "loss": 0.0723,
      "step": 3601
    },
    {
      "epoch": 0.5048353188507358,
      "grad_norm": 0.2963375449180603,
      "learning_rate": 0.00025555608706051184,
      "loss": 0.0776,
      "step": 3602
    },
    {
      "epoch": 0.5049754730203223,
      "grad_norm": 0.36004024744033813,
      "learning_rate": 0.00025554173642669217,
      "loss": 0.0857,
      "step": 3603
    },
    {
      "epoch": 0.5051156271899089,
      "grad_norm": 0.36819007992744446,
      "learning_rate": 0.0002555273857928725,
      "loss": 0.0936,
      "step": 3604
    },
    {
      "epoch": 0.5052557813594954,
      "grad_norm": 0.48663437366485596,
      "learning_rate": 0.0002555130351590528,
      "loss": 0.1217,
      "step": 3605
    },
    {
      "epoch": 0.505395935529082,
      "grad_norm": 0.2748851478099823,
      "learning_rate": 0.00025549868452523315,
      "loss": 0.0806,
      "step": 3606
    },
    {
      "epoch": 0.5055360896986686,
      "grad_norm": 0.405549019575119,
      "learning_rate": 0.00025548433389141354,
      "loss": 0.1514,
      "step": 3607
    },
    {
      "epoch": 0.505676243868255,
      "grad_norm": 0.531527042388916,
      "learning_rate": 0.00025546998325759386,
      "loss": 0.1428,
      "step": 3608
    },
    {
      "epoch": 0.5058163980378416,
      "grad_norm": 0.4019516706466675,
      "learning_rate": 0.0002554556326237742,
      "loss": 0.1675,
      "step": 3609
    },
    {
      "epoch": 0.5059565522074282,
      "grad_norm": 0.27313774824142456,
      "learning_rate": 0.0002554412819899546,
      "loss": 0.0527,
      "step": 3610
    },
    {
      "epoch": 0.5060967063770148,
      "grad_norm": 0.3598933517932892,
      "learning_rate": 0.0002554269313561349,
      "loss": 0.0526,
      "step": 3611
    },
    {
      "epoch": 0.5062368605466012,
      "grad_norm": 0.6425841450691223,
      "learning_rate": 0.00025541258072231523,
      "loss": 0.0658,
      "step": 3612
    },
    {
      "epoch": 0.5063770147161878,
      "grad_norm": 0.2769329845905304,
      "learning_rate": 0.00025539823008849556,
      "loss": 0.1075,
      "step": 3613
    },
    {
      "epoch": 0.5065171688857744,
      "grad_norm": 0.06827756017446518,
      "learning_rate": 0.0002553838794546759,
      "loss": 0.0117,
      "step": 3614
    },
    {
      "epoch": 0.5066573230553609,
      "grad_norm": 0.34457892179489136,
      "learning_rate": 0.0002553695288208562,
      "loss": 0.0481,
      "step": 3615
    },
    {
      "epoch": 0.5067974772249474,
      "grad_norm": 0.6185826063156128,
      "learning_rate": 0.00025535517818703655,
      "loss": 0.1218,
      "step": 3616
    },
    {
      "epoch": 0.506937631394534,
      "grad_norm": 0.3842414915561676,
      "learning_rate": 0.00025534082755321693,
      "loss": 0.0677,
      "step": 3617
    },
    {
      "epoch": 0.5070777855641205,
      "grad_norm": 0.7708631753921509,
      "learning_rate": 0.00025532647691939726,
      "loss": 0.1292,
      "step": 3618
    },
    {
      "epoch": 0.5072179397337071,
      "grad_norm": 0.4499856233596802,
      "learning_rate": 0.0002553121262855776,
      "loss": 0.0821,
      "step": 3619
    },
    {
      "epoch": 0.5073580939032937,
      "grad_norm": 0.4605194330215454,
      "learning_rate": 0.0002552977756517579,
      "loss": 0.0744,
      "step": 3620
    },
    {
      "epoch": 0.5074982480728801,
      "grad_norm": 0.2140355408191681,
      "learning_rate": 0.00025528342501793824,
      "loss": 0.0762,
      "step": 3621
    },
    {
      "epoch": 0.5076384022424667,
      "grad_norm": 0.3987409770488739,
      "learning_rate": 0.0002552690743841186,
      "loss": 0.0593,
      "step": 3622
    },
    {
      "epoch": 0.5077785564120533,
      "grad_norm": 0.22239546477794647,
      "learning_rate": 0.00025525472375029896,
      "loss": 0.0357,
      "step": 3623
    },
    {
      "epoch": 0.5079187105816398,
      "grad_norm": 0.43409937620162964,
      "learning_rate": 0.0002552403731164793,
      "loss": 0.0626,
      "step": 3624
    },
    {
      "epoch": 0.5080588647512263,
      "grad_norm": 0.32065635919570923,
      "learning_rate": 0.0002552260224826596,
      "loss": 0.0379,
      "step": 3625
    },
    {
      "epoch": 0.5081990189208129,
      "grad_norm": 0.5844007730484009,
      "learning_rate": 0.00025521167184884,
      "loss": 0.1395,
      "step": 3626
    },
    {
      "epoch": 0.5083391730903994,
      "grad_norm": 0.424710214138031,
      "learning_rate": 0.0002551973212150203,
      "loss": 0.0891,
      "step": 3627
    },
    {
      "epoch": 0.508479327259986,
      "grad_norm": 0.6183585524559021,
      "learning_rate": 0.00025518297058120065,
      "loss": 0.1645,
      "step": 3628
    },
    {
      "epoch": 0.5086194814295726,
      "grad_norm": 0.8181776404380798,
      "learning_rate": 0.000255168619947381,
      "loss": 0.1303,
      "step": 3629
    },
    {
      "epoch": 0.508759635599159,
      "grad_norm": 0.581883430480957,
      "learning_rate": 0.0002551542693135613,
      "loss": 0.0545,
      "step": 3630
    },
    {
      "epoch": 0.5088997897687456,
      "grad_norm": 1.1187405586242676,
      "learning_rate": 0.00025513991867974164,
      "loss": 0.2164,
      "step": 3631
    },
    {
      "epoch": 0.5090399439383322,
      "grad_norm": 0.4822360575199127,
      "learning_rate": 0.00025512556804592197,
      "loss": 0.1014,
      "step": 3632
    },
    {
      "epoch": 0.5091800981079188,
      "grad_norm": 0.23033207654953003,
      "learning_rate": 0.00025511121741210235,
      "loss": 0.0555,
      "step": 3633
    },
    {
      "epoch": 0.5093202522775052,
      "grad_norm": 0.3463762402534485,
      "learning_rate": 0.0002550968667782827,
      "loss": 0.079,
      "step": 3634
    },
    {
      "epoch": 0.5094604064470918,
      "grad_norm": 0.7652960419654846,
      "learning_rate": 0.000255082516144463,
      "loss": 0.0931,
      "step": 3635
    },
    {
      "epoch": 0.5096005606166784,
      "grad_norm": 0.2973748445510864,
      "learning_rate": 0.0002550681655106434,
      "loss": 0.0567,
      "step": 3636
    },
    {
      "epoch": 0.5097407147862649,
      "grad_norm": 0.660727858543396,
      "learning_rate": 0.0002550538148768237,
      "loss": 0.0993,
      "step": 3637
    },
    {
      "epoch": 0.5098808689558514,
      "grad_norm": 0.504641592502594,
      "learning_rate": 0.00025503946424300405,
      "loss": 0.0945,
      "step": 3638
    },
    {
      "epoch": 0.510021023125438,
      "grad_norm": 0.31552356481552124,
      "learning_rate": 0.0002550251136091844,
      "loss": 0.1125,
      "step": 3639
    },
    {
      "epoch": 0.5101611772950245,
      "grad_norm": 0.5281522870063782,
      "learning_rate": 0.0002550107629753647,
      "loss": 0.134,
      "step": 3640
    },
    {
      "epoch": 0.5103013314646111,
      "grad_norm": 0.2179023176431656,
      "learning_rate": 0.00025499641234154503,
      "loss": 0.0381,
      "step": 3641
    },
    {
      "epoch": 0.5104414856341977,
      "grad_norm": 0.24178653955459595,
      "learning_rate": 0.0002549820617077254,
      "loss": 0.0687,
      "step": 3642
    },
    {
      "epoch": 0.5105816398037841,
      "grad_norm": 0.40493133664131165,
      "learning_rate": 0.00025496771107390574,
      "loss": 0.0782,
      "step": 3643
    },
    {
      "epoch": 0.5107217939733707,
      "grad_norm": 0.26079076528549194,
      "learning_rate": 0.00025495336044008607,
      "loss": 0.0619,
      "step": 3644
    },
    {
      "epoch": 0.5108619481429573,
      "grad_norm": 0.4745108187198639,
      "learning_rate": 0.00025493900980626646,
      "loss": 0.1587,
      "step": 3645
    },
    {
      "epoch": 0.5110021023125438,
      "grad_norm": 0.32317304611206055,
      "learning_rate": 0.0002549246591724468,
      "loss": 0.0521,
      "step": 3646
    },
    {
      "epoch": 0.5111422564821303,
      "grad_norm": 0.9599839448928833,
      "learning_rate": 0.0002549103085386271,
      "loss": 0.2165,
      "step": 3647
    },
    {
      "epoch": 0.5112824106517169,
      "grad_norm": 0.37458568811416626,
      "learning_rate": 0.00025489595790480744,
      "loss": 0.0778,
      "step": 3648
    },
    {
      "epoch": 0.5114225648213034,
      "grad_norm": 0.9396136999130249,
      "learning_rate": 0.00025488160727098777,
      "loss": 0.2227,
      "step": 3649
    },
    {
      "epoch": 0.51156271899089,
      "grad_norm": 3.522144317626953,
      "learning_rate": 0.0002548672566371681,
      "loss": 0.6797,
      "step": 3650
    },
    {
      "epoch": 0.5117028731604766,
      "grad_norm": 0.5679569840431213,
      "learning_rate": 0.0002548529060033484,
      "loss": 0.0692,
      "step": 3651
    },
    {
      "epoch": 0.511843027330063,
      "grad_norm": 0.2392069697380066,
      "learning_rate": 0.0002548385553695288,
      "loss": 0.0502,
      "step": 3652
    },
    {
      "epoch": 0.5119831814996496,
      "grad_norm": 0.44987377524375916,
      "learning_rate": 0.00025482420473570914,
      "loss": 0.1429,
      "step": 3653
    },
    {
      "epoch": 0.5121233356692362,
      "grad_norm": 0.2524642050266266,
      "learning_rate": 0.00025480985410188947,
      "loss": 0.0616,
      "step": 3654
    },
    {
      "epoch": 0.5122634898388227,
      "grad_norm": 0.23557986319065094,
      "learning_rate": 0.00025479550346806985,
      "loss": 0.0783,
      "step": 3655
    },
    {
      "epoch": 0.5124036440084092,
      "grad_norm": 0.3249613046646118,
      "learning_rate": 0.0002547811528342502,
      "loss": 0.0685,
      "step": 3656
    },
    {
      "epoch": 0.5125437981779958,
      "grad_norm": 0.5398142337799072,
      "learning_rate": 0.0002547668022004305,
      "loss": 0.1051,
      "step": 3657
    },
    {
      "epoch": 0.5126839523475824,
      "grad_norm": 0.45770299434661865,
      "learning_rate": 0.00025475245156661084,
      "loss": 0.188,
      "step": 3658
    },
    {
      "epoch": 0.5128241065171689,
      "grad_norm": 0.4042181670665741,
      "learning_rate": 0.00025473810093279116,
      "loss": 0.0848,
      "step": 3659
    },
    {
      "epoch": 0.5129642606867554,
      "grad_norm": 0.26377978920936584,
      "learning_rate": 0.0002547237502989715,
      "loss": 0.0508,
      "step": 3660
    },
    {
      "epoch": 0.513104414856342,
      "grad_norm": 0.41191262006759644,
      "learning_rate": 0.0002547093996651519,
      "loss": 0.0947,
      "step": 3661
    },
    {
      "epoch": 0.5132445690259285,
      "grad_norm": 0.475039005279541,
      "learning_rate": 0.0002546950490313322,
      "loss": 0.0907,
      "step": 3662
    },
    {
      "epoch": 0.5133847231955151,
      "grad_norm": 0.5674811601638794,
      "learning_rate": 0.00025468069839751253,
      "loss": 0.1717,
      "step": 3663
    },
    {
      "epoch": 0.5135248773651017,
      "grad_norm": 0.4234640300273895,
      "learning_rate": 0.0002546663477636929,
      "loss": 0.1178,
      "step": 3664
    },
    {
      "epoch": 0.5136650315346881,
      "grad_norm": 0.41077032685279846,
      "learning_rate": 0.00025465199712987324,
      "loss": 0.1458,
      "step": 3665
    },
    {
      "epoch": 0.5138051857042747,
      "grad_norm": 0.3074701130390167,
      "learning_rate": 0.00025463764649605357,
      "loss": 0.0943,
      "step": 3666
    },
    {
      "epoch": 0.5139453398738613,
      "grad_norm": 0.412673681974411,
      "learning_rate": 0.0002546232958622339,
      "loss": 0.0831,
      "step": 3667
    },
    {
      "epoch": 0.5140854940434478,
      "grad_norm": 0.4425532817840576,
      "learning_rate": 0.00025460894522841423,
      "loss": 0.0777,
      "step": 3668
    },
    {
      "epoch": 0.5142256482130343,
      "grad_norm": 0.48464179039001465,
      "learning_rate": 0.00025459459459459456,
      "loss": 0.0734,
      "step": 3669
    },
    {
      "epoch": 0.5143658023826209,
      "grad_norm": 0.5093382000923157,
      "learning_rate": 0.0002545802439607749,
      "loss": 0.0691,
      "step": 3670
    },
    {
      "epoch": 0.5145059565522074,
      "grad_norm": 0.31960538029670715,
      "learning_rate": 0.00025456589332695527,
      "loss": 0.048,
      "step": 3671
    },
    {
      "epoch": 0.514646110721794,
      "grad_norm": 0.5682074427604675,
      "learning_rate": 0.0002545515426931356,
      "loss": 0.1239,
      "step": 3672
    },
    {
      "epoch": 0.5147862648913806,
      "grad_norm": 0.6105221509933472,
      "learning_rate": 0.0002545371920593159,
      "loss": 0.1195,
      "step": 3673
    },
    {
      "epoch": 0.514926419060967,
      "grad_norm": 0.1577884405851364,
      "learning_rate": 0.0002545228414254963,
      "loss": 0.0249,
      "step": 3674
    },
    {
      "epoch": 0.5150665732305536,
      "grad_norm": 0.6720603108406067,
      "learning_rate": 0.00025450849079167664,
      "loss": 0.1105,
      "step": 3675
    },
    {
      "epoch": 0.5152067274001402,
      "grad_norm": 0.4378015995025635,
      "learning_rate": 0.00025449414015785697,
      "loss": 0.0671,
      "step": 3676
    },
    {
      "epoch": 0.5153468815697267,
      "grad_norm": 0.4326946437358856,
      "learning_rate": 0.0002544797895240373,
      "loss": 0.1354,
      "step": 3677
    },
    {
      "epoch": 0.5154870357393132,
      "grad_norm": 0.3068216145038605,
      "learning_rate": 0.0002544654388902176,
      "loss": 0.1346,
      "step": 3678
    },
    {
      "epoch": 0.5156271899088998,
      "grad_norm": 0.405368447303772,
      "learning_rate": 0.00025445108825639795,
      "loss": 0.1416,
      "step": 3679
    },
    {
      "epoch": 0.5157673440784863,
      "grad_norm": 0.6262340545654297,
      "learning_rate": 0.00025443673762257833,
      "loss": 0.1014,
      "step": 3680
    },
    {
      "epoch": 0.5159074982480729,
      "grad_norm": 0.38591116666793823,
      "learning_rate": 0.00025442238698875866,
      "loss": 0.0652,
      "step": 3681
    },
    {
      "epoch": 0.5160476524176594,
      "grad_norm": 0.3750234544277191,
      "learning_rate": 0.000254408036354939,
      "loss": 0.0707,
      "step": 3682
    },
    {
      "epoch": 0.516187806587246,
      "grad_norm": 0.7763251662254333,
      "learning_rate": 0.0002543936857211193,
      "loss": 0.1007,
      "step": 3683
    },
    {
      "epoch": 0.5163279607568325,
      "grad_norm": 0.5143107175827026,
      "learning_rate": 0.00025437933508729965,
      "loss": 0.0745,
      "step": 3684
    },
    {
      "epoch": 0.5164681149264191,
      "grad_norm": 0.684270977973938,
      "learning_rate": 0.00025436498445348,
      "loss": 0.0437,
      "step": 3685
    },
    {
      "epoch": 0.5166082690960057,
      "grad_norm": 0.7050482034683228,
      "learning_rate": 0.0002543506338196603,
      "loss": 0.1121,
      "step": 3686
    },
    {
      "epoch": 0.5167484232655921,
      "grad_norm": 0.8040270209312439,
      "learning_rate": 0.0002543362831858407,
      "loss": 0.0722,
      "step": 3687
    },
    {
      "epoch": 0.5168885774351787,
      "grad_norm": 0.36413830518722534,
      "learning_rate": 0.000254321932552021,
      "loss": 0.1103,
      "step": 3688
    },
    {
      "epoch": 0.5170287316047653,
      "grad_norm": 0.6621695756912231,
      "learning_rate": 0.00025430758191820135,
      "loss": 0.0611,
      "step": 3689
    },
    {
      "epoch": 0.5171688857743518,
      "grad_norm": 0.4166024923324585,
      "learning_rate": 0.00025429323128438173,
      "loss": 0.0965,
      "step": 3690
    },
    {
      "epoch": 0.5173090399439383,
      "grad_norm": 0.7858618497848511,
      "learning_rate": 0.00025427888065056206,
      "loss": 0.0901,
      "step": 3691
    },
    {
      "epoch": 0.5174491941135249,
      "grad_norm": 0.38334763050079346,
      "learning_rate": 0.0002542645300167424,
      "loss": 0.0873,
      "step": 3692
    },
    {
      "epoch": 0.5175893482831114,
      "grad_norm": 0.5154088735580444,
      "learning_rate": 0.0002542501793829227,
      "loss": 0.1209,
      "step": 3693
    },
    {
      "epoch": 0.517729502452698,
      "grad_norm": 0.5021282434463501,
      "learning_rate": 0.00025423582874910304,
      "loss": 0.1728,
      "step": 3694
    },
    {
      "epoch": 0.5178696566222845,
      "grad_norm": 0.9212284684181213,
      "learning_rate": 0.00025422147811528337,
      "loss": 0.1217,
      "step": 3695
    },
    {
      "epoch": 0.518009810791871,
      "grad_norm": 0.6294364929199219,
      "learning_rate": 0.00025420712748146375,
      "loss": 0.1127,
      "step": 3696
    },
    {
      "epoch": 0.5181499649614576,
      "grad_norm": 1.495061993598938,
      "learning_rate": 0.0002541927768476441,
      "loss": 0.0955,
      "step": 3697
    },
    {
      "epoch": 0.5182901191310442,
      "grad_norm": 0.6642441153526306,
      "learning_rate": 0.0002541784262138244,
      "loss": 0.0836,
      "step": 3698
    },
    {
      "epoch": 0.5184302733006307,
      "grad_norm": 1.2072820663452148,
      "learning_rate": 0.0002541640755800048,
      "loss": 0.1559,
      "step": 3699
    },
    {
      "epoch": 0.5185704274702172,
      "grad_norm": 1.116833209991455,
      "learning_rate": 0.0002541497249461851,
      "loss": 0.3666,
      "step": 3700
    },
    {
      "epoch": 0.5187105816398038,
      "grad_norm": 0.39377114176750183,
      "learning_rate": 0.00025413537431236545,
      "loss": 0.0762,
      "step": 3701
    },
    {
      "epoch": 0.5188507358093903,
      "grad_norm": 0.5531027317047119,
      "learning_rate": 0.0002541210236785458,
      "loss": 0.1778,
      "step": 3702
    },
    {
      "epoch": 0.5189908899789769,
      "grad_norm": 0.4501936137676239,
      "learning_rate": 0.0002541066730447261,
      "loss": 0.0614,
      "step": 3703
    },
    {
      "epoch": 0.5191310441485634,
      "grad_norm": 0.5126311779022217,
      "learning_rate": 0.00025409232241090644,
      "loss": 0.0613,
      "step": 3704
    },
    {
      "epoch": 0.5192711983181499,
      "grad_norm": 0.41879671812057495,
      "learning_rate": 0.00025407797177708677,
      "loss": 0.0498,
      "step": 3705
    },
    {
      "epoch": 0.5194113524877365,
      "grad_norm": 0.4197855591773987,
      "learning_rate": 0.00025406362114326715,
      "loss": 0.0974,
      "step": 3706
    },
    {
      "epoch": 0.5195515066573231,
      "grad_norm": 0.8715519309043884,
      "learning_rate": 0.0002540492705094475,
      "loss": 0.1314,
      "step": 3707
    },
    {
      "epoch": 0.5196916608269097,
      "grad_norm": 0.636552631855011,
      "learning_rate": 0.0002540349198756278,
      "loss": 0.1547,
      "step": 3708
    },
    {
      "epoch": 0.5198318149964961,
      "grad_norm": 0.44017305970191956,
      "learning_rate": 0.0002540205692418082,
      "loss": 0.087,
      "step": 3709
    },
    {
      "epoch": 0.5199719691660827,
      "grad_norm": 0.18093568086624146,
      "learning_rate": 0.0002540062186079885,
      "loss": 0.0374,
      "step": 3710
    },
    {
      "epoch": 0.5201121233356693,
      "grad_norm": 0.3853932321071625,
      "learning_rate": 0.00025399186797416885,
      "loss": 0.1017,
      "step": 3711
    },
    {
      "epoch": 0.5202522775052558,
      "grad_norm": 0.25976938009262085,
      "learning_rate": 0.0002539775173403492,
      "loss": 0.076,
      "step": 3712
    },
    {
      "epoch": 0.5203924316748423,
      "grad_norm": 0.35607048869132996,
      "learning_rate": 0.0002539631667065295,
      "loss": 0.0831,
      "step": 3713
    },
    {
      "epoch": 0.5205325858444289,
      "grad_norm": 0.4062139689922333,
      "learning_rate": 0.00025394881607270983,
      "loss": 0.1315,
      "step": 3714
    },
    {
      "epoch": 0.5206727400140154,
      "grad_norm": 0.8705641031265259,
      "learning_rate": 0.0002539344654388902,
      "loss": 0.0913,
      "step": 3715
    },
    {
      "epoch": 0.520812894183602,
      "grad_norm": 0.4616554081439972,
      "learning_rate": 0.00025392011480507054,
      "loss": 0.1012,
      "step": 3716
    },
    {
      "epoch": 0.5209530483531885,
      "grad_norm": 0.49104246497154236,
      "learning_rate": 0.00025390576417125087,
      "loss": 0.0935,
      "step": 3717
    },
    {
      "epoch": 0.521093202522775,
      "grad_norm": 0.4312751293182373,
      "learning_rate": 0.0002538914135374312,
      "loss": 0.0405,
      "step": 3718
    },
    {
      "epoch": 0.5212333566923616,
      "grad_norm": 0.36991086602211,
      "learning_rate": 0.0002538770629036116,
      "loss": 0.0865,
      "step": 3719
    },
    {
      "epoch": 0.5213735108619482,
      "grad_norm": 0.3838135600090027,
      "learning_rate": 0.0002538627122697919,
      "loss": 0.0939,
      "step": 3720
    },
    {
      "epoch": 0.5215136650315347,
      "grad_norm": 0.9714263677597046,
      "learning_rate": 0.00025384836163597224,
      "loss": 0.1767,
      "step": 3721
    },
    {
      "epoch": 0.5216538192011212,
      "grad_norm": 0.4344284236431122,
      "learning_rate": 0.00025383401100215257,
      "loss": 0.1585,
      "step": 3722
    },
    {
      "epoch": 0.5217939733707078,
      "grad_norm": 0.40689384937286377,
      "learning_rate": 0.0002538196603683329,
      "loss": 0.0677,
      "step": 3723
    },
    {
      "epoch": 0.5219341275402943,
      "grad_norm": 0.7999861836433411,
      "learning_rate": 0.0002538053097345132,
      "loss": 0.1,
      "step": 3724
    },
    {
      "epoch": 0.5220742817098809,
      "grad_norm": 0.5300484895706177,
      "learning_rate": 0.0002537909591006936,
      "loss": 0.1454,
      "step": 3725
    },
    {
      "epoch": 0.5222144358794674,
      "grad_norm": 0.5659837126731873,
      "learning_rate": 0.00025377660846687394,
      "loss": 0.1146,
      "step": 3726
    },
    {
      "epoch": 0.5223545900490539,
      "grad_norm": 0.5480971932411194,
      "learning_rate": 0.00025376225783305426,
      "loss": 0.081,
      "step": 3727
    },
    {
      "epoch": 0.5224947442186405,
      "grad_norm": 0.5467221736907959,
      "learning_rate": 0.00025374790719923465,
      "loss": 0.1481,
      "step": 3728
    },
    {
      "epoch": 0.5226348983882271,
      "grad_norm": 0.26411089301109314,
      "learning_rate": 0.000253733556565415,
      "loss": 0.0447,
      "step": 3729
    },
    {
      "epoch": 0.5227750525578136,
      "grad_norm": 0.4189164340496063,
      "learning_rate": 0.0002537192059315953,
      "loss": 0.1339,
      "step": 3730
    },
    {
      "epoch": 0.5229152067274001,
      "grad_norm": 0.4686305820941925,
      "learning_rate": 0.00025370485529777563,
      "loss": 0.0545,
      "step": 3731
    },
    {
      "epoch": 0.5230553608969867,
      "grad_norm": 0.6158311367034912,
      "learning_rate": 0.00025369050466395596,
      "loss": 0.1736,
      "step": 3732
    },
    {
      "epoch": 0.5231955150665732,
      "grad_norm": 0.48443883657455444,
      "learning_rate": 0.0002536761540301363,
      "loss": 0.0679,
      "step": 3733
    },
    {
      "epoch": 0.5233356692361598,
      "grad_norm": 0.45886072516441345,
      "learning_rate": 0.0002536618033963166,
      "loss": 0.0812,
      "step": 3734
    },
    {
      "epoch": 0.5234758234057463,
      "grad_norm": 0.34823763370513916,
      "learning_rate": 0.000253647452762497,
      "loss": 0.0827,
      "step": 3735
    },
    {
      "epoch": 0.5236159775753328,
      "grad_norm": 0.5238205194473267,
      "learning_rate": 0.00025363310212867733,
      "loss": 0.1046,
      "step": 3736
    },
    {
      "epoch": 0.5237561317449194,
      "grad_norm": 0.7899017930030823,
      "learning_rate": 0.00025361875149485766,
      "loss": 0.079,
      "step": 3737
    },
    {
      "epoch": 0.523896285914506,
      "grad_norm": 0.5810121893882751,
      "learning_rate": 0.00025360440086103804,
      "loss": 0.1014,
      "step": 3738
    },
    {
      "epoch": 0.5240364400840924,
      "grad_norm": 0.291208952665329,
      "learning_rate": 0.00025359005022721837,
      "loss": 0.0477,
      "step": 3739
    },
    {
      "epoch": 0.524176594253679,
      "grad_norm": 0.4190904498100281,
      "learning_rate": 0.0002535756995933987,
      "loss": 0.1305,
      "step": 3740
    },
    {
      "epoch": 0.5243167484232656,
      "grad_norm": 0.8952327370643616,
      "learning_rate": 0.00025356134895957903,
      "loss": 0.0828,
      "step": 3741
    },
    {
      "epoch": 0.5244569025928522,
      "grad_norm": 0.3364610970020294,
      "learning_rate": 0.00025354699832575936,
      "loss": 0.1618,
      "step": 3742
    },
    {
      "epoch": 0.5245970567624387,
      "grad_norm": 0.4337835907936096,
      "learning_rate": 0.0002535326476919397,
      "loss": 0.0525,
      "step": 3743
    },
    {
      "epoch": 0.5247372109320252,
      "grad_norm": 0.7692883610725403,
      "learning_rate": 0.00025351829705812007,
      "loss": 0.0859,
      "step": 3744
    },
    {
      "epoch": 0.5248773651016118,
      "grad_norm": 1.7545913457870483,
      "learning_rate": 0.0002535039464243004,
      "loss": 0.1859,
      "step": 3745
    },
    {
      "epoch": 0.5250175192711983,
      "grad_norm": 2.708228826522827,
      "learning_rate": 0.0002534895957904807,
      "loss": 0.7529,
      "step": 3746
    },
    {
      "epoch": 0.5251576734407849,
      "grad_norm": 1.251065969467163,
      "learning_rate": 0.00025347524515666105,
      "loss": 0.3325,
      "step": 3747
    },
    {
      "epoch": 0.5252978276103714,
      "grad_norm": 1.7777938842773438,
      "learning_rate": 0.0002534608945228414,
      "loss": 0.2494,
      "step": 3748
    },
    {
      "epoch": 0.5254379817799579,
      "grad_norm": 1.7284437417984009,
      "learning_rate": 0.0002534465438890217,
      "loss": 0.1349,
      "step": 3749
    },
    {
      "epoch": 0.5255781359495445,
      "grad_norm": 1.3511062860488892,
      "learning_rate": 0.0002534321932552021,
      "loss": 0.0857,
      "step": 3750
    },
    {
      "epoch": 0.5257182901191311,
      "grad_norm": 0.7963975071907043,
      "learning_rate": 0.0002534178426213824,
      "loss": 0.089,
      "step": 3751
    },
    {
      "epoch": 0.5258584442887176,
      "grad_norm": 0.2531070411205292,
      "learning_rate": 0.00025340349198756275,
      "loss": 0.0537,
      "step": 3752
    },
    {
      "epoch": 0.5259985984583041,
      "grad_norm": 0.5108381509780884,
      "learning_rate": 0.0002533891413537431,
      "loss": 0.117,
      "step": 3753
    },
    {
      "epoch": 0.5261387526278907,
      "grad_norm": 0.46662551164627075,
      "learning_rate": 0.00025337479071992346,
      "loss": 0.1447,
      "step": 3754
    },
    {
      "epoch": 0.5262789067974772,
      "grad_norm": 0.5900042057037354,
      "learning_rate": 0.0002533604400861038,
      "loss": 0.0514,
      "step": 3755
    },
    {
      "epoch": 0.5264190609670638,
      "grad_norm": 0.4459666609764099,
      "learning_rate": 0.0002533460894522841,
      "loss": 0.1248,
      "step": 3756
    },
    {
      "epoch": 0.5265592151366503,
      "grad_norm": 0.5731019377708435,
      "learning_rate": 0.00025333173881846445,
      "loss": 0.1944,
      "step": 3757
    },
    {
      "epoch": 0.5266993693062368,
      "grad_norm": 0.811814546585083,
      "learning_rate": 0.0002533173881846448,
      "loss": 0.1698,
      "step": 3758
    },
    {
      "epoch": 0.5268395234758234,
      "grad_norm": 0.2540329396724701,
      "learning_rate": 0.0002533030375508251,
      "loss": 0.0745,
      "step": 3759
    },
    {
      "epoch": 0.52697967764541,
      "grad_norm": 0.2398877888917923,
      "learning_rate": 0.0002532886869170055,
      "loss": 0.0481,
      "step": 3760
    },
    {
      "epoch": 0.5271198318149964,
      "grad_norm": 0.18241244554519653,
      "learning_rate": 0.0002532743362831858,
      "loss": 0.0255,
      "step": 3761
    },
    {
      "epoch": 0.527259985984583,
      "grad_norm": 0.5653223395347595,
      "learning_rate": 0.00025325998564936614,
      "loss": 0.1234,
      "step": 3762
    },
    {
      "epoch": 0.5274001401541696,
      "grad_norm": 0.2712160050868988,
      "learning_rate": 0.0002532456350155465,
      "loss": 0.079,
      "step": 3763
    },
    {
      "epoch": 0.5275402943237562,
      "grad_norm": 0.6309765577316284,
      "learning_rate": 0.00025323128438172686,
      "loss": 0.0835,
      "step": 3764
    },
    {
      "epoch": 0.5276804484933427,
      "grad_norm": 0.5536413192749023,
      "learning_rate": 0.0002532169337479072,
      "loss": 0.1897,
      "step": 3765
    },
    {
      "epoch": 0.5278206026629292,
      "grad_norm": 0.19743096828460693,
      "learning_rate": 0.0002532025831140875,
      "loss": 0.0493,
      "step": 3766
    },
    {
      "epoch": 0.5279607568325158,
      "grad_norm": 0.4978054165840149,
      "learning_rate": 0.00025318823248026784,
      "loss": 0.0894,
      "step": 3767
    },
    {
      "epoch": 0.5281009110021023,
      "grad_norm": 0.7236056923866272,
      "learning_rate": 0.00025317388184644817,
      "loss": 0.1684,
      "step": 3768
    },
    {
      "epoch": 0.5282410651716889,
      "grad_norm": 0.4210502803325653,
      "learning_rate": 0.0002531595312126285,
      "loss": 0.1019,
      "step": 3769
    },
    {
      "epoch": 0.5283812193412754,
      "grad_norm": 0.3556743860244751,
      "learning_rate": 0.0002531451805788089,
      "loss": 0.0692,
      "step": 3770
    },
    {
      "epoch": 0.5285213735108619,
      "grad_norm": 0.3373337984085083,
      "learning_rate": 0.0002531308299449892,
      "loss": 0.0596,
      "step": 3771
    },
    {
      "epoch": 0.5286615276804485,
      "grad_norm": 0.5512180924415588,
      "learning_rate": 0.00025311647931116954,
      "loss": 0.1073,
      "step": 3772
    },
    {
      "epoch": 0.5288016818500351,
      "grad_norm": 0.7351663708686829,
      "learning_rate": 0.0002531021286773499,
      "loss": 0.035,
      "step": 3773
    },
    {
      "epoch": 0.5289418360196216,
      "grad_norm": 0.414307564496994,
      "learning_rate": 0.00025308777804353025,
      "loss": 0.115,
      "step": 3774
    },
    {
      "epoch": 0.5290819901892081,
      "grad_norm": 0.8039710521697998,
      "learning_rate": 0.0002530734274097106,
      "loss": 0.2339,
      "step": 3775
    },
    {
      "epoch": 0.5292221443587947,
      "grad_norm": 0.3805621564388275,
      "learning_rate": 0.0002530590767758909,
      "loss": 0.0833,
      "step": 3776
    },
    {
      "epoch": 0.5293622985283812,
      "grad_norm": 0.3147873282432556,
      "learning_rate": 0.00025304472614207123,
      "loss": 0.0593,
      "step": 3777
    },
    {
      "epoch": 0.5295024526979678,
      "grad_norm": 0.43599745631217957,
      "learning_rate": 0.00025303037550825156,
      "loss": 0.086,
      "step": 3778
    },
    {
      "epoch": 0.5296426068675543,
      "grad_norm": 0.46452921628952026,
      "learning_rate": 0.00025301602487443195,
      "loss": 0.0752,
      "step": 3779
    },
    {
      "epoch": 0.5297827610371408,
      "grad_norm": 0.3302789330482483,
      "learning_rate": 0.0002530016742406123,
      "loss": 0.1245,
      "step": 3780
    },
    {
      "epoch": 0.5299229152067274,
      "grad_norm": 0.32080042362213135,
      "learning_rate": 0.0002529873236067926,
      "loss": 0.0673,
      "step": 3781
    },
    {
      "epoch": 0.530063069376314,
      "grad_norm": 0.480005145072937,
      "learning_rate": 0.000252972972972973,
      "loss": 0.0665,
      "step": 3782
    },
    {
      "epoch": 0.5302032235459004,
      "grad_norm": 0.4755353331565857,
      "learning_rate": 0.0002529586223391533,
      "loss": 0.0865,
      "step": 3783
    },
    {
      "epoch": 0.530343377715487,
      "grad_norm": 0.2493240088224411,
      "learning_rate": 0.00025294427170533364,
      "loss": 0.0357,
      "step": 3784
    },
    {
      "epoch": 0.5304835318850736,
      "grad_norm": 0.44424477219581604,
      "learning_rate": 0.00025292992107151397,
      "loss": 0.0698,
      "step": 3785
    },
    {
      "epoch": 0.5306236860546601,
      "grad_norm": 0.3517785966396332,
      "learning_rate": 0.0002529155704376943,
      "loss": 0.1085,
      "step": 3786
    },
    {
      "epoch": 0.5307638402242467,
      "grad_norm": 0.6620528101921082,
      "learning_rate": 0.00025290121980387463,
      "loss": 0.1684,
      "step": 3787
    },
    {
      "epoch": 0.5309039943938332,
      "grad_norm": 0.36682337522506714,
      "learning_rate": 0.00025288686917005496,
      "loss": 0.0473,
      "step": 3788
    },
    {
      "epoch": 0.5310441485634197,
      "grad_norm": 0.5843929648399353,
      "learning_rate": 0.00025287251853623534,
      "loss": 0.0641,
      "step": 3789
    },
    {
      "epoch": 0.5311843027330063,
      "grad_norm": 0.3687399923801422,
      "learning_rate": 0.00025285816790241567,
      "loss": 0.0766,
      "step": 3790
    },
    {
      "epoch": 0.5313244569025929,
      "grad_norm": 0.39288610219955444,
      "learning_rate": 0.000252843817268596,
      "loss": 0.1129,
      "step": 3791
    },
    {
      "epoch": 0.5314646110721793,
      "grad_norm": 0.48271477222442627,
      "learning_rate": 0.0002528294666347764,
      "loss": 0.1261,
      "step": 3792
    },
    {
      "epoch": 0.5316047652417659,
      "grad_norm": 0.6124277710914612,
      "learning_rate": 0.0002528151160009567,
      "loss": 0.2185,
      "step": 3793
    },
    {
      "epoch": 0.5317449194113525,
      "grad_norm": 1.0658903121948242,
      "learning_rate": 0.00025280076536713704,
      "loss": 0.1458,
      "step": 3794
    },
    {
      "epoch": 0.5318850735809391,
      "grad_norm": 0.4022054672241211,
      "learning_rate": 0.00025278641473331737,
      "loss": 0.0646,
      "step": 3795
    },
    {
      "epoch": 0.5320252277505256,
      "grad_norm": 0.6764711141586304,
      "learning_rate": 0.0002527720640994977,
      "loss": 0.1157,
      "step": 3796
    },
    {
      "epoch": 0.5321653819201121,
      "grad_norm": 1.4497134685516357,
      "learning_rate": 0.000252757713465678,
      "loss": 0.1317,
      "step": 3797
    },
    {
      "epoch": 0.5323055360896987,
      "grad_norm": 0.482147753238678,
      "learning_rate": 0.0002527433628318584,
      "loss": 0.0587,
      "step": 3798
    },
    {
      "epoch": 0.5324456902592852,
      "grad_norm": 1.0537500381469727,
      "learning_rate": 0.00025272901219803873,
      "loss": 0.289,
      "step": 3799
    },
    {
      "epoch": 0.5325858444288718,
      "grad_norm": 0.7479227185249329,
      "learning_rate": 0.00025271466156421906,
      "loss": 0.2757,
      "step": 3800
    },
    {
      "epoch": 0.5327259985984583,
      "grad_norm": 0.43164584040641785,
      "learning_rate": 0.00025270031093039945,
      "loss": 0.0546,
      "step": 3801
    },
    {
      "epoch": 0.5328661527680448,
      "grad_norm": 0.5817529559135437,
      "learning_rate": 0.0002526859602965798,
      "loss": 0.0922,
      "step": 3802
    },
    {
      "epoch": 0.5330063069376314,
      "grad_norm": 0.20474418997764587,
      "learning_rate": 0.0002526716096627601,
      "loss": 0.0376,
      "step": 3803
    },
    {
      "epoch": 0.533146461107218,
      "grad_norm": 0.14792661368846893,
      "learning_rate": 0.00025265725902894043,
      "loss": 0.0286,
      "step": 3804
    },
    {
      "epoch": 0.5332866152768044,
      "grad_norm": 0.485036700963974,
      "learning_rate": 0.00025264290839512076,
      "loss": 0.0606,
      "step": 3805
    },
    {
      "epoch": 0.533426769446391,
      "grad_norm": 1.491607666015625,
      "learning_rate": 0.0002526285577613011,
      "loss": 0.0684,
      "step": 3806
    },
    {
      "epoch": 0.5335669236159776,
      "grad_norm": 1.566177248954773,
      "learning_rate": 0.0002526142071274814,
      "loss": 0.1531,
      "step": 3807
    },
    {
      "epoch": 0.5337070777855641,
      "grad_norm": 0.17010587453842163,
      "learning_rate": 0.0002525998564936618,
      "loss": 0.0436,
      "step": 3808
    },
    {
      "epoch": 0.5338472319551507,
      "grad_norm": 0.42140185832977295,
      "learning_rate": 0.00025258550585984213,
      "loss": 0.0678,
      "step": 3809
    },
    {
      "epoch": 0.5339873861247372,
      "grad_norm": 0.33653318881988525,
      "learning_rate": 0.00025257115522602246,
      "loss": 0.1267,
      "step": 3810
    },
    {
      "epoch": 0.5341275402943237,
      "grad_norm": 0.2026250660419464,
      "learning_rate": 0.0002525568045922028,
      "loss": 0.0687,
      "step": 3811
    },
    {
      "epoch": 0.5342676944639103,
      "grad_norm": 0.34078654646873474,
      "learning_rate": 0.0002525424539583831,
      "loss": 0.0603,
      "step": 3812
    },
    {
      "epoch": 0.5344078486334969,
      "grad_norm": 0.47780469059944153,
      "learning_rate": 0.00025252810332456344,
      "loss": 0.1145,
      "step": 3813
    },
    {
      "epoch": 0.5345480028030833,
      "grad_norm": 0.1717555820941925,
      "learning_rate": 0.0002525137526907438,
      "loss": 0.0524,
      "step": 3814
    },
    {
      "epoch": 0.5346881569726699,
      "grad_norm": 0.4112985134124756,
      "learning_rate": 0.00025249940205692415,
      "loss": 0.0686,
      "step": 3815
    },
    {
      "epoch": 0.5348283111422565,
      "grad_norm": 0.696688711643219,
      "learning_rate": 0.0002524850514231045,
      "loss": 0.1908,
      "step": 3816
    },
    {
      "epoch": 0.534968465311843,
      "grad_norm": 0.5086672902107239,
      "learning_rate": 0.00025247070078928487,
      "loss": 0.1447,
      "step": 3817
    },
    {
      "epoch": 0.5351086194814296,
      "grad_norm": 0.43156856298446655,
      "learning_rate": 0.0002524563501554652,
      "loss": 0.0527,
      "step": 3818
    },
    {
      "epoch": 0.5352487736510161,
      "grad_norm": 0.6361207962036133,
      "learning_rate": 0.0002524419995216455,
      "loss": 0.1017,
      "step": 3819
    },
    {
      "epoch": 0.5353889278206027,
      "grad_norm": 0.4026937782764435,
      "learning_rate": 0.00025242764888782585,
      "loss": 0.0967,
      "step": 3820
    },
    {
      "epoch": 0.5355290819901892,
      "grad_norm": 0.28759562969207764,
      "learning_rate": 0.0002524132982540062,
      "loss": 0.062,
      "step": 3821
    },
    {
      "epoch": 0.5356692361597758,
      "grad_norm": 0.8487041592597961,
      "learning_rate": 0.0002523989476201865,
      "loss": 0.1786,
      "step": 3822
    },
    {
      "epoch": 0.5358093903293623,
      "grad_norm": 0.48422151803970337,
      "learning_rate": 0.00025238459698636684,
      "loss": 0.0814,
      "step": 3823
    },
    {
      "epoch": 0.5359495444989488,
      "grad_norm": 0.3119770586490631,
      "learning_rate": 0.0002523702463525472,
      "loss": 0.0449,
      "step": 3824
    },
    {
      "epoch": 0.5360896986685354,
      "grad_norm": 0.778696596622467,
      "learning_rate": 0.00025235589571872755,
      "loss": 0.064,
      "step": 3825
    },
    {
      "epoch": 0.536229852838122,
      "grad_norm": 0.7794495820999146,
      "learning_rate": 0.0002523415450849079,
      "loss": 0.1609,
      "step": 3826
    },
    {
      "epoch": 0.5363700070077084,
      "grad_norm": 0.40856996178627014,
      "learning_rate": 0.00025232719445108826,
      "loss": 0.1005,
      "step": 3827
    },
    {
      "epoch": 0.536510161177295,
      "grad_norm": 0.46720045804977417,
      "learning_rate": 0.0002523128438172686,
      "loss": 0.0853,
      "step": 3828
    },
    {
      "epoch": 0.5366503153468816,
      "grad_norm": 0.47222110629081726,
      "learning_rate": 0.0002522984931834489,
      "loss": 0.0675,
      "step": 3829
    },
    {
      "epoch": 0.5367904695164681,
      "grad_norm": 0.3303963243961334,
      "learning_rate": 0.00025228414254962924,
      "loss": 0.0611,
      "step": 3830
    },
    {
      "epoch": 0.5369306236860547,
      "grad_norm": 0.34422165155410767,
      "learning_rate": 0.0002522697919158096,
      "loss": 0.1263,
      "step": 3831
    },
    {
      "epoch": 0.5370707778556412,
      "grad_norm": 0.7670153975486755,
      "learning_rate": 0.0002522554412819899,
      "loss": 0.188,
      "step": 3832
    },
    {
      "epoch": 0.5372109320252277,
      "grad_norm": 0.7226948738098145,
      "learning_rate": 0.0002522410906481703,
      "loss": 0.1451,
      "step": 3833
    },
    {
      "epoch": 0.5373510861948143,
      "grad_norm": 0.5015245676040649,
      "learning_rate": 0.0002522267400143506,
      "loss": 0.0627,
      "step": 3834
    },
    {
      "epoch": 0.5374912403644009,
      "grad_norm": 0.8689934611320496,
      "learning_rate": 0.00025221238938053094,
      "loss": 0.1289,
      "step": 3835
    },
    {
      "epoch": 0.5376313945339873,
      "grad_norm": 0.5005568265914917,
      "learning_rate": 0.0002521980387467113,
      "loss": 0.1736,
      "step": 3836
    },
    {
      "epoch": 0.5377715487035739,
      "grad_norm": 0.4905838668346405,
      "learning_rate": 0.00025218368811289165,
      "loss": 0.0801,
      "step": 3837
    },
    {
      "epoch": 0.5379117028731605,
      "grad_norm": 0.6516584753990173,
      "learning_rate": 0.000252169337479072,
      "loss": 0.1032,
      "step": 3838
    },
    {
      "epoch": 0.538051857042747,
      "grad_norm": 0.5888903141021729,
      "learning_rate": 0.0002521549868452523,
      "loss": 0.0943,
      "step": 3839
    },
    {
      "epoch": 0.5381920112123336,
      "grad_norm": 0.7239446043968201,
      "learning_rate": 0.00025214063621143264,
      "loss": 0.0973,
      "step": 3840
    },
    {
      "epoch": 0.5383321653819201,
      "grad_norm": 0.46840357780456543,
      "learning_rate": 0.00025212628557761297,
      "loss": 0.1203,
      "step": 3841
    },
    {
      "epoch": 0.5384723195515067,
      "grad_norm": 0.5807044506072998,
      "learning_rate": 0.0002521119349437933,
      "loss": 0.1201,
      "step": 3842
    },
    {
      "epoch": 0.5386124737210932,
      "grad_norm": 0.3918488323688507,
      "learning_rate": 0.0002520975843099737,
      "loss": 0.1207,
      "step": 3843
    },
    {
      "epoch": 0.5387526278906798,
      "grad_norm": 0.4665324091911316,
      "learning_rate": 0.000252083233676154,
      "loss": 0.0871,
      "step": 3844
    },
    {
      "epoch": 0.5388927820602663,
      "grad_norm": 1.6179182529449463,
      "learning_rate": 0.00025206888304233434,
      "loss": 0.2216,
      "step": 3845
    },
    {
      "epoch": 0.5390329362298528,
      "grad_norm": 0.9539428353309631,
      "learning_rate": 0.0002520545324085147,
      "loss": 0.2424,
      "step": 3846
    },
    {
      "epoch": 0.5391730903994394,
      "grad_norm": 1.303011178970337,
      "learning_rate": 0.00025204018177469505,
      "loss": 0.2224,
      "step": 3847
    },
    {
      "epoch": 0.539313244569026,
      "grad_norm": 0.44303184747695923,
      "learning_rate": 0.0002520258311408754,
      "loss": 0.0357,
      "step": 3848
    },
    {
      "epoch": 0.5394533987386124,
      "grad_norm": 2.4665374755859375,
      "learning_rate": 0.0002520114805070557,
      "loss": 0.4088,
      "step": 3849
    },
    {
      "epoch": 0.539593552908199,
      "grad_norm": 0.34303978085517883,
      "learning_rate": 0.00025199712987323603,
      "loss": 0.017,
      "step": 3850
    },
    {
      "epoch": 0.5397337070777856,
      "grad_norm": 0.2562355697154999,
      "learning_rate": 0.00025198277923941636,
      "loss": 0.0653,
      "step": 3851
    },
    {
      "epoch": 0.5398738612473721,
      "grad_norm": 0.37615686655044556,
      "learning_rate": 0.00025196842860559674,
      "loss": 0.1234,
      "step": 3852
    },
    {
      "epoch": 0.5400140154169587,
      "grad_norm": 0.6752070188522339,
      "learning_rate": 0.00025195407797177707,
      "loss": 0.0934,
      "step": 3853
    },
    {
      "epoch": 0.5401541695865452,
      "grad_norm": 0.7379781603813171,
      "learning_rate": 0.0002519397273379574,
      "loss": 0.1235,
      "step": 3854
    },
    {
      "epoch": 0.5402943237561317,
      "grad_norm": 0.34804943203926086,
      "learning_rate": 0.0002519253767041378,
      "loss": 0.0816,
      "step": 3855
    },
    {
      "epoch": 0.5404344779257183,
      "grad_norm": 0.34477242827415466,
      "learning_rate": 0.0002519110260703181,
      "loss": 0.1409,
      "step": 3856
    },
    {
      "epoch": 0.5405746320953049,
      "grad_norm": 0.31011903285980225,
      "learning_rate": 0.00025189667543649844,
      "loss": 0.07,
      "step": 3857
    },
    {
      "epoch": 0.5407147862648913,
      "grad_norm": 0.6184453964233398,
      "learning_rate": 0.00025188232480267877,
      "loss": 0.0843,
      "step": 3858
    },
    {
      "epoch": 0.5408549404344779,
      "grad_norm": 0.5770922303199768,
      "learning_rate": 0.0002518679741688591,
      "loss": 0.1214,
      "step": 3859
    },
    {
      "epoch": 0.5409950946040645,
      "grad_norm": 0.5008496046066284,
      "learning_rate": 0.0002518536235350394,
      "loss": 0.1714,
      "step": 3860
    },
    {
      "epoch": 0.541135248773651,
      "grad_norm": 0.2567344307899475,
      "learning_rate": 0.00025183927290121976,
      "loss": 0.0547,
      "step": 3861
    },
    {
      "epoch": 0.5412754029432376,
      "grad_norm": 0.2598913013935089,
      "learning_rate": 0.00025182492226740014,
      "loss": 0.1308,
      "step": 3862
    },
    {
      "epoch": 0.5414155571128241,
      "grad_norm": 0.33479151129722595,
      "learning_rate": 0.00025181057163358047,
      "loss": 0.1038,
      "step": 3863
    },
    {
      "epoch": 0.5415557112824106,
      "grad_norm": 0.5118029713630676,
      "learning_rate": 0.0002517962209997608,
      "loss": 0.0638,
      "step": 3864
    },
    {
      "epoch": 0.5416958654519972,
      "grad_norm": 0.5354476571083069,
      "learning_rate": 0.0002517818703659412,
      "loss": 0.0702,
      "step": 3865
    },
    {
      "epoch": 0.5418360196215838,
      "grad_norm": 1.1596344709396362,
      "learning_rate": 0.0002517675197321215,
      "loss": 0.3,
      "step": 3866
    },
    {
      "epoch": 0.5419761737911702,
      "grad_norm": 0.25599274039268494,
      "learning_rate": 0.00025175316909830184,
      "loss": 0.0861,
      "step": 3867
    },
    {
      "epoch": 0.5421163279607568,
      "grad_norm": 0.6188617944717407,
      "learning_rate": 0.00025173881846448216,
      "loss": 0.1269,
      "step": 3868
    },
    {
      "epoch": 0.5422564821303434,
      "grad_norm": 0.3649226725101471,
      "learning_rate": 0.0002517244678306625,
      "loss": 0.1292,
      "step": 3869
    },
    {
      "epoch": 0.54239663629993,
      "grad_norm": 0.686793327331543,
      "learning_rate": 0.0002517101171968428,
      "loss": 0.3501,
      "step": 3870
    },
    {
      "epoch": 0.5425367904695164,
      "grad_norm": 0.25735345482826233,
      "learning_rate": 0.0002516957665630232,
      "loss": 0.1089,
      "step": 3871
    },
    {
      "epoch": 0.542676944639103,
      "grad_norm": 0.33736652135849,
      "learning_rate": 0.00025168141592920353,
      "loss": 0.108,
      "step": 3872
    },
    {
      "epoch": 0.5428170988086896,
      "grad_norm": 0.6944626569747925,
      "learning_rate": 0.00025166706529538386,
      "loss": 0.1366,
      "step": 3873
    },
    {
      "epoch": 0.5429572529782761,
      "grad_norm": 0.34482455253601074,
      "learning_rate": 0.0002516527146615642,
      "loss": 0.1214,
      "step": 3874
    },
    {
      "epoch": 0.5430974071478627,
      "grad_norm": 0.38985881209373474,
      "learning_rate": 0.0002516383640277445,
      "loss": 0.1173,
      "step": 3875
    },
    {
      "epoch": 0.5432375613174492,
      "grad_norm": 1.4831211566925049,
      "learning_rate": 0.00025162401339392485,
      "loss": 0.1882,
      "step": 3876
    },
    {
      "epoch": 0.5433777154870357,
      "grad_norm": 0.41491633653640747,
      "learning_rate": 0.0002516096627601052,
      "loss": 0.1124,
      "step": 3877
    },
    {
      "epoch": 0.5435178696566223,
      "grad_norm": 0.5769524574279785,
      "learning_rate": 0.00025159531212628556,
      "loss": 0.1211,
      "step": 3878
    },
    {
      "epoch": 0.5436580238262089,
      "grad_norm": 0.33956480026245117,
      "learning_rate": 0.0002515809614924659,
      "loss": 0.0898,
      "step": 3879
    },
    {
      "epoch": 0.5437981779957953,
      "grad_norm": 0.943324089050293,
      "learning_rate": 0.0002515666108586462,
      "loss": 0.1766,
      "step": 3880
    },
    {
      "epoch": 0.5439383321653819,
      "grad_norm": 0.618470311164856,
      "learning_rate": 0.0002515522602248266,
      "loss": 0.1246,
      "step": 3881
    },
    {
      "epoch": 0.5440784863349685,
      "grad_norm": 0.7491851449012756,
      "learning_rate": 0.0002515379095910069,
      "loss": 0.1039,
      "step": 3882
    },
    {
      "epoch": 0.544218640504555,
      "grad_norm": 0.38389891386032104,
      "learning_rate": 0.00025152355895718725,
      "loss": 0.1167,
      "step": 3883
    },
    {
      "epoch": 0.5443587946741416,
      "grad_norm": 0.3989713788032532,
      "learning_rate": 0.0002515092083233676,
      "loss": 0.0456,
      "step": 3884
    },
    {
      "epoch": 0.5444989488437281,
      "grad_norm": 0.6277985572814941,
      "learning_rate": 0.0002514948576895479,
      "loss": 0.0933,
      "step": 3885
    },
    {
      "epoch": 0.5446391030133146,
      "grad_norm": 0.40347084403038025,
      "learning_rate": 0.00025148050705572824,
      "loss": 0.0725,
      "step": 3886
    },
    {
      "epoch": 0.5447792571829012,
      "grad_norm": 0.5757158994674683,
      "learning_rate": 0.0002514661564219086,
      "loss": 0.1307,
      "step": 3887
    },
    {
      "epoch": 0.5449194113524878,
      "grad_norm": 0.6314204335212708,
      "learning_rate": 0.00025145180578808895,
      "loss": 0.2435,
      "step": 3888
    },
    {
      "epoch": 0.5450595655220742,
      "grad_norm": 0.34765008091926575,
      "learning_rate": 0.0002514374551542693,
      "loss": 0.0891,
      "step": 3889
    },
    {
      "epoch": 0.5451997196916608,
      "grad_norm": 0.2535039484500885,
      "learning_rate": 0.00025142310452044966,
      "loss": 0.0591,
      "step": 3890
    },
    {
      "epoch": 0.5453398738612474,
      "grad_norm": 2.5984435081481934,
      "learning_rate": 0.00025140875388663,
      "loss": 0.2619,
      "step": 3891
    },
    {
      "epoch": 0.545480028030834,
      "grad_norm": 0.5420457720756531,
      "learning_rate": 0.0002513944032528103,
      "loss": 0.07,
      "step": 3892
    },
    {
      "epoch": 0.5456201822004204,
      "grad_norm": 0.5296517014503479,
      "learning_rate": 0.00025138005261899065,
      "loss": 0.105,
      "step": 3893
    },
    {
      "epoch": 0.545760336370007,
      "grad_norm": 0.6605236530303955,
      "learning_rate": 0.000251365701985171,
      "loss": 0.1957,
      "step": 3894
    },
    {
      "epoch": 0.5459004905395936,
      "grad_norm": 0.8708084225654602,
      "learning_rate": 0.0002513513513513513,
      "loss": 0.0853,
      "step": 3895
    },
    {
      "epoch": 0.5460406447091801,
      "grad_norm": 0.8178973197937012,
      "learning_rate": 0.00025133700071753163,
      "loss": 0.1541,
      "step": 3896
    },
    {
      "epoch": 0.5461807988787667,
      "grad_norm": 1.6926020383834839,
      "learning_rate": 0.000251322650083712,
      "loss": 0.2722,
      "step": 3897
    },
    {
      "epoch": 0.5463209530483532,
      "grad_norm": 0.3676035404205322,
      "learning_rate": 0.00025130829944989235,
      "loss": 0.0206,
      "step": 3898
    },
    {
      "epoch": 0.5464611072179397,
      "grad_norm": 0.8457313179969788,
      "learning_rate": 0.0002512939488160727,
      "loss": 0.117,
      "step": 3899
    },
    {
      "epoch": 0.5466012613875263,
      "grad_norm": 2.192903757095337,
      "learning_rate": 0.00025127959818225306,
      "loss": 0.5269,
      "step": 3900
    },
    {
      "epoch": 0.5467414155571129,
      "grad_norm": 0.48463279008865356,
      "learning_rate": 0.0002512652475484334,
      "loss": 0.1266,
      "step": 3901
    },
    {
      "epoch": 0.5468815697266993,
      "grad_norm": 0.3723304569721222,
      "learning_rate": 0.0002512508969146137,
      "loss": 0.1186,
      "step": 3902
    },
    {
      "epoch": 0.5470217238962859,
      "grad_norm": 0.5379852056503296,
      "learning_rate": 0.00025123654628079404,
      "loss": 0.1144,
      "step": 3903
    },
    {
      "epoch": 0.5471618780658725,
      "grad_norm": 0.24861934781074524,
      "learning_rate": 0.00025122219564697437,
      "loss": 0.0879,
      "step": 3904
    },
    {
      "epoch": 0.547302032235459,
      "grad_norm": 0.4434549808502197,
      "learning_rate": 0.0002512078450131547,
      "loss": 0.1819,
      "step": 3905
    },
    {
      "epoch": 0.5474421864050456,
      "grad_norm": 0.5221505761146545,
      "learning_rate": 0.0002511934943793351,
      "loss": 0.1046,
      "step": 3906
    },
    {
      "epoch": 0.5475823405746321,
      "grad_norm": 0.6591808795928955,
      "learning_rate": 0.0002511791437455154,
      "loss": 0.1118,
      "step": 3907
    },
    {
      "epoch": 0.5477224947442186,
      "grad_norm": 1.1716365814208984,
      "learning_rate": 0.00025116479311169574,
      "loss": 0.119,
      "step": 3908
    },
    {
      "epoch": 0.5478626489138052,
      "grad_norm": 0.5857277512550354,
      "learning_rate": 0.00025115044247787607,
      "loss": 0.1747,
      "step": 3909
    },
    {
      "epoch": 0.5480028030833918,
      "grad_norm": 0.29456159472465515,
      "learning_rate": 0.00025113609184405645,
      "loss": 0.0642,
      "step": 3910
    },
    {
      "epoch": 0.5481429572529782,
      "grad_norm": 0.45870885252952576,
      "learning_rate": 0.0002511217412102368,
      "loss": 0.0503,
      "step": 3911
    },
    {
      "epoch": 0.5482831114225648,
      "grad_norm": 0.45464789867401123,
      "learning_rate": 0.0002511073905764171,
      "loss": 0.0872,
      "step": 3912
    },
    {
      "epoch": 0.5484232655921514,
      "grad_norm": 0.40002551674842834,
      "learning_rate": 0.00025109303994259744,
      "loss": 0.076,
      "step": 3913
    },
    {
      "epoch": 0.548563419761738,
      "grad_norm": 0.4472227990627289,
      "learning_rate": 0.00025107868930877777,
      "loss": 0.0843,
      "step": 3914
    },
    {
      "epoch": 0.5487035739313244,
      "grad_norm": 1.1812400817871094,
      "learning_rate": 0.0002510643386749581,
      "loss": 0.3072,
      "step": 3915
    },
    {
      "epoch": 0.548843728100911,
      "grad_norm": 0.6882503628730774,
      "learning_rate": 0.0002510499880411385,
      "loss": 0.2001,
      "step": 3916
    },
    {
      "epoch": 0.5489838822704975,
      "grad_norm": 0.5562772750854492,
      "learning_rate": 0.0002510356374073188,
      "loss": 0.0554,
      "step": 3917
    },
    {
      "epoch": 0.5491240364400841,
      "grad_norm": 0.38235458731651306,
      "learning_rate": 0.00025102128677349913,
      "loss": 0.1116,
      "step": 3918
    },
    {
      "epoch": 0.5492641906096707,
      "grad_norm": 0.3034262955188751,
      "learning_rate": 0.0002510069361396795,
      "loss": 0.0751,
      "step": 3919
    },
    {
      "epoch": 0.5494043447792571,
      "grad_norm": 0.36013907194137573,
      "learning_rate": 0.00025099258550585985,
      "loss": 0.0735,
      "step": 3920
    },
    {
      "epoch": 0.5495444989488437,
      "grad_norm": 0.378262996673584,
      "learning_rate": 0.0002509782348720402,
      "loss": 0.0862,
      "step": 3921
    },
    {
      "epoch": 0.5496846531184303,
      "grad_norm": 0.21409563720226288,
      "learning_rate": 0.0002509638842382205,
      "loss": 0.0461,
      "step": 3922
    },
    {
      "epoch": 0.5498248072880169,
      "grad_norm": 0.7046626210212708,
      "learning_rate": 0.00025094953360440083,
      "loss": 0.1074,
      "step": 3923
    },
    {
      "epoch": 0.5499649614576033,
      "grad_norm": 0.16295380890369415,
      "learning_rate": 0.00025093518297058116,
      "loss": 0.0396,
      "step": 3924
    },
    {
      "epoch": 0.5501051156271899,
      "grad_norm": 0.56124347448349,
      "learning_rate": 0.00025092083233676154,
      "loss": 0.1418,
      "step": 3925
    },
    {
      "epoch": 0.5502452697967765,
      "grad_norm": 0.6420082449913025,
      "learning_rate": 0.00025090648170294187,
      "loss": 0.1873,
      "step": 3926
    },
    {
      "epoch": 0.550385423966363,
      "grad_norm": 0.488039493560791,
      "learning_rate": 0.0002508921310691222,
      "loss": 0.1087,
      "step": 3927
    },
    {
      "epoch": 0.5505255781359495,
      "grad_norm": 0.7287023067474365,
      "learning_rate": 0.00025087778043530253,
      "loss": 0.1071,
      "step": 3928
    },
    {
      "epoch": 0.5506657323055361,
      "grad_norm": 0.34821316599845886,
      "learning_rate": 0.0002508634298014829,
      "loss": 0.0925,
      "step": 3929
    },
    {
      "epoch": 0.5508058864751226,
      "grad_norm": 0.8709393739700317,
      "learning_rate": 0.00025084907916766324,
      "loss": 0.1277,
      "step": 3930
    },
    {
      "epoch": 0.5509460406447092,
      "grad_norm": 0.6097391247749329,
      "learning_rate": 0.00025083472853384357,
      "loss": 0.0714,
      "step": 3931
    },
    {
      "epoch": 0.5510861948142958,
      "grad_norm": 0.5883945822715759,
      "learning_rate": 0.0002508203779000239,
      "loss": 0.1688,
      "step": 3932
    },
    {
      "epoch": 0.5512263489838822,
      "grad_norm": 0.4826839566230774,
      "learning_rate": 0.0002508060272662042,
      "loss": 0.0724,
      "step": 3933
    },
    {
      "epoch": 0.5513665031534688,
      "grad_norm": 0.7326350808143616,
      "learning_rate": 0.00025079167663238455,
      "loss": 0.0539,
      "step": 3934
    },
    {
      "epoch": 0.5515066573230554,
      "grad_norm": 0.35331735014915466,
      "learning_rate": 0.00025077732599856494,
      "loss": 0.0556,
      "step": 3935
    },
    {
      "epoch": 0.5516468114926419,
      "grad_norm": 0.4192046821117401,
      "learning_rate": 0.00025076297536474526,
      "loss": 0.0539,
      "step": 3936
    },
    {
      "epoch": 0.5517869656622284,
      "grad_norm": 0.6254921555519104,
      "learning_rate": 0.0002507486247309256,
      "loss": 0.0825,
      "step": 3937
    },
    {
      "epoch": 0.551927119831815,
      "grad_norm": 0.7162885665893555,
      "learning_rate": 0.0002507342740971059,
      "loss": 0.0957,
      "step": 3938
    },
    {
      "epoch": 0.5520672740014015,
      "grad_norm": 0.5629931092262268,
      "learning_rate": 0.00025071992346328625,
      "loss": 0.0493,
      "step": 3939
    },
    {
      "epoch": 0.5522074281709881,
      "grad_norm": 0.6187030076980591,
      "learning_rate": 0.0002507055728294666,
      "loss": 0.0646,
      "step": 3940
    },
    {
      "epoch": 0.5523475823405747,
      "grad_norm": 0.6613685488700867,
      "learning_rate": 0.00025069122219564696,
      "loss": 0.1355,
      "step": 3941
    },
    {
      "epoch": 0.5524877365101611,
      "grad_norm": 0.7194153666496277,
      "learning_rate": 0.0002506768715618273,
      "loss": 0.0883,
      "step": 3942
    },
    {
      "epoch": 0.5526278906797477,
      "grad_norm": 0.7751923203468323,
      "learning_rate": 0.0002506625209280076,
      "loss": 0.1585,
      "step": 3943
    },
    {
      "epoch": 0.5527680448493343,
      "grad_norm": 0.4823018014431,
      "learning_rate": 0.00025064817029418795,
      "loss": 0.136,
      "step": 3944
    },
    {
      "epoch": 0.5529081990189209,
      "grad_norm": 0.5525204539299011,
      "learning_rate": 0.00025063381966036833,
      "loss": 0.0681,
      "step": 3945
    },
    {
      "epoch": 0.5530483531885073,
      "grad_norm": 0.7681614756584167,
      "learning_rate": 0.00025061946902654866,
      "loss": 0.0997,
      "step": 3946
    },
    {
      "epoch": 0.5531885073580939,
      "grad_norm": 0.36747393012046814,
      "learning_rate": 0.000250605118392729,
      "loss": 0.087,
      "step": 3947
    },
    {
      "epoch": 0.5533286615276805,
      "grad_norm": 0.6003926992416382,
      "learning_rate": 0.0002505907677589093,
      "loss": 0.1021,
      "step": 3948
    },
    {
      "epoch": 0.553468815697267,
      "grad_norm": 0.624787449836731,
      "learning_rate": 0.00025057641712508964,
      "loss": 0.1031,
      "step": 3949
    },
    {
      "epoch": 0.5536089698668535,
      "grad_norm": 2.234396457672119,
      "learning_rate": 0.00025056206649127,
      "loss": 0.6069,
      "step": 3950
    },
    {
      "epoch": 0.55374912403644,
      "grad_norm": 0.2522006928920746,
      "learning_rate": 0.00025054771585745036,
      "loss": 0.0566,
      "step": 3951
    },
    {
      "epoch": 0.5538892782060266,
      "grad_norm": 0.40768468379974365,
      "learning_rate": 0.0002505333652236307,
      "loss": 0.0913,
      "step": 3952
    },
    {
      "epoch": 0.5540294323756132,
      "grad_norm": 0.2317289113998413,
      "learning_rate": 0.000250519014589811,
      "loss": 0.0453,
      "step": 3953
    },
    {
      "epoch": 0.5541695865451998,
      "grad_norm": 0.63600093126297,
      "learning_rate": 0.0002505046639559914,
      "loss": 0.1177,
      "step": 3954
    },
    {
      "epoch": 0.5543097407147862,
      "grad_norm": 0.5825256705284119,
      "learning_rate": 0.0002504903133221717,
      "loss": 0.1402,
      "step": 3955
    },
    {
      "epoch": 0.5544498948843728,
      "grad_norm": 0.3389924466609955,
      "learning_rate": 0.00025047596268835205,
      "loss": 0.1094,
      "step": 3956
    },
    {
      "epoch": 0.5545900490539594,
      "grad_norm": 0.6104356646537781,
      "learning_rate": 0.0002504616120545324,
      "loss": 0.1691,
      "step": 3957
    },
    {
      "epoch": 0.5547302032235459,
      "grad_norm": 0.1497412919998169,
      "learning_rate": 0.0002504472614207127,
      "loss": 0.0296,
      "step": 3958
    },
    {
      "epoch": 0.5548703573931324,
      "grad_norm": 0.4638878405094147,
      "learning_rate": 0.00025043291078689304,
      "loss": 0.0746,
      "step": 3959
    },
    {
      "epoch": 0.555010511562719,
      "grad_norm": 0.4135279953479767,
      "learning_rate": 0.0002504185601530734,
      "loss": 0.1493,
      "step": 3960
    },
    {
      "epoch": 0.5551506657323055,
      "grad_norm": 0.3829207122325897,
      "learning_rate": 0.00025040420951925375,
      "loss": 0.1188,
      "step": 3961
    },
    {
      "epoch": 0.5552908199018921,
      "grad_norm": 0.2558688819408417,
      "learning_rate": 0.0002503898588854341,
      "loss": 0.0628,
      "step": 3962
    },
    {
      "epoch": 0.5554309740714787,
      "grad_norm": 0.4334415793418884,
      "learning_rate": 0.0002503755082516144,
      "loss": 0.0449,
      "step": 3963
    },
    {
      "epoch": 0.5555711282410651,
      "grad_norm": 0.46161124110221863,
      "learning_rate": 0.0002503611576177948,
      "loss": 0.08,
      "step": 3964
    },
    {
      "epoch": 0.5557112824106517,
      "grad_norm": 0.6392189860343933,
      "learning_rate": 0.0002503468069839751,
      "loss": 0.1077,
      "step": 3965
    },
    {
      "epoch": 0.5558514365802383,
      "grad_norm": 0.6778882741928101,
      "learning_rate": 0.00025033245635015545,
      "loss": 0.1234,
      "step": 3966
    },
    {
      "epoch": 0.5559915907498248,
      "grad_norm": 0.3571174144744873,
      "learning_rate": 0.0002503181057163358,
      "loss": 0.0796,
      "step": 3967
    },
    {
      "epoch": 0.5561317449194113,
      "grad_norm": 0.5710231065750122,
      "learning_rate": 0.0002503037550825161,
      "loss": 0.1053,
      "step": 3968
    },
    {
      "epoch": 0.5562718990889979,
      "grad_norm": 0.4297938048839569,
      "learning_rate": 0.00025028940444869643,
      "loss": 0.1299,
      "step": 3969
    },
    {
      "epoch": 0.5564120532585844,
      "grad_norm": 0.34932661056518555,
      "learning_rate": 0.0002502750538148768,
      "loss": 0.0847,
      "step": 3970
    },
    {
      "epoch": 0.556552207428171,
      "grad_norm": 0.82928866147995,
      "learning_rate": 0.00025026070318105714,
      "loss": 0.076,
      "step": 3971
    },
    {
      "epoch": 0.5566923615977575,
      "grad_norm": 0.36403122544288635,
      "learning_rate": 0.00025024635254723747,
      "loss": 0.1072,
      "step": 3972
    },
    {
      "epoch": 0.556832515767344,
      "grad_norm": 0.6261738538742065,
      "learning_rate": 0.00025023200191341786,
      "loss": 0.0855,
      "step": 3973
    },
    {
      "epoch": 0.5569726699369306,
      "grad_norm": 0.39374426007270813,
      "learning_rate": 0.0002502176512795982,
      "loss": 0.117,
      "step": 3974
    },
    {
      "epoch": 0.5571128241065172,
      "grad_norm": 0.26938101649284363,
      "learning_rate": 0.0002502033006457785,
      "loss": 0.1074,
      "step": 3975
    },
    {
      "epoch": 0.5572529782761038,
      "grad_norm": 0.2874991297721863,
      "learning_rate": 0.00025018895001195884,
      "loss": 0.0434,
      "step": 3976
    },
    {
      "epoch": 0.5573931324456902,
      "grad_norm": 0.19152790307998657,
      "learning_rate": 0.00025017459937813917,
      "loss": 0.0195,
      "step": 3977
    },
    {
      "epoch": 0.5575332866152768,
      "grad_norm": 0.7606933116912842,
      "learning_rate": 0.0002501602487443195,
      "loss": 0.117,
      "step": 3978
    },
    {
      "epoch": 0.5576734407848634,
      "grad_norm": 0.3154715299606323,
      "learning_rate": 0.0002501458981104998,
      "loss": 0.0322,
      "step": 3979
    },
    {
      "epoch": 0.5578135949544499,
      "grad_norm": 0.6483924388885498,
      "learning_rate": 0.0002501315474766802,
      "loss": 0.0402,
      "step": 3980
    },
    {
      "epoch": 0.5579537491240364,
      "grad_norm": 0.35352852940559387,
      "learning_rate": 0.00025011719684286054,
      "loss": 0.0186,
      "step": 3981
    },
    {
      "epoch": 0.558093903293623,
      "grad_norm": 0.646351158618927,
      "learning_rate": 0.00025010284620904087,
      "loss": 0.1501,
      "step": 3982
    },
    {
      "epoch": 0.5582340574632095,
      "grad_norm": 0.28428393602371216,
      "learning_rate": 0.00025008849557522125,
      "loss": 0.0779,
      "step": 3983
    },
    {
      "epoch": 0.5583742116327961,
      "grad_norm": 0.24462680518627167,
      "learning_rate": 0.0002500741449414016,
      "loss": 0.0289,
      "step": 3984
    },
    {
      "epoch": 0.5585143658023827,
      "grad_norm": 0.8907936811447144,
      "learning_rate": 0.0002500597943075819,
      "loss": 0.1823,
      "step": 3985
    },
    {
      "epoch": 0.5586545199719691,
      "grad_norm": 0.3714198172092438,
      "learning_rate": 0.00025004544367376223,
      "loss": 0.0796,
      "step": 3986
    },
    {
      "epoch": 0.5587946741415557,
      "grad_norm": 0.26748713850975037,
      "learning_rate": 0.00025003109303994256,
      "loss": 0.0749,
      "step": 3987
    },
    {
      "epoch": 0.5589348283111423,
      "grad_norm": 0.3552827537059784,
      "learning_rate": 0.0002500167424061229,
      "loss": 0.0674,
      "step": 3988
    },
    {
      "epoch": 0.5590749824807288,
      "grad_norm": 0.30290699005126953,
      "learning_rate": 0.0002500023917723033,
      "loss": 0.1022,
      "step": 3989
    },
    {
      "epoch": 0.5592151366503153,
      "grad_norm": 0.38541123270988464,
      "learning_rate": 0.0002499880411384836,
      "loss": 0.0574,
      "step": 3990
    },
    {
      "epoch": 0.5593552908199019,
      "grad_norm": 0.8959506154060364,
      "learning_rate": 0.00024997369050466393,
      "loss": 0.1114,
      "step": 3991
    },
    {
      "epoch": 0.5594954449894884,
      "grad_norm": 0.46974635124206543,
      "learning_rate": 0.0002499593398708443,
      "loss": 0.0544,
      "step": 3992
    },
    {
      "epoch": 0.559635599159075,
      "grad_norm": 0.5642286539077759,
      "learning_rate": 0.00024994498923702464,
      "loss": 0.1147,
      "step": 3993
    },
    {
      "epoch": 0.5597757533286615,
      "grad_norm": 0.6111794114112854,
      "learning_rate": 0.00024993063860320497,
      "loss": 0.1448,
      "step": 3994
    },
    {
      "epoch": 0.559915907498248,
      "grad_norm": 0.3942194879055023,
      "learning_rate": 0.0002499162879693853,
      "loss": 0.0425,
      "step": 3995
    },
    {
      "epoch": 0.5600560616678346,
      "grad_norm": 0.3252815008163452,
      "learning_rate": 0.00024990193733556563,
      "loss": 0.0609,
      "step": 3996
    },
    {
      "epoch": 0.5601962158374212,
      "grad_norm": 1.5041699409484863,
      "learning_rate": 0.00024988758670174596,
      "loss": 0.1565,
      "step": 3997
    },
    {
      "epoch": 0.5603363700070078,
      "grad_norm": 2.7567338943481445,
      "learning_rate": 0.0002498732360679263,
      "loss": 0.3778,
      "step": 3998
    },
    {
      "epoch": 0.5604765241765942,
      "grad_norm": 1.8308205604553223,
      "learning_rate": 0.00024985888543410667,
      "loss": 0.2435,
      "step": 3999
    },
    {
      "epoch": 0.5606166783461808,
      "grad_norm": 1.7873958349227905,
      "learning_rate": 0.000249844534800287,
      "loss": 0.364,
      "step": 4000
    },
    {
      "epoch": 0.5607568325157674,
      "grad_norm": 0.4975601136684418,
      "learning_rate": 0.0002498301841664673,
      "loss": 0.1152,
      "step": 4001
    },
    {
      "epoch": 0.5608969866853539,
      "grad_norm": 0.3295333683490753,
      "learning_rate": 0.00024981583353264765,
      "loss": 0.081,
      "step": 4002
    },
    {
      "epoch": 0.5610371408549404,
      "grad_norm": 0.3756139874458313,
      "learning_rate": 0.000249801482898828,
      "loss": 0.1317,
      "step": 4003
    },
    {
      "epoch": 0.561177295024527,
      "grad_norm": 0.48621007800102234,
      "learning_rate": 0.0002497871322650083,
      "loss": 0.1518,
      "step": 4004
    },
    {
      "epoch": 0.5613174491941135,
      "grad_norm": 0.3367638885974884,
      "learning_rate": 0.0002497727816311887,
      "loss": 0.1321,
      "step": 4005
    },
    {
      "epoch": 0.5614576033637001,
      "grad_norm": 0.2516014277935028,
      "learning_rate": 0.000249758430997369,
      "loss": 0.0399,
      "step": 4006
    },
    {
      "epoch": 0.5615977575332867,
      "grad_norm": 0.44831156730651855,
      "learning_rate": 0.00024974408036354935,
      "loss": 0.0791,
      "step": 4007
    },
    {
      "epoch": 0.5617379117028731,
      "grad_norm": 0.5114452242851257,
      "learning_rate": 0.00024972972972972973,
      "loss": 0.1112,
      "step": 4008
    },
    {
      "epoch": 0.5618780658724597,
      "grad_norm": 0.21150460839271545,
      "learning_rate": 0.00024971537909591006,
      "loss": 0.0429,
      "step": 4009
    },
    {
      "epoch": 0.5620182200420463,
      "grad_norm": 0.3199412226676941,
      "learning_rate": 0.0002497010284620904,
      "loss": 0.0751,
      "step": 4010
    },
    {
      "epoch": 0.5621583742116328,
      "grad_norm": 0.6559122800827026,
      "learning_rate": 0.0002496866778282707,
      "loss": 0.1154,
      "step": 4011
    },
    {
      "epoch": 0.5622985283812193,
      "grad_norm": 0.37633848190307617,
      "learning_rate": 0.00024967232719445105,
      "loss": 0.0768,
      "step": 4012
    },
    {
      "epoch": 0.5624386825508059,
      "grad_norm": 0.27470019459724426,
      "learning_rate": 0.0002496579765606314,
      "loss": 0.1047,
      "step": 4013
    },
    {
      "epoch": 0.5625788367203924,
      "grad_norm": 0.4732481837272644,
      "learning_rate": 0.0002496436259268117,
      "loss": 0.1782,
      "step": 4014
    },
    {
      "epoch": 0.562718990889979,
      "grad_norm": 0.3817456364631653,
      "learning_rate": 0.0002496292752929921,
      "loss": 0.1026,
      "step": 4015
    },
    {
      "epoch": 0.5628591450595655,
      "grad_norm": 0.6931123733520508,
      "learning_rate": 0.0002496149246591724,
      "loss": 0.1483,
      "step": 4016
    },
    {
      "epoch": 0.562999299229152,
      "grad_norm": 0.3439079225063324,
      "learning_rate": 0.00024960057402535275,
      "loss": 0.1594,
      "step": 4017
    },
    {
      "epoch": 0.5631394533987386,
      "grad_norm": 0.47241100668907166,
      "learning_rate": 0.00024958622339153313,
      "loss": 0.0952,
      "step": 4018
    },
    {
      "epoch": 0.5632796075683252,
      "grad_norm": 0.2326921820640564,
      "learning_rate": 0.00024957187275771346,
      "loss": 0.0464,
      "step": 4019
    },
    {
      "epoch": 0.5634197617379118,
      "grad_norm": 0.4949483275413513,
      "learning_rate": 0.0002495575221238938,
      "loss": 0.1201,
      "step": 4020
    },
    {
      "epoch": 0.5635599159074982,
      "grad_norm": 0.26340755820274353,
      "learning_rate": 0.0002495431714900741,
      "loss": 0.0397,
      "step": 4021
    },
    {
      "epoch": 0.5637000700770848,
      "grad_norm": 0.42154282331466675,
      "learning_rate": 0.00024952882085625444,
      "loss": 0.0497,
      "step": 4022
    },
    {
      "epoch": 0.5638402242466714,
      "grad_norm": 0.42211347818374634,
      "learning_rate": 0.00024951447022243477,
      "loss": 0.063,
      "step": 4023
    },
    {
      "epoch": 0.5639803784162579,
      "grad_norm": 0.40765026211738586,
      "learning_rate": 0.00024950011958861515,
      "loss": 0.1137,
      "step": 4024
    },
    {
      "epoch": 0.5641205325858444,
      "grad_norm": 0.5014655590057373,
      "learning_rate": 0.0002494857689547955,
      "loss": 0.1083,
      "step": 4025
    },
    {
      "epoch": 0.564260686755431,
      "grad_norm": 0.5584135055541992,
      "learning_rate": 0.0002494714183209758,
      "loss": 0.1316,
      "step": 4026
    },
    {
      "epoch": 0.5644008409250175,
      "grad_norm": 0.49080297350883484,
      "learning_rate": 0.0002494570676871562,
      "loss": 0.0619,
      "step": 4027
    },
    {
      "epoch": 0.5645409950946041,
      "grad_norm": 0.5462132096290588,
      "learning_rate": 0.0002494427170533365,
      "loss": 0.1033,
      "step": 4028
    },
    {
      "epoch": 0.5646811492641907,
      "grad_norm": 0.7864841818809509,
      "learning_rate": 0.00024942836641951685,
      "loss": 0.1765,
      "step": 4029
    },
    {
      "epoch": 0.5648213034337771,
      "grad_norm": 0.19623635709285736,
      "learning_rate": 0.0002494140157856972,
      "loss": 0.0295,
      "step": 4030
    },
    {
      "epoch": 0.5649614576033637,
      "grad_norm": 1.5547457933425903,
      "learning_rate": 0.0002493996651518775,
      "loss": 0.1645,
      "step": 4031
    },
    {
      "epoch": 0.5651016117729503,
      "grad_norm": 0.25827813148498535,
      "learning_rate": 0.00024938531451805784,
      "loss": 0.0641,
      "step": 4032
    },
    {
      "epoch": 0.5652417659425368,
      "grad_norm": 0.5804337859153748,
      "learning_rate": 0.00024937096388423817,
      "loss": 0.0906,
      "step": 4033
    },
    {
      "epoch": 0.5653819201121233,
      "grad_norm": 0.21657679975032806,
      "learning_rate": 0.00024935661325041855,
      "loss": 0.0441,
      "step": 4034
    },
    {
      "epoch": 0.5655220742817099,
      "grad_norm": 0.41447126865386963,
      "learning_rate": 0.0002493422626165989,
      "loss": 0.1223,
      "step": 4035
    },
    {
      "epoch": 0.5656622284512964,
      "grad_norm": 0.5161694884300232,
      "learning_rate": 0.0002493279119827792,
      "loss": 0.131,
      "step": 4036
    },
    {
      "epoch": 0.565802382620883,
      "grad_norm": 0.36643195152282715,
      "learning_rate": 0.0002493135613489596,
      "loss": 0.056,
      "step": 4037
    },
    {
      "epoch": 0.5659425367904695,
      "grad_norm": 0.4670483469963074,
      "learning_rate": 0.0002492992107151399,
      "loss": 0.1201,
      "step": 4038
    },
    {
      "epoch": 0.566082690960056,
      "grad_norm": 0.5224334001541138,
      "learning_rate": 0.00024928486008132024,
      "loss": 0.103,
      "step": 4039
    },
    {
      "epoch": 0.5662228451296426,
      "grad_norm": 0.42706790566444397,
      "learning_rate": 0.0002492705094475006,
      "loss": 0.0639,
      "step": 4040
    },
    {
      "epoch": 0.5663629992992292,
      "grad_norm": 0.2869316339492798,
      "learning_rate": 0.0002492561588136809,
      "loss": 0.0215,
      "step": 4041
    },
    {
      "epoch": 0.5665031534688157,
      "grad_norm": 0.40846318006515503,
      "learning_rate": 0.00024924180817986123,
      "loss": 0.0555,
      "step": 4042
    },
    {
      "epoch": 0.5666433076384022,
      "grad_norm": 0.26569992303848267,
      "learning_rate": 0.0002492274575460416,
      "loss": 0.0293,
      "step": 4043
    },
    {
      "epoch": 0.5667834618079888,
      "grad_norm": 0.8932842016220093,
      "learning_rate": 0.00024921310691222194,
      "loss": 0.0543,
      "step": 4044
    },
    {
      "epoch": 0.5669236159775753,
      "grad_norm": 0.5756842494010925,
      "learning_rate": 0.00024919875627840227,
      "loss": 0.0882,
      "step": 4045
    },
    {
      "epoch": 0.5670637701471619,
      "grad_norm": 0.3436899185180664,
      "learning_rate": 0.00024918440564458265,
      "loss": 0.0415,
      "step": 4046
    },
    {
      "epoch": 0.5672039243167484,
      "grad_norm": 2.1847753524780273,
      "learning_rate": 0.000249170055010763,
      "loss": 0.5203,
      "step": 4047
    },
    {
      "epoch": 0.567344078486335,
      "grad_norm": 1.7849920988082886,
      "learning_rate": 0.0002491557043769433,
      "loss": 0.3395,
      "step": 4048
    },
    {
      "epoch": 0.5674842326559215,
      "grad_norm": 1.6180174350738525,
      "learning_rate": 0.00024914135374312364,
      "loss": 0.0328,
      "step": 4049
    },
    {
      "epoch": 0.5676243868255081,
      "grad_norm": 3.5772085189819336,
      "learning_rate": 0.00024912700310930397,
      "loss": 0.6015,
      "step": 4050
    },
    {
      "epoch": 0.5677645409950947,
      "grad_norm": 0.35565653443336487,
      "learning_rate": 0.0002491126524754843,
      "loss": 0.0595,
      "step": 4051
    },
    {
      "epoch": 0.5679046951646811,
      "grad_norm": 0.6185441613197327,
      "learning_rate": 0.0002490983018416646,
      "loss": 0.1611,
      "step": 4052
    },
    {
      "epoch": 0.5680448493342677,
      "grad_norm": 0.38222718238830566,
      "learning_rate": 0.000249083951207845,
      "loss": 0.1053,
      "step": 4053
    },
    {
      "epoch": 0.5681850035038543,
      "grad_norm": 0.6351040601730347,
      "learning_rate": 0.00024906960057402534,
      "loss": 0.0941,
      "step": 4054
    },
    {
      "epoch": 0.5683251576734408,
      "grad_norm": 0.21364746987819672,
      "learning_rate": 0.00024905524994020566,
      "loss": 0.0187,
      "step": 4055
    },
    {
      "epoch": 0.5684653118430273,
      "grad_norm": 0.15619692206382751,
      "learning_rate": 0.00024904089930638605,
      "loss": 0.0398,
      "step": 4056
    },
    {
      "epoch": 0.5686054660126139,
      "grad_norm": 0.536992609500885,
      "learning_rate": 0.0002490265486725664,
      "loss": 0.1884,
      "step": 4057
    },
    {
      "epoch": 0.5687456201822004,
      "grad_norm": 0.5837216377258301,
      "learning_rate": 0.0002490121980387467,
      "loss": 0.196,
      "step": 4058
    },
    {
      "epoch": 0.568885774351787,
      "grad_norm": 0.34991973638534546,
      "learning_rate": 0.00024899784740492703,
      "loss": 0.1003,
      "step": 4059
    },
    {
      "epoch": 0.5690259285213735,
      "grad_norm": 0.42867881059646606,
      "learning_rate": 0.00024898349677110736,
      "loss": 0.0864,
      "step": 4060
    },
    {
      "epoch": 0.56916608269096,
      "grad_norm": 0.5674730539321899,
      "learning_rate": 0.0002489691461372877,
      "loss": 0.0794,
      "step": 4061
    },
    {
      "epoch": 0.5693062368605466,
      "grad_norm": 1.384779930114746,
      "learning_rate": 0.00024895479550346807,
      "loss": 0.1561,
      "step": 4062
    },
    {
      "epoch": 0.5694463910301332,
      "grad_norm": 0.6737813949584961,
      "learning_rate": 0.0002489404448696484,
      "loss": 0.1744,
      "step": 4063
    },
    {
      "epoch": 0.5695865451997197,
      "grad_norm": 0.24421334266662598,
      "learning_rate": 0.00024892609423582873,
      "loss": 0.0753,
      "step": 4064
    },
    {
      "epoch": 0.5697266993693062,
      "grad_norm": 0.7380961775779724,
      "learning_rate": 0.00024891174360200906,
      "loss": 0.2025,
      "step": 4065
    },
    {
      "epoch": 0.5698668535388928,
      "grad_norm": 0.39453673362731934,
      "learning_rate": 0.0002488973929681894,
      "loss": 0.0693,
      "step": 4066
    },
    {
      "epoch": 0.5700070077084793,
      "grad_norm": 0.5544195771217346,
      "learning_rate": 0.0002488830423343697,
      "loss": 0.105,
      "step": 4067
    },
    {
      "epoch": 0.5701471618780659,
      "grad_norm": 0.35562485456466675,
      "learning_rate": 0.0002488686917005501,
      "loss": 0.0881,
      "step": 4068
    },
    {
      "epoch": 0.5702873160476524,
      "grad_norm": 0.3294217586517334,
      "learning_rate": 0.0002488543410667304,
      "loss": 0.086,
      "step": 4069
    },
    {
      "epoch": 0.5704274702172389,
      "grad_norm": 0.22964254021644592,
      "learning_rate": 0.00024883999043291076,
      "loss": 0.0487,
      "step": 4070
    },
    {
      "epoch": 0.5705676243868255,
      "grad_norm": 0.34469130635261536,
      "learning_rate": 0.0002488256397990911,
      "loss": 0.1562,
      "step": 4071
    },
    {
      "epoch": 0.5707077785564121,
      "grad_norm": 0.5755208134651184,
      "learning_rate": 0.00024881128916527147,
      "loss": 0.0813,
      "step": 4072
    },
    {
      "epoch": 0.5708479327259987,
      "grad_norm": 0.412884384393692,
      "learning_rate": 0.0002487969385314518,
      "loss": 0.0746,
      "step": 4073
    },
    {
      "epoch": 0.5709880868955851,
      "grad_norm": 0.5088204741477966,
      "learning_rate": 0.0002487825878976321,
      "loss": 0.0871,
      "step": 4074
    },
    {
      "epoch": 0.5711282410651717,
      "grad_norm": 0.4229466915130615,
      "learning_rate": 0.00024876823726381245,
      "loss": 0.1073,
      "step": 4075
    },
    {
      "epoch": 0.5712683952347583,
      "grad_norm": 0.5490579605102539,
      "learning_rate": 0.0002487538866299928,
      "loss": 0.1335,
      "step": 4076
    },
    {
      "epoch": 0.5714085494043448,
      "grad_norm": 0.5359660387039185,
      "learning_rate": 0.0002487395359961731,
      "loss": 0.1896,
      "step": 4077
    },
    {
      "epoch": 0.5715487035739313,
      "grad_norm": 0.45330074429512024,
      "learning_rate": 0.0002487251853623535,
      "loss": 0.1007,
      "step": 4078
    },
    {
      "epoch": 0.5716888577435179,
      "grad_norm": 0.35710084438323975,
      "learning_rate": 0.0002487108347285338,
      "loss": 0.0946,
      "step": 4079
    },
    {
      "epoch": 0.5718290119131044,
      "grad_norm": 0.712883710861206,
      "learning_rate": 0.00024869648409471415,
      "loss": 0.0575,
      "step": 4080
    },
    {
      "epoch": 0.571969166082691,
      "grad_norm": 1.0127742290496826,
      "learning_rate": 0.00024868213346089453,
      "loss": 0.2128,
      "step": 4081
    },
    {
      "epoch": 0.5721093202522775,
      "grad_norm": 0.5833719968795776,
      "learning_rate": 0.00024866778282707486,
      "loss": 0.1671,
      "step": 4082
    },
    {
      "epoch": 0.572249474421864,
      "grad_norm": 0.8719350695610046,
      "learning_rate": 0.0002486534321932552,
      "loss": 0.0533,
      "step": 4083
    },
    {
      "epoch": 0.5723896285914506,
      "grad_norm": 0.271806538105011,
      "learning_rate": 0.0002486390815594355,
      "loss": 0.0453,
      "step": 4084
    },
    {
      "epoch": 0.5725297827610372,
      "grad_norm": 0.3668370246887207,
      "learning_rate": 0.00024862473092561585,
      "loss": 0.0832,
      "step": 4085
    },
    {
      "epoch": 0.5726699369306237,
      "grad_norm": 0.8421066403388977,
      "learning_rate": 0.0002486103802917962,
      "loss": 0.093,
      "step": 4086
    },
    {
      "epoch": 0.5728100911002102,
      "grad_norm": 0.32390454411506653,
      "learning_rate": 0.0002485960296579765,
      "loss": 0.0328,
      "step": 4087
    },
    {
      "epoch": 0.5729502452697968,
      "grad_norm": 0.5096945762634277,
      "learning_rate": 0.0002485816790241569,
      "loss": 0.0521,
      "step": 4088
    },
    {
      "epoch": 0.5730903994393833,
      "grad_norm": 0.30692732334136963,
      "learning_rate": 0.0002485673283903372,
      "loss": 0.0471,
      "step": 4089
    },
    {
      "epoch": 0.5732305536089699,
      "grad_norm": 0.4252639710903168,
      "learning_rate": 0.00024855297775651754,
      "loss": 0.1007,
      "step": 4090
    },
    {
      "epoch": 0.5733707077785564,
      "grad_norm": 0.24195390939712524,
      "learning_rate": 0.0002485386271226979,
      "loss": 0.0239,
      "step": 4091
    },
    {
      "epoch": 0.5735108619481429,
      "grad_norm": 0.6424862146377563,
      "learning_rate": 0.00024852427648887825,
      "loss": 0.083,
      "step": 4092
    },
    {
      "epoch": 0.5736510161177295,
      "grad_norm": 1.4497126340866089,
      "learning_rate": 0.0002485099258550586,
      "loss": 0.1553,
      "step": 4093
    },
    {
      "epoch": 0.5737911702873161,
      "grad_norm": 0.23256167769432068,
      "learning_rate": 0.0002484955752212389,
      "loss": 0.048,
      "step": 4094
    },
    {
      "epoch": 0.5739313244569026,
      "grad_norm": 0.904105007648468,
      "learning_rate": 0.00024848122458741924,
      "loss": 0.0256,
      "step": 4095
    },
    {
      "epoch": 0.5740714786264891,
      "grad_norm": 1.0455588102340698,
      "learning_rate": 0.00024846687395359957,
      "loss": 0.101,
      "step": 4096
    },
    {
      "epoch": 0.5742116327960757,
      "grad_norm": 0.6662237048149109,
      "learning_rate": 0.00024845252331977995,
      "loss": 0.1583,
      "step": 4097
    },
    {
      "epoch": 0.5743517869656622,
      "grad_norm": 0.3509131073951721,
      "learning_rate": 0.0002484381726859603,
      "loss": 0.1066,
      "step": 4098
    },
    {
      "epoch": 0.5744919411352488,
      "grad_norm": 2.289806365966797,
      "learning_rate": 0.0002484238220521406,
      "loss": 0.4292,
      "step": 4099
    },
    {
      "epoch": 0.5746320953048353,
      "grad_norm": 3.5445199012756348,
      "learning_rate": 0.00024840947141832094,
      "loss": 0.2655,
      "step": 4100
    },
    {
      "epoch": 0.5747722494744218,
      "grad_norm": 0.46635720133781433,
      "learning_rate": 0.0002483951207845013,
      "loss": 0.0838,
      "step": 4101
    },
    {
      "epoch": 0.5749124036440084,
      "grad_norm": 0.5635322332382202,
      "learning_rate": 0.00024838077015068165,
      "loss": 0.1241,
      "step": 4102
    },
    {
      "epoch": 0.575052557813595,
      "grad_norm": 0.29454296827316284,
      "learning_rate": 0.000248366419516862,
      "loss": 0.0616,
      "step": 4103
    },
    {
      "epoch": 0.5751927119831814,
      "grad_norm": 0.6339954733848572,
      "learning_rate": 0.0002483520688830423,
      "loss": 0.0947,
      "step": 4104
    },
    {
      "epoch": 0.575332866152768,
      "grad_norm": 1.1464428901672363,
      "learning_rate": 0.00024833771824922263,
      "loss": 0.1619,
      "step": 4105
    },
    {
      "epoch": 0.5754730203223546,
      "grad_norm": 0.35010188817977905,
      "learning_rate": 0.00024832336761540296,
      "loss": 0.0692,
      "step": 4106
    },
    {
      "epoch": 0.5756131744919412,
      "grad_norm": 0.6388195753097534,
      "learning_rate": 0.00024830901698158335,
      "loss": 0.1135,
      "step": 4107
    },
    {
      "epoch": 0.5757533286615277,
      "grad_norm": 0.4957703948020935,
      "learning_rate": 0.0002482946663477637,
      "loss": 0.1491,
      "step": 4108
    },
    {
      "epoch": 0.5758934828311142,
      "grad_norm": 0.5899906158447266,
      "learning_rate": 0.000248280315713944,
      "loss": 0.1167,
      "step": 4109
    },
    {
      "epoch": 0.5760336370007008,
      "grad_norm": 0.5695058703422546,
      "learning_rate": 0.0002482659650801244,
      "loss": 0.0461,
      "step": 4110
    },
    {
      "epoch": 0.5761737911702873,
      "grad_norm": 0.4779747724533081,
      "learning_rate": 0.0002482516144463047,
      "loss": 0.1222,
      "step": 4111
    },
    {
      "epoch": 0.5763139453398739,
      "grad_norm": 0.46240243315696716,
      "learning_rate": 0.00024823726381248504,
      "loss": 0.1495,
      "step": 4112
    },
    {
      "epoch": 0.5764540995094604,
      "grad_norm": 0.5702279806137085,
      "learning_rate": 0.00024822291317866537,
      "loss": 0.071,
      "step": 4113
    },
    {
      "epoch": 0.5765942536790469,
      "grad_norm": 0.3322455585002899,
      "learning_rate": 0.0002482085625448457,
      "loss": 0.0652,
      "step": 4114
    },
    {
      "epoch": 0.5767344078486335,
      "grad_norm": 0.16973331570625305,
      "learning_rate": 0.00024819421191102603,
      "loss": 0.0563,
      "step": 4115
    },
    {
      "epoch": 0.5768745620182201,
      "grad_norm": 0.290131539106369,
      "learning_rate": 0.0002481798612772064,
      "loss": 0.0844,
      "step": 4116
    },
    {
      "epoch": 0.5770147161878066,
      "grad_norm": 0.5400785207748413,
      "learning_rate": 0.00024816551064338674,
      "loss": 0.1359,
      "step": 4117
    },
    {
      "epoch": 0.5771548703573931,
      "grad_norm": 0.3293546438217163,
      "learning_rate": 0.00024815116000956707,
      "loss": 0.0684,
      "step": 4118
    },
    {
      "epoch": 0.5772950245269797,
      "grad_norm": 0.2727906107902527,
      "learning_rate": 0.0002481368093757474,
      "loss": 0.0525,
      "step": 4119
    },
    {
      "epoch": 0.5774351786965662,
      "grad_norm": 0.6346110105514526,
      "learning_rate": 0.0002481224587419278,
      "loss": 0.0634,
      "step": 4120
    },
    {
      "epoch": 0.5775753328661528,
      "grad_norm": 0.7491541504859924,
      "learning_rate": 0.0002481081081081081,
      "loss": 0.0766,
      "step": 4121
    },
    {
      "epoch": 0.5777154870357393,
      "grad_norm": 0.15558725595474243,
      "learning_rate": 0.00024809375747428844,
      "loss": 0.036,
      "step": 4122
    },
    {
      "epoch": 0.5778556412053258,
      "grad_norm": 0.6777389645576477,
      "learning_rate": 0.00024807940684046877,
      "loss": 0.054,
      "step": 4123
    },
    {
      "epoch": 0.5779957953749124,
      "grad_norm": 0.6266332864761353,
      "learning_rate": 0.0002480650562066491,
      "loss": 0.1201,
      "step": 4124
    },
    {
      "epoch": 0.578135949544499,
      "grad_norm": 0.40791377425193787,
      "learning_rate": 0.0002480507055728294,
      "loss": 0.0563,
      "step": 4125
    },
    {
      "epoch": 0.5782761037140854,
      "grad_norm": 1.1017180681228638,
      "learning_rate": 0.0002480363549390098,
      "loss": 0.2022,
      "step": 4126
    },
    {
      "epoch": 0.578416257883672,
      "grad_norm": 0.44981738924980164,
      "learning_rate": 0.00024802200430519013,
      "loss": 0.0508,
      "step": 4127
    },
    {
      "epoch": 0.5785564120532586,
      "grad_norm": 0.7006047368049622,
      "learning_rate": 0.00024800765367137046,
      "loss": 0.1259,
      "step": 4128
    },
    {
      "epoch": 0.5786965662228452,
      "grad_norm": 0.3415127992630005,
      "learning_rate": 0.0002479933030375508,
      "loss": 0.0794,
      "step": 4129
    },
    {
      "epoch": 0.5788367203924317,
      "grad_norm": 0.5654847025871277,
      "learning_rate": 0.0002479789524037311,
      "loss": 0.1126,
      "step": 4130
    },
    {
      "epoch": 0.5789768745620182,
      "grad_norm": 0.4290890097618103,
      "learning_rate": 0.00024796460176991145,
      "loss": 0.0566,
      "step": 4131
    },
    {
      "epoch": 0.5791170287316048,
      "grad_norm": 1.3742039203643799,
      "learning_rate": 0.00024795025113609183,
      "loss": 0.1048,
      "step": 4132
    },
    {
      "epoch": 0.5792571829011913,
      "grad_norm": 0.4091750383377075,
      "learning_rate": 0.00024793590050227216,
      "loss": 0.0553,
      "step": 4133
    },
    {
      "epoch": 0.5793973370707779,
      "grad_norm": 0.2758570909500122,
      "learning_rate": 0.0002479215498684525,
      "loss": 0.0683,
      "step": 4134
    },
    {
      "epoch": 0.5795374912403644,
      "grad_norm": 0.4589548707008362,
      "learning_rate": 0.0002479071992346328,
      "loss": 0.0775,
      "step": 4135
    },
    {
      "epoch": 0.5796776454099509,
      "grad_norm": 0.5256208181381226,
      "learning_rate": 0.0002478928486008132,
      "loss": 0.0917,
      "step": 4136
    },
    {
      "epoch": 0.5798177995795375,
      "grad_norm": 0.6021117568016052,
      "learning_rate": 0.00024787849796699353,
      "loss": 0.0724,
      "step": 4137
    },
    {
      "epoch": 0.5799579537491241,
      "grad_norm": 0.37503284215927124,
      "learning_rate": 0.00024786414733317386,
      "loss": 0.0459,
      "step": 4138
    },
    {
      "epoch": 0.5800981079187106,
      "grad_norm": 1.1888717412948608,
      "learning_rate": 0.0002478497966993542,
      "loss": 0.1642,
      "step": 4139
    },
    {
      "epoch": 0.5802382620882971,
      "grad_norm": 0.5471056699752808,
      "learning_rate": 0.0002478354460655345,
      "loss": 0.1095,
      "step": 4140
    },
    {
      "epoch": 0.5803784162578837,
      "grad_norm": 0.42414405941963196,
      "learning_rate": 0.00024782109543171484,
      "loss": 0.0668,
      "step": 4141
    },
    {
      "epoch": 0.5805185704274702,
      "grad_norm": 0.565610945224762,
      "learning_rate": 0.0002478067447978952,
      "loss": 0.0669,
      "step": 4142
    },
    {
      "epoch": 0.5806587245970568,
      "grad_norm": 1.8306336402893066,
      "learning_rate": 0.00024779239416407555,
      "loss": 0.099,
      "step": 4143
    },
    {
      "epoch": 0.5807988787666433,
      "grad_norm": 0.6975414752960205,
      "learning_rate": 0.0002477780435302559,
      "loss": 0.0682,
      "step": 4144
    },
    {
      "epoch": 0.5809390329362298,
      "grad_norm": 0.38210222125053406,
      "learning_rate": 0.00024776369289643626,
      "loss": 0.09,
      "step": 4145
    },
    {
      "epoch": 0.5810791871058164,
      "grad_norm": 0.931677520275116,
      "learning_rate": 0.0002477493422626166,
      "loss": 0.0767,
      "step": 4146
    },
    {
      "epoch": 0.581219341275403,
      "grad_norm": 0.7239224314689636,
      "learning_rate": 0.0002477349916287969,
      "loss": 0.2869,
      "step": 4147
    },
    {
      "epoch": 0.5813594954449894,
      "grad_norm": 0.3559267520904541,
      "learning_rate": 0.00024772064099497725,
      "loss": 0.0428,
      "step": 4148
    },
    {
      "epoch": 0.581499649614576,
      "grad_norm": 2.6591994762420654,
      "learning_rate": 0.0002477062903611576,
      "loss": 0.3192,
      "step": 4149
    },
    {
      "epoch": 0.5816398037841626,
      "grad_norm": 1.7101879119873047,
      "learning_rate": 0.0002476919397273379,
      "loss": 0.3687,
      "step": 4150
    },
    {
      "epoch": 0.5817799579537491,
      "grad_norm": 0.552477240562439,
      "learning_rate": 0.0002476775890935183,
      "loss": 0.0997,
      "step": 4151
    },
    {
      "epoch": 0.5819201121233357,
      "grad_norm": 0.8018133640289307,
      "learning_rate": 0.0002476632384596986,
      "loss": 0.1595,
      "step": 4152
    },
    {
      "epoch": 0.5820602662929222,
      "grad_norm": 0.46562767028808594,
      "learning_rate": 0.00024764888782587895,
      "loss": 0.1232,
      "step": 4153
    },
    {
      "epoch": 0.5822004204625087,
      "grad_norm": 0.5799255967140198,
      "learning_rate": 0.0002476345371920593,
      "loss": 0.0703,
      "step": 4154
    },
    {
      "epoch": 0.5823405746320953,
      "grad_norm": 0.5444428324699402,
      "learning_rate": 0.00024762018655823966,
      "loss": 0.1088,
      "step": 4155
    },
    {
      "epoch": 0.5824807288016819,
      "grad_norm": 0.44830042123794556,
      "learning_rate": 0.00024760583592442,
      "loss": 0.1067,
      "step": 4156
    },
    {
      "epoch": 0.5826208829712684,
      "grad_norm": 0.29601743817329407,
      "learning_rate": 0.0002475914852906003,
      "loss": 0.0516,
      "step": 4157
    },
    {
      "epoch": 0.5827610371408549,
      "grad_norm": 0.5629777312278748,
      "learning_rate": 0.00024757713465678064,
      "loss": 0.0786,
      "step": 4158
    },
    {
      "epoch": 0.5829011913104415,
      "grad_norm": 0.5988829135894775,
      "learning_rate": 0.000247562784022961,
      "loss": 0.1206,
      "step": 4159
    },
    {
      "epoch": 0.5830413454800281,
      "grad_norm": 0.35152924060821533,
      "learning_rate": 0.0002475484333891413,
      "loss": 0.0503,
      "step": 4160
    },
    {
      "epoch": 0.5831814996496145,
      "grad_norm": 0.4479272663593292,
      "learning_rate": 0.0002475340827553217,
      "loss": 0.0567,
      "step": 4161
    },
    {
      "epoch": 0.5833216538192011,
      "grad_norm": 0.3762878179550171,
      "learning_rate": 0.000247519732121502,
      "loss": 0.0888,
      "step": 4162
    },
    {
      "epoch": 0.5834618079887877,
      "grad_norm": 0.36845239996910095,
      "learning_rate": 0.00024750538148768234,
      "loss": 0.0681,
      "step": 4163
    },
    {
      "epoch": 0.5836019621583742,
      "grad_norm": 0.6185504794120789,
      "learning_rate": 0.0002474910308538627,
      "loss": 0.166,
      "step": 4164
    },
    {
      "epoch": 0.5837421163279608,
      "grad_norm": 0.39327865839004517,
      "learning_rate": 0.00024747668022004305,
      "loss": 0.1149,
      "step": 4165
    },
    {
      "epoch": 0.5838822704975473,
      "grad_norm": 0.33189845085144043,
      "learning_rate": 0.0002474623295862234,
      "loss": 0.1345,
      "step": 4166
    },
    {
      "epoch": 0.5840224246671338,
      "grad_norm": 0.5765977501869202,
      "learning_rate": 0.0002474479789524037,
      "loss": 0.1014,
      "step": 4167
    },
    {
      "epoch": 0.5841625788367204,
      "grad_norm": 0.18811741471290588,
      "learning_rate": 0.00024743362831858404,
      "loss": 0.0539,
      "step": 4168
    },
    {
      "epoch": 0.584302733006307,
      "grad_norm": 0.3931630253791809,
      "learning_rate": 0.00024741927768476437,
      "loss": 0.081,
      "step": 4169
    },
    {
      "epoch": 0.5844428871758934,
      "grad_norm": 0.3942629098892212,
      "learning_rate": 0.0002474049270509447,
      "loss": 0.0766,
      "step": 4170
    },
    {
      "epoch": 0.58458304134548,
      "grad_norm": 0.3167996108531952,
      "learning_rate": 0.0002473905764171251,
      "loss": 0.0731,
      "step": 4171
    },
    {
      "epoch": 0.5847231955150666,
      "grad_norm": 0.3922867774963379,
      "learning_rate": 0.0002473762257833054,
      "loss": 0.0663,
      "step": 4172
    },
    {
      "epoch": 0.5848633496846531,
      "grad_norm": 0.40867912769317627,
      "learning_rate": 0.00024736187514948574,
      "loss": 0.0992,
      "step": 4173
    },
    {
      "epoch": 0.5850035038542397,
      "grad_norm": 0.41855743527412415,
      "learning_rate": 0.0002473475245156661,
      "loss": 0.091,
      "step": 4174
    },
    {
      "epoch": 0.5851436580238262,
      "grad_norm": 0.28520187735557556,
      "learning_rate": 0.00024733317388184645,
      "loss": 0.1134,
      "step": 4175
    },
    {
      "epoch": 0.5852838121934127,
      "grad_norm": 0.7383517026901245,
      "learning_rate": 0.0002473188232480268,
      "loss": 0.091,
      "step": 4176
    },
    {
      "epoch": 0.5854239663629993,
      "grad_norm": 0.6429657936096191,
      "learning_rate": 0.0002473044726142071,
      "loss": 0.1305,
      "step": 4177
    },
    {
      "epoch": 0.5855641205325859,
      "grad_norm": 0.9740548133850098,
      "learning_rate": 0.00024729012198038743,
      "loss": 0.1399,
      "step": 4178
    },
    {
      "epoch": 0.5857042747021723,
      "grad_norm": 0.7544068098068237,
      "learning_rate": 0.00024727577134656776,
      "loss": 0.087,
      "step": 4179
    },
    {
      "epoch": 0.5858444288717589,
      "grad_norm": 0.30386796593666077,
      "learning_rate": 0.00024726142071274814,
      "loss": 0.0606,
      "step": 4180
    },
    {
      "epoch": 0.5859845830413455,
      "grad_norm": 0.3910101354122162,
      "learning_rate": 0.00024724707007892847,
      "loss": 0.0623,
      "step": 4181
    },
    {
      "epoch": 0.5861247372109321,
      "grad_norm": 0.3714311420917511,
      "learning_rate": 0.0002472327194451088,
      "loss": 0.1029,
      "step": 4182
    },
    {
      "epoch": 0.5862648913805185,
      "grad_norm": 0.3898859918117523,
      "learning_rate": 0.0002472183688112892,
      "loss": 0.0612,
      "step": 4183
    },
    {
      "epoch": 0.5864050455501051,
      "grad_norm": 1.8018828630447388,
      "learning_rate": 0.0002472040181774695,
      "loss": 0.076,
      "step": 4184
    },
    {
      "epoch": 0.5865451997196917,
      "grad_norm": 0.35695692896842957,
      "learning_rate": 0.00024718966754364984,
      "loss": 0.1092,
      "step": 4185
    },
    {
      "epoch": 0.5866853538892782,
      "grad_norm": 0.3678378462791443,
      "learning_rate": 0.00024717531690983017,
      "loss": 0.0538,
      "step": 4186
    },
    {
      "epoch": 0.5868255080588648,
      "grad_norm": 0.2878757417201996,
      "learning_rate": 0.0002471609662760105,
      "loss": 0.1163,
      "step": 4187
    },
    {
      "epoch": 0.5869656622284513,
      "grad_norm": 0.6163018941879272,
      "learning_rate": 0.0002471466156421908,
      "loss": 0.1305,
      "step": 4188
    },
    {
      "epoch": 0.5871058163980378,
      "grad_norm": 0.28597933053970337,
      "learning_rate": 0.00024713226500837116,
      "loss": 0.0423,
      "step": 4189
    },
    {
      "epoch": 0.5872459705676244,
      "grad_norm": 0.41139349341392517,
      "learning_rate": 0.00024711791437455154,
      "loss": 0.0457,
      "step": 4190
    },
    {
      "epoch": 0.587386124737211,
      "grad_norm": 0.6114638447761536,
      "learning_rate": 0.00024710356374073187,
      "loss": 0.0901,
      "step": 4191
    },
    {
      "epoch": 0.5875262789067974,
      "grad_norm": 0.24156561493873596,
      "learning_rate": 0.0002470892131069122,
      "loss": 0.0479,
      "step": 4192
    },
    {
      "epoch": 0.587666433076384,
      "grad_norm": 0.660129964351654,
      "learning_rate": 0.0002470748624730925,
      "loss": 0.0646,
      "step": 4193
    },
    {
      "epoch": 0.5878065872459706,
      "grad_norm": 0.6134944558143616,
      "learning_rate": 0.00024706051183927285,
      "loss": 0.0892,
      "step": 4194
    },
    {
      "epoch": 0.5879467414155571,
      "grad_norm": 0.3676808178424835,
      "learning_rate": 0.00024704616120545323,
      "loss": 0.1095,
      "step": 4195
    },
    {
      "epoch": 0.5880868955851437,
      "grad_norm": 0.47366175055503845,
      "learning_rate": 0.00024703181057163356,
      "loss": 0.0537,
      "step": 4196
    },
    {
      "epoch": 0.5882270497547302,
      "grad_norm": 0.8181947469711304,
      "learning_rate": 0.0002470174599378139,
      "loss": 0.1164,
      "step": 4197
    },
    {
      "epoch": 0.5883672039243167,
      "grad_norm": 0.4179781973361969,
      "learning_rate": 0.0002470031093039942,
      "loss": 0.0861,
      "step": 4198
    },
    {
      "epoch": 0.5885073580939033,
      "grad_norm": 0.7941940426826477,
      "learning_rate": 0.0002469887586701746,
      "loss": 0.1255,
      "step": 4199
    },
    {
      "epoch": 0.5886475122634899,
      "grad_norm": 1.4141478538513184,
      "learning_rate": 0.00024697440803635493,
      "loss": 0.2585,
      "step": 4200
    },
    {
      "epoch": 0.5887876664330763,
      "grad_norm": 0.6340208649635315,
      "learning_rate": 0.00024696005740253526,
      "loss": 0.0661,
      "step": 4201
    },
    {
      "epoch": 0.5889278206026629,
      "grad_norm": 0.6301118731498718,
      "learning_rate": 0.0002469457067687156,
      "loss": 0.1244,
      "step": 4202
    },
    {
      "epoch": 0.5890679747722495,
      "grad_norm": 0.3215789794921875,
      "learning_rate": 0.0002469313561348959,
      "loss": 0.0701,
      "step": 4203
    },
    {
      "epoch": 0.589208128941836,
      "grad_norm": 0.2909778952598572,
      "learning_rate": 0.00024691700550107625,
      "loss": 0.0475,
      "step": 4204
    },
    {
      "epoch": 0.5893482831114225,
      "grad_norm": 0.49275925755500793,
      "learning_rate": 0.0002469026548672566,
      "loss": 0.1673,
      "step": 4205
    },
    {
      "epoch": 0.5894884372810091,
      "grad_norm": 0.3579420745372772,
      "learning_rate": 0.00024688830423343696,
      "loss": 0.1337,
      "step": 4206
    },
    {
      "epoch": 0.5896285914505957,
      "grad_norm": 0.453367680311203,
      "learning_rate": 0.0002468739535996173,
      "loss": 0.0597,
      "step": 4207
    },
    {
      "epoch": 0.5897687456201822,
      "grad_norm": 0.9823862314224243,
      "learning_rate": 0.0002468596029657976,
      "loss": 0.0816,
      "step": 4208
    },
    {
      "epoch": 0.5899088997897688,
      "grad_norm": 0.43889015913009644,
      "learning_rate": 0.000246845252331978,
      "loss": 0.1123,
      "step": 4209
    },
    {
      "epoch": 0.5900490539593553,
      "grad_norm": 0.16822578012943268,
      "learning_rate": 0.0002468309016981583,
      "loss": 0.0227,
      "step": 4210
    },
    {
      "epoch": 0.5901892081289418,
      "grad_norm": 0.7074646949768066,
      "learning_rate": 0.00024681655106433865,
      "loss": 0.092,
      "step": 4211
    },
    {
      "epoch": 0.5903293622985284,
      "grad_norm": 0.35537704825401306,
      "learning_rate": 0.000246802200430519,
      "loss": 0.0877,
      "step": 4212
    },
    {
      "epoch": 0.590469516468115,
      "grad_norm": 0.28335681557655334,
      "learning_rate": 0.0002467878497966993,
      "loss": 0.1035,
      "step": 4213
    },
    {
      "epoch": 0.5906096706377014,
      "grad_norm": 0.17370456457138062,
      "learning_rate": 0.00024677349916287964,
      "loss": 0.0659,
      "step": 4214
    },
    {
      "epoch": 0.590749824807288,
      "grad_norm": 0.4285907745361328,
      "learning_rate": 0.00024675914852906,
      "loss": 0.0802,
      "step": 4215
    },
    {
      "epoch": 0.5908899789768746,
      "grad_norm": 0.36889031529426575,
      "learning_rate": 0.00024674479789524035,
      "loss": 0.046,
      "step": 4216
    },
    {
      "epoch": 0.5910301331464611,
      "grad_norm": 0.3619019091129303,
      "learning_rate": 0.0002467304472614207,
      "loss": 0.1123,
      "step": 4217
    },
    {
      "epoch": 0.5911702873160477,
      "grad_norm": 0.3716450333595276,
      "learning_rate": 0.00024671609662760106,
      "loss": 0.118,
      "step": 4218
    },
    {
      "epoch": 0.5913104414856342,
      "grad_norm": 0.5450120568275452,
      "learning_rate": 0.0002467017459937814,
      "loss": 0.0497,
      "step": 4219
    },
    {
      "epoch": 0.5914505956552207,
      "grad_norm": 0.43615761399269104,
      "learning_rate": 0.0002466873953599617,
      "loss": 0.092,
      "step": 4220
    },
    {
      "epoch": 0.5915907498248073,
      "grad_norm": 0.6440950632095337,
      "learning_rate": 0.00024667304472614205,
      "loss": 0.0817,
      "step": 4221
    },
    {
      "epoch": 0.5917309039943939,
      "grad_norm": 0.34157589077949524,
      "learning_rate": 0.0002466586940923224,
      "loss": 0.0612,
      "step": 4222
    },
    {
      "epoch": 0.5918710581639803,
      "grad_norm": 0.29978910088539124,
      "learning_rate": 0.0002466443434585027,
      "loss": 0.0744,
      "step": 4223
    },
    {
      "epoch": 0.5920112123335669,
      "grad_norm": 0.3031977415084839,
      "learning_rate": 0.00024662999282468303,
      "loss": 0.122,
      "step": 4224
    },
    {
      "epoch": 0.5921513665031535,
      "grad_norm": 0.2998000681400299,
      "learning_rate": 0.0002466156421908634,
      "loss": 0.0563,
      "step": 4225
    },
    {
      "epoch": 0.59229152067274,
      "grad_norm": 0.288450688123703,
      "learning_rate": 0.00024660129155704375,
      "loss": 0.065,
      "step": 4226
    },
    {
      "epoch": 0.5924316748423265,
      "grad_norm": 0.5636252164840698,
      "learning_rate": 0.0002465869409232241,
      "loss": 0.2363,
      "step": 4227
    },
    {
      "epoch": 0.5925718290119131,
      "grad_norm": 0.7118433117866516,
      "learning_rate": 0.00024657259028940446,
      "loss": 0.0742,
      "step": 4228
    },
    {
      "epoch": 0.5927119831814996,
      "grad_norm": 0.31525370478630066,
      "learning_rate": 0.0002465582396555848,
      "loss": 0.0596,
      "step": 4229
    },
    {
      "epoch": 0.5928521373510862,
      "grad_norm": 0.36143434047698975,
      "learning_rate": 0.0002465438890217651,
      "loss": 0.0937,
      "step": 4230
    },
    {
      "epoch": 0.5929922915206728,
      "grad_norm": 0.7440447211265564,
      "learning_rate": 0.00024652953838794544,
      "loss": 0.1241,
      "step": 4231
    },
    {
      "epoch": 0.5931324456902592,
      "grad_norm": 0.38406074047088623,
      "learning_rate": 0.00024651518775412577,
      "loss": 0.1021,
      "step": 4232
    },
    {
      "epoch": 0.5932725998598458,
      "grad_norm": 0.33659985661506653,
      "learning_rate": 0.0002465008371203061,
      "loss": 0.0783,
      "step": 4233
    },
    {
      "epoch": 0.5934127540294324,
      "grad_norm": 0.4978031516075134,
      "learning_rate": 0.0002464864864864865,
      "loss": 0.0741,
      "step": 4234
    },
    {
      "epoch": 0.593552908199019,
      "grad_norm": 0.4831142723560333,
      "learning_rate": 0.0002464721358526668,
      "loss": 0.1532,
      "step": 4235
    },
    {
      "epoch": 0.5936930623686054,
      "grad_norm": 0.5020435452461243,
      "learning_rate": 0.00024645778521884714,
      "loss": 0.1099,
      "step": 4236
    },
    {
      "epoch": 0.593833216538192,
      "grad_norm": 0.3171883821487427,
      "learning_rate": 0.0002464434345850275,
      "loss": 0.0446,
      "step": 4237
    },
    {
      "epoch": 0.5939733707077786,
      "grad_norm": 0.3256409466266632,
      "learning_rate": 0.00024642908395120785,
      "loss": 0.1188,
      "step": 4238
    },
    {
      "epoch": 0.5941135248773651,
      "grad_norm": 0.5025069713592529,
      "learning_rate": 0.0002464147333173882,
      "loss": 0.23,
      "step": 4239
    },
    {
      "epoch": 0.5942536790469517,
      "grad_norm": 1.327749490737915,
      "learning_rate": 0.0002464003826835685,
      "loss": 0.0435,
      "step": 4240
    },
    {
      "epoch": 0.5943938332165382,
      "grad_norm": 1.1041979789733887,
      "learning_rate": 0.00024638603204974884,
      "loss": 0.1733,
      "step": 4241
    },
    {
      "epoch": 0.5945339873861247,
      "grad_norm": 0.6252986788749695,
      "learning_rate": 0.00024637168141592917,
      "loss": 0.1012,
      "step": 4242
    },
    {
      "epoch": 0.5946741415557113,
      "grad_norm": 0.4027010202407837,
      "learning_rate": 0.0002463573307821095,
      "loss": 0.1079,
      "step": 4243
    },
    {
      "epoch": 0.5948142957252979,
      "grad_norm": 0.6175240278244019,
      "learning_rate": 0.0002463429801482899,
      "loss": 0.1914,
      "step": 4244
    },
    {
      "epoch": 0.5949544498948843,
      "grad_norm": 0.3476596474647522,
      "learning_rate": 0.0002463286295144702,
      "loss": 0.1145,
      "step": 4245
    },
    {
      "epoch": 0.5950946040644709,
      "grad_norm": 0.364168256521225,
      "learning_rate": 0.00024631427888065053,
      "loss": 0.0803,
      "step": 4246
    },
    {
      "epoch": 0.5952347582340575,
      "grad_norm": 0.9181652665138245,
      "learning_rate": 0.0002462999282468309,
      "loss": 0.4212,
      "step": 4247
    },
    {
      "epoch": 0.595374912403644,
      "grad_norm": 0.6801532506942749,
      "learning_rate": 0.00024628557761301124,
      "loss": 0.3116,
      "step": 4248
    },
    {
      "epoch": 0.5955150665732305,
      "grad_norm": 0.6569100022315979,
      "learning_rate": 0.0002462712269791916,
      "loss": 0.1415,
      "step": 4249
    },
    {
      "epoch": 0.5956552207428171,
      "grad_norm": 0.5212199091911316,
      "learning_rate": 0.0002462568763453719,
      "loss": 0.0629,
      "step": 4250
    },
    {
      "epoch": 0.5957953749124036,
      "grad_norm": 0.48831114172935486,
      "learning_rate": 0.00024624252571155223,
      "loss": 0.1779,
      "step": 4251
    },
    {
      "epoch": 0.5959355290819902,
      "grad_norm": 0.1731216460466385,
      "learning_rate": 0.00024622817507773256,
      "loss": 0.0329,
      "step": 4252
    },
    {
      "epoch": 0.5960756832515768,
      "grad_norm": 0.3428407609462738,
      "learning_rate": 0.00024621382444391294,
      "loss": 0.0972,
      "step": 4253
    },
    {
      "epoch": 0.5962158374211632,
      "grad_norm": 0.3357507586479187,
      "learning_rate": 0.00024619947381009327,
      "loss": 0.0578,
      "step": 4254
    },
    {
      "epoch": 0.5963559915907498,
      "grad_norm": 0.27311787009239197,
      "learning_rate": 0.0002461851231762736,
      "loss": 0.0872,
      "step": 4255
    },
    {
      "epoch": 0.5964961457603364,
      "grad_norm": 0.34937262535095215,
      "learning_rate": 0.00024617077254245393,
      "loss": 0.0711,
      "step": 4256
    },
    {
      "epoch": 0.596636299929923,
      "grad_norm": 0.2117593139410019,
      "learning_rate": 0.00024615642190863426,
      "loss": 0.0533,
      "step": 4257
    },
    {
      "epoch": 0.5967764540995094,
      "grad_norm": 0.3577021062374115,
      "learning_rate": 0.0002461420712748146,
      "loss": 0.158,
      "step": 4258
    },
    {
      "epoch": 0.596916608269096,
      "grad_norm": 0.19032466411590576,
      "learning_rate": 0.00024612772064099497,
      "loss": 0.0586,
      "step": 4259
    },
    {
      "epoch": 0.5970567624386826,
      "grad_norm": 0.7568800449371338,
      "learning_rate": 0.0002461133700071753,
      "loss": 0.0689,
      "step": 4260
    },
    {
      "epoch": 0.5971969166082691,
      "grad_norm": 0.8177509903907776,
      "learning_rate": 0.0002460990193733556,
      "loss": 0.0727,
      "step": 4261
    },
    {
      "epoch": 0.5973370707778557,
      "grad_norm": 0.41709256172180176,
      "learning_rate": 0.00024608466873953595,
      "loss": 0.0788,
      "step": 4262
    },
    {
      "epoch": 0.5974772249474422,
      "grad_norm": 0.3124392628669739,
      "learning_rate": 0.00024607031810571634,
      "loss": 0.0566,
      "step": 4263
    },
    {
      "epoch": 0.5976173791170287,
      "grad_norm": 0.16209626197814941,
      "learning_rate": 0.00024605596747189666,
      "loss": 0.0379,
      "step": 4264
    },
    {
      "epoch": 0.5977575332866153,
      "grad_norm": 1.0296924114227295,
      "learning_rate": 0.000246041616838077,
      "loss": 0.0919,
      "step": 4265
    },
    {
      "epoch": 0.5978976874562019,
      "grad_norm": 0.3040546178817749,
      "learning_rate": 0.0002460272662042573,
      "loss": 0.0817,
      "step": 4266
    },
    {
      "epoch": 0.5980378416257883,
      "grad_norm": 0.9100993275642395,
      "learning_rate": 0.00024601291557043765,
      "loss": 0.1831,
      "step": 4267
    },
    {
      "epoch": 0.5981779957953749,
      "grad_norm": 0.2755889892578125,
      "learning_rate": 0.000245998564936618,
      "loss": 0.0284,
      "step": 4268
    },
    {
      "epoch": 0.5983181499649615,
      "grad_norm": 0.3427288234233856,
      "learning_rate": 0.00024598421430279836,
      "loss": 0.0644,
      "step": 4269
    },
    {
      "epoch": 0.598458304134548,
      "grad_norm": 0.5269371867179871,
      "learning_rate": 0.0002459698636689787,
      "loss": 0.0607,
      "step": 4270
    },
    {
      "epoch": 0.5985984583041345,
      "grad_norm": 0.4627378582954407,
      "learning_rate": 0.000245955513035159,
      "loss": 0.0983,
      "step": 4271
    },
    {
      "epoch": 0.5987386124737211,
      "grad_norm": 0.3181998133659363,
      "learning_rate": 0.0002459411624013394,
      "loss": 0.0462,
      "step": 4272
    },
    {
      "epoch": 0.5988787666433076,
      "grad_norm": 0.30405065417289734,
      "learning_rate": 0.00024592681176751973,
      "loss": 0.0772,
      "step": 4273
    },
    {
      "epoch": 0.5990189208128942,
      "grad_norm": 0.3996593952178955,
      "learning_rate": 0.00024591246113370006,
      "loss": 0.1246,
      "step": 4274
    },
    {
      "epoch": 0.5991590749824808,
      "grad_norm": 0.468820720911026,
      "learning_rate": 0.0002458981104998804,
      "loss": 0.1081,
      "step": 4275
    },
    {
      "epoch": 0.5992992291520672,
      "grad_norm": 0.45422908663749695,
      "learning_rate": 0.0002458837598660607,
      "loss": 0.1535,
      "step": 4276
    },
    {
      "epoch": 0.5994393833216538,
      "grad_norm": 0.5360621213912964,
      "learning_rate": 0.00024586940923224104,
      "loss": 0.1074,
      "step": 4277
    },
    {
      "epoch": 0.5995795374912404,
      "grad_norm": 0.4938092529773712,
      "learning_rate": 0.00024585505859842137,
      "loss": 0.0698,
      "step": 4278
    },
    {
      "epoch": 0.599719691660827,
      "grad_norm": 0.35549435019493103,
      "learning_rate": 0.00024584070796460176,
      "loss": 0.034,
      "step": 4279
    },
    {
      "epoch": 0.5998598458304134,
      "grad_norm": 0.24330860376358032,
      "learning_rate": 0.0002458263573307821,
      "loss": 0.0921,
      "step": 4280
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.4356054663658142,
      "learning_rate": 0.0002458120066969624,
      "loss": 0.2351,
      "step": 4281
    },
    {
      "epoch": 0.6001401541695865,
      "grad_norm": 0.4803878664970398,
      "learning_rate": 0.0002457976560631428,
      "loss": 0.1899,
      "step": 4282
    },
    {
      "epoch": 0.6002803083391731,
      "grad_norm": 0.5061917901039124,
      "learning_rate": 0.0002457833054293231,
      "loss": 0.0495,
      "step": 4283
    },
    {
      "epoch": 0.6004204625087597,
      "grad_norm": 0.4666076898574829,
      "learning_rate": 0.00024576895479550345,
      "loss": 0.1593,
      "step": 4284
    },
    {
      "epoch": 0.6005606166783461,
      "grad_norm": 0.8980773091316223,
      "learning_rate": 0.0002457546041616838,
      "loss": 0.2015,
      "step": 4285
    },
    {
      "epoch": 0.6007007708479327,
      "grad_norm": 0.5777843594551086,
      "learning_rate": 0.0002457402535278641,
      "loss": 0.0676,
      "step": 4286
    },
    {
      "epoch": 0.6008409250175193,
      "grad_norm": 0.993579626083374,
      "learning_rate": 0.00024572590289404444,
      "loss": 0.0721,
      "step": 4287
    },
    {
      "epoch": 0.6009810791871059,
      "grad_norm": 0.5573502779006958,
      "learning_rate": 0.0002457115522602248,
      "loss": 0.1125,
      "step": 4288
    },
    {
      "epoch": 0.6011212333566923,
      "grad_norm": 0.2585557699203491,
      "learning_rate": 0.00024569720162640515,
      "loss": 0.0643,
      "step": 4289
    },
    {
      "epoch": 0.6012613875262789,
      "grad_norm": 0.46761780977249146,
      "learning_rate": 0.0002456828509925855,
      "loss": 0.0649,
      "step": 4290
    },
    {
      "epoch": 0.6014015416958655,
      "grad_norm": 0.7123151421546936,
      "learning_rate": 0.0002456685003587658,
      "loss": 0.1156,
      "step": 4291
    },
    {
      "epoch": 0.601541695865452,
      "grad_norm": 0.6322957277297974,
      "learning_rate": 0.0002456541497249462,
      "loss": 0.085,
      "step": 4292
    },
    {
      "epoch": 0.6016818500350385,
      "grad_norm": 0.7120218873023987,
      "learning_rate": 0.0002456397990911265,
      "loss": 0.1058,
      "step": 4293
    },
    {
      "epoch": 0.6018220042046251,
      "grad_norm": 0.6270059943199158,
      "learning_rate": 0.00024562544845730685,
      "loss": 0.1428,
      "step": 4294
    },
    {
      "epoch": 0.6019621583742116,
      "grad_norm": 0.2194758653640747,
      "learning_rate": 0.0002456110978234872,
      "loss": 0.041,
      "step": 4295
    },
    {
      "epoch": 0.6021023125437982,
      "grad_norm": 0.5218814015388489,
      "learning_rate": 0.0002455967471896675,
      "loss": 0.066,
      "step": 4296
    },
    {
      "epoch": 0.6022424667133848,
      "grad_norm": 0.644420862197876,
      "learning_rate": 0.00024558239655584783,
      "loss": 0.0599,
      "step": 4297
    },
    {
      "epoch": 0.6023826208829712,
      "grad_norm": 0.9728002548217773,
      "learning_rate": 0.0002455680459220282,
      "loss": 0.0512,
      "step": 4298
    },
    {
      "epoch": 0.6025227750525578,
      "grad_norm": 1.218588948249817,
      "learning_rate": 0.00024555369528820854,
      "loss": 0.1672,
      "step": 4299
    },
    {
      "epoch": 0.6026629292221444,
      "grad_norm": 1.6218665838241577,
      "learning_rate": 0.00024553934465438887,
      "loss": 0.1723,
      "step": 4300
    },
    {
      "epoch": 0.6028030833917309,
      "grad_norm": 0.27974990010261536,
      "learning_rate": 0.00024552499402056925,
      "loss": 0.0727,
      "step": 4301
    },
    {
      "epoch": 0.6029432375613174,
      "grad_norm": 0.3434129059314728,
      "learning_rate": 0.0002455106433867496,
      "loss": 0.1364,
      "step": 4302
    },
    {
      "epoch": 0.603083391730904,
      "grad_norm": 0.2227613776922226,
      "learning_rate": 0.0002454962927529299,
      "loss": 0.0634,
      "step": 4303
    },
    {
      "epoch": 0.6032235459004905,
      "grad_norm": 0.3121270537376404,
      "learning_rate": 0.00024548194211911024,
      "loss": 0.0351,
      "step": 4304
    },
    {
      "epoch": 0.6033637000700771,
      "grad_norm": 0.21233882009983063,
      "learning_rate": 0.00024546759148529057,
      "loss": 0.0408,
      "step": 4305
    },
    {
      "epoch": 0.6035038542396637,
      "grad_norm": 0.3764982521533966,
      "learning_rate": 0.0002454532408514709,
      "loss": 0.0698,
      "step": 4306
    },
    {
      "epoch": 0.6036440084092501,
      "grad_norm": 0.7671663761138916,
      "learning_rate": 0.0002454388902176513,
      "loss": 0.0777,
      "step": 4307
    },
    {
      "epoch": 0.6037841625788367,
      "grad_norm": 0.3349737823009491,
      "learning_rate": 0.0002454245395838316,
      "loss": 0.1156,
      "step": 4308
    },
    {
      "epoch": 0.6039243167484233,
      "grad_norm": 0.49075812101364136,
      "learning_rate": 0.00024541018895001194,
      "loss": 0.0966,
      "step": 4309
    },
    {
      "epoch": 0.6040644709180099,
      "grad_norm": 0.7330748438835144,
      "learning_rate": 0.00024539583831619227,
      "loss": 0.1238,
      "step": 4310
    },
    {
      "epoch": 0.6042046250875963,
      "grad_norm": 0.2991892099380493,
      "learning_rate": 0.00024538148768237265,
      "loss": 0.0803,
      "step": 4311
    },
    {
      "epoch": 0.6043447792571829,
      "grad_norm": 0.7433737516403198,
      "learning_rate": 0.000245367137048553,
      "loss": 0.0734,
      "step": 4312
    },
    {
      "epoch": 0.6044849334267695,
      "grad_norm": 0.495500385761261,
      "learning_rate": 0.0002453527864147333,
      "loss": 0.0713,
      "step": 4313
    },
    {
      "epoch": 0.604625087596356,
      "grad_norm": 0.4748336970806122,
      "learning_rate": 0.00024533843578091363,
      "loss": 0.1337,
      "step": 4314
    },
    {
      "epoch": 0.6047652417659425,
      "grad_norm": 0.46583452820777893,
      "learning_rate": 0.00024532408514709396,
      "loss": 0.0955,
      "step": 4315
    },
    {
      "epoch": 0.604905395935529,
      "grad_norm": 0.4464094042778015,
      "learning_rate": 0.0002453097345132743,
      "loss": 0.0996,
      "step": 4316
    },
    {
      "epoch": 0.6050455501051156,
      "grad_norm": 0.4841650426387787,
      "learning_rate": 0.0002452953838794547,
      "loss": 0.1073,
      "step": 4317
    },
    {
      "epoch": 0.6051857042747022,
      "grad_norm": 0.8044472932815552,
      "learning_rate": 0.000245281033245635,
      "loss": 0.0814,
      "step": 4318
    },
    {
      "epoch": 0.6053258584442888,
      "grad_norm": 0.7630645632743835,
      "learning_rate": 0.00024526668261181533,
      "loss": 0.154,
      "step": 4319
    },
    {
      "epoch": 0.6054660126138752,
      "grad_norm": 0.4888552725315094,
      "learning_rate": 0.00024525233197799566,
      "loss": 0.087,
      "step": 4320
    },
    {
      "epoch": 0.6056061667834618,
      "grad_norm": 0.44488105177879333,
      "learning_rate": 0.000245237981344176,
      "loss": 0.0751,
      "step": 4321
    },
    {
      "epoch": 0.6057463209530484,
      "grad_norm": 0.7497250437736511,
      "learning_rate": 0.00024522363071035637,
      "loss": 0.1415,
      "step": 4322
    },
    {
      "epoch": 0.6058864751226349,
      "grad_norm": 0.47417131066322327,
      "learning_rate": 0.0002452092800765367,
      "loss": 0.0748,
      "step": 4323
    },
    {
      "epoch": 0.6060266292922214,
      "grad_norm": 0.269990473985672,
      "learning_rate": 0.00024519492944271703,
      "loss": 0.0704,
      "step": 4324
    },
    {
      "epoch": 0.606166783461808,
      "grad_norm": 0.2408597618341446,
      "learning_rate": 0.00024518057880889736,
      "loss": 0.0674,
      "step": 4325
    },
    {
      "epoch": 0.6063069376313945,
      "grad_norm": 0.5705625414848328,
      "learning_rate": 0.0002451662281750777,
      "loss": 0.1137,
      "step": 4326
    },
    {
      "epoch": 0.6064470918009811,
      "grad_norm": 0.2860572040081024,
      "learning_rate": 0.00024515187754125807,
      "loss": 0.0821,
      "step": 4327
    },
    {
      "epoch": 0.6065872459705677,
      "grad_norm": 0.9140095114707947,
      "learning_rate": 0.0002451375269074384,
      "loss": 0.1939,
      "step": 4328
    },
    {
      "epoch": 0.6067274001401541,
      "grad_norm": 0.32056695222854614,
      "learning_rate": 0.0002451231762736187,
      "loss": 0.0553,
      "step": 4329
    },
    {
      "epoch": 0.6068675543097407,
      "grad_norm": 0.38588041067123413,
      "learning_rate": 0.00024510882563979905,
      "loss": 0.1023,
      "step": 4330
    },
    {
      "epoch": 0.6070077084793273,
      "grad_norm": 0.4503379762172699,
      "learning_rate": 0.0002450944750059794,
      "loss": 0.0866,
      "step": 4331
    },
    {
      "epoch": 0.6071478626489138,
      "grad_norm": 0.48218831419944763,
      "learning_rate": 0.0002450801243721597,
      "loss": 0.075,
      "step": 4332
    },
    {
      "epoch": 0.6072880168185003,
      "grad_norm": 0.7645440697669983,
      "learning_rate": 0.0002450657737383401,
      "loss": 0.1228,
      "step": 4333
    },
    {
      "epoch": 0.6074281709880869,
      "grad_norm": 0.28834107518196106,
      "learning_rate": 0.0002450514231045204,
      "loss": 0.0961,
      "step": 4334
    },
    {
      "epoch": 0.6075683251576735,
      "grad_norm": 0.8012709617614746,
      "learning_rate": 0.00024503707247070075,
      "loss": 0.1517,
      "step": 4335
    },
    {
      "epoch": 0.60770847932726,
      "grad_norm": 0.3561074435710907,
      "learning_rate": 0.00024502272183688113,
      "loss": 0.08,
      "step": 4336
    },
    {
      "epoch": 0.6078486334968465,
      "grad_norm": 0.6639081239700317,
      "learning_rate": 0.00024500837120306146,
      "loss": 0.0664,
      "step": 4337
    },
    {
      "epoch": 0.607988787666433,
      "grad_norm": 0.7292010188102722,
      "learning_rate": 0.0002449940205692418,
      "loss": 0.0639,
      "step": 4338
    },
    {
      "epoch": 0.6081289418360196,
      "grad_norm": 0.5321303606033325,
      "learning_rate": 0.0002449796699354221,
      "loss": 0.1339,
      "step": 4339
    },
    {
      "epoch": 0.6082690960056062,
      "grad_norm": 1.0508623123168945,
      "learning_rate": 0.00024496531930160245,
      "loss": 0.1713,
      "step": 4340
    },
    {
      "epoch": 0.6084092501751928,
      "grad_norm": 0.8318646550178528,
      "learning_rate": 0.0002449509686677828,
      "loss": 0.0727,
      "step": 4341
    },
    {
      "epoch": 0.6085494043447792,
      "grad_norm": 0.3840405344963074,
      "learning_rate": 0.00024493661803396316,
      "loss": 0.0468,
      "step": 4342
    },
    {
      "epoch": 0.6086895585143658,
      "grad_norm": 0.3322233557701111,
      "learning_rate": 0.0002449222674001435,
      "loss": 0.0427,
      "step": 4343
    },
    {
      "epoch": 0.6088297126839524,
      "grad_norm": 0.5010483264923096,
      "learning_rate": 0.0002449079167663238,
      "loss": 0.096,
      "step": 4344
    },
    {
      "epoch": 0.6089698668535389,
      "grad_norm": 1.0068780183792114,
      "learning_rate": 0.00024489356613250415,
      "loss": 0.0412,
      "step": 4345
    },
    {
      "epoch": 0.6091100210231254,
      "grad_norm": 0.8120301365852356,
      "learning_rate": 0.00024487921549868453,
      "loss": 0.0981,
      "step": 4346
    },
    {
      "epoch": 0.609250175192712,
      "grad_norm": 0.9413319826126099,
      "learning_rate": 0.00024486486486486486,
      "loss": 0.058,
      "step": 4347
    },
    {
      "epoch": 0.6093903293622985,
      "grad_norm": 2.320493221282959,
      "learning_rate": 0.0002448505142310452,
      "loss": 0.0763,
      "step": 4348
    },
    {
      "epoch": 0.6095304835318851,
      "grad_norm": 1.0113309621810913,
      "learning_rate": 0.0002448361635972255,
      "loss": 0.2882,
      "step": 4349
    },
    {
      "epoch": 0.6096706377014717,
      "grad_norm": 3.857027292251587,
      "learning_rate": 0.00024482181296340584,
      "loss": 0.3451,
      "step": 4350
    },
    {
      "epoch": 0.6098107918710581,
      "grad_norm": 0.29157987236976624,
      "learning_rate": 0.00024480746232958617,
      "loss": 0.0332,
      "step": 4351
    },
    {
      "epoch": 0.6099509460406447,
      "grad_norm": 0.3212697505950928,
      "learning_rate": 0.00024479311169576655,
      "loss": 0.0384,
      "step": 4352
    },
    {
      "epoch": 0.6100911002102313,
      "grad_norm": 0.25363045930862427,
      "learning_rate": 0.0002447787610619469,
      "loss": 0.0756,
      "step": 4353
    },
    {
      "epoch": 0.6102312543798178,
      "grad_norm": 0.3170822858810425,
      "learning_rate": 0.0002447644104281272,
      "loss": 0.0425,
      "step": 4354
    },
    {
      "epoch": 0.6103714085494043,
      "grad_norm": 0.3066902756690979,
      "learning_rate": 0.0002447500597943076,
      "loss": 0.0878,
      "step": 4355
    },
    {
      "epoch": 0.6105115627189909,
      "grad_norm": 0.3292471468448639,
      "learning_rate": 0.0002447357091604879,
      "loss": 0.081,
      "step": 4356
    },
    {
      "epoch": 0.6106517168885774,
      "grad_norm": 0.4762860834598541,
      "learning_rate": 0.00024472135852666825,
      "loss": 0.1618,
      "step": 4357
    },
    {
      "epoch": 0.610791871058164,
      "grad_norm": 0.35840871930122375,
      "learning_rate": 0.0002447070078928486,
      "loss": 0.0835,
      "step": 4358
    },
    {
      "epoch": 0.6109320252277505,
      "grad_norm": 0.7974569201469421,
      "learning_rate": 0.0002446926572590289,
      "loss": 0.0971,
      "step": 4359
    },
    {
      "epoch": 0.611072179397337,
      "grad_norm": 0.1929108202457428,
      "learning_rate": 0.00024467830662520924,
      "loss": 0.0505,
      "step": 4360
    },
    {
      "epoch": 0.6112123335669236,
      "grad_norm": 0.7492509484291077,
      "learning_rate": 0.00024466395599138956,
      "loss": 0.0611,
      "step": 4361
    },
    {
      "epoch": 0.6113524877365102,
      "grad_norm": 0.3383336663246155,
      "learning_rate": 0.00024464960535756995,
      "loss": 0.0606,
      "step": 4362
    },
    {
      "epoch": 0.6114926419060968,
      "grad_norm": 0.3851757049560547,
      "learning_rate": 0.0002446352547237503,
      "loss": 0.0903,
      "step": 4363
    },
    {
      "epoch": 0.6116327960756832,
      "grad_norm": 0.797488272190094,
      "learning_rate": 0.0002446209040899306,
      "loss": 0.1297,
      "step": 4364
    },
    {
      "epoch": 0.6117729502452698,
      "grad_norm": 0.487002968788147,
      "learning_rate": 0.000244606553456111,
      "loss": 0.0707,
      "step": 4365
    },
    {
      "epoch": 0.6119131044148564,
      "grad_norm": 0.39489510655403137,
      "learning_rate": 0.0002445922028222913,
      "loss": 0.0665,
      "step": 4366
    },
    {
      "epoch": 0.6120532585844429,
      "grad_norm": 0.49685123562812805,
      "learning_rate": 0.00024457785218847164,
      "loss": 0.1061,
      "step": 4367
    },
    {
      "epoch": 0.6121934127540294,
      "grad_norm": 0.3775698244571686,
      "learning_rate": 0.000244563501554652,
      "loss": 0.1033,
      "step": 4368
    },
    {
      "epoch": 0.612333566923616,
      "grad_norm": 0.29300031065940857,
      "learning_rate": 0.0002445491509208323,
      "loss": 0.0468,
      "step": 4369
    },
    {
      "epoch": 0.6124737210932025,
      "grad_norm": 0.20870809257030487,
      "learning_rate": 0.00024453480028701263,
      "loss": 0.0246,
      "step": 4370
    },
    {
      "epoch": 0.6126138752627891,
      "grad_norm": 0.6352375745773315,
      "learning_rate": 0.000244520449653193,
      "loss": 0.1485,
      "step": 4371
    },
    {
      "epoch": 0.6127540294323757,
      "grad_norm": 0.28585219383239746,
      "learning_rate": 0.00024450609901937334,
      "loss": 0.0896,
      "step": 4372
    },
    {
      "epoch": 0.6128941836019621,
      "grad_norm": 0.3606799244880676,
      "learning_rate": 0.00024449174838555367,
      "loss": 0.0567,
      "step": 4373
    },
    {
      "epoch": 0.6130343377715487,
      "grad_norm": 0.46540823578834534,
      "learning_rate": 0.00024447739775173405,
      "loss": 0.0842,
      "step": 4374
    },
    {
      "epoch": 0.6131744919411353,
      "grad_norm": 0.7603920102119446,
      "learning_rate": 0.0002444630471179144,
      "loss": 0.086,
      "step": 4375
    },
    {
      "epoch": 0.6133146461107218,
      "grad_norm": 0.36296015977859497,
      "learning_rate": 0.0002444486964840947,
      "loss": 0.0768,
      "step": 4376
    },
    {
      "epoch": 0.6134548002803083,
      "grad_norm": 0.28165048360824585,
      "learning_rate": 0.00024443434585027504,
      "loss": 0.0734,
      "step": 4377
    },
    {
      "epoch": 0.6135949544498949,
      "grad_norm": 0.37743857502937317,
      "learning_rate": 0.00024441999521645537,
      "loss": 0.071,
      "step": 4378
    },
    {
      "epoch": 0.6137351086194814,
      "grad_norm": 0.37869128584861755,
      "learning_rate": 0.0002444056445826357,
      "loss": 0.076,
      "step": 4379
    },
    {
      "epoch": 0.613875262789068,
      "grad_norm": 0.1750558465719223,
      "learning_rate": 0.000244391293948816,
      "loss": 0.0248,
      "step": 4380
    },
    {
      "epoch": 0.6140154169586545,
      "grad_norm": 0.2464558631181717,
      "learning_rate": 0.0002443769433149964,
      "loss": 0.0504,
      "step": 4381
    },
    {
      "epoch": 0.614155571128241,
      "grad_norm": 0.1975264698266983,
      "learning_rate": 0.00024436259268117674,
      "loss": 0.0429,
      "step": 4382
    },
    {
      "epoch": 0.6142957252978276,
      "grad_norm": 0.39335569739341736,
      "learning_rate": 0.00024434824204735706,
      "loss": 0.0813,
      "step": 4383
    },
    {
      "epoch": 0.6144358794674142,
      "grad_norm": 0.3592526316642761,
      "learning_rate": 0.0002443338914135374,
      "loss": 0.0644,
      "step": 4384
    },
    {
      "epoch": 0.6145760336370008,
      "grad_norm": 0.7763746976852417,
      "learning_rate": 0.0002443195407797177,
      "loss": 0.1568,
      "step": 4385
    },
    {
      "epoch": 0.6147161878065872,
      "grad_norm": 1.4262676239013672,
      "learning_rate": 0.0002443051901458981,
      "loss": 0.412,
      "step": 4386
    },
    {
      "epoch": 0.6148563419761738,
      "grad_norm": 0.6510946750640869,
      "learning_rate": 0.00024429083951207843,
      "loss": 0.1386,
      "step": 4387
    },
    {
      "epoch": 0.6149964961457604,
      "grad_norm": 0.3391013443470001,
      "learning_rate": 0.00024427648887825876,
      "loss": 0.1077,
      "step": 4388
    },
    {
      "epoch": 0.6151366503153469,
      "grad_norm": 0.827523410320282,
      "learning_rate": 0.0002442621382444391,
      "loss": 0.1334,
      "step": 4389
    },
    {
      "epoch": 0.6152768044849334,
      "grad_norm": 0.3507126569747925,
      "learning_rate": 0.00024424778761061947,
      "loss": 0.1238,
      "step": 4390
    },
    {
      "epoch": 0.61541695865452,
      "grad_norm": 0.8287339210510254,
      "learning_rate": 0.0002442334369767998,
      "loss": 0.1035,
      "step": 4391
    },
    {
      "epoch": 0.6155571128241065,
      "grad_norm": 0.7334623336791992,
      "learning_rate": 0.00024421908634298013,
      "loss": 0.1257,
      "step": 4392
    },
    {
      "epoch": 0.6156972669936931,
      "grad_norm": 0.8060169816017151,
      "learning_rate": 0.00024420473570916046,
      "loss": 0.0632,
      "step": 4393
    },
    {
      "epoch": 0.6158374211632796,
      "grad_norm": 0.49323779344558716,
      "learning_rate": 0.0002441903850753408,
      "loss": 0.1475,
      "step": 4394
    },
    {
      "epoch": 0.6159775753328661,
      "grad_norm": 1.107265830039978,
      "learning_rate": 0.0002441760344415211,
      "loss": 0.1472,
      "step": 4395
    },
    {
      "epoch": 0.6161177295024527,
      "grad_norm": 0.4396919310092926,
      "learning_rate": 0.00024416168380770144,
      "loss": 0.0765,
      "step": 4396
    },
    {
      "epoch": 0.6162578836720393,
      "grad_norm": 0.8279666900634766,
      "learning_rate": 0.0002441473331738818,
      "loss": 0.1933,
      "step": 4397
    },
    {
      "epoch": 0.6163980378416258,
      "grad_norm": 2.0156943798065186,
      "learning_rate": 0.00024413298254006216,
      "loss": 0.1694,
      "step": 4398
    },
    {
      "epoch": 0.6165381920112123,
      "grad_norm": 1.5239709615707397,
      "learning_rate": 0.00024411863190624248,
      "loss": 0.3152,
      "step": 4399
    },
    {
      "epoch": 0.6166783461807989,
      "grad_norm": 2.478585720062256,
      "learning_rate": 0.00024410428127242284,
      "loss": 0.3303,
      "step": 4400
    },
    {
      "epoch": 0.6168185003503854,
      "grad_norm": 0.2714337110519409,
      "learning_rate": 0.0002440899306386032,
      "loss": 0.111,
      "step": 4401
    },
    {
      "epoch": 0.616958654519972,
      "grad_norm": 0.4110240042209625,
      "learning_rate": 0.00024407558000478352,
      "loss": 0.1119,
      "step": 4402
    },
    {
      "epoch": 0.6170988086895585,
      "grad_norm": 0.31719252467155457,
      "learning_rate": 0.00024406122937096388,
      "loss": 0.0832,
      "step": 4403
    },
    {
      "epoch": 0.617238962859145,
      "grad_norm": 0.3928353786468506,
      "learning_rate": 0.0002440468787371442,
      "loss": 0.1329,
      "step": 4404
    },
    {
      "epoch": 0.6173791170287316,
      "grad_norm": 0.46870750188827515,
      "learning_rate": 0.00024403252810332454,
      "loss": 0.0995,
      "step": 4405
    },
    {
      "epoch": 0.6175192711983182,
      "grad_norm": 0.3520171642303467,
      "learning_rate": 0.0002440181774695049,
      "loss": 0.1214,
      "step": 4406
    },
    {
      "epoch": 0.6176594253679047,
      "grad_norm": 0.30453261733055115,
      "learning_rate": 0.00024400382683568522,
      "loss": 0.0791,
      "step": 4407
    },
    {
      "epoch": 0.6177995795374912,
      "grad_norm": 0.25287672877311707,
      "learning_rate": 0.00024398947620186555,
      "loss": 0.0835,
      "step": 4408
    },
    {
      "epoch": 0.6179397337070778,
      "grad_norm": 0.39686262607574463,
      "learning_rate": 0.0002439751255680459,
      "loss": 0.1069,
      "step": 4409
    },
    {
      "epoch": 0.6180798878766643,
      "grad_norm": 0.38980939984321594,
      "learning_rate": 0.00024396077493422623,
      "loss": 0.1512,
      "step": 4410
    },
    {
      "epoch": 0.6182200420462509,
      "grad_norm": 0.11704090237617493,
      "learning_rate": 0.00024394642430040656,
      "loss": 0.0191,
      "step": 4411
    },
    {
      "epoch": 0.6183601962158374,
      "grad_norm": 0.23291423916816711,
      "learning_rate": 0.00024393207366658694,
      "loss": 0.0843,
      "step": 4412
    },
    {
      "epoch": 0.618500350385424,
      "grad_norm": 0.19935207068920135,
      "learning_rate": 0.00024391772303276727,
      "loss": 0.0629,
      "step": 4413
    },
    {
      "epoch": 0.6186405045550105,
      "grad_norm": 0.5691215395927429,
      "learning_rate": 0.0002439033723989476,
      "loss": 0.0903,
      "step": 4414
    },
    {
      "epoch": 0.6187806587245971,
      "grad_norm": 0.22797924280166626,
      "learning_rate": 0.00024388902176512793,
      "loss": 0.0646,
      "step": 4415
    },
    {
      "epoch": 0.6189208128941835,
      "grad_norm": 0.23018260300159454,
      "learning_rate": 0.00024387467113130829,
      "loss": 0.0932,
      "step": 4416
    },
    {
      "epoch": 0.6190609670637701,
      "grad_norm": 0.6204971671104431,
      "learning_rate": 0.00024386032049748861,
      "loss": 0.1055,
      "step": 4417
    },
    {
      "epoch": 0.6192011212333567,
      "grad_norm": 0.4983014464378357,
      "learning_rate": 0.00024384596986366894,
      "loss": 0.1327,
      "step": 4418
    },
    {
      "epoch": 0.6193412754029433,
      "grad_norm": 0.5641348958015442,
      "learning_rate": 0.0002438316192298493,
      "loss": 0.1468,
      "step": 4419
    },
    {
      "epoch": 0.6194814295725298,
      "grad_norm": 0.31819379329681396,
      "learning_rate": 0.00024381726859602963,
      "loss": 0.0976,
      "step": 4420
    },
    {
      "epoch": 0.6196215837421163,
      "grad_norm": 0.36068475246429443,
      "learning_rate": 0.00024380291796220996,
      "loss": 0.0673,
      "step": 4421
    },
    {
      "epoch": 0.6197617379117029,
      "grad_norm": 0.6764531135559082,
      "learning_rate": 0.00024378856732839034,
      "loss": 0.1174,
      "step": 4422
    },
    {
      "epoch": 0.6199018920812894,
      "grad_norm": 0.3103175461292267,
      "learning_rate": 0.00024377421669457067,
      "loss": 0.0481,
      "step": 4423
    },
    {
      "epoch": 0.620042046250876,
      "grad_norm": 0.4179884195327759,
      "learning_rate": 0.000243759866060751,
      "loss": 0.2048,
      "step": 4424
    },
    {
      "epoch": 0.6201822004204625,
      "grad_norm": 0.4337652027606964,
      "learning_rate": 0.00024374551542693135,
      "loss": 0.0958,
      "step": 4425
    },
    {
      "epoch": 0.620322354590049,
      "grad_norm": 0.4864049255847931,
      "learning_rate": 0.00024373116479311168,
      "loss": 0.0633,
      "step": 4426
    },
    {
      "epoch": 0.6204625087596356,
      "grad_norm": 0.5863664150238037,
      "learning_rate": 0.000243716814159292,
      "loss": 0.1467,
      "step": 4427
    },
    {
      "epoch": 0.6206026629292222,
      "grad_norm": 0.2796110510826111,
      "learning_rate": 0.00024370246352547236,
      "loss": 0.0379,
      "step": 4428
    },
    {
      "epoch": 0.6207428170988087,
      "grad_norm": 0.5027040243148804,
      "learning_rate": 0.0002436881128916527,
      "loss": 0.0806,
      "step": 4429
    },
    {
      "epoch": 0.6208829712683952,
      "grad_norm": 0.534468948841095,
      "learning_rate": 0.00024367376225783302,
      "loss": 0.1218,
      "step": 4430
    },
    {
      "epoch": 0.6210231254379818,
      "grad_norm": 0.510786771774292,
      "learning_rate": 0.00024365941162401335,
      "loss": 0.184,
      "step": 4431
    },
    {
      "epoch": 0.6211632796075683,
      "grad_norm": 0.41986384987831116,
      "learning_rate": 0.0002436450609901937,
      "loss": 0.0406,
      "step": 4432
    },
    {
      "epoch": 0.6213034337771549,
      "grad_norm": 0.34594154357910156,
      "learning_rate": 0.00024363071035637406,
      "loss": 0.0668,
      "step": 4433
    },
    {
      "epoch": 0.6214435879467414,
      "grad_norm": 0.3428405523300171,
      "learning_rate": 0.0002436163597225544,
      "loss": 0.0431,
      "step": 4434
    },
    {
      "epoch": 0.6215837421163279,
      "grad_norm": 0.4442746341228485,
      "learning_rate": 0.00024360200908873475,
      "loss": 0.0855,
      "step": 4435
    },
    {
      "epoch": 0.6217238962859145,
      "grad_norm": 0.36158838868141174,
      "learning_rate": 0.00024358765845491507,
      "loss": 0.0686,
      "step": 4436
    },
    {
      "epoch": 0.6218640504555011,
      "grad_norm": 0.41118520498275757,
      "learning_rate": 0.0002435733078210954,
      "loss": 0.094,
      "step": 4437
    },
    {
      "epoch": 0.6220042046250875,
      "grad_norm": 0.7667691707611084,
      "learning_rate": 0.00024355895718727576,
      "loss": 0.1374,
      "step": 4438
    },
    {
      "epoch": 0.6221443587946741,
      "grad_norm": 0.3761586546897888,
      "learning_rate": 0.0002435446065534561,
      "loss": 0.0448,
      "step": 4439
    },
    {
      "epoch": 0.6222845129642607,
      "grad_norm": 0.5398222208023071,
      "learning_rate": 0.00024353025591963642,
      "loss": 0.0936,
      "step": 4440
    },
    {
      "epoch": 0.6224246671338473,
      "grad_norm": 0.4447212517261505,
      "learning_rate": 0.00024351590528581677,
      "loss": 0.1139,
      "step": 4441
    },
    {
      "epoch": 0.6225648213034338,
      "grad_norm": 0.36371704936027527,
      "learning_rate": 0.0002435015546519971,
      "loss": 0.0634,
      "step": 4442
    },
    {
      "epoch": 0.6227049754730203,
      "grad_norm": 0.560509204864502,
      "learning_rate": 0.00024348720401817743,
      "loss": 0.0689,
      "step": 4443
    },
    {
      "epoch": 0.6228451296426069,
      "grad_norm": 0.4575555920600891,
      "learning_rate": 0.0002434728533843578,
      "loss": 0.0301,
      "step": 4444
    },
    {
      "epoch": 0.6229852838121934,
      "grad_norm": 0.3562361001968384,
      "learning_rate": 0.00024345850275053814,
      "loss": 0.091,
      "step": 4445
    },
    {
      "epoch": 0.62312543798178,
      "grad_norm": 1.1659826040267944,
      "learning_rate": 0.00024344415211671847,
      "loss": 0.257,
      "step": 4446
    },
    {
      "epoch": 0.6232655921513665,
      "grad_norm": 0.4810757040977478,
      "learning_rate": 0.00024342980148289882,
      "loss": 0.1023,
      "step": 4447
    },
    {
      "epoch": 0.623405746320953,
      "grad_norm": 2.876710891723633,
      "learning_rate": 0.00024341545084907915,
      "loss": 0.0887,
      "step": 4448
    },
    {
      "epoch": 0.6235459004905396,
      "grad_norm": 0.4492841362953186,
      "learning_rate": 0.00024340110021525948,
      "loss": 0.0571,
      "step": 4449
    },
    {
      "epoch": 0.6236860546601262,
      "grad_norm": 2.8683910369873047,
      "learning_rate": 0.0002433867495814398,
      "loss": 0.2872,
      "step": 4450
    },
    {
      "epoch": 0.6238262088297127,
      "grad_norm": 0.38978391885757446,
      "learning_rate": 0.00024337239894762017,
      "loss": 0.0873,
      "step": 4451
    },
    {
      "epoch": 0.6239663629992992,
      "grad_norm": 0.6746708750724792,
      "learning_rate": 0.0002433580483138005,
      "loss": 0.1101,
      "step": 4452
    },
    {
      "epoch": 0.6241065171688858,
      "grad_norm": 0.8837757110595703,
      "learning_rate": 0.00024334369767998082,
      "loss": 0.066,
      "step": 4453
    },
    {
      "epoch": 0.6242466713384723,
      "grad_norm": 0.6266793608665466,
      "learning_rate": 0.0002433293470461612,
      "loss": 0.1132,
      "step": 4454
    },
    {
      "epoch": 0.6243868255080589,
      "grad_norm": 0.65644371509552,
      "learning_rate": 0.00024331499641234153,
      "loss": 0.1313,
      "step": 4455
    },
    {
      "epoch": 0.6245269796776454,
      "grad_norm": 0.3828071355819702,
      "learning_rate": 0.00024330064577852186,
      "loss": 0.0594,
      "step": 4456
    },
    {
      "epoch": 0.6246671338472319,
      "grad_norm": 0.2612523138523102,
      "learning_rate": 0.00024328629514470222,
      "loss": 0.085,
      "step": 4457
    },
    {
      "epoch": 0.6248072880168185,
      "grad_norm": 0.26661962270736694,
      "learning_rate": 0.00024327194451088255,
      "loss": 0.063,
      "step": 4458
    },
    {
      "epoch": 0.6249474421864051,
      "grad_norm": 0.383114755153656,
      "learning_rate": 0.00024325759387706288,
      "loss": 0.0802,
      "step": 4459
    },
    {
      "epoch": 0.6250875963559915,
      "grad_norm": 0.3588640093803406,
      "learning_rate": 0.00024324324324324323,
      "loss": 0.0834,
      "step": 4460
    },
    {
      "epoch": 0.6252277505255781,
      "grad_norm": 0.14527222514152527,
      "learning_rate": 0.00024322889260942356,
      "loss": 0.0262,
      "step": 4461
    },
    {
      "epoch": 0.6253679046951647,
      "grad_norm": 0.3797816336154938,
      "learning_rate": 0.0002432145419756039,
      "loss": 0.0754,
      "step": 4462
    },
    {
      "epoch": 0.6255080588647512,
      "grad_norm": 0.42258310317993164,
      "learning_rate": 0.00024320019134178424,
      "loss": 0.0364,
      "step": 4463
    },
    {
      "epoch": 0.6256482130343378,
      "grad_norm": 0.5546248555183411,
      "learning_rate": 0.00024318584070796457,
      "loss": 0.0668,
      "step": 4464
    },
    {
      "epoch": 0.6257883672039243,
      "grad_norm": 0.2608524560928345,
      "learning_rate": 0.00024317149007414493,
      "loss": 0.0336,
      "step": 4465
    },
    {
      "epoch": 0.6259285213735108,
      "grad_norm": 0.5189929008483887,
      "learning_rate": 0.00024315713944032526,
      "loss": 0.0721,
      "step": 4466
    },
    {
      "epoch": 0.6260686755430974,
      "grad_norm": 0.40395602583885193,
      "learning_rate": 0.0002431427888065056,
      "loss": 0.0432,
      "step": 4467
    },
    {
      "epoch": 0.626208829712684,
      "grad_norm": 1.1061335802078247,
      "learning_rate": 0.00024312843817268594,
      "loss": 0.0513,
      "step": 4468
    },
    {
      "epoch": 0.6263489838822704,
      "grad_norm": 0.3939414322376251,
      "learning_rate": 0.00024311408753886627,
      "loss": 0.0877,
      "step": 4469
    },
    {
      "epoch": 0.626489138051857,
      "grad_norm": 0.6109133958816528,
      "learning_rate": 0.00024309973690504662,
      "loss": 0.1487,
      "step": 4470
    },
    {
      "epoch": 0.6266292922214436,
      "grad_norm": 1.2329773902893066,
      "learning_rate": 0.00024308538627122695,
      "loss": 0.1253,
      "step": 4471
    },
    {
      "epoch": 0.6267694463910302,
      "grad_norm": 0.4583156704902649,
      "learning_rate": 0.00024307103563740728,
      "loss": 0.0603,
      "step": 4472
    },
    {
      "epoch": 0.6269096005606167,
      "grad_norm": 0.7002905011177063,
      "learning_rate": 0.00024305668500358764,
      "loss": 0.3109,
      "step": 4473
    },
    {
      "epoch": 0.6270497547302032,
      "grad_norm": 0.5552048683166504,
      "learning_rate": 0.00024304233436976797,
      "loss": 0.2018,
      "step": 4474
    },
    {
      "epoch": 0.6271899088997898,
      "grad_norm": 0.5952927470207214,
      "learning_rate": 0.0002430279837359483,
      "loss": 0.1302,
      "step": 4475
    },
    {
      "epoch": 0.6273300630693763,
      "grad_norm": 0.32670772075653076,
      "learning_rate": 0.00024301363310212868,
      "loss": 0.0501,
      "step": 4476
    },
    {
      "epoch": 0.6274702172389629,
      "grad_norm": 0.27814510464668274,
      "learning_rate": 0.000242999282468309,
      "loss": 0.0888,
      "step": 4477
    },
    {
      "epoch": 0.6276103714085494,
      "grad_norm": 0.4076862931251526,
      "learning_rate": 0.00024298493183448933,
      "loss": 0.0935,
      "step": 4478
    },
    {
      "epoch": 0.6277505255781359,
      "grad_norm": 0.3227880597114563,
      "learning_rate": 0.0002429705812006697,
      "loss": 0.0814,
      "step": 4479
    },
    {
      "epoch": 0.6278906797477225,
      "grad_norm": 0.7244734764099121,
      "learning_rate": 0.00024295623056685002,
      "loss": 0.145,
      "step": 4480
    },
    {
      "epoch": 0.6280308339173091,
      "grad_norm": 0.30963683128356934,
      "learning_rate": 0.00024294187993303035,
      "loss": 0.059,
      "step": 4481
    },
    {
      "epoch": 0.6281709880868955,
      "grad_norm": 0.3892687261104584,
      "learning_rate": 0.0002429275292992107,
      "loss": 0.0925,
      "step": 4482
    },
    {
      "epoch": 0.6283111422564821,
      "grad_norm": 0.45980051159858704,
      "learning_rate": 0.00024291317866539103,
      "loss": 0.0857,
      "step": 4483
    },
    {
      "epoch": 0.6284512964260687,
      "grad_norm": 0.22275343537330627,
      "learning_rate": 0.00024289882803157136,
      "loss": 0.0815,
      "step": 4484
    },
    {
      "epoch": 0.6285914505956552,
      "grad_norm": 0.08691888302564621,
      "learning_rate": 0.0002428844773977517,
      "loss": 0.0202,
      "step": 4485
    },
    {
      "epoch": 0.6287316047652418,
      "grad_norm": 0.498346745967865,
      "learning_rate": 0.00024287012676393207,
      "loss": 0.1034,
      "step": 4486
    },
    {
      "epoch": 0.6288717589348283,
      "grad_norm": 0.3046766221523285,
      "learning_rate": 0.0002428557761301124,
      "loss": 0.0921,
      "step": 4487
    },
    {
      "epoch": 0.6290119131044148,
      "grad_norm": 0.3708658516407013,
      "learning_rate": 0.00024284142549629273,
      "loss": 0.0756,
      "step": 4488
    },
    {
      "epoch": 0.6291520672740014,
      "grad_norm": 0.49132239818573,
      "learning_rate": 0.00024282707486247308,
      "loss": 0.1751,
      "step": 4489
    },
    {
      "epoch": 0.629292221443588,
      "grad_norm": 0.37744471430778503,
      "learning_rate": 0.0002428127242286534,
      "loss": 0.0941,
      "step": 4490
    },
    {
      "epoch": 0.6294323756131744,
      "grad_norm": 0.4208647608757019,
      "learning_rate": 0.00024279837359483374,
      "loss": 0.0893,
      "step": 4491
    },
    {
      "epoch": 0.629572529782761,
      "grad_norm": 0.928225040435791,
      "learning_rate": 0.0002427840229610141,
      "loss": 0.1669,
      "step": 4492
    },
    {
      "epoch": 0.6297126839523476,
      "grad_norm": 0.12094986438751221,
      "learning_rate": 0.00024276967232719443,
      "loss": 0.0171,
      "step": 4493
    },
    {
      "epoch": 0.6298528381219342,
      "grad_norm": 0.6277962327003479,
      "learning_rate": 0.00024275532169337475,
      "loss": 0.1133,
      "step": 4494
    },
    {
      "epoch": 0.6299929922915207,
      "grad_norm": 0.4300065338611603,
      "learning_rate": 0.0002427409710595551,
      "loss": 0.0739,
      "step": 4495
    },
    {
      "epoch": 0.6301331464611072,
      "grad_norm": 0.383602112531662,
      "learning_rate": 0.00024272662042573544,
      "loss": 0.0612,
      "step": 4496
    },
    {
      "epoch": 0.6302733006306938,
      "grad_norm": 0.8136420845985413,
      "learning_rate": 0.0002427122697919158,
      "loss": 0.4323,
      "step": 4497
    },
    {
      "epoch": 0.6304134548002803,
      "grad_norm": 2.779512882232666,
      "learning_rate": 0.00024269791915809615,
      "loss": 0.2818,
      "step": 4498
    },
    {
      "epoch": 0.6305536089698669,
      "grad_norm": 1.7227791547775269,
      "learning_rate": 0.00024268356852427648,
      "loss": 0.4441,
      "step": 4499
    },
    {
      "epoch": 0.6306937631394534,
      "grad_norm": 3.11232852935791,
      "learning_rate": 0.0002426692178904568,
      "loss": 0.2778,
      "step": 4500
    },
    {
      "epoch": 0.6308339173090399,
      "grad_norm": 0.5326541662216187,
      "learning_rate": 0.00024265486725663714,
      "loss": 0.1259,
      "step": 4501
    },
    {
      "epoch": 0.6309740714786265,
      "grad_norm": 0.3800836205482483,
      "learning_rate": 0.0002426405166228175,
      "loss": 0.1182,
      "step": 4502
    },
    {
      "epoch": 0.6311142256482131,
      "grad_norm": 0.3213178217411041,
      "learning_rate": 0.00024262616598899782,
      "loss": 0.0744,
      "step": 4503
    },
    {
      "epoch": 0.6312543798177995,
      "grad_norm": 0.2859538793563843,
      "learning_rate": 0.00024261181535517815,
      "loss": 0.1095,
      "step": 4504
    },
    {
      "epoch": 0.6313945339873861,
      "grad_norm": 0.3558676242828369,
      "learning_rate": 0.0002425974647213585,
      "loss": 0.0637,
      "step": 4505
    },
    {
      "epoch": 0.6315346881569727,
      "grad_norm": 0.45185384154319763,
      "learning_rate": 0.00024258311408753883,
      "loss": 0.058,
      "step": 4506
    },
    {
      "epoch": 0.6316748423265592,
      "grad_norm": 0.3520965874195099,
      "learning_rate": 0.00024256876345371916,
      "loss": 0.0732,
      "step": 4507
    },
    {
      "epoch": 0.6318149964961458,
      "grad_norm": 0.7532474398612976,
      "learning_rate": 0.00024255441281989954,
      "loss": 0.0447,
      "step": 4508
    },
    {
      "epoch": 0.6319551506657323,
      "grad_norm": 0.25900688767433167,
      "learning_rate": 0.00024254006218607987,
      "loss": 0.0613,
      "step": 4509
    },
    {
      "epoch": 0.6320953048353188,
      "grad_norm": 0.7063947916030884,
      "learning_rate": 0.0002425257115522602,
      "loss": 0.067,
      "step": 4510
    },
    {
      "epoch": 0.6322354590049054,
      "grad_norm": 0.5420425534248352,
      "learning_rate": 0.00024251136091844056,
      "loss": 0.1357,
      "step": 4511
    },
    {
      "epoch": 0.632375613174492,
      "grad_norm": 0.4076794683933258,
      "learning_rate": 0.00024249701028462089,
      "loss": 0.0741,
      "step": 4512
    },
    {
      "epoch": 0.6325157673440784,
      "grad_norm": 0.2640433609485626,
      "learning_rate": 0.00024248265965080121,
      "loss": 0.0798,
      "step": 4513
    },
    {
      "epoch": 0.632655921513665,
      "grad_norm": 0.48513099551200867,
      "learning_rate": 0.00024246830901698157,
      "loss": 0.0627,
      "step": 4514
    },
    {
      "epoch": 0.6327960756832516,
      "grad_norm": 0.4062906503677368,
      "learning_rate": 0.0002424539583831619,
      "loss": 0.0986,
      "step": 4515
    },
    {
      "epoch": 0.6329362298528382,
      "grad_norm": 0.398327499628067,
      "learning_rate": 0.00024243960774934223,
      "loss": 0.1177,
      "step": 4516
    },
    {
      "epoch": 0.6330763840224247,
      "grad_norm": 0.5958709716796875,
      "learning_rate": 0.0002424252571155226,
      "loss": 0.1484,
      "step": 4517
    },
    {
      "epoch": 0.6332165381920112,
      "grad_norm": 0.4138696789741516,
      "learning_rate": 0.00024241090648170294,
      "loss": 0.0476,
      "step": 4518
    },
    {
      "epoch": 0.6333566923615978,
      "grad_norm": 0.3402014672756195,
      "learning_rate": 0.00024239655584788327,
      "loss": 0.0467,
      "step": 4519
    },
    {
      "epoch": 0.6334968465311843,
      "grad_norm": 0.9100338220596313,
      "learning_rate": 0.0002423822052140636,
      "loss": 0.0972,
      "step": 4520
    },
    {
      "epoch": 0.6336370007007709,
      "grad_norm": 0.23704035580158234,
      "learning_rate": 0.00024236785458024395,
      "loss": 0.0418,
      "step": 4521
    },
    {
      "epoch": 0.6337771548703574,
      "grad_norm": 0.8553435802459717,
      "learning_rate": 0.00024235350394642428,
      "loss": 0.0573,
      "step": 4522
    },
    {
      "epoch": 0.6339173090399439,
      "grad_norm": 0.48949527740478516,
      "learning_rate": 0.0002423391533126046,
      "loss": 0.0934,
      "step": 4523
    },
    {
      "epoch": 0.6340574632095305,
      "grad_norm": 1.2994810342788696,
      "learning_rate": 0.00024232480267878496,
      "loss": 0.1191,
      "step": 4524
    },
    {
      "epoch": 0.6341976173791171,
      "grad_norm": 0.16651922464370728,
      "learning_rate": 0.0002423104520449653,
      "loss": 0.0198,
      "step": 4525
    },
    {
      "epoch": 0.6343377715487035,
      "grad_norm": 0.38856351375579834,
      "learning_rate": 0.00024229610141114562,
      "loss": 0.0726,
      "step": 4526
    },
    {
      "epoch": 0.6344779257182901,
      "grad_norm": 0.22474133968353271,
      "learning_rate": 0.00024228175077732598,
      "loss": 0.0545,
      "step": 4527
    },
    {
      "epoch": 0.6346180798878767,
      "grad_norm": 0.6340205669403076,
      "learning_rate": 0.00024226740014350633,
      "loss": 0.1177,
      "step": 4528
    },
    {
      "epoch": 0.6347582340574632,
      "grad_norm": 0.5469197034835815,
      "learning_rate": 0.00024225304950968666,
      "loss": 0.1142,
      "step": 4529
    },
    {
      "epoch": 0.6348983882270498,
      "grad_norm": 0.24882544577121735,
      "learning_rate": 0.00024223869887586702,
      "loss": 0.0206,
      "step": 4530
    },
    {
      "epoch": 0.6350385423966363,
      "grad_norm": 0.4992559552192688,
      "learning_rate": 0.00024222434824204734,
      "loss": 0.0444,
      "step": 4531
    },
    {
      "epoch": 0.6351786965662228,
      "grad_norm": 0.38363006711006165,
      "learning_rate": 0.00024220999760822767,
      "loss": 0.1327,
      "step": 4532
    },
    {
      "epoch": 0.6353188507358094,
      "grad_norm": 0.47136613726615906,
      "learning_rate": 0.00024219564697440803,
      "loss": 0.0883,
      "step": 4533
    },
    {
      "epoch": 0.635459004905396,
      "grad_norm": 0.38705387711524963,
      "learning_rate": 0.00024218129634058836,
      "loss": 0.0336,
      "step": 4534
    },
    {
      "epoch": 0.6355991590749824,
      "grad_norm": 0.44826316833496094,
      "learning_rate": 0.00024216694570676869,
      "loss": 0.1161,
      "step": 4535
    },
    {
      "epoch": 0.635739313244569,
      "grad_norm": 0.8910202383995056,
      "learning_rate": 0.00024215259507294901,
      "loss": 0.0834,
      "step": 4536
    },
    {
      "epoch": 0.6358794674141556,
      "grad_norm": 0.3817906379699707,
      "learning_rate": 0.00024213824443912937,
      "loss": 0.0529,
      "step": 4537
    },
    {
      "epoch": 0.6360196215837421,
      "grad_norm": 0.46518051624298096,
      "learning_rate": 0.0002421238938053097,
      "loss": 0.102,
      "step": 4538
    },
    {
      "epoch": 0.6361597757533287,
      "grad_norm": 0.18469616770744324,
      "learning_rate": 0.00024210954317149003,
      "loss": 0.0309,
      "step": 4539
    },
    {
      "epoch": 0.6362999299229152,
      "grad_norm": 0.6559878587722778,
      "learning_rate": 0.0002420951925376704,
      "loss": 0.0748,
      "step": 4540
    },
    {
      "epoch": 0.6364400840925017,
      "grad_norm": 0.24291583895683289,
      "learning_rate": 0.00024208084190385074,
      "loss": 0.0147,
      "step": 4541
    },
    {
      "epoch": 0.6365802382620883,
      "grad_norm": 0.3506116271018982,
      "learning_rate": 0.00024206649127003107,
      "loss": 0.092,
      "step": 4542
    },
    {
      "epoch": 0.6367203924316749,
      "grad_norm": 0.7008859515190125,
      "learning_rate": 0.00024205214063621142,
      "loss": 0.0799,
      "step": 4543
    },
    {
      "epoch": 0.6368605466012613,
      "grad_norm": 0.36548927426338196,
      "learning_rate": 0.00024203779000239175,
      "loss": 0.0581,
      "step": 4544
    },
    {
      "epoch": 0.6370007007708479,
      "grad_norm": 1.1923069953918457,
      "learning_rate": 0.00024202343936857208,
      "loss": 0.2091,
      "step": 4545
    },
    {
      "epoch": 0.6371408549404345,
      "grad_norm": 0.2648469805717468,
      "learning_rate": 0.00024200908873475244,
      "loss": 0.0088,
      "step": 4546
    },
    {
      "epoch": 0.6372810091100211,
      "grad_norm": 0.5329459309577942,
      "learning_rate": 0.00024199473810093276,
      "loss": 0.0616,
      "step": 4547
    },
    {
      "epoch": 0.6374211632796075,
      "grad_norm": 2.047420024871826,
      "learning_rate": 0.0002419803874671131,
      "loss": 0.1895,
      "step": 4548
    },
    {
      "epoch": 0.6375613174491941,
      "grad_norm": 11.075716018676758,
      "learning_rate": 0.00024196603683329348,
      "loss": 0.336,
      "step": 4549
    },
    {
      "epoch": 0.6377014716187807,
      "grad_norm": 2.1514623165130615,
      "learning_rate": 0.0002419516861994738,
      "loss": 0.1376,
      "step": 4550
    },
    {
      "epoch": 0.6378416257883672,
      "grad_norm": 0.6289103031158447,
      "learning_rate": 0.00024193733556565413,
      "loss": 0.0803,
      "step": 4551
    },
    {
      "epoch": 0.6379817799579538,
      "grad_norm": 0.3676750659942627,
      "learning_rate": 0.00024192298493183446,
      "loss": 0.0837,
      "step": 4552
    },
    {
      "epoch": 0.6381219341275403,
      "grad_norm": 0.46150317788124084,
      "learning_rate": 0.00024190863429801482,
      "loss": 0.1248,
      "step": 4553
    },
    {
      "epoch": 0.6382620882971268,
      "grad_norm": 0.46485984325408936,
      "learning_rate": 0.00024189428366419515,
      "loss": 0.0984,
      "step": 4554
    },
    {
      "epoch": 0.6384022424667134,
      "grad_norm": 0.5445472598075867,
      "learning_rate": 0.00024187993303037547,
      "loss": 0.1094,
      "step": 4555
    },
    {
      "epoch": 0.6385423966363,
      "grad_norm": 0.4286209046840668,
      "learning_rate": 0.00024186558239655583,
      "loss": 0.1566,
      "step": 4556
    },
    {
      "epoch": 0.6386825508058864,
      "grad_norm": 0.45852240920066833,
      "learning_rate": 0.00024185123176273616,
      "loss": 0.0801,
      "step": 4557
    },
    {
      "epoch": 0.638822704975473,
      "grad_norm": 0.27761003375053406,
      "learning_rate": 0.0002418368811289165,
      "loss": 0.037,
      "step": 4558
    },
    {
      "epoch": 0.6389628591450596,
      "grad_norm": 0.23453634977340698,
      "learning_rate": 0.00024182253049509684,
      "loss": 0.0557,
      "step": 4559
    },
    {
      "epoch": 0.6391030133146461,
      "grad_norm": 0.2455267310142517,
      "learning_rate": 0.0002418081798612772,
      "loss": 0.0303,
      "step": 4560
    },
    {
      "epoch": 0.6392431674842327,
      "grad_norm": 0.5576555132865906,
      "learning_rate": 0.00024179382922745753,
      "loss": 0.0907,
      "step": 4561
    },
    {
      "epoch": 0.6393833216538192,
      "grad_norm": 0.36986711621284485,
      "learning_rate": 0.00024177947859363788,
      "loss": 0.0467,
      "step": 4562
    },
    {
      "epoch": 0.6395234758234057,
      "grad_norm": 0.5336852073669434,
      "learning_rate": 0.0002417651279598182,
      "loss": 0.0908,
      "step": 4563
    },
    {
      "epoch": 0.6396636299929923,
      "grad_norm": 0.754698634147644,
      "learning_rate": 0.00024175077732599854,
      "loss": 0.1407,
      "step": 4564
    },
    {
      "epoch": 0.6398037841625789,
      "grad_norm": 0.3857322335243225,
      "learning_rate": 0.0002417364266921789,
      "loss": 0.0639,
      "step": 4565
    },
    {
      "epoch": 0.6399439383321653,
      "grad_norm": 0.8895483613014221,
      "learning_rate": 0.00024172207605835922,
      "loss": 0.0936,
      "step": 4566
    },
    {
      "epoch": 0.6400840925017519,
      "grad_norm": 0.3353564739227295,
      "learning_rate": 0.00024170772542453955,
      "loss": 0.1098,
      "step": 4567
    },
    {
      "epoch": 0.6402242466713385,
      "grad_norm": 0.4150405824184418,
      "learning_rate": 0.0002416933747907199,
      "loss": 0.0609,
      "step": 4568
    },
    {
      "epoch": 0.640364400840925,
      "grad_norm": 0.20065344870090485,
      "learning_rate": 0.00024167902415690024,
      "loss": 0.0339,
      "step": 4569
    },
    {
      "epoch": 0.6405045550105115,
      "grad_norm": 0.3569844365119934,
      "learning_rate": 0.00024166467352308056,
      "loss": 0.0917,
      "step": 4570
    },
    {
      "epoch": 0.6406447091800981,
      "grad_norm": 0.4130513072013855,
      "learning_rate": 0.0002416503228892609,
      "loss": 0.1194,
      "step": 4571
    },
    {
      "epoch": 0.6407848633496847,
      "grad_norm": 0.22559505701065063,
      "learning_rate": 0.00024163597225544128,
      "loss": 0.0839,
      "step": 4572
    },
    {
      "epoch": 0.6409250175192712,
      "grad_norm": 0.18988323211669922,
      "learning_rate": 0.0002416216216216216,
      "loss": 0.028,
      "step": 4573
    },
    {
      "epoch": 0.6410651716888578,
      "grad_norm": 0.49719610810279846,
      "learning_rate": 0.00024160727098780193,
      "loss": 0.0615,
      "step": 4574
    },
    {
      "epoch": 0.6412053258584443,
      "grad_norm": 0.2265058010816574,
      "learning_rate": 0.0002415929203539823,
      "loss": 0.0342,
      "step": 4575
    },
    {
      "epoch": 0.6413454800280308,
      "grad_norm": 0.4475458562374115,
      "learning_rate": 0.00024157856972016262,
      "loss": 0.0708,
      "step": 4576
    },
    {
      "epoch": 0.6414856341976174,
      "grad_norm": 0.2926807403564453,
      "learning_rate": 0.00024156421908634295,
      "loss": 0.0429,
      "step": 4577
    },
    {
      "epoch": 0.641625788367204,
      "grad_norm": 0.16132262349128723,
      "learning_rate": 0.0002415498684525233,
      "loss": 0.032,
      "step": 4578
    },
    {
      "epoch": 0.6417659425367904,
      "grad_norm": 0.5955548286437988,
      "learning_rate": 0.00024153551781870363,
      "loss": 0.065,
      "step": 4579
    },
    {
      "epoch": 0.641906096706377,
      "grad_norm": 0.413908451795578,
      "learning_rate": 0.00024152116718488396,
      "loss": 0.0698,
      "step": 4580
    },
    {
      "epoch": 0.6420462508759636,
      "grad_norm": 0.27634891867637634,
      "learning_rate": 0.00024150681655106434,
      "loss": 0.1072,
      "step": 4581
    },
    {
      "epoch": 0.6421864050455501,
      "grad_norm": 0.4600512385368347,
      "learning_rate": 0.00024149246591724467,
      "loss": 0.0664,
      "step": 4582
    },
    {
      "epoch": 0.6423265592151367,
      "grad_norm": 0.27367252111434937,
      "learning_rate": 0.000241478115283425,
      "loss": 0.0354,
      "step": 4583
    },
    {
      "epoch": 0.6424667133847232,
      "grad_norm": 0.44200485944747925,
      "learning_rate": 0.00024146376464960535,
      "loss": 0.0658,
      "step": 4584
    },
    {
      "epoch": 0.6426068675543097,
      "grad_norm": 0.9172531366348267,
      "learning_rate": 0.00024144941401578568,
      "loss": 0.1618,
      "step": 4585
    },
    {
      "epoch": 0.6427470217238963,
      "grad_norm": 0.4246320128440857,
      "learning_rate": 0.000241435063381966,
      "loss": 0.0349,
      "step": 4586
    },
    {
      "epoch": 0.6428871758934829,
      "grad_norm": 0.7287715077400208,
      "learning_rate": 0.00024142071274814634,
      "loss": 0.1912,
      "step": 4587
    },
    {
      "epoch": 0.6430273300630693,
      "grad_norm": 0.25295770168304443,
      "learning_rate": 0.0002414063621143267,
      "loss": 0.0534,
      "step": 4588
    },
    {
      "epoch": 0.6431674842326559,
      "grad_norm": 0.959119975566864,
      "learning_rate": 0.00024139201148050702,
      "loss": 0.1385,
      "step": 4589
    },
    {
      "epoch": 0.6433076384022425,
      "grad_norm": 0.5495650172233582,
      "learning_rate": 0.00024137766084668735,
      "loss": 0.1598,
      "step": 4590
    },
    {
      "epoch": 0.643447792571829,
      "grad_norm": 0.36835548281669617,
      "learning_rate": 0.0002413633102128677,
      "loss": 0.0543,
      "step": 4591
    },
    {
      "epoch": 0.6435879467414155,
      "grad_norm": 0.6164745092391968,
      "learning_rate": 0.00024134895957904806,
      "loss": 0.0457,
      "step": 4592
    },
    {
      "epoch": 0.6437281009110021,
      "grad_norm": 0.3202774226665497,
      "learning_rate": 0.0002413346089452284,
      "loss": 0.049,
      "step": 4593
    },
    {
      "epoch": 0.6438682550805886,
      "grad_norm": 0.38736215233802795,
      "learning_rate": 0.00024132025831140875,
      "loss": 0.2323,
      "step": 4594
    },
    {
      "epoch": 0.6440084092501752,
      "grad_norm": 0.7943335771560669,
      "learning_rate": 0.00024130590767758908,
      "loss": 0.0683,
      "step": 4595
    },
    {
      "epoch": 0.6441485634197618,
      "grad_norm": 0.36901378631591797,
      "learning_rate": 0.0002412915570437694,
      "loss": 0.0815,
      "step": 4596
    },
    {
      "epoch": 0.6442887175893482,
      "grad_norm": 0.9580649733543396,
      "learning_rate": 0.00024127720640994976,
      "loss": 0.2632,
      "step": 4597
    },
    {
      "epoch": 0.6444288717589348,
      "grad_norm": 0.7454104423522949,
      "learning_rate": 0.0002412628557761301,
      "loss": 0.1209,
      "step": 4598
    },
    {
      "epoch": 0.6445690259285214,
      "grad_norm": 2.1465260982513428,
      "learning_rate": 0.00024124850514231042,
      "loss": 0.1245,
      "step": 4599
    },
    {
      "epoch": 0.644709180098108,
      "grad_norm": 1.2225797176361084,
      "learning_rate": 0.00024123415450849077,
      "loss": 0.356,
      "step": 4600
    },
    {
      "epoch": 0.6448493342676944,
      "grad_norm": 0.29114487767219543,
      "learning_rate": 0.0002412198038746711,
      "loss": 0.1,
      "step": 4601
    },
    {
      "epoch": 0.644989488437281,
      "grad_norm": 0.3677774667739868,
      "learning_rate": 0.00024120545324085143,
      "loss": 0.0582,
      "step": 4602
    },
    {
      "epoch": 0.6451296426068676,
      "grad_norm": 0.22189921140670776,
      "learning_rate": 0.00024119110260703181,
      "loss": 0.0756,
      "step": 4603
    },
    {
      "epoch": 0.6452697967764541,
      "grad_norm": 0.4159783124923706,
      "learning_rate": 0.00024117675197321214,
      "loss": 0.1066,
      "step": 4604
    },
    {
      "epoch": 0.6454099509460406,
      "grad_norm": 0.3475879728794098,
      "learning_rate": 0.00024116240133939247,
      "loss": 0.0708,
      "step": 4605
    },
    {
      "epoch": 0.6455501051156272,
      "grad_norm": 0.2770099341869354,
      "learning_rate": 0.0002411480507055728,
      "loss": 0.121,
      "step": 4606
    },
    {
      "epoch": 0.6456902592852137,
      "grad_norm": 0.9238994121551514,
      "learning_rate": 0.00024113370007175316,
      "loss": 0.0834,
      "step": 4607
    },
    {
      "epoch": 0.6458304134548003,
      "grad_norm": 0.29952025413513184,
      "learning_rate": 0.00024111934943793348,
      "loss": 0.1106,
      "step": 4608
    },
    {
      "epoch": 0.6459705676243869,
      "grad_norm": 0.2555348575115204,
      "learning_rate": 0.0002411049988041138,
      "loss": 0.0633,
      "step": 4609
    },
    {
      "epoch": 0.6461107217939733,
      "grad_norm": 0.42149806022644043,
      "learning_rate": 0.00024109064817029417,
      "loss": 0.1139,
      "step": 4610
    },
    {
      "epoch": 0.6462508759635599,
      "grad_norm": 0.5292149782180786,
      "learning_rate": 0.0002410762975364745,
      "loss": 0.0853,
      "step": 4611
    },
    {
      "epoch": 0.6463910301331465,
      "grad_norm": 0.6884406805038452,
      "learning_rate": 0.00024106194690265483,
      "loss": 0.1194,
      "step": 4612
    },
    {
      "epoch": 0.646531184302733,
      "grad_norm": 0.37258827686309814,
      "learning_rate": 0.0002410475962688352,
      "loss": 0.0624,
      "step": 4613
    },
    {
      "epoch": 0.6466713384723195,
      "grad_norm": 0.8067600727081299,
      "learning_rate": 0.00024103324563501554,
      "loss": 0.0473,
      "step": 4614
    },
    {
      "epoch": 0.6468114926419061,
      "grad_norm": 0.33629342913627625,
      "learning_rate": 0.00024101889500119587,
      "loss": 0.1091,
      "step": 4615
    },
    {
      "epoch": 0.6469516468114926,
      "grad_norm": 0.572068452835083,
      "learning_rate": 0.00024100454436737622,
      "loss": 0.1286,
      "step": 4616
    },
    {
      "epoch": 0.6470918009810792,
      "grad_norm": 0.3866429924964905,
      "learning_rate": 0.00024099019373355655,
      "loss": 0.0723,
      "step": 4617
    },
    {
      "epoch": 0.6472319551506658,
      "grad_norm": 0.23980627954006195,
      "learning_rate": 0.00024097584309973688,
      "loss": 0.0529,
      "step": 4618
    },
    {
      "epoch": 0.6473721093202522,
      "grad_norm": 0.7995784878730774,
      "learning_rate": 0.00024096149246591723,
      "loss": 0.16,
      "step": 4619
    },
    {
      "epoch": 0.6475122634898388,
      "grad_norm": 1.1471000909805298,
      "learning_rate": 0.00024094714183209756,
      "loss": 0.1222,
      "step": 4620
    },
    {
      "epoch": 0.6476524176594254,
      "grad_norm": 0.42206305265426636,
      "learning_rate": 0.0002409327911982779,
      "loss": 0.0744,
      "step": 4621
    },
    {
      "epoch": 0.647792571829012,
      "grad_norm": 0.48790743947029114,
      "learning_rate": 0.00024091844056445822,
      "loss": 0.0645,
      "step": 4622
    },
    {
      "epoch": 0.6479327259985984,
      "grad_norm": 0.3827284574508667,
      "learning_rate": 0.0002409040899306386,
      "loss": 0.0614,
      "step": 4623
    },
    {
      "epoch": 0.648072880168185,
      "grad_norm": 0.5078491568565369,
      "learning_rate": 0.00024088973929681893,
      "loss": 0.06,
      "step": 4624
    },
    {
      "epoch": 0.6482130343377716,
      "grad_norm": 0.13038191199302673,
      "learning_rate": 0.00024087538866299926,
      "loss": 0.009,
      "step": 4625
    },
    {
      "epoch": 0.6483531885073581,
      "grad_norm": 0.3788299560546875,
      "learning_rate": 0.00024086103802917961,
      "loss": 0.0476,
      "step": 4626
    },
    {
      "epoch": 0.6484933426769446,
      "grad_norm": 0.2752673029899597,
      "learning_rate": 0.00024084668739535994,
      "loss": 0.0922,
      "step": 4627
    },
    {
      "epoch": 0.6486334968465312,
      "grad_norm": 0.28419870138168335,
      "learning_rate": 0.00024083233676154027,
      "loss": 0.0317,
      "step": 4628
    },
    {
      "epoch": 0.6487736510161177,
      "grad_norm": 0.7177084684371948,
      "learning_rate": 0.00024081798612772063,
      "loss": 0.1128,
      "step": 4629
    },
    {
      "epoch": 0.6489138051857043,
      "grad_norm": 0.2612219750881195,
      "learning_rate": 0.00024080363549390096,
      "loss": 0.0531,
      "step": 4630
    },
    {
      "epoch": 0.6490539593552909,
      "grad_norm": 0.41845041513442993,
      "learning_rate": 0.00024078928486008128,
      "loss": 0.05,
      "step": 4631
    },
    {
      "epoch": 0.6491941135248773,
      "grad_norm": 0.3815973103046417,
      "learning_rate": 0.00024077493422626164,
      "loss": 0.094,
      "step": 4632
    },
    {
      "epoch": 0.6493342676944639,
      "grad_norm": 0.3043394684791565,
      "learning_rate": 0.00024076058359244197,
      "loss": 0.0848,
      "step": 4633
    },
    {
      "epoch": 0.6494744218640505,
      "grad_norm": 0.3352712094783783,
      "learning_rate": 0.0002407462329586223,
      "loss": 0.0247,
      "step": 4634
    },
    {
      "epoch": 0.649614576033637,
      "grad_norm": 0.7556617856025696,
      "learning_rate": 0.00024073188232480268,
      "loss": 0.1307,
      "step": 4635
    },
    {
      "epoch": 0.6497547302032235,
      "grad_norm": 0.6536099314689636,
      "learning_rate": 0.000240717531690983,
      "loss": 0.028,
      "step": 4636
    },
    {
      "epoch": 0.6498948843728101,
      "grad_norm": 0.49741828441619873,
      "learning_rate": 0.00024070318105716334,
      "loss": 0.0838,
      "step": 4637
    },
    {
      "epoch": 0.6500350385423966,
      "grad_norm": 0.8072445392608643,
      "learning_rate": 0.0002406888304233437,
      "loss": 0.0669,
      "step": 4638
    },
    {
      "epoch": 0.6501751927119832,
      "grad_norm": 0.6338628530502319,
      "learning_rate": 0.00024067447978952402,
      "loss": 0.1524,
      "step": 4639
    },
    {
      "epoch": 0.6503153468815698,
      "grad_norm": 0.24803027510643005,
      "learning_rate": 0.00024066012915570435,
      "loss": 0.076,
      "step": 4640
    },
    {
      "epoch": 0.6504555010511562,
      "grad_norm": 0.2887865900993347,
      "learning_rate": 0.00024064577852188468,
      "loss": 0.0583,
      "step": 4641
    },
    {
      "epoch": 0.6505956552207428,
      "grad_norm": 0.3580440580844879,
      "learning_rate": 0.00024063142788806503,
      "loss": 0.0844,
      "step": 4642
    },
    {
      "epoch": 0.6507358093903294,
      "grad_norm": 1.5213966369628906,
      "learning_rate": 0.00024061707725424536,
      "loss": 0.1455,
      "step": 4643
    },
    {
      "epoch": 0.650875963559916,
      "grad_norm": 0.40449780225753784,
      "learning_rate": 0.0002406027266204257,
      "loss": 0.0878,
      "step": 4644
    },
    {
      "epoch": 0.6510161177295024,
      "grad_norm": 0.5398775339126587,
      "learning_rate": 0.00024058837598660607,
      "loss": 0.0664,
      "step": 4645
    },
    {
      "epoch": 0.651156271899089,
      "grad_norm": 1.061957836151123,
      "learning_rate": 0.0002405740253527864,
      "loss": 0.1866,
      "step": 4646
    },
    {
      "epoch": 0.6512964260686755,
      "grad_norm": 1.5541675090789795,
      "learning_rate": 0.00024055967471896673,
      "loss": 0.2731,
      "step": 4647
    },
    {
      "epoch": 0.6514365802382621,
      "grad_norm": 1.4449336528778076,
      "learning_rate": 0.0002405453240851471,
      "loss": 0.2413,
      "step": 4648
    },
    {
      "epoch": 0.6515767344078486,
      "grad_norm": 1.1806188821792603,
      "learning_rate": 0.00024053097345132742,
      "loss": 0.088,
      "step": 4649
    },
    {
      "epoch": 0.6517168885774351,
      "grad_norm": 1.8111268281936646,
      "learning_rate": 0.00024051662281750774,
      "loss": 0.1807,
      "step": 4650
    },
    {
      "epoch": 0.6518570427470217,
      "grad_norm": 0.18268246948719025,
      "learning_rate": 0.0002405022721836881,
      "loss": 0.0353,
      "step": 4651
    },
    {
      "epoch": 0.6519971969166083,
      "grad_norm": 0.2123505026102066,
      "learning_rate": 0.00024048792154986843,
      "loss": 0.042,
      "step": 4652
    },
    {
      "epoch": 0.6521373510861949,
      "grad_norm": 0.32514774799346924,
      "learning_rate": 0.00024047357091604876,
      "loss": 0.0827,
      "step": 4653
    },
    {
      "epoch": 0.6522775052557813,
      "grad_norm": 0.18817712366580963,
      "learning_rate": 0.0002404592202822291,
      "loss": 0.0414,
      "step": 4654
    },
    {
      "epoch": 0.6524176594253679,
      "grad_norm": 0.4933124780654907,
      "learning_rate": 0.00024044486964840947,
      "loss": 0.1173,
      "step": 4655
    },
    {
      "epoch": 0.6525578135949545,
      "grad_norm": 0.5564436316490173,
      "learning_rate": 0.0002404305190145898,
      "loss": 0.0875,
      "step": 4656
    },
    {
      "epoch": 0.652697967764541,
      "grad_norm": 0.5919475555419922,
      "learning_rate": 0.00024041616838077013,
      "loss": 0.1728,
      "step": 4657
    },
    {
      "epoch": 0.6528381219341275,
      "grad_norm": 0.28961148858070374,
      "learning_rate": 0.00024040181774695048,
      "loss": 0.1137,
      "step": 4658
    },
    {
      "epoch": 0.6529782761037141,
      "grad_norm": 0.36180225014686584,
      "learning_rate": 0.0002403874671131308,
      "loss": 0.1185,
      "step": 4659
    },
    {
      "epoch": 0.6531184302733006,
      "grad_norm": 0.46975764632225037,
      "learning_rate": 0.00024037311647931114,
      "loss": 0.0932,
      "step": 4660
    },
    {
      "epoch": 0.6532585844428872,
      "grad_norm": 0.46723005175590515,
      "learning_rate": 0.0002403587658454915,
      "loss": 0.1139,
      "step": 4661
    },
    {
      "epoch": 0.6533987386124738,
      "grad_norm": 0.36402279138565063,
      "learning_rate": 0.00024034441521167182,
      "loss": 0.0828,
      "step": 4662
    },
    {
      "epoch": 0.6535388927820602,
      "grad_norm": 0.3752870559692383,
      "learning_rate": 0.00024033006457785215,
      "loss": 0.0681,
      "step": 4663
    },
    {
      "epoch": 0.6536790469516468,
      "grad_norm": 0.313428670167923,
      "learning_rate": 0.0002403157139440325,
      "loss": 0.0622,
      "step": 4664
    },
    {
      "epoch": 0.6538192011212334,
      "grad_norm": 0.5038762092590332,
      "learning_rate": 0.00024030136331021284,
      "loss": 0.0642,
      "step": 4665
    },
    {
      "epoch": 0.6539593552908199,
      "grad_norm": 0.26616787910461426,
      "learning_rate": 0.00024028701267639316,
      "loss": 0.0495,
      "step": 4666
    },
    {
      "epoch": 0.6540995094604064,
      "grad_norm": 0.30009329319000244,
      "learning_rate": 0.00024027266204257355,
      "loss": 0.04,
      "step": 4667
    },
    {
      "epoch": 0.654239663629993,
      "grad_norm": 0.4667293131351471,
      "learning_rate": 0.00024025831140875388,
      "loss": 0.1864,
      "step": 4668
    },
    {
      "epoch": 0.6543798177995795,
      "grad_norm": 0.48057129979133606,
      "learning_rate": 0.0002402439607749342,
      "loss": 0.1088,
      "step": 4669
    },
    {
      "epoch": 0.6545199719691661,
      "grad_norm": 0.5876733064651489,
      "learning_rate": 0.00024022961014111456,
      "loss": 0.0927,
      "step": 4670
    },
    {
      "epoch": 0.6546601261387526,
      "grad_norm": 0.6569058895111084,
      "learning_rate": 0.0002402152595072949,
      "loss": 0.043,
      "step": 4671
    },
    {
      "epoch": 0.6548002803083391,
      "grad_norm": 0.7138683795928955,
      "learning_rate": 0.00024020090887347522,
      "loss": 0.14,
      "step": 4672
    },
    {
      "epoch": 0.6549404344779257,
      "grad_norm": 0.5391191244125366,
      "learning_rate": 0.00024018655823965557,
      "loss": 0.1094,
      "step": 4673
    },
    {
      "epoch": 0.6550805886475123,
      "grad_norm": 0.3928048610687256,
      "learning_rate": 0.0002401722076058359,
      "loss": 0.0854,
      "step": 4674
    },
    {
      "epoch": 0.6552207428170989,
      "grad_norm": 0.605776309967041,
      "learning_rate": 0.00024015785697201623,
      "loss": 0.1096,
      "step": 4675
    },
    {
      "epoch": 0.6553608969866853,
      "grad_norm": 0.22983290255069733,
      "learning_rate": 0.00024014350633819656,
      "loss": 0.0677,
      "step": 4676
    },
    {
      "epoch": 0.6555010511562719,
      "grad_norm": 0.2627416253089905,
      "learning_rate": 0.00024012915570437694,
      "loss": 0.0863,
      "step": 4677
    },
    {
      "epoch": 0.6556412053258585,
      "grad_norm": 0.25304800271987915,
      "learning_rate": 0.00024011480507055727,
      "loss": 0.0625,
      "step": 4678
    },
    {
      "epoch": 0.655781359495445,
      "grad_norm": 0.7146767377853394,
      "learning_rate": 0.0002401004544367376,
      "loss": 0.1076,
      "step": 4679
    },
    {
      "epoch": 0.6559215136650315,
      "grad_norm": 0.8054242134094238,
      "learning_rate": 0.00024008610380291795,
      "loss": 0.1975,
      "step": 4680
    },
    {
      "epoch": 0.6560616678346181,
      "grad_norm": 0.3236759305000305,
      "learning_rate": 0.00024007175316909828,
      "loss": 0.0466,
      "step": 4681
    },
    {
      "epoch": 0.6562018220042046,
      "grad_norm": 0.28964611887931824,
      "learning_rate": 0.0002400574025352786,
      "loss": 0.0635,
      "step": 4682
    },
    {
      "epoch": 0.6563419761737912,
      "grad_norm": 0.48297131061553955,
      "learning_rate": 0.00024004305190145897,
      "loss": 0.0837,
      "step": 4683
    },
    {
      "epoch": 0.6564821303433778,
      "grad_norm": 0.16519542038440704,
      "learning_rate": 0.0002400287012676393,
      "loss": 0.0458,
      "step": 4684
    },
    {
      "epoch": 0.6566222845129642,
      "grad_norm": 0.4170892536640167,
      "learning_rate": 0.00024001435063381962,
      "loss": 0.0595,
      "step": 4685
    },
    {
      "epoch": 0.6567624386825508,
      "grad_norm": 0.31840410828590393,
      "learning_rate": 0.00023999999999999998,
      "loss": 0.0551,
      "step": 4686
    },
    {
      "epoch": 0.6569025928521374,
      "grad_norm": 0.4003826677799225,
      "learning_rate": 0.00023998564936618033,
      "loss": 0.0468,
      "step": 4687
    },
    {
      "epoch": 0.6570427470217239,
      "grad_norm": 0.5897322297096252,
      "learning_rate": 0.00023997129873236066,
      "loss": 0.1636,
      "step": 4688
    },
    {
      "epoch": 0.6571829011913104,
      "grad_norm": 0.3908691704273224,
      "learning_rate": 0.00023995694809854102,
      "loss": 0.0128,
      "step": 4689
    },
    {
      "epoch": 0.657323055360897,
      "grad_norm": 0.3724898397922516,
      "learning_rate": 0.00023994259746472135,
      "loss": 0.0766,
      "step": 4690
    },
    {
      "epoch": 0.6574632095304835,
      "grad_norm": 0.40407121181488037,
      "learning_rate": 0.00023992824683090168,
      "loss": 0.0506,
      "step": 4691
    },
    {
      "epoch": 0.6576033637000701,
      "grad_norm": 0.3172832131385803,
      "learning_rate": 0.000239913896197082,
      "loss": 0.0763,
      "step": 4692
    },
    {
      "epoch": 0.6577435178696566,
      "grad_norm": 0.32272204756736755,
      "learning_rate": 0.00023989954556326236,
      "loss": 0.043,
      "step": 4693
    },
    {
      "epoch": 0.6578836720392431,
      "grad_norm": 0.6650473475456238,
      "learning_rate": 0.0002398851949294427,
      "loss": 0.1666,
      "step": 4694
    },
    {
      "epoch": 0.6580238262088297,
      "grad_norm": 0.5213014483451843,
      "learning_rate": 0.00023987084429562302,
      "loss": 0.1701,
      "step": 4695
    },
    {
      "epoch": 0.6581639803784163,
      "grad_norm": 0.49516984820365906,
      "learning_rate": 0.00023985649366180337,
      "loss": 0.1059,
      "step": 4696
    },
    {
      "epoch": 0.6583041345480029,
      "grad_norm": 0.4641109108924866,
      "learning_rate": 0.0002398421430279837,
      "loss": 0.0857,
      "step": 4697
    },
    {
      "epoch": 0.6584442887175893,
      "grad_norm": 4.261027812957764,
      "learning_rate": 0.00023982779239416403,
      "loss": 0.1845,
      "step": 4698
    },
    {
      "epoch": 0.6585844428871759,
      "grad_norm": 0.8818010091781616,
      "learning_rate": 0.0002398134417603444,
      "loss": 0.1359,
      "step": 4699
    },
    {
      "epoch": 0.6587245970567625,
      "grad_norm": 2.034306049346924,
      "learning_rate": 0.00023979909112652474,
      "loss": 0.1046,
      "step": 4700
    },
    {
      "epoch": 0.658864751226349,
      "grad_norm": 0.5487805604934692,
      "learning_rate": 0.00023978474049270507,
      "loss": 0.1404,
      "step": 4701
    },
    {
      "epoch": 0.6590049053959355,
      "grad_norm": 0.8866620659828186,
      "learning_rate": 0.00023977038985888543,
      "loss": 0.1678,
      "step": 4702
    },
    {
      "epoch": 0.659145059565522,
      "grad_norm": 0.3447425365447998,
      "learning_rate": 0.00023975603922506575,
      "loss": 0.0572,
      "step": 4703
    },
    {
      "epoch": 0.6592852137351086,
      "grad_norm": 0.31762513518333435,
      "learning_rate": 0.00023974168859124608,
      "loss": 0.0557,
      "step": 4704
    },
    {
      "epoch": 0.6594253679046952,
      "grad_norm": 0.6727063059806824,
      "learning_rate": 0.00023972733795742644,
      "loss": 0.1323,
      "step": 4705
    },
    {
      "epoch": 0.6595655220742818,
      "grad_norm": 0.323579877614975,
      "learning_rate": 0.00023971298732360677,
      "loss": 0.0524,
      "step": 4706
    },
    {
      "epoch": 0.6597056762438682,
      "grad_norm": 0.3233967423439026,
      "learning_rate": 0.0002396986366897871,
      "loss": 0.0605,
      "step": 4707
    },
    {
      "epoch": 0.6598458304134548,
      "grad_norm": 0.2196207493543625,
      "learning_rate": 0.00023968428605596748,
      "loss": 0.0484,
      "step": 4708
    },
    {
      "epoch": 0.6599859845830414,
      "grad_norm": 0.2329094111919403,
      "learning_rate": 0.0002396699354221478,
      "loss": 0.085,
      "step": 4709
    },
    {
      "epoch": 0.6601261387526279,
      "grad_norm": 0.43465256690979004,
      "learning_rate": 0.00023965558478832814,
      "loss": 0.1119,
      "step": 4710
    },
    {
      "epoch": 0.6602662929222144,
      "grad_norm": 0.2841493785381317,
      "learning_rate": 0.00023964123415450846,
      "loss": 0.0525,
      "step": 4711
    },
    {
      "epoch": 0.660406447091801,
      "grad_norm": 0.47998476028442383,
      "learning_rate": 0.00023962688352068882,
      "loss": 0.0859,
      "step": 4712
    },
    {
      "epoch": 0.6605466012613875,
      "grad_norm": 0.399516761302948,
      "learning_rate": 0.00023961253288686915,
      "loss": 0.0758,
      "step": 4713
    },
    {
      "epoch": 0.6606867554309741,
      "grad_norm": 0.33252906799316406,
      "learning_rate": 0.00023959818225304948,
      "loss": 0.0659,
      "step": 4714
    },
    {
      "epoch": 0.6608269096005606,
      "grad_norm": 0.2603592276573181,
      "learning_rate": 0.00023958383161922983,
      "loss": 0.0851,
      "step": 4715
    },
    {
      "epoch": 0.6609670637701471,
      "grad_norm": 0.23905311524868011,
      "learning_rate": 0.00023956948098541016,
      "loss": 0.0539,
      "step": 4716
    },
    {
      "epoch": 0.6611072179397337,
      "grad_norm": 0.5367918014526367,
      "learning_rate": 0.0002395551303515905,
      "loss": 0.0949,
      "step": 4717
    },
    {
      "epoch": 0.6612473721093203,
      "grad_norm": 0.5777433514595032,
      "learning_rate": 0.00023954077971777085,
      "loss": 0.1243,
      "step": 4718
    },
    {
      "epoch": 0.6613875262789068,
      "grad_norm": 0.5530431866645813,
      "learning_rate": 0.0002395264290839512,
      "loss": 0.068,
      "step": 4719
    },
    {
      "epoch": 0.6615276804484933,
      "grad_norm": 0.33826640248298645,
      "learning_rate": 0.00023951207845013153,
      "loss": 0.0496,
      "step": 4720
    },
    {
      "epoch": 0.6616678346180799,
      "grad_norm": 0.432858943939209,
      "learning_rate": 0.00023949772781631189,
      "loss": 0.1,
      "step": 4721
    },
    {
      "epoch": 0.6618079887876664,
      "grad_norm": 0.34647655487060547,
      "learning_rate": 0.00023948337718249221,
      "loss": 0.0985,
      "step": 4722
    },
    {
      "epoch": 0.661948142957253,
      "grad_norm": 0.48452380299568176,
      "learning_rate": 0.00023946902654867254,
      "loss": 0.1269,
      "step": 4723
    },
    {
      "epoch": 0.6620882971268395,
      "grad_norm": 0.2846594452857971,
      "learning_rate": 0.0002394546759148529,
      "loss": 0.0564,
      "step": 4724
    },
    {
      "epoch": 0.662228451296426,
      "grad_norm": 0.41883227229118347,
      "learning_rate": 0.00023944032528103323,
      "loss": 0.0809,
      "step": 4725
    },
    {
      "epoch": 0.6623686054660126,
      "grad_norm": 0.5125524997711182,
      "learning_rate": 0.00023942597464721356,
      "loss": 0.0947,
      "step": 4726
    },
    {
      "epoch": 0.6625087596355992,
      "grad_norm": 0.6130154132843018,
      "learning_rate": 0.00023941162401339388,
      "loss": 0.0821,
      "step": 4727
    },
    {
      "epoch": 0.6626489138051858,
      "grad_norm": 0.7663873434066772,
      "learning_rate": 0.00023939727337957424,
      "loss": 0.1176,
      "step": 4728
    },
    {
      "epoch": 0.6627890679747722,
      "grad_norm": 0.8150898814201355,
      "learning_rate": 0.00023938292274575457,
      "loss": 0.0406,
      "step": 4729
    },
    {
      "epoch": 0.6629292221443588,
      "grad_norm": 0.34998074173927307,
      "learning_rate": 0.0002393685721119349,
      "loss": 0.0529,
      "step": 4730
    },
    {
      "epoch": 0.6630693763139454,
      "grad_norm": 0.28779566287994385,
      "learning_rate": 0.00023935422147811528,
      "loss": 0.0688,
      "step": 4731
    },
    {
      "epoch": 0.6632095304835319,
      "grad_norm": 0.5959911942481995,
      "learning_rate": 0.0002393398708442956,
      "loss": 0.0879,
      "step": 4732
    },
    {
      "epoch": 0.6633496846531184,
      "grad_norm": 0.47114622592926025,
      "learning_rate": 0.00023932552021047594,
      "loss": 0.0975,
      "step": 4733
    },
    {
      "epoch": 0.663489838822705,
      "grad_norm": 0.45995211601257324,
      "learning_rate": 0.0002393111695766563,
      "loss": 0.1008,
      "step": 4734
    },
    {
      "epoch": 0.6636299929922915,
      "grad_norm": 0.604130744934082,
      "learning_rate": 0.00023929681894283662,
      "loss": 0.108,
      "step": 4735
    },
    {
      "epoch": 0.6637701471618781,
      "grad_norm": 0.5945006012916565,
      "learning_rate": 0.00023928246830901695,
      "loss": 0.1285,
      "step": 4736
    },
    {
      "epoch": 0.6639103013314646,
      "grad_norm": 0.43148139119148254,
      "learning_rate": 0.0002392681176751973,
      "loss": 0.1201,
      "step": 4737
    },
    {
      "epoch": 0.6640504555010511,
      "grad_norm": 0.4388241767883301,
      "learning_rate": 0.00023925376704137763,
      "loss": 0.0963,
      "step": 4738
    },
    {
      "epoch": 0.6641906096706377,
      "grad_norm": 0.7865726351737976,
      "learning_rate": 0.00023923941640755796,
      "loss": 0.0858,
      "step": 4739
    },
    {
      "epoch": 0.6643307638402243,
      "grad_norm": 0.3987134099006653,
      "learning_rate": 0.00023922506577373834,
      "loss": 0.1214,
      "step": 4740
    },
    {
      "epoch": 0.6644709180098108,
      "grad_norm": 0.3827250897884369,
      "learning_rate": 0.00023921071513991867,
      "loss": 0.06,
      "step": 4741
    },
    {
      "epoch": 0.6646110721793973,
      "grad_norm": 0.554088830947876,
      "learning_rate": 0.000239196364506099,
      "loss": 0.1411,
      "step": 4742
    },
    {
      "epoch": 0.6647512263489839,
      "grad_norm": 0.623368501663208,
      "learning_rate": 0.00023918201387227936,
      "loss": 0.1916,
      "step": 4743
    },
    {
      "epoch": 0.6648913805185704,
      "grad_norm": 0.5339826941490173,
      "learning_rate": 0.00023916766323845969,
      "loss": 0.0997,
      "step": 4744
    },
    {
      "epoch": 0.665031534688157,
      "grad_norm": 1.4671242237091064,
      "learning_rate": 0.00023915331260464001,
      "loss": 0.1852,
      "step": 4745
    },
    {
      "epoch": 0.6651716888577435,
      "grad_norm": 1.4832878112792969,
      "learning_rate": 0.00023913896197082034,
      "loss": 0.1548,
      "step": 4746
    },
    {
      "epoch": 0.66531184302733,
      "grad_norm": 0.9231269955635071,
      "learning_rate": 0.0002391246113370007,
      "loss": 0.1329,
      "step": 4747
    },
    {
      "epoch": 0.6654519971969166,
      "grad_norm": 0.6684432029724121,
      "learning_rate": 0.00023911026070318103,
      "loss": 0.1273,
      "step": 4748
    },
    {
      "epoch": 0.6655921513665032,
      "grad_norm": 0.8273610472679138,
      "learning_rate": 0.00023909591006936136,
      "loss": 0.1433,
      "step": 4749
    },
    {
      "epoch": 0.6657323055360898,
      "grad_norm": 0.9958446621894836,
      "learning_rate": 0.00023908155943554174,
      "loss": 0.2772,
      "step": 4750
    },
    {
      "epoch": 0.6658724597056762,
      "grad_norm": 0.2235517054796219,
      "learning_rate": 0.00023906720880172207,
      "loss": 0.0532,
      "step": 4751
    },
    {
      "epoch": 0.6660126138752628,
      "grad_norm": 0.40702909231185913,
      "learning_rate": 0.0002390528581679024,
      "loss": 0.1269,
      "step": 4752
    },
    {
      "epoch": 0.6661527680448494,
      "grad_norm": 0.25221917033195496,
      "learning_rate": 0.00023903850753408275,
      "loss": 0.0645,
      "step": 4753
    },
    {
      "epoch": 0.6662929222144359,
      "grad_norm": 0.36963382363319397,
      "learning_rate": 0.00023902415690026308,
      "loss": 0.0915,
      "step": 4754
    },
    {
      "epoch": 0.6664330763840224,
      "grad_norm": 0.3213574290275574,
      "learning_rate": 0.0002390098062664434,
      "loss": 0.1093,
      "step": 4755
    },
    {
      "epoch": 0.666573230553609,
      "grad_norm": 0.6365625858306885,
      "learning_rate": 0.00023899545563262376,
      "loss": 0.1657,
      "step": 4756
    },
    {
      "epoch": 0.6667133847231955,
      "grad_norm": 0.2171628326177597,
      "learning_rate": 0.0002389811049988041,
      "loss": 0.0622,
      "step": 4757
    },
    {
      "epoch": 0.6668535388927821,
      "grad_norm": 0.6924606561660767,
      "learning_rate": 0.00023896675436498442,
      "loss": 0.0916,
      "step": 4758
    },
    {
      "epoch": 0.6669936930623686,
      "grad_norm": 0.3006793260574341,
      "learning_rate": 0.00023895240373116478,
      "loss": 0.0835,
      "step": 4759
    },
    {
      "epoch": 0.6671338472319551,
      "grad_norm": 0.2549748420715332,
      "learning_rate": 0.0002389380530973451,
      "loss": 0.0573,
      "step": 4760
    },
    {
      "epoch": 0.6672740014015417,
      "grad_norm": 0.35190248489379883,
      "learning_rate": 0.00023892370246352543,
      "loss": 0.0804,
      "step": 4761
    },
    {
      "epoch": 0.6674141555711283,
      "grad_norm": 0.39118215441703796,
      "learning_rate": 0.00023890935182970576,
      "loss": 0.0443,
      "step": 4762
    },
    {
      "epoch": 0.6675543097407148,
      "grad_norm": 0.4683927595615387,
      "learning_rate": 0.00023889500119588615,
      "loss": 0.0437,
      "step": 4763
    },
    {
      "epoch": 0.6676944639103013,
      "grad_norm": 0.7677825689315796,
      "learning_rate": 0.00023888065056206647,
      "loss": 0.1322,
      "step": 4764
    },
    {
      "epoch": 0.6678346180798879,
      "grad_norm": 0.40390464663505554,
      "learning_rate": 0.0002388662999282468,
      "loss": 0.0959,
      "step": 4765
    },
    {
      "epoch": 0.6679747722494744,
      "grad_norm": 0.6339204907417297,
      "learning_rate": 0.00023885194929442716,
      "loss": 0.0531,
      "step": 4766
    },
    {
      "epoch": 0.668114926419061,
      "grad_norm": 0.5264211893081665,
      "learning_rate": 0.0002388375986606075,
      "loss": 0.1239,
      "step": 4767
    },
    {
      "epoch": 0.6682550805886475,
      "grad_norm": 0.4180755615234375,
      "learning_rate": 0.00023882324802678782,
      "loss": 0.0635,
      "step": 4768
    },
    {
      "epoch": 0.668395234758234,
      "grad_norm": 1.2219139337539673,
      "learning_rate": 0.00023880889739296817,
      "loss": 0.1783,
      "step": 4769
    },
    {
      "epoch": 0.6685353889278206,
      "grad_norm": 0.42783322930336,
      "learning_rate": 0.0002387945467591485,
      "loss": 0.1009,
      "step": 4770
    },
    {
      "epoch": 0.6686755430974072,
      "grad_norm": 0.3901127278804779,
      "learning_rate": 0.00023878019612532883,
      "loss": 0.0391,
      "step": 4771
    },
    {
      "epoch": 0.6688156972669937,
      "grad_norm": 0.30918020009994507,
      "learning_rate": 0.0002387658454915092,
      "loss": 0.1143,
      "step": 4772
    },
    {
      "epoch": 0.6689558514365802,
      "grad_norm": 0.4193136394023895,
      "learning_rate": 0.00023875149485768954,
      "loss": 0.0811,
      "step": 4773
    },
    {
      "epoch": 0.6690960056061668,
      "grad_norm": 0.2928798496723175,
      "learning_rate": 0.00023873714422386987,
      "loss": 0.0325,
      "step": 4774
    },
    {
      "epoch": 0.6692361597757533,
      "grad_norm": 0.5497828125953674,
      "learning_rate": 0.00023872279359005022,
      "loss": 0.0795,
      "step": 4775
    },
    {
      "epoch": 0.6693763139453399,
      "grad_norm": 0.3676680326461792,
      "learning_rate": 0.00023870844295623055,
      "loss": 0.0625,
      "step": 4776
    },
    {
      "epoch": 0.6695164681149264,
      "grad_norm": 0.31779536604881287,
      "learning_rate": 0.00023869409232241088,
      "loss": 0.0331,
      "step": 4777
    },
    {
      "epoch": 0.669656622284513,
      "grad_norm": 1.0345954895019531,
      "learning_rate": 0.00023867974168859124,
      "loss": 0.1395,
      "step": 4778
    },
    {
      "epoch": 0.6697967764540995,
      "grad_norm": 0.6910383701324463,
      "learning_rate": 0.00023866539105477157,
      "loss": 0.0505,
      "step": 4779
    },
    {
      "epoch": 0.6699369306236861,
      "grad_norm": 0.38107091188430786,
      "learning_rate": 0.0002386510404209519,
      "loss": 0.086,
      "step": 4780
    },
    {
      "epoch": 0.6700770847932725,
      "grad_norm": 0.4565012454986572,
      "learning_rate": 0.00023863668978713222,
      "loss": 0.1188,
      "step": 4781
    },
    {
      "epoch": 0.6702172389628591,
      "grad_norm": 0.40541872382164,
      "learning_rate": 0.0002386223391533126,
      "loss": 0.1722,
      "step": 4782
    },
    {
      "epoch": 0.6703573931324457,
      "grad_norm": 0.42854994535446167,
      "learning_rate": 0.00023860798851949293,
      "loss": 0.1306,
      "step": 4783
    },
    {
      "epoch": 0.6704975473020323,
      "grad_norm": 0.9302937984466553,
      "learning_rate": 0.00023859363788567326,
      "loss": 0.1103,
      "step": 4784
    },
    {
      "epoch": 0.6706377014716188,
      "grad_norm": 0.23054715991020203,
      "learning_rate": 0.00023857928725185362,
      "loss": 0.0615,
      "step": 4785
    },
    {
      "epoch": 0.6707778556412053,
      "grad_norm": 0.56084805727005,
      "learning_rate": 0.00023856493661803395,
      "loss": 0.1594,
      "step": 4786
    },
    {
      "epoch": 0.6709180098107919,
      "grad_norm": 0.5259038209915161,
      "learning_rate": 0.00023855058598421427,
      "loss": 0.1421,
      "step": 4787
    },
    {
      "epoch": 0.6710581639803784,
      "grad_norm": 0.22728697955608368,
      "learning_rate": 0.00023853623535039463,
      "loss": 0.0463,
      "step": 4788
    },
    {
      "epoch": 0.671198318149965,
      "grad_norm": 0.5576725006103516,
      "learning_rate": 0.00023852188471657496,
      "loss": 0.1141,
      "step": 4789
    },
    {
      "epoch": 0.6713384723195515,
      "grad_norm": 0.3839956223964691,
      "learning_rate": 0.0002385075340827553,
      "loss": 0.0439,
      "step": 4790
    },
    {
      "epoch": 0.671478626489138,
      "grad_norm": 1.3236631155014038,
      "learning_rate": 0.00023849318344893564,
      "loss": 0.1423,
      "step": 4791
    },
    {
      "epoch": 0.6716187806587246,
      "grad_norm": 1.4242830276489258,
      "learning_rate": 0.00023847883281511597,
      "loss": 0.1033,
      "step": 4792
    },
    {
      "epoch": 0.6717589348283112,
      "grad_norm": 0.7619696855545044,
      "learning_rate": 0.0002384644821812963,
      "loss": 0.0761,
      "step": 4793
    },
    {
      "epoch": 0.6718990889978977,
      "grad_norm": 1.0667437314987183,
      "learning_rate": 0.00023845013154747668,
      "loss": 0.3012,
      "step": 4794
    },
    {
      "epoch": 0.6720392431674842,
      "grad_norm": 0.4316199719905853,
      "learning_rate": 0.000238435780913657,
      "loss": 0.1149,
      "step": 4795
    },
    {
      "epoch": 0.6721793973370708,
      "grad_norm": 0.636736273765564,
      "learning_rate": 0.00023842143027983734,
      "loss": 0.0896,
      "step": 4796
    },
    {
      "epoch": 0.6723195515066573,
      "grad_norm": 1.3274928331375122,
      "learning_rate": 0.00023840707964601767,
      "loss": 0.0921,
      "step": 4797
    },
    {
      "epoch": 0.6724597056762439,
      "grad_norm": 1.606554388999939,
      "learning_rate": 0.00023839272901219802,
      "loss": 0.153,
      "step": 4798
    },
    {
      "epoch": 0.6725998598458304,
      "grad_norm": 0.6033066511154175,
      "learning_rate": 0.00023837837837837835,
      "loss": 0.0965,
      "step": 4799
    },
    {
      "epoch": 0.6727400140154169,
      "grad_norm": 0.1443236917257309,
      "learning_rate": 0.00023836402774455868,
      "loss": 0.0135,
      "step": 4800
    },
    {
      "epoch": 0.6728801681850035,
      "grad_norm": 0.2583099901676178,
      "learning_rate": 0.00023834967711073904,
      "loss": 0.0416,
      "step": 4801
    },
    {
      "epoch": 0.6730203223545901,
      "grad_norm": 0.43310612440109253,
      "learning_rate": 0.00023833532647691937,
      "loss": 0.0387,
      "step": 4802
    },
    {
      "epoch": 0.6731604765241765,
      "grad_norm": 0.39550283551216125,
      "learning_rate": 0.0002383209758430997,
      "loss": 0.0933,
      "step": 4803
    },
    {
      "epoch": 0.6733006306937631,
      "grad_norm": 0.8012619614601135,
      "learning_rate": 0.00023830662520928008,
      "loss": 0.143,
      "step": 4804
    },
    {
      "epoch": 0.6734407848633497,
      "grad_norm": 0.31968745589256287,
      "learning_rate": 0.0002382922745754604,
      "loss": 0.0795,
      "step": 4805
    },
    {
      "epoch": 0.6735809390329363,
      "grad_norm": 0.3046531677246094,
      "learning_rate": 0.00023827792394164073,
      "loss": 0.0901,
      "step": 4806
    },
    {
      "epoch": 0.6737210932025228,
      "grad_norm": 0.629946768283844,
      "learning_rate": 0.0002382635733078211,
      "loss": 0.0539,
      "step": 4807
    },
    {
      "epoch": 0.6738612473721093,
      "grad_norm": 0.23978684842586517,
      "learning_rate": 0.00023824922267400142,
      "loss": 0.0628,
      "step": 4808
    },
    {
      "epoch": 0.6740014015416959,
      "grad_norm": 0.36997658014297485,
      "learning_rate": 0.00023823487204018175,
      "loss": 0.1285,
      "step": 4809
    },
    {
      "epoch": 0.6741415557112824,
      "grad_norm": 0.21985359489917755,
      "learning_rate": 0.0002382205214063621,
      "loss": 0.0638,
      "step": 4810
    },
    {
      "epoch": 0.674281709880869,
      "grad_norm": 0.305414617061615,
      "learning_rate": 0.00023820617077254243,
      "loss": 0.081,
      "step": 4811
    },
    {
      "epoch": 0.6744218640504555,
      "grad_norm": 0.4753924608230591,
      "learning_rate": 0.00023819182013872276,
      "loss": 0.0756,
      "step": 4812
    },
    {
      "epoch": 0.674562018220042,
      "grad_norm": 0.37053221464157104,
      "learning_rate": 0.00023817746950490312,
      "loss": 0.0663,
      "step": 4813
    },
    {
      "epoch": 0.6747021723896286,
      "grad_norm": 0.5205087065696716,
      "learning_rate": 0.00023816311887108347,
      "loss": 0.0875,
      "step": 4814
    },
    {
      "epoch": 0.6748423265592152,
      "grad_norm": 0.3467317223548889,
      "learning_rate": 0.0002381487682372638,
      "loss": 0.0612,
      "step": 4815
    },
    {
      "epoch": 0.6749824807288017,
      "grad_norm": 0.20218747854232788,
      "learning_rate": 0.00023813441760344413,
      "loss": 0.0481,
      "step": 4816
    },
    {
      "epoch": 0.6751226348983882,
      "grad_norm": 0.14742673933506012,
      "learning_rate": 0.00023812006696962448,
      "loss": 0.0291,
      "step": 4817
    },
    {
      "epoch": 0.6752627890679748,
      "grad_norm": 0.5972278714179993,
      "learning_rate": 0.0002381057163358048,
      "loss": 0.0621,
      "step": 4818
    },
    {
      "epoch": 0.6754029432375613,
      "grad_norm": 0.43682608008384705,
      "learning_rate": 0.00023809136570198514,
      "loss": 0.0953,
      "step": 4819
    },
    {
      "epoch": 0.6755430974071479,
      "grad_norm": 1.4963914155960083,
      "learning_rate": 0.0002380770150681655,
      "loss": 0.2019,
      "step": 4820
    },
    {
      "epoch": 0.6756832515767344,
      "grad_norm": 0.315142959356308,
      "learning_rate": 0.00023806266443434583,
      "loss": 0.0256,
      "step": 4821
    },
    {
      "epoch": 0.6758234057463209,
      "grad_norm": 0.48459675908088684,
      "learning_rate": 0.00023804831380052615,
      "loss": 0.0789,
      "step": 4822
    },
    {
      "epoch": 0.6759635599159075,
      "grad_norm": 1.2829340696334839,
      "learning_rate": 0.0002380339631667065,
      "loss": 0.1706,
      "step": 4823
    },
    {
      "epoch": 0.6761037140854941,
      "grad_norm": 0.4187527000904083,
      "learning_rate": 0.00023801961253288684,
      "loss": 0.0577,
      "step": 4824
    },
    {
      "epoch": 0.6762438682550805,
      "grad_norm": 0.5584620237350464,
      "learning_rate": 0.00023800526189906717,
      "loss": 0.1846,
      "step": 4825
    },
    {
      "epoch": 0.6763840224246671,
      "grad_norm": 0.26539307832717896,
      "learning_rate": 0.00023799091126524755,
      "loss": 0.017,
      "step": 4826
    },
    {
      "epoch": 0.6765241765942537,
      "grad_norm": 0.3480900228023529,
      "learning_rate": 0.00023797656063142788,
      "loss": 0.0231,
      "step": 4827
    },
    {
      "epoch": 0.6766643307638402,
      "grad_norm": 0.6160892844200134,
      "learning_rate": 0.0002379622099976082,
      "loss": 0.1144,
      "step": 4828
    },
    {
      "epoch": 0.6768044849334268,
      "grad_norm": 0.10382203757762909,
      "learning_rate": 0.00023794785936378856,
      "loss": 0.02,
      "step": 4829
    },
    {
      "epoch": 0.6769446391030133,
      "grad_norm": 0.3970741331577301,
      "learning_rate": 0.0002379335087299689,
      "loss": 0.0416,
      "step": 4830
    },
    {
      "epoch": 0.6770847932725998,
      "grad_norm": 0.33715978264808655,
      "learning_rate": 0.00023791915809614922,
      "loss": 0.0909,
      "step": 4831
    },
    {
      "epoch": 0.6772249474421864,
      "grad_norm": 0.292333722114563,
      "learning_rate": 0.00023790480746232955,
      "loss": 0.0417,
      "step": 4832
    },
    {
      "epoch": 0.677365101611773,
      "grad_norm": 0.33026257157325745,
      "learning_rate": 0.0002378904568285099,
      "loss": 0.078,
      "step": 4833
    },
    {
      "epoch": 0.6775052557813595,
      "grad_norm": 0.3665729761123657,
      "learning_rate": 0.00023787610619469023,
      "loss": 0.0648,
      "step": 4834
    },
    {
      "epoch": 0.677645409950946,
      "grad_norm": 0.40561768412590027,
      "learning_rate": 0.00023786175556087056,
      "loss": 0.0338,
      "step": 4835
    },
    {
      "epoch": 0.6777855641205326,
      "grad_norm": 0.6484575867652893,
      "learning_rate": 0.00023784740492705094,
      "loss": 0.0898,
      "step": 4836
    },
    {
      "epoch": 0.6779257182901192,
      "grad_norm": 0.4605390429496765,
      "learning_rate": 0.00023783305429323127,
      "loss": 0.0819,
      "step": 4837
    },
    {
      "epoch": 0.6780658724597056,
      "grad_norm": 0.5616440176963806,
      "learning_rate": 0.0002378187036594116,
      "loss": 0.086,
      "step": 4838
    },
    {
      "epoch": 0.6782060266292922,
      "grad_norm": 0.44194066524505615,
      "learning_rate": 0.00023780435302559196,
      "loss": 0.0987,
      "step": 4839
    },
    {
      "epoch": 0.6783461807988788,
      "grad_norm": 0.7424604892730713,
      "learning_rate": 0.00023779000239177228,
      "loss": 0.1066,
      "step": 4840
    },
    {
      "epoch": 0.6784863349684653,
      "grad_norm": 0.4017472565174103,
      "learning_rate": 0.0002377756517579526,
      "loss": 0.0896,
      "step": 4841
    },
    {
      "epoch": 0.6786264891380519,
      "grad_norm": 0.2393740564584732,
      "learning_rate": 0.00023776130112413297,
      "loss": 0.0822,
      "step": 4842
    },
    {
      "epoch": 0.6787666433076384,
      "grad_norm": 0.26493439078330994,
      "learning_rate": 0.0002377469504903133,
      "loss": 0.045,
      "step": 4843
    },
    {
      "epoch": 0.6789067974772249,
      "grad_norm": 0.82822185754776,
      "learning_rate": 0.00023773259985649363,
      "loss": 0.0863,
      "step": 4844
    },
    {
      "epoch": 0.6790469516468115,
      "grad_norm": 0.40902775526046753,
      "learning_rate": 0.000237718249222674,
      "loss": 0.0507,
      "step": 4845
    },
    {
      "epoch": 0.6791871058163981,
      "grad_norm": 2.1106908321380615,
      "learning_rate": 0.00023770389858885434,
      "loss": 0.2549,
      "step": 4846
    },
    {
      "epoch": 0.6793272599859845,
      "grad_norm": 0.7432109117507935,
      "learning_rate": 0.00023768954795503467,
      "loss": 0.0946,
      "step": 4847
    },
    {
      "epoch": 0.6794674141555711,
      "grad_norm": 0.15034587681293488,
      "learning_rate": 0.000237675197321215,
      "loss": 0.0112,
      "step": 4848
    },
    {
      "epoch": 0.6796075683251577,
      "grad_norm": 0.6746618151664734,
      "learning_rate": 0.00023766084668739535,
      "loss": 0.0554,
      "step": 4849
    },
    {
      "epoch": 0.6797477224947442,
      "grad_norm": 1.48973548412323,
      "learning_rate": 0.00023764649605357568,
      "loss": 0.2704,
      "step": 4850
    },
    {
      "epoch": 0.6798878766643308,
      "grad_norm": 0.29558584094047546,
      "learning_rate": 0.000237632145419756,
      "loss": 0.047,
      "step": 4851
    },
    {
      "epoch": 0.6800280308339173,
      "grad_norm": 0.29801949858665466,
      "learning_rate": 0.00023761779478593636,
      "loss": 0.0794,
      "step": 4852
    },
    {
      "epoch": 0.6801681850035038,
      "grad_norm": 0.33727243542671204,
      "learning_rate": 0.0002376034441521167,
      "loss": 0.0582,
      "step": 4853
    },
    {
      "epoch": 0.6803083391730904,
      "grad_norm": 0.358218252658844,
      "learning_rate": 0.00023758909351829702,
      "loss": 0.0495,
      "step": 4854
    },
    {
      "epoch": 0.680448493342677,
      "grad_norm": 0.7154395580291748,
      "learning_rate": 0.00023757474288447738,
      "loss": 0.0896,
      "step": 4855
    },
    {
      "epoch": 0.6805886475122634,
      "grad_norm": 0.42337000370025635,
      "learning_rate": 0.0002375603922506577,
      "loss": 0.0421,
      "step": 4856
    },
    {
      "epoch": 0.68072880168185,
      "grad_norm": 0.40004295110702515,
      "learning_rate": 0.00023754604161683803,
      "loss": 0.1212,
      "step": 4857
    },
    {
      "epoch": 0.6808689558514366,
      "grad_norm": 0.6850024461746216,
      "learning_rate": 0.00023753169098301842,
      "loss": 0.0956,
      "step": 4858
    },
    {
      "epoch": 0.6810091100210232,
      "grad_norm": 0.5558722019195557,
      "learning_rate": 0.00023751734034919874,
      "loss": 0.0788,
      "step": 4859
    },
    {
      "epoch": 0.6811492641906096,
      "grad_norm": 0.2649074196815491,
      "learning_rate": 0.00023750298971537907,
      "loss": 0.0384,
      "step": 4860
    },
    {
      "epoch": 0.6812894183601962,
      "grad_norm": 0.3291005790233612,
      "learning_rate": 0.00023748863908155943,
      "loss": 0.0307,
      "step": 4861
    },
    {
      "epoch": 0.6814295725297828,
      "grad_norm": 0.33707088232040405,
      "learning_rate": 0.00023747428844773976,
      "loss": 0.046,
      "step": 4862
    },
    {
      "epoch": 0.6815697266993693,
      "grad_norm": 0.5303124189376831,
      "learning_rate": 0.00023745993781392009,
      "loss": 0.094,
      "step": 4863
    },
    {
      "epoch": 0.6817098808689559,
      "grad_norm": 0.6422899961471558,
      "learning_rate": 0.00023744558718010044,
      "loss": 0.1121,
      "step": 4864
    },
    {
      "epoch": 0.6818500350385424,
      "grad_norm": 0.6890474557876587,
      "learning_rate": 0.00023743123654628077,
      "loss": 0.1075,
      "step": 4865
    },
    {
      "epoch": 0.6819901892081289,
      "grad_norm": 0.2981206774711609,
      "learning_rate": 0.0002374168859124611,
      "loss": 0.0573,
      "step": 4866
    },
    {
      "epoch": 0.6821303433777155,
      "grad_norm": 0.37856847047805786,
      "learning_rate": 0.00023740253527864143,
      "loss": 0.0746,
      "step": 4867
    },
    {
      "epoch": 0.6822704975473021,
      "grad_norm": 0.19851186871528625,
      "learning_rate": 0.0002373881846448218,
      "loss": 0.0917,
      "step": 4868
    },
    {
      "epoch": 0.6824106517168885,
      "grad_norm": 0.2003358155488968,
      "learning_rate": 0.00023737383401100214,
      "loss": 0.0579,
      "step": 4869
    },
    {
      "epoch": 0.6825508058864751,
      "grad_norm": 0.44520801305770874,
      "learning_rate": 0.00023735948337718247,
      "loss": 0.0659,
      "step": 4870
    },
    {
      "epoch": 0.6826909600560617,
      "grad_norm": 1.0162887573242188,
      "learning_rate": 0.00023734513274336282,
      "loss": 0.0932,
      "step": 4871
    },
    {
      "epoch": 0.6828311142256482,
      "grad_norm": 0.6490917205810547,
      "learning_rate": 0.00023733078210954315,
      "loss": 0.122,
      "step": 4872
    },
    {
      "epoch": 0.6829712683952348,
      "grad_norm": 0.44667473435401917,
      "learning_rate": 0.00023731643147572348,
      "loss": 0.103,
      "step": 4873
    },
    {
      "epoch": 0.6831114225648213,
      "grad_norm": 0.8786742687225342,
      "learning_rate": 0.00023730208084190384,
      "loss": 0.123,
      "step": 4874
    },
    {
      "epoch": 0.6832515767344078,
      "grad_norm": 0.4250861406326294,
      "learning_rate": 0.00023728773020808416,
      "loss": 0.1716,
      "step": 4875
    },
    {
      "epoch": 0.6833917309039944,
      "grad_norm": 0.5772868394851685,
      "learning_rate": 0.0002372733795742645,
      "loss": 0.1228,
      "step": 4876
    },
    {
      "epoch": 0.683531885073581,
      "grad_norm": 0.4319916367530823,
      "learning_rate": 0.00023725902894044488,
      "loss": 0.0863,
      "step": 4877
    },
    {
      "epoch": 0.6836720392431674,
      "grad_norm": 0.28940045833587646,
      "learning_rate": 0.0002372446783066252,
      "loss": 0.0567,
      "step": 4878
    },
    {
      "epoch": 0.683812193412754,
      "grad_norm": 0.6128058433532715,
      "learning_rate": 0.00023723032767280553,
      "loss": 0.0551,
      "step": 4879
    },
    {
      "epoch": 0.6839523475823406,
      "grad_norm": 0.7274519801139832,
      "learning_rate": 0.0002372159770389859,
      "loss": 0.1297,
      "step": 4880
    },
    {
      "epoch": 0.6840925017519272,
      "grad_norm": 0.23557989299297333,
      "learning_rate": 0.00023720162640516622,
      "loss": 0.0243,
      "step": 4881
    },
    {
      "epoch": 0.6842326559215136,
      "grad_norm": 0.16109292209148407,
      "learning_rate": 0.00023718727577134655,
      "loss": 0.0561,
      "step": 4882
    },
    {
      "epoch": 0.6843728100911002,
      "grad_norm": 0.4620678424835205,
      "learning_rate": 0.00023717292513752687,
      "loss": 0.0739,
      "step": 4883
    },
    {
      "epoch": 0.6845129642606868,
      "grad_norm": 0.72284334897995,
      "learning_rate": 0.00023715857450370723,
      "loss": 0.0977,
      "step": 4884
    },
    {
      "epoch": 0.6846531184302733,
      "grad_norm": 0.6425224542617798,
      "learning_rate": 0.00023714422386988756,
      "loss": 0.1987,
      "step": 4885
    },
    {
      "epoch": 0.6847932725998599,
      "grad_norm": 0.7455645799636841,
      "learning_rate": 0.0002371298732360679,
      "loss": 0.1035,
      "step": 4886
    },
    {
      "epoch": 0.6849334267694464,
      "grad_norm": 0.4513646960258484,
      "learning_rate": 0.00023711552260224824,
      "loss": 0.0863,
      "step": 4887
    },
    {
      "epoch": 0.6850735809390329,
      "grad_norm": 0.24903219938278198,
      "learning_rate": 0.00023710117196842857,
      "loss": 0.0683,
      "step": 4888
    },
    {
      "epoch": 0.6852137351086195,
      "grad_norm": 1.014467477798462,
      "learning_rate": 0.0002370868213346089,
      "loss": 0.1482,
      "step": 4889
    },
    {
      "epoch": 0.6853538892782061,
      "grad_norm": 0.8224611282348633,
      "learning_rate": 0.00023707247070078928,
      "loss": 0.0922,
      "step": 4890
    },
    {
      "epoch": 0.6854940434477925,
      "grad_norm": 1.4471269845962524,
      "learning_rate": 0.0002370581200669696,
      "loss": 0.1765,
      "step": 4891
    },
    {
      "epoch": 0.6856341976173791,
      "grad_norm": 0.2019975632429123,
      "learning_rate": 0.00023704376943314994,
      "loss": 0.0409,
      "step": 4892
    },
    {
      "epoch": 0.6857743517869657,
      "grad_norm": 0.7212958931922913,
      "learning_rate": 0.0002370294187993303,
      "loss": 0.0952,
      "step": 4893
    },
    {
      "epoch": 0.6859145059565522,
      "grad_norm": 0.7052573561668396,
      "learning_rate": 0.00023701506816551062,
      "loss": 0.1361,
      "step": 4894
    },
    {
      "epoch": 0.6860546601261388,
      "grad_norm": 0.8826233148574829,
      "learning_rate": 0.00023700071753169095,
      "loss": 0.4416,
      "step": 4895
    },
    {
      "epoch": 0.6861948142957253,
      "grad_norm": 0.6817396283149719,
      "learning_rate": 0.0002369863668978713,
      "loss": 0.0894,
      "step": 4896
    },
    {
      "epoch": 0.6863349684653118,
      "grad_norm": 2.287858724594116,
      "learning_rate": 0.00023697201626405164,
      "loss": 0.1601,
      "step": 4897
    },
    {
      "epoch": 0.6864751226348984,
      "grad_norm": 0.5211647748947144,
      "learning_rate": 0.00023695766563023196,
      "loss": 0.027,
      "step": 4898
    },
    {
      "epoch": 0.686615276804485,
      "grad_norm": 2.598926067352295,
      "learning_rate": 0.00023694331499641235,
      "loss": 0.1217,
      "step": 4899
    },
    {
      "epoch": 0.6867554309740714,
      "grad_norm": 3.092555284500122,
      "learning_rate": 0.00023692896436259268,
      "loss": 0.3096,
      "step": 4900
    },
    {
      "epoch": 0.686895585143658,
      "grad_norm": 0.3360291123390198,
      "learning_rate": 0.000236914613728773,
      "loss": 0.0823,
      "step": 4901
    },
    {
      "epoch": 0.6870357393132446,
      "grad_norm": 0.47223180532455444,
      "learning_rate": 0.00023690026309495333,
      "loss": 0.0685,
      "step": 4902
    },
    {
      "epoch": 0.6871758934828311,
      "grad_norm": 0.5508928894996643,
      "learning_rate": 0.0002368859124611337,
      "loss": 0.0657,
      "step": 4903
    },
    {
      "epoch": 0.6873160476524176,
      "grad_norm": 0.34525009989738464,
      "learning_rate": 0.00023687156182731402,
      "loss": 0.1164,
      "step": 4904
    },
    {
      "epoch": 0.6874562018220042,
      "grad_norm": 0.4783444106578827,
      "learning_rate": 0.00023685721119349435,
      "loss": 0.121,
      "step": 4905
    },
    {
      "epoch": 0.6875963559915907,
      "grad_norm": 0.22666792571544647,
      "learning_rate": 0.0002368428605596747,
      "loss": 0.0461,
      "step": 4906
    },
    {
      "epoch": 0.6877365101611773,
      "grad_norm": 0.5437456369400024,
      "learning_rate": 0.00023682850992585503,
      "loss": 0.1436,
      "step": 4907
    },
    {
      "epoch": 0.6878766643307639,
      "grad_norm": 0.3524503707885742,
      "learning_rate": 0.00023681415929203536,
      "loss": 0.113,
      "step": 4908
    },
    {
      "epoch": 0.6880168185003503,
      "grad_norm": 0.6922323703765869,
      "learning_rate": 0.00023679980865821574,
      "loss": 0.0983,
      "step": 4909
    },
    {
      "epoch": 0.6881569726699369,
      "grad_norm": 0.3491920232772827,
      "learning_rate": 0.00023678545802439607,
      "loss": 0.0875,
      "step": 4910
    },
    {
      "epoch": 0.6882971268395235,
      "grad_norm": 0.6609526872634888,
      "learning_rate": 0.0002367711073905764,
      "loss": 0.0898,
      "step": 4911
    },
    {
      "epoch": 0.6884372810091101,
      "grad_norm": 0.40448012948036194,
      "learning_rate": 0.00023675675675675675,
      "loss": 0.0804,
      "step": 4912
    },
    {
      "epoch": 0.6885774351786965,
      "grad_norm": 0.7104846835136414,
      "learning_rate": 0.00023674240612293708,
      "loss": 0.1181,
      "step": 4913
    },
    {
      "epoch": 0.6887175893482831,
      "grad_norm": 0.4720240831375122,
      "learning_rate": 0.0002367280554891174,
      "loss": 0.096,
      "step": 4914
    },
    {
      "epoch": 0.6888577435178697,
      "grad_norm": 0.1936282068490982,
      "learning_rate": 0.00023671370485529777,
      "loss": 0.0398,
      "step": 4915
    },
    {
      "epoch": 0.6889978976874562,
      "grad_norm": 0.3752985894680023,
      "learning_rate": 0.0002366993542214781,
      "loss": 0.0801,
      "step": 4916
    },
    {
      "epoch": 0.6891380518570428,
      "grad_norm": 0.4167878329753876,
      "learning_rate": 0.00023668500358765842,
      "loss": 0.1084,
      "step": 4917
    },
    {
      "epoch": 0.6892782060266293,
      "grad_norm": 0.47314736247062683,
      "learning_rate": 0.00023667065295383875,
      "loss": 0.0604,
      "step": 4918
    },
    {
      "epoch": 0.6894183601962158,
      "grad_norm": 0.8354405164718628,
      "learning_rate": 0.0002366563023200191,
      "loss": 0.0739,
      "step": 4919
    },
    {
      "epoch": 0.6895585143658024,
      "grad_norm": 0.5121402144432068,
      "learning_rate": 0.00023664195168619944,
      "loss": 0.0541,
      "step": 4920
    },
    {
      "epoch": 0.689698668535389,
      "grad_norm": 0.4273848235607147,
      "learning_rate": 0.00023662760105237977,
      "loss": 0.0501,
      "step": 4921
    },
    {
      "epoch": 0.6898388227049754,
      "grad_norm": 0.6261603832244873,
      "learning_rate": 0.00023661325041856015,
      "loss": 0.1467,
      "step": 4922
    },
    {
      "epoch": 0.689978976874562,
      "grad_norm": 0.46040821075439453,
      "learning_rate": 0.00023659889978474048,
      "loss": 0.0941,
      "step": 4923
    },
    {
      "epoch": 0.6901191310441486,
      "grad_norm": 0.7623235583305359,
      "learning_rate": 0.0002365845491509208,
      "loss": 0.1054,
      "step": 4924
    },
    {
      "epoch": 0.6902592852137351,
      "grad_norm": 0.2708573043346405,
      "learning_rate": 0.00023657019851710116,
      "loss": 0.0632,
      "step": 4925
    },
    {
      "epoch": 0.6903994393833216,
      "grad_norm": 0.16677004098892212,
      "learning_rate": 0.0002365558478832815,
      "loss": 0.0364,
      "step": 4926
    },
    {
      "epoch": 0.6905395935529082,
      "grad_norm": 0.2472703903913498,
      "learning_rate": 0.00023654149724946182,
      "loss": 0.0342,
      "step": 4927
    },
    {
      "epoch": 0.6906797477224947,
      "grad_norm": 0.16047130525112152,
      "learning_rate": 0.00023652714661564217,
      "loss": 0.0274,
      "step": 4928
    },
    {
      "epoch": 0.6908199018920813,
      "grad_norm": 0.134731262922287,
      "learning_rate": 0.0002365127959818225,
      "loss": 0.0236,
      "step": 4929
    },
    {
      "epoch": 0.6909600560616679,
      "grad_norm": 0.2427656054496765,
      "learning_rate": 0.00023649844534800283,
      "loss": 0.076,
      "step": 4930
    },
    {
      "epoch": 0.6911002102312543,
      "grad_norm": 0.41598305106163025,
      "learning_rate": 0.00023648409471418321,
      "loss": 0.11,
      "step": 4931
    },
    {
      "epoch": 0.6912403644008409,
      "grad_norm": 0.48773691058158875,
      "learning_rate": 0.00023646974408036354,
      "loss": 0.1063,
      "step": 4932
    },
    {
      "epoch": 0.6913805185704275,
      "grad_norm": 0.3838995397090912,
      "learning_rate": 0.00023645539344654387,
      "loss": 0.0862,
      "step": 4933
    },
    {
      "epoch": 0.691520672740014,
      "grad_norm": 0.8167698383331299,
      "learning_rate": 0.00023644104281272423,
      "loss": 0.1032,
      "step": 4934
    },
    {
      "epoch": 0.6916608269096005,
      "grad_norm": 0.32955455780029297,
      "learning_rate": 0.00023642669217890456,
      "loss": 0.1011,
      "step": 4935
    },
    {
      "epoch": 0.6918009810791871,
      "grad_norm": 0.23160520195960999,
      "learning_rate": 0.00023641234154508488,
      "loss": 0.051,
      "step": 4936
    },
    {
      "epoch": 0.6919411352487737,
      "grad_norm": 0.17174088954925537,
      "learning_rate": 0.0002363979909112652,
      "loss": 0.0662,
      "step": 4937
    },
    {
      "epoch": 0.6920812894183602,
      "grad_norm": 0.345893919467926,
      "learning_rate": 0.00023638364027744557,
      "loss": 0.0329,
      "step": 4938
    },
    {
      "epoch": 0.6922214435879468,
      "grad_norm": 0.4581599831581116,
      "learning_rate": 0.0002363692896436259,
      "loss": 0.1207,
      "step": 4939
    },
    {
      "epoch": 0.6923615977575333,
      "grad_norm": 0.6067915558815002,
      "learning_rate": 0.00023635493900980623,
      "loss": 0.0954,
      "step": 4940
    },
    {
      "epoch": 0.6925017519271198,
      "grad_norm": 0.6950394511222839,
      "learning_rate": 0.0002363405883759866,
      "loss": 0.083,
      "step": 4941
    },
    {
      "epoch": 0.6926419060967064,
      "grad_norm": 0.47026780247688293,
      "learning_rate": 0.00023632623774216694,
      "loss": 0.073,
      "step": 4942
    },
    {
      "epoch": 0.692782060266293,
      "grad_norm": 0.25256428122520447,
      "learning_rate": 0.00023631188710834726,
      "loss": 0.0817,
      "step": 4943
    },
    {
      "epoch": 0.6929222144358794,
      "grad_norm": 0.3354937732219696,
      "learning_rate": 0.00023629753647452762,
      "loss": 0.0655,
      "step": 4944
    },
    {
      "epoch": 0.693062368605466,
      "grad_norm": 0.27554529905319214,
      "learning_rate": 0.00023628318584070795,
      "loss": 0.0711,
      "step": 4945
    },
    {
      "epoch": 0.6932025227750526,
      "grad_norm": 0.4474349915981293,
      "learning_rate": 0.00023626883520688828,
      "loss": 0.111,
      "step": 4946
    },
    {
      "epoch": 0.6933426769446391,
      "grad_norm": 0.5785543918609619,
      "learning_rate": 0.00023625448457306863,
      "loss": 0.0614,
      "step": 4947
    },
    {
      "epoch": 0.6934828311142256,
      "grad_norm": 0.6148270964622498,
      "learning_rate": 0.00023624013393924896,
      "loss": 0.1167,
      "step": 4948
    },
    {
      "epoch": 0.6936229852838122,
      "grad_norm": 4.919109344482422,
      "learning_rate": 0.0002362257833054293,
      "loss": 0.224,
      "step": 4949
    },
    {
      "epoch": 0.6937631394533987,
      "grad_norm": 1.4899595975875854,
      "learning_rate": 0.00023621143267160965,
      "loss": 0.2789,
      "step": 4950
    },
    {
      "epoch": 0.6939032936229853,
      "grad_norm": 0.45124882459640503,
      "learning_rate": 0.00023619708203778997,
      "loss": 0.1184,
      "step": 4951
    },
    {
      "epoch": 0.6940434477925719,
      "grad_norm": 0.4265531003475189,
      "learning_rate": 0.0002361827314039703,
      "loss": 0.106,
      "step": 4952
    },
    {
      "epoch": 0.6941836019621583,
      "grad_norm": 0.6305046677589417,
      "learning_rate": 0.00023616838077015063,
      "loss": 0.1031,
      "step": 4953
    },
    {
      "epoch": 0.6943237561317449,
      "grad_norm": 0.32009822130203247,
      "learning_rate": 0.00023615403013633101,
      "loss": 0.1061,
      "step": 4954
    },
    {
      "epoch": 0.6944639103013315,
      "grad_norm": 0.6415496468544006,
      "learning_rate": 0.00023613967950251134,
      "loss": 0.0869,
      "step": 4955
    },
    {
      "epoch": 0.694604064470918,
      "grad_norm": 0.21832266449928284,
      "learning_rate": 0.00023612532886869167,
      "loss": 0.0788,
      "step": 4956
    },
    {
      "epoch": 0.6947442186405045,
      "grad_norm": 0.49858248233795166,
      "learning_rate": 0.00023611097823487203,
      "loss": 0.0797,
      "step": 4957
    },
    {
      "epoch": 0.6948843728100911,
      "grad_norm": 0.32161790132522583,
      "learning_rate": 0.00023609662760105236,
      "loss": 0.0563,
      "step": 4958
    },
    {
      "epoch": 0.6950245269796776,
      "grad_norm": 0.33466893434524536,
      "learning_rate": 0.00023608227696723268,
      "loss": 0.1185,
      "step": 4959
    },
    {
      "epoch": 0.6951646811492642,
      "grad_norm": 0.39162588119506836,
      "learning_rate": 0.00023606792633341304,
      "loss": 0.0955,
      "step": 4960
    },
    {
      "epoch": 0.6953048353188508,
      "grad_norm": 0.3379994332790375,
      "learning_rate": 0.00023605357569959337,
      "loss": 0.0731,
      "step": 4961
    },
    {
      "epoch": 0.6954449894884372,
      "grad_norm": 0.6227350831031799,
      "learning_rate": 0.0002360392250657737,
      "loss": 0.1651,
      "step": 4962
    },
    {
      "epoch": 0.6955851436580238,
      "grad_norm": 0.35392624139785767,
      "learning_rate": 0.00023602487443195408,
      "loss": 0.0525,
      "step": 4963
    },
    {
      "epoch": 0.6957252978276104,
      "grad_norm": 0.4256768226623535,
      "learning_rate": 0.0002360105237981344,
      "loss": 0.0613,
      "step": 4964
    },
    {
      "epoch": 0.695865451997197,
      "grad_norm": 0.7919667363166809,
      "learning_rate": 0.00023599617316431474,
      "loss": 0.0527,
      "step": 4965
    },
    {
      "epoch": 0.6960056061667834,
      "grad_norm": 0.30182307958602905,
      "learning_rate": 0.0002359818225304951,
      "loss": 0.104,
      "step": 4966
    },
    {
      "epoch": 0.69614576033637,
      "grad_norm": 0.40462568402290344,
      "learning_rate": 0.00023596747189667542,
      "loss": 0.1069,
      "step": 4967
    },
    {
      "epoch": 0.6962859145059566,
      "grad_norm": 0.23089641332626343,
      "learning_rate": 0.00023595312126285575,
      "loss": 0.0253,
      "step": 4968
    },
    {
      "epoch": 0.6964260686755431,
      "grad_norm": 0.5238849520683289,
      "learning_rate": 0.0002359387706290361,
      "loss": 0.1258,
      "step": 4969
    },
    {
      "epoch": 0.6965662228451296,
      "grad_norm": 0.37786635756492615,
      "learning_rate": 0.00023592441999521643,
      "loss": 0.1281,
      "step": 4970
    },
    {
      "epoch": 0.6967063770147162,
      "grad_norm": 0.36394861340522766,
      "learning_rate": 0.00023591006936139676,
      "loss": 0.0422,
      "step": 4971
    },
    {
      "epoch": 0.6968465311843027,
      "grad_norm": 0.4449184536933899,
      "learning_rate": 0.0002358957187275771,
      "loss": 0.0794,
      "step": 4972
    },
    {
      "epoch": 0.6969866853538893,
      "grad_norm": 0.20255087316036224,
      "learning_rate": 0.00023588136809375747,
      "loss": 0.0457,
      "step": 4973
    },
    {
      "epoch": 0.6971268395234759,
      "grad_norm": 0.36113888025283813,
      "learning_rate": 0.0002358670174599378,
      "loss": 0.0499,
      "step": 4974
    },
    {
      "epoch": 0.6972669936930623,
      "grad_norm": 0.5287073254585266,
      "learning_rate": 0.00023585266682611813,
      "loss": 0.0872,
      "step": 4975
    },
    {
      "epoch": 0.6974071478626489,
      "grad_norm": 0.3566041886806488,
      "learning_rate": 0.0002358383161922985,
      "loss": 0.0504,
      "step": 4976
    },
    {
      "epoch": 0.6975473020322355,
      "grad_norm": 0.2616063356399536,
      "learning_rate": 0.00023582396555847882,
      "loss": 0.0506,
      "step": 4977
    },
    {
      "epoch": 0.697687456201822,
      "grad_norm": 0.22803056240081787,
      "learning_rate": 0.00023580961492465914,
      "loss": 0.0651,
      "step": 4978
    },
    {
      "epoch": 0.6978276103714085,
      "grad_norm": 0.4278799593448639,
      "learning_rate": 0.0002357952642908395,
      "loss": 0.0518,
      "step": 4979
    },
    {
      "epoch": 0.6979677645409951,
      "grad_norm": 0.43411722779273987,
      "learning_rate": 0.00023578091365701983,
      "loss": 0.0796,
      "step": 4980
    },
    {
      "epoch": 0.6981079187105816,
      "grad_norm": 0.7161135673522949,
      "learning_rate": 0.00023576656302320016,
      "loss": 0.1399,
      "step": 4981
    },
    {
      "epoch": 0.6982480728801682,
      "grad_norm": 0.2713644206523895,
      "learning_rate": 0.0002357522123893805,
      "loss": 0.0722,
      "step": 4982
    },
    {
      "epoch": 0.6983882270497548,
      "grad_norm": 0.6444911360740662,
      "learning_rate": 0.00023573786175556084,
      "loss": 0.0518,
      "step": 4983
    },
    {
      "epoch": 0.6985283812193412,
      "grad_norm": 1.2718766927719116,
      "learning_rate": 0.00023572351112174117,
      "loss": 0.1725,
      "step": 4984
    },
    {
      "epoch": 0.6986685353889278,
      "grad_norm": 0.6072478890419006,
      "learning_rate": 0.00023570916048792155,
      "loss": 0.0987,
      "step": 4985
    },
    {
      "epoch": 0.6988086895585144,
      "grad_norm": 0.34491369128227234,
      "learning_rate": 0.00023569480985410188,
      "loss": 0.1027,
      "step": 4986
    },
    {
      "epoch": 0.698948843728101,
      "grad_norm": 0.6384333372116089,
      "learning_rate": 0.0002356804592202822,
      "loss": 0.1637,
      "step": 4987
    },
    {
      "epoch": 0.6990889978976874,
      "grad_norm": 0.16771025955677032,
      "learning_rate": 0.00023566610858646254,
      "loss": 0.0764,
      "step": 4988
    },
    {
      "epoch": 0.699229152067274,
      "grad_norm": 0.28696170449256897,
      "learning_rate": 0.0002356517579526429,
      "loss": 0.0302,
      "step": 4989
    },
    {
      "epoch": 0.6993693062368606,
      "grad_norm": 0.20444737374782562,
      "learning_rate": 0.00023563740731882322,
      "loss": 0.0472,
      "step": 4990
    },
    {
      "epoch": 0.6995094604064471,
      "grad_norm": 0.2923547923564911,
      "learning_rate": 0.00023562305668500355,
      "loss": 0.0642,
      "step": 4991
    },
    {
      "epoch": 0.6996496145760336,
      "grad_norm": 0.21817873418331146,
      "learning_rate": 0.0002356087060511839,
      "loss": 0.027,
      "step": 4992
    },
    {
      "epoch": 0.6997897687456202,
      "grad_norm": 0.6990349292755127,
      "learning_rate": 0.00023559435541736424,
      "loss": 0.1423,
      "step": 4993
    },
    {
      "epoch": 0.6999299229152067,
      "grad_norm": 0.4087603986263275,
      "learning_rate": 0.00023558000478354456,
      "loss": 0.0811,
      "step": 4994
    },
    {
      "epoch": 0.7000700770847933,
      "grad_norm": 0.8099572062492371,
      "learning_rate": 0.00023556565414972495,
      "loss": 0.1,
      "step": 4995
    },
    {
      "epoch": 0.7002102312543799,
      "grad_norm": 0.4728637933731079,
      "learning_rate": 0.00023555130351590527,
      "loss": 0.0905,
      "step": 4996
    },
    {
      "epoch": 0.7003503854239663,
      "grad_norm": 1.3188045024871826,
      "learning_rate": 0.0002355369528820856,
      "loss": 0.1761,
      "step": 4997
    },
    {
      "epoch": 0.7004905395935529,
      "grad_norm": 0.553395688533783,
      "learning_rate": 0.00023552260224826596,
      "loss": 0.1177,
      "step": 4998
    },
    {
      "epoch": 0.7006306937631395,
      "grad_norm": 0.948553204536438,
      "learning_rate": 0.0002355082516144463,
      "loss": 0.1856,
      "step": 4999
    },
    {
      "epoch": 0.700770847932726,
      "grad_norm": 2.8646082878112793,
      "learning_rate": 0.00023549390098062662,
      "loss": 0.0974,
      "step": 5000
    },
    {
      "epoch": 0.7009110021023125,
      "grad_norm": 0.45448389649391174,
      "learning_rate": 0.00023547955034680697,
      "loss": 0.1466,
      "step": 5001
    },
    {
      "epoch": 0.7010511562718991,
      "grad_norm": 0.3053547441959381,
      "learning_rate": 0.0002354651997129873,
      "loss": 0.0757,
      "step": 5002
    },
    {
      "epoch": 0.7011913104414856,
      "grad_norm": 0.4656880497932434,
      "learning_rate": 0.00023545084907916763,
      "loss": 0.0518,
      "step": 5003
    },
    {
      "epoch": 0.7013314646110722,
      "grad_norm": 0.259682297706604,
      "learning_rate": 0.000235436498445348,
      "loss": 0.0618,
      "step": 5004
    },
    {
      "epoch": 0.7014716187806588,
      "grad_norm": 0.23479433357715607,
      "learning_rate": 0.00023542214781152834,
      "loss": 0.0435,
      "step": 5005
    },
    {
      "epoch": 0.7016117729502452,
      "grad_norm": 0.4558795392513275,
      "learning_rate": 0.00023540779717770867,
      "loss": 0.1396,
      "step": 5006
    },
    {
      "epoch": 0.7017519271198318,
      "grad_norm": 0.2944003641605377,
      "learning_rate": 0.000235393446543889,
      "loss": 0.0601,
      "step": 5007
    },
    {
      "epoch": 0.7018920812894184,
      "grad_norm": 0.34318599104881287,
      "learning_rate": 0.00023537909591006935,
      "loss": 0.0883,
      "step": 5008
    },
    {
      "epoch": 0.702032235459005,
      "grad_norm": 0.29112017154693604,
      "learning_rate": 0.00023536474527624968,
      "loss": 0.075,
      "step": 5009
    },
    {
      "epoch": 0.7021723896285914,
      "grad_norm": 0.491881787776947,
      "learning_rate": 0.00023535039464243,
      "loss": 0.0856,
      "step": 5010
    },
    {
      "epoch": 0.702312543798178,
      "grad_norm": 0.6046550869941711,
      "learning_rate": 0.00023533604400861037,
      "loss": 0.0669,
      "step": 5011
    },
    {
      "epoch": 0.7024526979677645,
      "grad_norm": 0.23227007687091827,
      "learning_rate": 0.0002353216933747907,
      "loss": 0.0888,
      "step": 5012
    },
    {
      "epoch": 0.7025928521373511,
      "grad_norm": 0.431506872177124,
      "learning_rate": 0.00023530734274097102,
      "loss": 0.0754,
      "step": 5013
    },
    {
      "epoch": 0.7027330063069376,
      "grad_norm": 0.36266499757766724,
      "learning_rate": 0.00023529299210715138,
      "loss": 0.0808,
      "step": 5014
    },
    {
      "epoch": 0.7028731604765242,
      "grad_norm": 1.0934547185897827,
      "learning_rate": 0.0002352786414733317,
      "loss": 0.0843,
      "step": 5015
    },
    {
      "epoch": 0.7030133146461107,
      "grad_norm": 0.47969672083854675,
      "learning_rate": 0.00023526429083951204,
      "loss": 0.0544,
      "step": 5016
    },
    {
      "epoch": 0.7031534688156973,
      "grad_norm": 0.27149614691734314,
      "learning_rate": 0.00023524994020569242,
      "loss": 0.0763,
      "step": 5017
    },
    {
      "epoch": 0.7032936229852839,
      "grad_norm": 0.3926200270652771,
      "learning_rate": 0.00023523558957187275,
      "loss": 0.0546,
      "step": 5018
    },
    {
      "epoch": 0.7034337771548703,
      "grad_norm": 0.5547006726264954,
      "learning_rate": 0.00023522123893805308,
      "loss": 0.1948,
      "step": 5019
    },
    {
      "epoch": 0.7035739313244569,
      "grad_norm": 0.31557026505470276,
      "learning_rate": 0.00023520688830423343,
      "loss": 0.0392,
      "step": 5020
    },
    {
      "epoch": 0.7037140854940435,
      "grad_norm": 0.7667475342750549,
      "learning_rate": 0.00023519253767041376,
      "loss": 0.1638,
      "step": 5021
    },
    {
      "epoch": 0.70385423966363,
      "grad_norm": 0.6007963418960571,
      "learning_rate": 0.0002351781870365941,
      "loss": 0.1742,
      "step": 5022
    },
    {
      "epoch": 0.7039943938332165,
      "grad_norm": 0.4213850796222687,
      "learning_rate": 0.00023516383640277442,
      "loss": 0.1667,
      "step": 5023
    },
    {
      "epoch": 0.7041345480028031,
      "grad_norm": 0.7641958594322205,
      "learning_rate": 0.00023514948576895477,
      "loss": 0.1815,
      "step": 5024
    },
    {
      "epoch": 0.7042747021723896,
      "grad_norm": 0.3817320764064789,
      "learning_rate": 0.0002351351351351351,
      "loss": 0.0594,
      "step": 5025
    },
    {
      "epoch": 0.7044148563419762,
      "grad_norm": 0.2560789883136749,
      "learning_rate": 0.00023512078450131543,
      "loss": 0.0742,
      "step": 5026
    },
    {
      "epoch": 0.7045550105115628,
      "grad_norm": 0.3542046844959259,
      "learning_rate": 0.0002351064338674958,
      "loss": 0.0963,
      "step": 5027
    },
    {
      "epoch": 0.7046951646811492,
      "grad_norm": 0.46601173281669617,
      "learning_rate": 0.00023509208323367614,
      "loss": 0.0895,
      "step": 5028
    },
    {
      "epoch": 0.7048353188507358,
      "grad_norm": 0.20284810662269592,
      "learning_rate": 0.00023507773259985647,
      "loss": 0.0612,
      "step": 5029
    },
    {
      "epoch": 0.7049754730203224,
      "grad_norm": 0.3436676263809204,
      "learning_rate": 0.00023506338196603683,
      "loss": 0.0413,
      "step": 5030
    },
    {
      "epoch": 0.705115627189909,
      "grad_norm": 0.2958710491657257,
      "learning_rate": 0.00023504903133221715,
      "loss": 0.0863,
      "step": 5031
    },
    {
      "epoch": 0.7052557813594954,
      "grad_norm": 0.2852146625518799,
      "learning_rate": 0.00023503468069839748,
      "loss": 0.035,
      "step": 5032
    },
    {
      "epoch": 0.705395935529082,
      "grad_norm": 0.3974468410015106,
      "learning_rate": 0.00023502033006457784,
      "loss": 0.0533,
      "step": 5033
    },
    {
      "epoch": 0.7055360896986685,
      "grad_norm": 0.24293284118175507,
      "learning_rate": 0.00023500597943075817,
      "loss": 0.0673,
      "step": 5034
    },
    {
      "epoch": 0.7056762438682551,
      "grad_norm": 0.1357775181531906,
      "learning_rate": 0.0002349916287969385,
      "loss": 0.0368,
      "step": 5035
    },
    {
      "epoch": 0.7058163980378416,
      "grad_norm": 0.6605488657951355,
      "learning_rate": 0.00023497727816311888,
      "loss": 0.065,
      "step": 5036
    },
    {
      "epoch": 0.7059565522074281,
      "grad_norm": 0.4690210819244385,
      "learning_rate": 0.0002349629275292992,
      "loss": 0.1224,
      "step": 5037
    },
    {
      "epoch": 0.7060967063770147,
      "grad_norm": 0.35564124584198,
      "learning_rate": 0.00023494857689547954,
      "loss": 0.069,
      "step": 5038
    },
    {
      "epoch": 0.7062368605466013,
      "grad_norm": 0.5142439007759094,
      "learning_rate": 0.0002349342262616599,
      "loss": 0.0803,
      "step": 5039
    },
    {
      "epoch": 0.7063770147161879,
      "grad_norm": 1.554359793663025,
      "learning_rate": 0.00023491987562784022,
      "loss": 0.3452,
      "step": 5040
    },
    {
      "epoch": 0.7065171688857743,
      "grad_norm": 0.35481011867523193,
      "learning_rate": 0.00023490552499402055,
      "loss": 0.0686,
      "step": 5041
    },
    {
      "epoch": 0.7066573230553609,
      "grad_norm": 0.5887169241905212,
      "learning_rate": 0.00023489117436020088,
      "loss": 0.0713,
      "step": 5042
    },
    {
      "epoch": 0.7067974772249475,
      "grad_norm": 0.30480802059173584,
      "learning_rate": 0.00023487682372638123,
      "loss": 0.054,
      "step": 5043
    },
    {
      "epoch": 0.706937631394534,
      "grad_norm": 0.4054400324821472,
      "learning_rate": 0.00023486247309256156,
      "loss": 0.0867,
      "step": 5044
    },
    {
      "epoch": 0.7070777855641205,
      "grad_norm": 1.2069053649902344,
      "learning_rate": 0.0002348481224587419,
      "loss": 0.1405,
      "step": 5045
    },
    {
      "epoch": 0.7072179397337071,
      "grad_norm": 0.5618757009506226,
      "learning_rate": 0.00023483377182492225,
      "loss": 0.1636,
      "step": 5046
    },
    {
      "epoch": 0.7073580939032936,
      "grad_norm": 0.7865987420082092,
      "learning_rate": 0.00023481942119110257,
      "loss": 0.1725,
      "step": 5047
    },
    {
      "epoch": 0.7074982480728802,
      "grad_norm": 0.838504433631897,
      "learning_rate": 0.0002348050705572829,
      "loss": 0.1417,
      "step": 5048
    },
    {
      "epoch": 0.7076384022424668,
      "grad_norm": 0.31529176235198975,
      "learning_rate": 0.00023479071992346328,
      "loss": 0.0328,
      "step": 5049
    },
    {
      "epoch": 0.7077785564120532,
      "grad_norm": 1.9494835138320923,
      "learning_rate": 0.0002347763692896436,
      "loss": 0.1889,
      "step": 5050
    },
    {
      "epoch": 0.7079187105816398,
      "grad_norm": 0.4203592538833618,
      "learning_rate": 0.00023476201865582394,
      "loss": 0.0623,
      "step": 5051
    },
    {
      "epoch": 0.7080588647512264,
      "grad_norm": 0.13881684839725494,
      "learning_rate": 0.0002347476680220043,
      "loss": 0.0333,
      "step": 5052
    },
    {
      "epoch": 0.7081990189208129,
      "grad_norm": 1.0938949584960938,
      "learning_rate": 0.00023473331738818463,
      "loss": 0.1527,
      "step": 5053
    },
    {
      "epoch": 0.7083391730903994,
      "grad_norm": 0.33289945125579834,
      "learning_rate": 0.00023471896675436495,
      "loss": 0.0307,
      "step": 5054
    },
    {
      "epoch": 0.708479327259986,
      "grad_norm": 0.4040643870830536,
      "learning_rate": 0.0002347046161205453,
      "loss": 0.0535,
      "step": 5055
    },
    {
      "epoch": 0.7086194814295725,
      "grad_norm": 0.45840442180633545,
      "learning_rate": 0.00023469026548672564,
      "loss": 0.036,
      "step": 5056
    },
    {
      "epoch": 0.7087596355991591,
      "grad_norm": 0.7804120182991028,
      "learning_rate": 0.00023467591485290597,
      "loss": 0.1064,
      "step": 5057
    },
    {
      "epoch": 0.7088997897687456,
      "grad_norm": 0.8684229850769043,
      "learning_rate": 0.0002346615642190863,
      "loss": 0.1196,
      "step": 5058
    },
    {
      "epoch": 0.7090399439383321,
      "grad_norm": 0.32349488139152527,
      "learning_rate": 0.00023464721358526668,
      "loss": 0.0824,
      "step": 5059
    },
    {
      "epoch": 0.7091800981079187,
      "grad_norm": 0.7354174852371216,
      "learning_rate": 0.000234632862951447,
      "loss": 0.0835,
      "step": 5060
    },
    {
      "epoch": 0.7093202522775053,
      "grad_norm": 0.3127652108669281,
      "learning_rate": 0.00023461851231762734,
      "loss": 0.0551,
      "step": 5061
    },
    {
      "epoch": 0.7094604064470919,
      "grad_norm": 0.4412612318992615,
      "learning_rate": 0.0002346041616838077,
      "loss": 0.0596,
      "step": 5062
    },
    {
      "epoch": 0.7096005606166783,
      "grad_norm": 0.24782206118106842,
      "learning_rate": 0.00023458981104998802,
      "loss": 0.0743,
      "step": 5063
    },
    {
      "epoch": 0.7097407147862649,
      "grad_norm": 0.23947393894195557,
      "learning_rate": 0.00023457546041616835,
      "loss": 0.0633,
      "step": 5064
    },
    {
      "epoch": 0.7098808689558515,
      "grad_norm": 0.6765064597129822,
      "learning_rate": 0.0002345611097823487,
      "loss": 0.1379,
      "step": 5065
    },
    {
      "epoch": 0.710021023125438,
      "grad_norm": 0.5635762810707092,
      "learning_rate": 0.00023454675914852903,
      "loss": 0.0675,
      "step": 5066
    },
    {
      "epoch": 0.7101611772950245,
      "grad_norm": 0.3269973397254944,
      "learning_rate": 0.00023453240851470936,
      "loss": 0.0606,
      "step": 5067
    },
    {
      "epoch": 0.710301331464611,
      "grad_norm": 1.3347358703613281,
      "learning_rate": 0.00023451805788088974,
      "loss": 0.2575,
      "step": 5068
    },
    {
      "epoch": 0.7104414856341976,
      "grad_norm": 0.6788570284843445,
      "learning_rate": 0.00023450370724707007,
      "loss": 0.0829,
      "step": 5069
    },
    {
      "epoch": 0.7105816398037842,
      "grad_norm": 0.27341267466545105,
      "learning_rate": 0.0002344893566132504,
      "loss": 0.0454,
      "step": 5070
    },
    {
      "epoch": 0.7107217939733707,
      "grad_norm": 0.4122808873653412,
      "learning_rate": 0.00023447500597943076,
      "loss": 0.1219,
      "step": 5071
    },
    {
      "epoch": 0.7108619481429572,
      "grad_norm": 0.1932440847158432,
      "learning_rate": 0.00023446065534561109,
      "loss": 0.0332,
      "step": 5072
    },
    {
      "epoch": 0.7110021023125438,
      "grad_norm": 0.12211096286773682,
      "learning_rate": 0.00023444630471179141,
      "loss": 0.0307,
      "step": 5073
    },
    {
      "epoch": 0.7111422564821304,
      "grad_norm": 0.26029810309410095,
      "learning_rate": 0.00023443195407797177,
      "loss": 0.0327,
      "step": 5074
    },
    {
      "epoch": 0.7112824106517169,
      "grad_norm": 0.3763386011123657,
      "learning_rate": 0.0002344176034441521,
      "loss": 0.0847,
      "step": 5075
    },
    {
      "epoch": 0.7114225648213034,
      "grad_norm": 0.42975911498069763,
      "learning_rate": 0.00023440325281033243,
      "loss": 0.1221,
      "step": 5076
    },
    {
      "epoch": 0.71156271899089,
      "grad_norm": 0.40714436769485474,
      "learning_rate": 0.00023438890217651276,
      "loss": 0.1232,
      "step": 5077
    },
    {
      "epoch": 0.7117028731604765,
      "grad_norm": 0.5231318473815918,
      "learning_rate": 0.0002343745515426931,
      "loss": 0.0901,
      "step": 5078
    },
    {
      "epoch": 0.7118430273300631,
      "grad_norm": 0.48449090123176575,
      "learning_rate": 0.00023436020090887344,
      "loss": 0.0566,
      "step": 5079
    },
    {
      "epoch": 0.7119831814996496,
      "grad_norm": 0.26782605051994324,
      "learning_rate": 0.00023434585027505377,
      "loss": 0.0951,
      "step": 5080
    },
    {
      "epoch": 0.7121233356692361,
      "grad_norm": 0.43498867750167847,
      "learning_rate": 0.00023433149964123415,
      "loss": 0.0453,
      "step": 5081
    },
    {
      "epoch": 0.7122634898388227,
      "grad_norm": 0.3748285472393036,
      "learning_rate": 0.00023431714900741448,
      "loss": 0.1466,
      "step": 5082
    },
    {
      "epoch": 0.7124036440084093,
      "grad_norm": 0.5433907508850098,
      "learning_rate": 0.0002343027983735948,
      "loss": 0.1037,
      "step": 5083
    },
    {
      "epoch": 0.7125437981779958,
      "grad_norm": 0.5655926465988159,
      "learning_rate": 0.00023428844773977516,
      "loss": 0.0599,
      "step": 5084
    },
    {
      "epoch": 0.7126839523475823,
      "grad_norm": 0.5901606678962708,
      "learning_rate": 0.0002342740971059555,
      "loss": 0.1123,
      "step": 5085
    },
    {
      "epoch": 0.7128241065171689,
      "grad_norm": 0.4195170998573303,
      "learning_rate": 0.00023425974647213582,
      "loss": 0.0329,
      "step": 5086
    },
    {
      "epoch": 0.7129642606867554,
      "grad_norm": 0.33596286177635193,
      "learning_rate": 0.00023424539583831618,
      "loss": 0.0846,
      "step": 5087
    },
    {
      "epoch": 0.713104414856342,
      "grad_norm": 0.3595917224884033,
      "learning_rate": 0.0002342310452044965,
      "loss": 0.1196,
      "step": 5088
    },
    {
      "epoch": 0.7132445690259285,
      "grad_norm": 0.39344146847724915,
      "learning_rate": 0.00023421669457067683,
      "loss": 0.1846,
      "step": 5089
    },
    {
      "epoch": 0.713384723195515,
      "grad_norm": 0.5979328751564026,
      "learning_rate": 0.00023420234393685722,
      "loss": 0.069,
      "step": 5090
    },
    {
      "epoch": 0.7135248773651016,
      "grad_norm": 0.5251967906951904,
      "learning_rate": 0.00023418799330303755,
      "loss": 0.0864,
      "step": 5091
    },
    {
      "epoch": 0.7136650315346882,
      "grad_norm": 0.3856375217437744,
      "learning_rate": 0.00023417364266921787,
      "loss": 0.0441,
      "step": 5092
    },
    {
      "epoch": 0.7138051857042746,
      "grad_norm": 0.6210995316505432,
      "learning_rate": 0.0002341592920353982,
      "loss": 0.0538,
      "step": 5093
    },
    {
      "epoch": 0.7139453398738612,
      "grad_norm": 0.7057554125785828,
      "learning_rate": 0.00023414494140157856,
      "loss": 0.1129,
      "step": 5094
    },
    {
      "epoch": 0.7140854940434478,
      "grad_norm": 0.9471320509910583,
      "learning_rate": 0.0002341305907677589,
      "loss": 0.1692,
      "step": 5095
    },
    {
      "epoch": 0.7142256482130344,
      "grad_norm": 0.7931409478187561,
      "learning_rate": 0.00023411624013393922,
      "loss": 0.1032,
      "step": 5096
    },
    {
      "epoch": 0.7143658023826209,
      "grad_norm": 0.9453210830688477,
      "learning_rate": 0.00023410188950011957,
      "loss": 0.0951,
      "step": 5097
    },
    {
      "epoch": 0.7145059565522074,
      "grad_norm": 1.0313867330551147,
      "learning_rate": 0.0002340875388662999,
      "loss": 0.1764,
      "step": 5098
    },
    {
      "epoch": 0.714646110721794,
      "grad_norm": 4.681178569793701,
      "learning_rate": 0.00023407318823248023,
      "loss": 0.3094,
      "step": 5099
    },
    {
      "epoch": 0.7147862648913805,
      "grad_norm": 6.18013334274292,
      "learning_rate": 0.0002340588375986606,
      "loss": 0.3564,
      "step": 5100
    },
    {
      "epoch": 0.7149264190609671,
      "grad_norm": 0.24321196973323822,
      "learning_rate": 0.00023404448696484094,
      "loss": 0.0374,
      "step": 5101
    },
    {
      "epoch": 0.7150665732305536,
      "grad_norm": 0.2917323708534241,
      "learning_rate": 0.00023403013633102127,
      "loss": 0.0666,
      "step": 5102
    },
    {
      "epoch": 0.7152067274001401,
      "grad_norm": 0.5228332281112671,
      "learning_rate": 0.00023401578569720162,
      "loss": 0.0598,
      "step": 5103
    },
    {
      "epoch": 0.7153468815697267,
      "grad_norm": 0.34571489691734314,
      "learning_rate": 0.00023400143506338195,
      "loss": 0.0819,
      "step": 5104
    },
    {
      "epoch": 0.7154870357393133,
      "grad_norm": 0.3090171813964844,
      "learning_rate": 0.00023398708442956228,
      "loss": 0.0714,
      "step": 5105
    },
    {
      "epoch": 0.7156271899088998,
      "grad_norm": 0.31017976999282837,
      "learning_rate": 0.00023397273379574264,
      "loss": 0.0563,
      "step": 5106
    },
    {
      "epoch": 0.7157673440784863,
      "grad_norm": 0.39233192801475525,
      "learning_rate": 0.00023395838316192296,
      "loss": 0.0514,
      "step": 5107
    },
    {
      "epoch": 0.7159074982480729,
      "grad_norm": 0.27979087829589844,
      "learning_rate": 0.0002339440325281033,
      "loss": 0.0648,
      "step": 5108
    },
    {
      "epoch": 0.7160476524176594,
      "grad_norm": 0.306318461894989,
      "learning_rate": 0.00023392968189428362,
      "loss": 0.0567,
      "step": 5109
    },
    {
      "epoch": 0.716187806587246,
      "grad_norm": 0.3515383303165436,
      "learning_rate": 0.00023391533126046398,
      "loss": 0.0696,
      "step": 5110
    },
    {
      "epoch": 0.7163279607568325,
      "grad_norm": 0.5803670883178711,
      "learning_rate": 0.0002339009806266443,
      "loss": 0.1081,
      "step": 5111
    },
    {
      "epoch": 0.716468114926419,
      "grad_norm": 0.3434430956840515,
      "learning_rate": 0.00023388662999282463,
      "loss": 0.1214,
      "step": 5112
    },
    {
      "epoch": 0.7166082690960056,
      "grad_norm": 0.3058526813983917,
      "learning_rate": 0.00023387227935900502,
      "loss": 0.1162,
      "step": 5113
    },
    {
      "epoch": 0.7167484232655922,
      "grad_norm": 0.46661266684532166,
      "learning_rate": 0.00023385792872518535,
      "loss": 0.0828,
      "step": 5114
    },
    {
      "epoch": 0.7168885774351786,
      "grad_norm": 0.2832973599433899,
      "learning_rate": 0.00023384357809136567,
      "loss": 0.0665,
      "step": 5115
    },
    {
      "epoch": 0.7170287316047652,
      "grad_norm": 0.23683227598667145,
      "learning_rate": 0.00023382922745754603,
      "loss": 0.0444,
      "step": 5116
    },
    {
      "epoch": 0.7171688857743518,
      "grad_norm": 0.5688052773475647,
      "learning_rate": 0.00023381487682372636,
      "loss": 0.1125,
      "step": 5117
    },
    {
      "epoch": 0.7173090399439384,
      "grad_norm": 0.46640726923942566,
      "learning_rate": 0.0002338005261899067,
      "loss": 0.0886,
      "step": 5118
    },
    {
      "epoch": 0.7174491941135249,
      "grad_norm": 0.3015604317188263,
      "learning_rate": 0.00023378617555608704,
      "loss": 0.1054,
      "step": 5119
    },
    {
      "epoch": 0.7175893482831114,
      "grad_norm": 0.3122844696044922,
      "learning_rate": 0.00023377182492226737,
      "loss": 0.0781,
      "step": 5120
    },
    {
      "epoch": 0.717729502452698,
      "grad_norm": 0.2697398066520691,
      "learning_rate": 0.0002337574742884477,
      "loss": 0.0436,
      "step": 5121
    },
    {
      "epoch": 0.7178696566222845,
      "grad_norm": 0.2917141318321228,
      "learning_rate": 0.00023374312365462808,
      "loss": 0.0739,
      "step": 5122
    },
    {
      "epoch": 0.7180098107918711,
      "grad_norm": 0.39828020334243774,
      "learning_rate": 0.0002337287730208084,
      "loss": 0.0256,
      "step": 5123
    },
    {
      "epoch": 0.7181499649614576,
      "grad_norm": 0.46709543466567993,
      "learning_rate": 0.00023371442238698874,
      "loss": 0.0858,
      "step": 5124
    },
    {
      "epoch": 0.7182901191310441,
      "grad_norm": 0.5850250720977783,
      "learning_rate": 0.0002337000717531691,
      "loss": 0.1303,
      "step": 5125
    },
    {
      "epoch": 0.7184302733006307,
      "grad_norm": 0.819942057132721,
      "learning_rate": 0.00023368572111934942,
      "loss": 0.0603,
      "step": 5126
    },
    {
      "epoch": 0.7185704274702173,
      "grad_norm": 0.4037875533103943,
      "learning_rate": 0.00023367137048552975,
      "loss": 0.0698,
      "step": 5127
    },
    {
      "epoch": 0.7187105816398038,
      "grad_norm": 0.9957055449485779,
      "learning_rate": 0.00023365701985171008,
      "loss": 0.1372,
      "step": 5128
    },
    {
      "epoch": 0.7188507358093903,
      "grad_norm": 0.5481537580490112,
      "learning_rate": 0.00023364266921789044,
      "loss": 0.0973,
      "step": 5129
    },
    {
      "epoch": 0.7189908899789769,
      "grad_norm": 0.49370571970939636,
      "learning_rate": 0.00023362831858407077,
      "loss": 0.097,
      "step": 5130
    },
    {
      "epoch": 0.7191310441485634,
      "grad_norm": 0.3298933506011963,
      "learning_rate": 0.0002336139679502511,
      "loss": 0.0642,
      "step": 5131
    },
    {
      "epoch": 0.71927119831815,
      "grad_norm": 0.2864142060279846,
      "learning_rate": 0.00023359961731643148,
      "loss": 0.0354,
      "step": 5132
    },
    {
      "epoch": 0.7194113524877365,
      "grad_norm": 1.5379772186279297,
      "learning_rate": 0.0002335852666826118,
      "loss": 0.1454,
      "step": 5133
    },
    {
      "epoch": 0.719551506657323,
      "grad_norm": 0.28490015864372253,
      "learning_rate": 0.00023357091604879213,
      "loss": 0.0524,
      "step": 5134
    },
    {
      "epoch": 0.7196916608269096,
      "grad_norm": 0.3085574209690094,
      "learning_rate": 0.0002335565654149725,
      "loss": 0.0255,
      "step": 5135
    },
    {
      "epoch": 0.7198318149964962,
      "grad_norm": 0.35978496074676514,
      "learning_rate": 0.00023354221478115282,
      "loss": 0.0837,
      "step": 5136
    },
    {
      "epoch": 0.7199719691660826,
      "grad_norm": 0.35980701446533203,
      "learning_rate": 0.00023352786414733315,
      "loss": 0.0541,
      "step": 5137
    },
    {
      "epoch": 0.7201121233356692,
      "grad_norm": 0.38387787342071533,
      "learning_rate": 0.0002335135135135135,
      "loss": 0.0497,
      "step": 5138
    },
    {
      "epoch": 0.7202522775052558,
      "grad_norm": 0.7311360239982605,
      "learning_rate": 0.00023349916287969383,
      "loss": 0.1206,
      "step": 5139
    },
    {
      "epoch": 0.7203924316748423,
      "grad_norm": 0.41189852356910706,
      "learning_rate": 0.00023348481224587416,
      "loss": 0.0404,
      "step": 5140
    },
    {
      "epoch": 0.7205325858444289,
      "grad_norm": 0.7187962532043457,
      "learning_rate": 0.00023347046161205452,
      "loss": 0.1048,
      "step": 5141
    },
    {
      "epoch": 0.7206727400140154,
      "grad_norm": 0.2527622580528259,
      "learning_rate": 0.00023345611097823484,
      "loss": 0.0506,
      "step": 5142
    },
    {
      "epoch": 0.720812894183602,
      "grad_norm": 0.2305736094713211,
      "learning_rate": 0.00023344176034441517,
      "loss": 0.0769,
      "step": 5143
    },
    {
      "epoch": 0.7209530483531885,
      "grad_norm": 0.3243862986564636,
      "learning_rate": 0.0002334274097105955,
      "loss": 0.0439,
      "step": 5144
    },
    {
      "epoch": 0.7210932025227751,
      "grad_norm": 0.4699042737483978,
      "learning_rate": 0.00023341305907677588,
      "loss": 0.1077,
      "step": 5145
    },
    {
      "epoch": 0.7212333566923615,
      "grad_norm": 0.24194759130477905,
      "learning_rate": 0.0002333987084429562,
      "loss": 0.042,
      "step": 5146
    },
    {
      "epoch": 0.7213735108619481,
      "grad_norm": 0.802369236946106,
      "learning_rate": 0.00023338435780913654,
      "loss": 0.1197,
      "step": 5147
    },
    {
      "epoch": 0.7215136650315347,
      "grad_norm": 2.8443851470947266,
      "learning_rate": 0.0002333700071753169,
      "loss": 0.3762,
      "step": 5148
    },
    {
      "epoch": 0.7216538192011213,
      "grad_norm": 0.6536649465560913,
      "learning_rate": 0.00023335565654149723,
      "loss": 0.0937,
      "step": 5149
    },
    {
      "epoch": 0.7217939733707078,
      "grad_norm": 1.9151217937469482,
      "learning_rate": 0.00023334130590767755,
      "loss": 0.4609,
      "step": 5150
    },
    {
      "epoch": 0.7219341275402943,
      "grad_norm": 0.5031387805938721,
      "learning_rate": 0.0002333269552738579,
      "loss": 0.1766,
      "step": 5151
    },
    {
      "epoch": 0.7220742817098809,
      "grad_norm": 0.641347348690033,
      "learning_rate": 0.00023331260464003824,
      "loss": 0.1777,
      "step": 5152
    },
    {
      "epoch": 0.7222144358794674,
      "grad_norm": 0.2836684286594391,
      "learning_rate": 0.00023329825400621857,
      "loss": 0.051,
      "step": 5153
    },
    {
      "epoch": 0.722354590049054,
      "grad_norm": 0.8412325978279114,
      "learning_rate": 0.00023328390337239895,
      "loss": 0.2053,
      "step": 5154
    },
    {
      "epoch": 0.7224947442186405,
      "grad_norm": 0.26968541741371155,
      "learning_rate": 0.00023326955273857928,
      "loss": 0.0308,
      "step": 5155
    },
    {
      "epoch": 0.722634898388227,
      "grad_norm": 0.2725106477737427,
      "learning_rate": 0.0002332552021047596,
      "loss": 0.0686,
      "step": 5156
    },
    {
      "epoch": 0.7227750525578136,
      "grad_norm": 0.17711910605430603,
      "learning_rate": 0.00023324085147093996,
      "loss": 0.035,
      "step": 5157
    },
    {
      "epoch": 0.7229152067274002,
      "grad_norm": 0.3863382041454315,
      "learning_rate": 0.0002332265008371203,
      "loss": 0.1225,
      "step": 5158
    },
    {
      "epoch": 0.7230553608969866,
      "grad_norm": 0.2858009338378906,
      "learning_rate": 0.00023321215020330062,
      "loss": 0.058,
      "step": 5159
    },
    {
      "epoch": 0.7231955150665732,
      "grad_norm": 0.22802037000656128,
      "learning_rate": 0.00023319779956948097,
      "loss": 0.0554,
      "step": 5160
    },
    {
      "epoch": 0.7233356692361598,
      "grad_norm": 0.18020832538604736,
      "learning_rate": 0.0002331834489356613,
      "loss": 0.0665,
      "step": 5161
    },
    {
      "epoch": 0.7234758234057463,
      "grad_norm": 0.4329358637332916,
      "learning_rate": 0.00023316909830184163,
      "loss": 0.0692,
      "step": 5162
    },
    {
      "epoch": 0.7236159775753329,
      "grad_norm": 0.6758047938346863,
      "learning_rate": 0.00023315474766802196,
      "loss": 0.1063,
      "step": 5163
    },
    {
      "epoch": 0.7237561317449194,
      "grad_norm": 0.3999052941799164,
      "learning_rate": 0.00023314039703420234,
      "loss": 0.0778,
      "step": 5164
    },
    {
      "epoch": 0.7238962859145059,
      "grad_norm": 0.29249122738838196,
      "learning_rate": 0.00023312604640038267,
      "loss": 0.071,
      "step": 5165
    },
    {
      "epoch": 0.7240364400840925,
      "grad_norm": 0.3687693774700165,
      "learning_rate": 0.000233111695766563,
      "loss": 0.0963,
      "step": 5166
    },
    {
      "epoch": 0.7241765942536791,
      "grad_norm": 0.5186643600463867,
      "learning_rate": 0.00023309734513274336,
      "loss": 0.0781,
      "step": 5167
    },
    {
      "epoch": 0.7243167484232655,
      "grad_norm": 0.527142345905304,
      "learning_rate": 0.00023308299449892368,
      "loss": 0.2136,
      "step": 5168
    },
    {
      "epoch": 0.7244569025928521,
      "grad_norm": 0.265505313873291,
      "learning_rate": 0.000233068643865104,
      "loss": 0.0562,
      "step": 5169
    },
    {
      "epoch": 0.7245970567624387,
      "grad_norm": 0.6605541110038757,
      "learning_rate": 0.00023305429323128437,
      "loss": 0.087,
      "step": 5170
    },
    {
      "epoch": 0.7247372109320253,
      "grad_norm": 0.44356611371040344,
      "learning_rate": 0.0002330399425974647,
      "loss": 0.1449,
      "step": 5171
    },
    {
      "epoch": 0.7248773651016118,
      "grad_norm": 0.13862387835979462,
      "learning_rate": 0.00023302559196364503,
      "loss": 0.0244,
      "step": 5172
    },
    {
      "epoch": 0.7250175192711983,
      "grad_norm": 0.16604040563106537,
      "learning_rate": 0.00023301124132982538,
      "loss": 0.057,
      "step": 5173
    },
    {
      "epoch": 0.7251576734407849,
      "grad_norm": 0.8485308289527893,
      "learning_rate": 0.0002329968906960057,
      "loss": 0.1922,
      "step": 5174
    },
    {
      "epoch": 0.7252978276103714,
      "grad_norm": 0.45553362369537354,
      "learning_rate": 0.00023298254006218604,
      "loss": 0.0928,
      "step": 5175
    },
    {
      "epoch": 0.725437981779958,
      "grad_norm": 1.0841885805130005,
      "learning_rate": 0.00023296818942836642,
      "loss": 0.0822,
      "step": 5176
    },
    {
      "epoch": 0.7255781359495445,
      "grad_norm": 0.26723960041999817,
      "learning_rate": 0.00023295383879454675,
      "loss": 0.0839,
      "step": 5177
    },
    {
      "epoch": 0.725718290119131,
      "grad_norm": 0.21551693975925446,
      "learning_rate": 0.00023293948816072708,
      "loss": 0.0494,
      "step": 5178
    },
    {
      "epoch": 0.7258584442887176,
      "grad_norm": 0.5043818950653076,
      "learning_rate": 0.0002329251375269074,
      "loss": 0.1181,
      "step": 5179
    },
    {
      "epoch": 0.7259985984583042,
      "grad_norm": 0.3727339208126068,
      "learning_rate": 0.00023291078689308776,
      "loss": 0.0975,
      "step": 5180
    },
    {
      "epoch": 0.7261387526278906,
      "grad_norm": 0.2844492197036743,
      "learning_rate": 0.0002328964362592681,
      "loss": 0.0666,
      "step": 5181
    },
    {
      "epoch": 0.7262789067974772,
      "grad_norm": 0.4735451936721802,
      "learning_rate": 0.00023288208562544842,
      "loss": 0.1129,
      "step": 5182
    },
    {
      "epoch": 0.7264190609670638,
      "grad_norm": 0.40668800473213196,
      "learning_rate": 0.00023286773499162878,
      "loss": 0.0897,
      "step": 5183
    },
    {
      "epoch": 0.7265592151366503,
      "grad_norm": 0.42613571882247925,
      "learning_rate": 0.0002328533843578091,
      "loss": 0.2022,
      "step": 5184
    },
    {
      "epoch": 0.7266993693062369,
      "grad_norm": 0.3661457598209381,
      "learning_rate": 0.00023283903372398943,
      "loss": 0.0436,
      "step": 5185
    },
    {
      "epoch": 0.7268395234758234,
      "grad_norm": 0.3932240307331085,
      "learning_rate": 0.00023282468309016982,
      "loss": 0.1641,
      "step": 5186
    },
    {
      "epoch": 0.7269796776454099,
      "grad_norm": 0.7341818809509277,
      "learning_rate": 0.00023281033245635014,
      "loss": 0.0786,
      "step": 5187
    },
    {
      "epoch": 0.7271198318149965,
      "grad_norm": 0.5837560296058655,
      "learning_rate": 0.00023279598182253047,
      "loss": 0.0862,
      "step": 5188
    },
    {
      "epoch": 0.7272599859845831,
      "grad_norm": 0.16939066350460052,
      "learning_rate": 0.00023278163118871083,
      "loss": 0.0389,
      "step": 5189
    },
    {
      "epoch": 0.7274001401541695,
      "grad_norm": 0.39616861939430237,
      "learning_rate": 0.00023276728055489116,
      "loss": 0.0592,
      "step": 5190
    },
    {
      "epoch": 0.7275402943237561,
      "grad_norm": 0.36377978324890137,
      "learning_rate": 0.00023275292992107149,
      "loss": 0.0481,
      "step": 5191
    },
    {
      "epoch": 0.7276804484933427,
      "grad_norm": 0.4941023290157318,
      "learning_rate": 0.00023273857928725184,
      "loss": 0.079,
      "step": 5192
    },
    {
      "epoch": 0.7278206026629293,
      "grad_norm": 0.5489917397499084,
      "learning_rate": 0.00023272422865343217,
      "loss": 0.0712,
      "step": 5193
    },
    {
      "epoch": 0.7279607568325158,
      "grad_norm": 0.45941728353500366,
      "learning_rate": 0.0002327098780196125,
      "loss": 0.0605,
      "step": 5194
    },
    {
      "epoch": 0.7281009110021023,
      "grad_norm": 1.1872835159301758,
      "learning_rate": 0.00023269552738579288,
      "loss": 0.1343,
      "step": 5195
    },
    {
      "epoch": 0.7282410651716889,
      "grad_norm": 1.267572283744812,
      "learning_rate": 0.0002326811767519732,
      "loss": 0.1438,
      "step": 5196
    },
    {
      "epoch": 0.7283812193412754,
      "grad_norm": 2.0092835426330566,
      "learning_rate": 0.00023266682611815354,
      "loss": 0.2441,
      "step": 5197
    },
    {
      "epoch": 0.728521373510862,
      "grad_norm": 1.744664192199707,
      "learning_rate": 0.00023265247548433387,
      "loss": 0.1901,
      "step": 5198
    },
    {
      "epoch": 0.7286615276804485,
      "grad_norm": 1.714289665222168,
      "learning_rate": 0.00023263812485051422,
      "loss": 0.1596,
      "step": 5199
    },
    {
      "epoch": 0.728801681850035,
      "grad_norm": 1.742221713066101,
      "learning_rate": 0.00023262377421669455,
      "loss": 0.0317,
      "step": 5200
    },
    {
      "epoch": 0.7289418360196216,
      "grad_norm": 0.38922059535980225,
      "learning_rate": 0.00023260942358287488,
      "loss": 0.1005,
      "step": 5201
    },
    {
      "epoch": 0.7290819901892082,
      "grad_norm": 0.3114355206489563,
      "learning_rate": 0.00023259507294905524,
      "loss": 0.0453,
      "step": 5202
    },
    {
      "epoch": 0.7292221443587946,
      "grad_norm": 0.14320124685764313,
      "learning_rate": 0.00023258072231523556,
      "loss": 0.0184,
      "step": 5203
    },
    {
      "epoch": 0.7293622985283812,
      "grad_norm": 0.2641476094722748,
      "learning_rate": 0.0002325663716814159,
      "loss": 0.0489,
      "step": 5204
    },
    {
      "epoch": 0.7295024526979678,
      "grad_norm": 0.3066056966781616,
      "learning_rate": 0.00023255202104759625,
      "loss": 0.0626,
      "step": 5205
    },
    {
      "epoch": 0.7296426068675543,
      "grad_norm": 0.2508927285671234,
      "learning_rate": 0.00023253767041377658,
      "loss": 0.0669,
      "step": 5206
    },
    {
      "epoch": 0.7297827610371409,
      "grad_norm": 0.4991329312324524,
      "learning_rate": 0.0002325233197799569,
      "loss": 0.1117,
      "step": 5207
    },
    {
      "epoch": 0.7299229152067274,
      "grad_norm": 0.4104723036289215,
      "learning_rate": 0.0002325089691461373,
      "loss": 0.0921,
      "step": 5208
    },
    {
      "epoch": 0.7300630693763139,
      "grad_norm": 0.4139060080051422,
      "learning_rate": 0.00023249461851231762,
      "loss": 0.0487,
      "step": 5209
    },
    {
      "epoch": 0.7302032235459005,
      "grad_norm": 0.2513093054294586,
      "learning_rate": 0.00023248026787849794,
      "loss": 0.0474,
      "step": 5210
    },
    {
      "epoch": 0.7303433777154871,
      "grad_norm": 0.23586632311344147,
      "learning_rate": 0.0002324659172446783,
      "loss": 0.0431,
      "step": 5211
    },
    {
      "epoch": 0.7304835318850735,
      "grad_norm": 0.29323136806488037,
      "learning_rate": 0.00023245156661085863,
      "loss": 0.0724,
      "step": 5212
    },
    {
      "epoch": 0.7306236860546601,
      "grad_norm": 0.37320569157600403,
      "learning_rate": 0.00023243721597703896,
      "loss": 0.0551,
      "step": 5213
    },
    {
      "epoch": 0.7307638402242467,
      "grad_norm": 0.37209105491638184,
      "learning_rate": 0.00023242286534321929,
      "loss": 0.0716,
      "step": 5214
    },
    {
      "epoch": 0.7309039943938332,
      "grad_norm": 0.2746914029121399,
      "learning_rate": 0.00023240851470939964,
      "loss": 0.0581,
      "step": 5215
    },
    {
      "epoch": 0.7310441485634198,
      "grad_norm": 0.3433951437473297,
      "learning_rate": 0.00023239416407557997,
      "loss": 0.0885,
      "step": 5216
    },
    {
      "epoch": 0.7311843027330063,
      "grad_norm": 1.1792762279510498,
      "learning_rate": 0.0002323798134417603,
      "loss": 0.115,
      "step": 5217
    },
    {
      "epoch": 0.7313244569025928,
      "grad_norm": 0.339902400970459,
      "learning_rate": 0.00023236546280794068,
      "loss": 0.0783,
      "step": 5218
    },
    {
      "epoch": 0.7314646110721794,
      "grad_norm": 0.4927713871002197,
      "learning_rate": 0.000232351112174121,
      "loss": 0.069,
      "step": 5219
    },
    {
      "epoch": 0.731604765241766,
      "grad_norm": 0.6773391366004944,
      "learning_rate": 0.00023233676154030134,
      "loss": 0.1087,
      "step": 5220
    },
    {
      "epoch": 0.7317449194113524,
      "grad_norm": 0.37482526898384094,
      "learning_rate": 0.0002323224109064817,
      "loss": 0.0865,
      "step": 5221
    },
    {
      "epoch": 0.731885073580939,
      "grad_norm": 0.2873540222644806,
      "learning_rate": 0.00023230806027266202,
      "loss": 0.0574,
      "step": 5222
    },
    {
      "epoch": 0.7320252277505256,
      "grad_norm": 0.2761348783969879,
      "learning_rate": 0.00023229370963884235,
      "loss": 0.0577,
      "step": 5223
    },
    {
      "epoch": 0.7321653819201122,
      "grad_norm": 0.8928381204605103,
      "learning_rate": 0.0002322793590050227,
      "loss": 0.2217,
      "step": 5224
    },
    {
      "epoch": 0.7323055360896986,
      "grad_norm": 0.3465178906917572,
      "learning_rate": 0.00023226500837120304,
      "loss": 0.1032,
      "step": 5225
    },
    {
      "epoch": 0.7324456902592852,
      "grad_norm": 0.3352315127849579,
      "learning_rate": 0.00023225065773738336,
      "loss": 0.1259,
      "step": 5226
    },
    {
      "epoch": 0.7325858444288718,
      "grad_norm": 0.17192241549491882,
      "learning_rate": 0.00023223630710356375,
      "loss": 0.0459,
      "step": 5227
    },
    {
      "epoch": 0.7327259985984583,
      "grad_norm": 0.22748470306396484,
      "learning_rate": 0.00023222195646974408,
      "loss": 0.0428,
      "step": 5228
    },
    {
      "epoch": 0.7328661527680449,
      "grad_norm": 0.4069025218486786,
      "learning_rate": 0.0002322076058359244,
      "loss": 0.1006,
      "step": 5229
    },
    {
      "epoch": 0.7330063069376314,
      "grad_norm": 0.24883797764778137,
      "learning_rate": 0.00023219325520210476,
      "loss": 0.0569,
      "step": 5230
    },
    {
      "epoch": 0.7331464611072179,
      "grad_norm": 0.26614904403686523,
      "learning_rate": 0.0002321789045682851,
      "loss": 0.0874,
      "step": 5231
    },
    {
      "epoch": 0.7332866152768045,
      "grad_norm": 0.30397799611091614,
      "learning_rate": 0.00023216455393446542,
      "loss": 0.0634,
      "step": 5232
    },
    {
      "epoch": 0.7334267694463911,
      "grad_norm": 0.5899048447608948,
      "learning_rate": 0.00023215020330064575,
      "loss": 0.0459,
      "step": 5233
    },
    {
      "epoch": 0.7335669236159775,
      "grad_norm": 0.26289013028144836,
      "learning_rate": 0.0002321358526668261,
      "loss": 0.0306,
      "step": 5234
    },
    {
      "epoch": 0.7337070777855641,
      "grad_norm": 0.3717977702617645,
      "learning_rate": 0.00023212150203300643,
      "loss": 0.0959,
      "step": 5235
    },
    {
      "epoch": 0.7338472319551507,
      "grad_norm": 0.13371019065380096,
      "learning_rate": 0.00023210715139918676,
      "loss": 0.0447,
      "step": 5236
    },
    {
      "epoch": 0.7339873861247372,
      "grad_norm": 0.1784622073173523,
      "learning_rate": 0.00023209280076536711,
      "loss": 0.0331,
      "step": 5237
    },
    {
      "epoch": 0.7341275402943238,
      "grad_norm": 0.40873488783836365,
      "learning_rate": 0.00023207845013154744,
      "loss": 0.0861,
      "step": 5238
    },
    {
      "epoch": 0.7342676944639103,
      "grad_norm": 0.45802634954452515,
      "learning_rate": 0.00023206409949772777,
      "loss": 0.1016,
      "step": 5239
    },
    {
      "epoch": 0.7344078486334968,
      "grad_norm": 0.29189273715019226,
      "learning_rate": 0.00023204974886390815,
      "loss": 0.0551,
      "step": 5240
    },
    {
      "epoch": 0.7345480028030834,
      "grad_norm": 0.5538116097450256,
      "learning_rate": 0.00023203539823008848,
      "loss": 0.0858,
      "step": 5241
    },
    {
      "epoch": 0.73468815697267,
      "grad_norm": 1.4417089223861694,
      "learning_rate": 0.0002320210475962688,
      "loss": 0.3018,
      "step": 5242
    },
    {
      "epoch": 0.7348283111422564,
      "grad_norm": 0.3187911808490753,
      "learning_rate": 0.00023200669696244917,
      "loss": 0.0513,
      "step": 5243
    },
    {
      "epoch": 0.734968465311843,
      "grad_norm": 0.6681216359138489,
      "learning_rate": 0.0002319923463286295,
      "loss": 0.1883,
      "step": 5244
    },
    {
      "epoch": 0.7351086194814296,
      "grad_norm": 0.24455589056015015,
      "learning_rate": 0.00023197799569480982,
      "loss": 0.0391,
      "step": 5245
    },
    {
      "epoch": 0.7352487736510162,
      "grad_norm": 0.6566860675811768,
      "learning_rate": 0.00023196364506099018,
      "loss": 0.0577,
      "step": 5246
    },
    {
      "epoch": 0.7353889278206026,
      "grad_norm": 0.9834797382354736,
      "learning_rate": 0.0002319492944271705,
      "loss": 0.1566,
      "step": 5247
    },
    {
      "epoch": 0.7355290819901892,
      "grad_norm": 0.8370198011398315,
      "learning_rate": 0.00023193494379335084,
      "loss": 0.061,
      "step": 5248
    },
    {
      "epoch": 0.7356692361597758,
      "grad_norm": 1.8041143417358398,
      "learning_rate": 0.00023192059315953117,
      "loss": 0.1527,
      "step": 5249
    },
    {
      "epoch": 0.7358093903293623,
      "grad_norm": 3.4857828617095947,
      "learning_rate": 0.00023190624252571155,
      "loss": 0.2805,
      "step": 5250
    },
    {
      "epoch": 0.7359495444989489,
      "grad_norm": 0.11467088758945465,
      "learning_rate": 0.00023189189189189188,
      "loss": 0.0156,
      "step": 5251
    },
    {
      "epoch": 0.7360896986685354,
      "grad_norm": 0.4019100069999695,
      "learning_rate": 0.0002318775412580722,
      "loss": 0.0403,
      "step": 5252
    },
    {
      "epoch": 0.7362298528381219,
      "grad_norm": 0.46338367462158203,
      "learning_rate": 0.00023186319062425256,
      "loss": 0.0979,
      "step": 5253
    },
    {
      "epoch": 0.7363700070077085,
      "grad_norm": 0.26703977584838867,
      "learning_rate": 0.0002318488399904329,
      "loss": 0.0542,
      "step": 5254
    },
    {
      "epoch": 0.7365101611772951,
      "grad_norm": 0.3099108040332794,
      "learning_rate": 0.00023183448935661322,
      "loss": 0.0848,
      "step": 5255
    },
    {
      "epoch": 0.7366503153468815,
      "grad_norm": 0.19513411819934845,
      "learning_rate": 0.00023182013872279357,
      "loss": 0.0199,
      "step": 5256
    },
    {
      "epoch": 0.7367904695164681,
      "grad_norm": 0.7094303965568542,
      "learning_rate": 0.0002318057880889739,
      "loss": 0.063,
      "step": 5257
    },
    {
      "epoch": 0.7369306236860547,
      "grad_norm": 0.3076759874820709,
      "learning_rate": 0.00023179143745515423,
      "loss": 0.0928,
      "step": 5258
    },
    {
      "epoch": 0.7370707778556412,
      "grad_norm": 0.1262865662574768,
      "learning_rate": 0.00023177708682133461,
      "loss": 0.0434,
      "step": 5259
    },
    {
      "epoch": 0.7372109320252278,
      "grad_norm": 0.4564856290817261,
      "learning_rate": 0.00023176273618751494,
      "loss": 0.0648,
      "step": 5260
    },
    {
      "epoch": 0.7373510861948143,
      "grad_norm": 0.42936354875564575,
      "learning_rate": 0.00023174838555369527,
      "loss": 0.0749,
      "step": 5261
    },
    {
      "epoch": 0.7374912403644008,
      "grad_norm": 0.1837492734193802,
      "learning_rate": 0.00023173403491987563,
      "loss": 0.0267,
      "step": 5262
    },
    {
      "epoch": 0.7376313945339874,
      "grad_norm": 0.5880001783370972,
      "learning_rate": 0.00023171968428605595,
      "loss": 0.1105,
      "step": 5263
    },
    {
      "epoch": 0.737771548703574,
      "grad_norm": 0.4614347815513611,
      "learning_rate": 0.00023170533365223628,
      "loss": 0.049,
      "step": 5264
    },
    {
      "epoch": 0.7379117028731604,
      "grad_norm": 0.4977606236934662,
      "learning_rate": 0.00023169098301841664,
      "loss": 0.0915,
      "step": 5265
    },
    {
      "epoch": 0.738051857042747,
      "grad_norm": 0.2489607334136963,
      "learning_rate": 0.00023167663238459697,
      "loss": 0.0689,
      "step": 5266
    },
    {
      "epoch": 0.7381920112123336,
      "grad_norm": 0.35282954573631287,
      "learning_rate": 0.0002316622817507773,
      "loss": 0.0656,
      "step": 5267
    },
    {
      "epoch": 0.7383321653819201,
      "grad_norm": 0.4747740924358368,
      "learning_rate": 0.00023164793111695762,
      "loss": 0.1195,
      "step": 5268
    },
    {
      "epoch": 0.7384723195515066,
      "grad_norm": 0.5125481486320496,
      "learning_rate": 0.00023163358048313798,
      "loss": 0.1129,
      "step": 5269
    },
    {
      "epoch": 0.7386124737210932,
      "grad_norm": 0.3013273775577545,
      "learning_rate": 0.0002316192298493183,
      "loss": 0.0854,
      "step": 5270
    },
    {
      "epoch": 0.7387526278906797,
      "grad_norm": 0.4458228051662445,
      "learning_rate": 0.00023160487921549864,
      "loss": 0.0338,
      "step": 5271
    },
    {
      "epoch": 0.7388927820602663,
      "grad_norm": 0.8556686639785767,
      "learning_rate": 0.00023159052858167902,
      "loss": 0.1236,
      "step": 5272
    },
    {
      "epoch": 0.7390329362298529,
      "grad_norm": 0.5918168425559998,
      "learning_rate": 0.00023157617794785935,
      "loss": 0.1524,
      "step": 5273
    },
    {
      "epoch": 0.7391730903994393,
      "grad_norm": 0.45965802669525146,
      "learning_rate": 0.00023156182731403968,
      "loss": 0.1137,
      "step": 5274
    },
    {
      "epoch": 0.7393132445690259,
      "grad_norm": 0.42896077036857605,
      "learning_rate": 0.00023154747668022003,
      "loss": 0.064,
      "step": 5275
    },
    {
      "epoch": 0.7394533987386125,
      "grad_norm": 0.4948711395263672,
      "learning_rate": 0.00023153312604640036,
      "loss": 0.1047,
      "step": 5276
    },
    {
      "epoch": 0.7395935529081991,
      "grad_norm": 0.2017544060945511,
      "learning_rate": 0.0002315187754125807,
      "loss": 0.0255,
      "step": 5277
    },
    {
      "epoch": 0.7397337070777855,
      "grad_norm": 0.6741991639137268,
      "learning_rate": 0.00023150442477876105,
      "loss": 0.0827,
      "step": 5278
    },
    {
      "epoch": 0.7398738612473721,
      "grad_norm": 0.26342734694480896,
      "learning_rate": 0.00023149007414494137,
      "loss": 0.0558,
      "step": 5279
    },
    {
      "epoch": 0.7400140154169587,
      "grad_norm": 0.288428395986557,
      "learning_rate": 0.0002314757235111217,
      "loss": 0.054,
      "step": 5280
    },
    {
      "epoch": 0.7401541695865452,
      "grad_norm": 0.20877547562122345,
      "learning_rate": 0.00023146137287730209,
      "loss": 0.0288,
      "step": 5281
    },
    {
      "epoch": 0.7402943237561318,
      "grad_norm": 0.48135656118392944,
      "learning_rate": 0.00023144702224348241,
      "loss": 0.0815,
      "step": 5282
    },
    {
      "epoch": 0.7404344779257183,
      "grad_norm": 0.23357640206813812,
      "learning_rate": 0.00023143267160966274,
      "loss": 0.1051,
      "step": 5283
    },
    {
      "epoch": 0.7405746320953048,
      "grad_norm": 0.46737703680992126,
      "learning_rate": 0.00023141832097584307,
      "loss": 0.1595,
      "step": 5284
    },
    {
      "epoch": 0.7407147862648914,
      "grad_norm": 0.13414525985717773,
      "learning_rate": 0.00023140397034202343,
      "loss": 0.0157,
      "step": 5285
    },
    {
      "epoch": 0.740854940434478,
      "grad_norm": 1.2838695049285889,
      "learning_rate": 0.00023138961970820376,
      "loss": 0.1122,
      "step": 5286
    },
    {
      "epoch": 0.7409950946040644,
      "grad_norm": 0.4931929409503937,
      "learning_rate": 0.00023137526907438408,
      "loss": 0.1286,
      "step": 5287
    },
    {
      "epoch": 0.741135248773651,
      "grad_norm": 0.6304449439048767,
      "learning_rate": 0.00023136091844056444,
      "loss": 0.1813,
      "step": 5288
    },
    {
      "epoch": 0.7412754029432376,
      "grad_norm": 0.3236653208732605,
      "learning_rate": 0.00023134656780674477,
      "loss": 0.0727,
      "step": 5289
    },
    {
      "epoch": 0.7414155571128241,
      "grad_norm": 0.41367143392562866,
      "learning_rate": 0.0002313322171729251,
      "loss": 0.1643,
      "step": 5290
    },
    {
      "epoch": 0.7415557112824106,
      "grad_norm": 0.5044955015182495,
      "learning_rate": 0.00023131786653910548,
      "loss": 0.0814,
      "step": 5291
    },
    {
      "epoch": 0.7416958654519972,
      "grad_norm": 0.36461788415908813,
      "learning_rate": 0.0002313035159052858,
      "loss": 0.0483,
      "step": 5292
    },
    {
      "epoch": 0.7418360196215837,
      "grad_norm": 1.1738500595092773,
      "learning_rate": 0.00023128916527146614,
      "loss": 0.1703,
      "step": 5293
    },
    {
      "epoch": 0.7419761737911703,
      "grad_norm": 0.4505654573440552,
      "learning_rate": 0.0002312748146376465,
      "loss": 0.041,
      "step": 5294
    },
    {
      "epoch": 0.7421163279607569,
      "grad_norm": 1.0805988311767578,
      "learning_rate": 0.00023126046400382682,
      "loss": 0.145,
      "step": 5295
    },
    {
      "epoch": 0.7422564821303433,
      "grad_norm": 0.17228636145591736,
      "learning_rate": 0.00023124611337000715,
      "loss": 0.0308,
      "step": 5296
    },
    {
      "epoch": 0.7423966362999299,
      "grad_norm": 0.41310766339302063,
      "learning_rate": 0.0002312317627361875,
      "loss": 0.0923,
      "step": 5297
    },
    {
      "epoch": 0.7425367904695165,
      "grad_norm": 0.2606530785560608,
      "learning_rate": 0.00023121741210236783,
      "loss": 0.0339,
      "step": 5298
    },
    {
      "epoch": 0.742676944639103,
      "grad_norm": 1.6337460279464722,
      "learning_rate": 0.00023120306146854816,
      "loss": 0.4906,
      "step": 5299
    },
    {
      "epoch": 0.7428170988086895,
      "grad_norm": 0.632016122341156,
      "learning_rate": 0.00023118871083472852,
      "loss": 0.0742,
      "step": 5300
    },
    {
      "epoch": 0.7429572529782761,
      "grad_norm": 0.1908397525548935,
      "learning_rate": 0.00023117436020090885,
      "loss": 0.0519,
      "step": 5301
    },
    {
      "epoch": 0.7430974071478627,
      "grad_norm": 0.5935701727867126,
      "learning_rate": 0.00023116000956708918,
      "loss": 0.0747,
      "step": 5302
    },
    {
      "epoch": 0.7432375613174492,
      "grad_norm": 0.17464745044708252,
      "learning_rate": 0.0002311456589332695,
      "loss": 0.048,
      "step": 5303
    },
    {
      "epoch": 0.7433777154870357,
      "grad_norm": 0.3084712028503418,
      "learning_rate": 0.0002311313082994499,
      "loss": 0.112,
      "step": 5304
    },
    {
      "epoch": 0.7435178696566223,
      "grad_norm": 0.19376736879348755,
      "learning_rate": 0.00023111695766563022,
      "loss": 0.0314,
      "step": 5305
    },
    {
      "epoch": 0.7436580238262088,
      "grad_norm": 0.2405315488576889,
      "learning_rate": 0.00023110260703181054,
      "loss": 0.0623,
      "step": 5306
    },
    {
      "epoch": 0.7437981779957954,
      "grad_norm": 0.6986135244369507,
      "learning_rate": 0.0002310882563979909,
      "loss": 0.1593,
      "step": 5307
    },
    {
      "epoch": 0.743938332165382,
      "grad_norm": 0.4332660138607025,
      "learning_rate": 0.00023107390576417123,
      "loss": 0.0267,
      "step": 5308
    },
    {
      "epoch": 0.7440784863349684,
      "grad_norm": 0.14004847407341003,
      "learning_rate": 0.00023105955513035156,
      "loss": 0.0421,
      "step": 5309
    },
    {
      "epoch": 0.744218640504555,
      "grad_norm": 0.4322187304496765,
      "learning_rate": 0.0002310452044965319,
      "loss": 0.1039,
      "step": 5310
    },
    {
      "epoch": 0.7443587946741416,
      "grad_norm": 0.21246327459812164,
      "learning_rate": 0.00023103085386271224,
      "loss": 0.0348,
      "step": 5311
    },
    {
      "epoch": 0.7444989488437281,
      "grad_norm": 0.428611695766449,
      "learning_rate": 0.00023101650322889257,
      "loss": 0.0697,
      "step": 5312
    },
    {
      "epoch": 0.7446391030133146,
      "grad_norm": 0.9361205697059631,
      "learning_rate": 0.00023100215259507295,
      "loss": 0.0534,
      "step": 5313
    },
    {
      "epoch": 0.7447792571829012,
      "grad_norm": 0.2423429787158966,
      "learning_rate": 0.00023098780196125328,
      "loss": 0.0695,
      "step": 5314
    },
    {
      "epoch": 0.7449194113524877,
      "grad_norm": 0.9934574365615845,
      "learning_rate": 0.0002309734513274336,
      "loss": 0.1319,
      "step": 5315
    },
    {
      "epoch": 0.7450595655220743,
      "grad_norm": 0.34104806184768677,
      "learning_rate": 0.00023095910069361396,
      "loss": 0.0749,
      "step": 5316
    },
    {
      "epoch": 0.7451997196916609,
      "grad_norm": 0.430503249168396,
      "learning_rate": 0.0002309447500597943,
      "loss": 0.058,
      "step": 5317
    },
    {
      "epoch": 0.7453398738612473,
      "grad_norm": 0.5390800833702087,
      "learning_rate": 0.00023093039942597462,
      "loss": 0.0757,
      "step": 5318
    },
    {
      "epoch": 0.7454800280308339,
      "grad_norm": 0.23335863649845123,
      "learning_rate": 0.00023091604879215495,
      "loss": 0.0495,
      "step": 5319
    },
    {
      "epoch": 0.7456201822004205,
      "grad_norm": 0.31612691283226013,
      "learning_rate": 0.0002309016981583353,
      "loss": 0.0575,
      "step": 5320
    },
    {
      "epoch": 0.745760336370007,
      "grad_norm": 0.9983062744140625,
      "learning_rate": 0.00023088734752451563,
      "loss": 0.0886,
      "step": 5321
    },
    {
      "epoch": 0.7459004905395935,
      "grad_norm": 1.0260056257247925,
      "learning_rate": 0.00023087299689069596,
      "loss": 0.1082,
      "step": 5322
    },
    {
      "epoch": 0.7460406447091801,
      "grad_norm": 0.3832590579986572,
      "learning_rate": 0.00023085864625687635,
      "loss": 0.0932,
      "step": 5323
    },
    {
      "epoch": 0.7461807988787666,
      "grad_norm": 0.3209840953350067,
      "learning_rate": 0.00023084429562305667,
      "loss": 0.0382,
      "step": 5324
    },
    {
      "epoch": 0.7463209530483532,
      "grad_norm": 0.616166353225708,
      "learning_rate": 0.000230829944989237,
      "loss": 0.1083,
      "step": 5325
    },
    {
      "epoch": 0.7464611072179397,
      "grad_norm": 0.26599881052970886,
      "learning_rate": 0.00023081559435541736,
      "loss": 0.0496,
      "step": 5326
    },
    {
      "epoch": 0.7466012613875262,
      "grad_norm": 0.27324193716049194,
      "learning_rate": 0.0002308012437215977,
      "loss": 0.0991,
      "step": 5327
    },
    {
      "epoch": 0.7467414155571128,
      "grad_norm": 0.4607287347316742,
      "learning_rate": 0.00023078689308777802,
      "loss": 0.0606,
      "step": 5328
    },
    {
      "epoch": 0.7468815697266994,
      "grad_norm": 0.3601208031177521,
      "learning_rate": 0.00023077254245395837,
      "loss": 0.0359,
      "step": 5329
    },
    {
      "epoch": 0.747021723896286,
      "grad_norm": 0.5417634844779968,
      "learning_rate": 0.0002307581918201387,
      "loss": 0.1147,
      "step": 5330
    },
    {
      "epoch": 0.7471618780658724,
      "grad_norm": 0.6748397946357727,
      "learning_rate": 0.00023074384118631903,
      "loss": 0.1519,
      "step": 5331
    },
    {
      "epoch": 0.747302032235459,
      "grad_norm": 0.5206419229507446,
      "learning_rate": 0.00023072949055249938,
      "loss": 0.1492,
      "step": 5332
    },
    {
      "epoch": 0.7474421864050456,
      "grad_norm": 0.29515206813812256,
      "learning_rate": 0.0002307151399186797,
      "loss": 0.0408,
      "step": 5333
    },
    {
      "epoch": 0.7475823405746321,
      "grad_norm": 0.5675029754638672,
      "learning_rate": 0.00023070078928486004,
      "loss": 0.0625,
      "step": 5334
    },
    {
      "epoch": 0.7477224947442186,
      "grad_norm": 0.37335205078125,
      "learning_rate": 0.00023068643865104042,
      "loss": 0.0673,
      "step": 5335
    },
    {
      "epoch": 0.7478626489138052,
      "grad_norm": 0.39389199018478394,
      "learning_rate": 0.00023067208801722075,
      "loss": 0.0939,
      "step": 5336
    },
    {
      "epoch": 0.7480028030833917,
      "grad_norm": 0.536999523639679,
      "learning_rate": 0.00023065773738340108,
      "loss": 0.1173,
      "step": 5337
    },
    {
      "epoch": 0.7481429572529783,
      "grad_norm": 0.23542311787605286,
      "learning_rate": 0.0002306433867495814,
      "loss": 0.0511,
      "step": 5338
    },
    {
      "epoch": 0.7482831114225649,
      "grad_norm": 0.45447593927383423,
      "learning_rate": 0.00023062903611576177,
      "loss": 0.1394,
      "step": 5339
    },
    {
      "epoch": 0.7484232655921513,
      "grad_norm": 0.35767289996147156,
      "learning_rate": 0.0002306146854819421,
      "loss": 0.0605,
      "step": 5340
    },
    {
      "epoch": 0.7485634197617379,
      "grad_norm": 0.6906814575195312,
      "learning_rate": 0.00023060033484812242,
      "loss": 0.1471,
      "step": 5341
    },
    {
      "epoch": 0.7487035739313245,
      "grad_norm": 0.4207620918750763,
      "learning_rate": 0.00023058598421430278,
      "loss": 0.0527,
      "step": 5342
    },
    {
      "epoch": 0.748843728100911,
      "grad_norm": 0.6929364204406738,
      "learning_rate": 0.0002305716335804831,
      "loss": 0.1169,
      "step": 5343
    },
    {
      "epoch": 0.7489838822704975,
      "grad_norm": 0.3982695937156677,
      "learning_rate": 0.00023055728294666344,
      "loss": 0.0811,
      "step": 5344
    },
    {
      "epoch": 0.7491240364400841,
      "grad_norm": 0.3504420518875122,
      "learning_rate": 0.00023054293231284382,
      "loss": 0.0756,
      "step": 5345
    },
    {
      "epoch": 0.7492641906096706,
      "grad_norm": 0.4439573287963867,
      "learning_rate": 0.00023052858167902415,
      "loss": 0.0796,
      "step": 5346
    },
    {
      "epoch": 0.7494043447792572,
      "grad_norm": 0.6658779978752136,
      "learning_rate": 0.00023051423104520448,
      "loss": 0.1295,
      "step": 5347
    },
    {
      "epoch": 0.7495444989488437,
      "grad_norm": 1.8031822443008423,
      "learning_rate": 0.00023049988041138483,
      "loss": 0.2205,
      "step": 5348
    },
    {
      "epoch": 0.7496846531184302,
      "grad_norm": 1.138259768486023,
      "learning_rate": 0.00023048552977756516,
      "loss": 0.0508,
      "step": 5349
    },
    {
      "epoch": 0.7498248072880168,
      "grad_norm": 2.118530511856079,
      "learning_rate": 0.0002304711791437455,
      "loss": 0.3368,
      "step": 5350
    },
    {
      "epoch": 0.7499649614576034,
      "grad_norm": 0.713670551776886,
      "learning_rate": 0.00023045682850992584,
      "loss": 0.1128,
      "step": 5351
    },
    {
      "epoch": 0.75010511562719,
      "grad_norm": 0.5301555395126343,
      "learning_rate": 0.00023044247787610617,
      "loss": 0.0871,
      "step": 5352
    },
    {
      "epoch": 0.7502452697967764,
      "grad_norm": 0.4175166189670563,
      "learning_rate": 0.0002304281272422865,
      "loss": 0.0838,
      "step": 5353
    },
    {
      "epoch": 0.750385423966363,
      "grad_norm": 0.4329306185245514,
      "learning_rate": 0.00023041377660846683,
      "loss": 0.1172,
      "step": 5354
    },
    {
      "epoch": 0.7505255781359496,
      "grad_norm": 0.5823131203651428,
      "learning_rate": 0.0002303994259746472,
      "loss": 0.0829,
      "step": 5355
    },
    {
      "epoch": 0.7506657323055361,
      "grad_norm": 0.43705475330352783,
      "learning_rate": 0.00023038507534082754,
      "loss": 0.0978,
      "step": 5356
    },
    {
      "epoch": 0.7508058864751226,
      "grad_norm": 0.3040735721588135,
      "learning_rate": 0.00023037072470700787,
      "loss": 0.0573,
      "step": 5357
    },
    {
      "epoch": 0.7509460406447092,
      "grad_norm": 0.35153764486312866,
      "learning_rate": 0.00023035637407318823,
      "loss": 0.1054,
      "step": 5358
    },
    {
      "epoch": 0.7510861948142957,
      "grad_norm": 0.3424082398414612,
      "learning_rate": 0.00023034202343936855,
      "loss": 0.0516,
      "step": 5359
    },
    {
      "epoch": 0.7512263489838823,
      "grad_norm": 0.2389257550239563,
      "learning_rate": 0.00023032767280554888,
      "loss": 0.065,
      "step": 5360
    },
    {
      "epoch": 0.7513665031534689,
      "grad_norm": 0.3950711190700531,
      "learning_rate": 0.00023031332217172924,
      "loss": 0.1427,
      "step": 5361
    },
    {
      "epoch": 0.7515066573230553,
      "grad_norm": 0.4938294589519501,
      "learning_rate": 0.00023029897153790957,
      "loss": 0.121,
      "step": 5362
    },
    {
      "epoch": 0.7516468114926419,
      "grad_norm": 0.3829919099807739,
      "learning_rate": 0.0002302846209040899,
      "loss": 0.0583,
      "step": 5363
    },
    {
      "epoch": 0.7517869656622285,
      "grad_norm": 0.3149269223213196,
      "learning_rate": 0.00023027027027027025,
      "loss": 0.0519,
      "step": 5364
    },
    {
      "epoch": 0.751927119831815,
      "grad_norm": 0.5110300183296204,
      "learning_rate": 0.00023025591963645058,
      "loss": 0.0546,
      "step": 5365
    },
    {
      "epoch": 0.7520672740014015,
      "grad_norm": 0.1330219954252243,
      "learning_rate": 0.0002302415690026309,
      "loss": 0.0249,
      "step": 5366
    },
    {
      "epoch": 0.7522074281709881,
      "grad_norm": 0.3704589307308197,
      "learning_rate": 0.0002302272183688113,
      "loss": 0.0626,
      "step": 5367
    },
    {
      "epoch": 0.7523475823405746,
      "grad_norm": 0.3934774398803711,
      "learning_rate": 0.00023021286773499162,
      "loss": 0.0594,
      "step": 5368
    },
    {
      "epoch": 0.7524877365101612,
      "grad_norm": 1.109682321548462,
      "learning_rate": 0.00023019851710117195,
      "loss": 0.0687,
      "step": 5369
    },
    {
      "epoch": 0.7526278906797477,
      "grad_norm": 0.37943321466445923,
      "learning_rate": 0.0002301841664673523,
      "loss": 0.0525,
      "step": 5370
    },
    {
      "epoch": 0.7527680448493342,
      "grad_norm": 1.3197342157363892,
      "learning_rate": 0.00023016981583353263,
      "loss": 0.1265,
      "step": 5371
    },
    {
      "epoch": 0.7529081990189208,
      "grad_norm": 0.31257906556129456,
      "learning_rate": 0.00023015546519971296,
      "loss": 0.0496,
      "step": 5372
    },
    {
      "epoch": 0.7530483531885074,
      "grad_norm": 0.23621530830860138,
      "learning_rate": 0.0002301411145658933,
      "loss": 0.0625,
      "step": 5373
    },
    {
      "epoch": 0.753188507358094,
      "grad_norm": 0.7305300235748291,
      "learning_rate": 0.00023012676393207364,
      "loss": 0.0657,
      "step": 5374
    },
    {
      "epoch": 0.7533286615276804,
      "grad_norm": 0.2700155973434448,
      "learning_rate": 0.00023011241329825397,
      "loss": 0.0505,
      "step": 5375
    },
    {
      "epoch": 0.753468815697267,
      "grad_norm": 0.597258985042572,
      "learning_rate": 0.0002300980626644343,
      "loss": 0.1141,
      "step": 5376
    },
    {
      "epoch": 0.7536089698668536,
      "grad_norm": 0.41933366656303406,
      "learning_rate": 0.00023008371203061468,
      "loss": 0.062,
      "step": 5377
    },
    {
      "epoch": 0.7537491240364401,
      "grad_norm": 0.49484869837760925,
      "learning_rate": 0.000230069361396795,
      "loss": 0.0897,
      "step": 5378
    },
    {
      "epoch": 0.7538892782060266,
      "grad_norm": 0.9962931871414185,
      "learning_rate": 0.00023005501076297534,
      "loss": 0.0775,
      "step": 5379
    },
    {
      "epoch": 0.7540294323756132,
      "grad_norm": 0.40183693170547485,
      "learning_rate": 0.0002300406601291557,
      "loss": 0.1059,
      "step": 5380
    },
    {
      "epoch": 0.7541695865451997,
      "grad_norm": 0.3148702085018158,
      "learning_rate": 0.00023002630949533603,
      "loss": 0.0391,
      "step": 5381
    },
    {
      "epoch": 0.7543097407147863,
      "grad_norm": 0.46502649784088135,
      "learning_rate": 0.00023001195886151635,
      "loss": 0.0662,
      "step": 5382
    },
    {
      "epoch": 0.7544498948843729,
      "grad_norm": 0.44700387120246887,
      "learning_rate": 0.0002299976082276967,
      "loss": 0.1109,
      "step": 5383
    },
    {
      "epoch": 0.7545900490539593,
      "grad_norm": 0.5130743384361267,
      "learning_rate": 0.00022998325759387704,
      "loss": 0.1,
      "step": 5384
    },
    {
      "epoch": 0.7547302032235459,
      "grad_norm": 0.5566855669021606,
      "learning_rate": 0.00022996890696005737,
      "loss": 0.1291,
      "step": 5385
    },
    {
      "epoch": 0.7548703573931325,
      "grad_norm": 0.7589330673217773,
      "learning_rate": 0.00022995455632623775,
      "loss": 0.0861,
      "step": 5386
    },
    {
      "epoch": 0.755010511562719,
      "grad_norm": 0.7593021988868713,
      "learning_rate": 0.00022994020569241808,
      "loss": 0.1528,
      "step": 5387
    },
    {
      "epoch": 0.7551506657323055,
      "grad_norm": 0.2339348942041397,
      "learning_rate": 0.0002299258550585984,
      "loss": 0.0634,
      "step": 5388
    },
    {
      "epoch": 0.7552908199018921,
      "grad_norm": 0.5440716743469238,
      "learning_rate": 0.00022991150442477874,
      "loss": 0.0472,
      "step": 5389
    },
    {
      "epoch": 0.7554309740714786,
      "grad_norm": 1.0329618453979492,
      "learning_rate": 0.0002298971537909591,
      "loss": 0.1092,
      "step": 5390
    },
    {
      "epoch": 0.7555711282410652,
      "grad_norm": 0.4245467483997345,
      "learning_rate": 0.00022988280315713942,
      "loss": 0.0733,
      "step": 5391
    },
    {
      "epoch": 0.7557112824106517,
      "grad_norm": 0.5720781683921814,
      "learning_rate": 0.00022986845252331975,
      "loss": 0.1889,
      "step": 5392
    },
    {
      "epoch": 0.7558514365802382,
      "grad_norm": 0.22144603729248047,
      "learning_rate": 0.0002298541018895001,
      "loss": 0.0294,
      "step": 5393
    },
    {
      "epoch": 0.7559915907498248,
      "grad_norm": 2.2244887351989746,
      "learning_rate": 0.00022983975125568043,
      "loss": 0.1697,
      "step": 5394
    },
    {
      "epoch": 0.7561317449194114,
      "grad_norm": 0.7043256759643555,
      "learning_rate": 0.00022982540062186076,
      "loss": 0.0496,
      "step": 5395
    },
    {
      "epoch": 0.756271899088998,
      "grad_norm": 1.7605684995651245,
      "learning_rate": 0.00022981104998804112,
      "loss": 0.0966,
      "step": 5396
    },
    {
      "epoch": 0.7564120532585844,
      "grad_norm": 1.0889674425125122,
      "learning_rate": 0.00022979669935422145,
      "loss": 0.0912,
      "step": 5397
    },
    {
      "epoch": 0.756552207428171,
      "grad_norm": 2.1754627227783203,
      "learning_rate": 0.00022978234872040177,
      "loss": 0.2401,
      "step": 5398
    },
    {
      "epoch": 0.7566923615977575,
      "grad_norm": 1.9431825876235962,
      "learning_rate": 0.00022976799808658216,
      "loss": 0.2546,
      "step": 5399
    },
    {
      "epoch": 0.7568325157673441,
      "grad_norm": 3.2252488136291504,
      "learning_rate": 0.00022975364745276249,
      "loss": 0.2302,
      "step": 5400
    },
    {
      "epoch": 0.7569726699369306,
      "grad_norm": 0.22456753253936768,
      "learning_rate": 0.00022973929681894281,
      "loss": 0.0507,
      "step": 5401
    },
    {
      "epoch": 0.7571128241065171,
      "grad_norm": 0.2638319432735443,
      "learning_rate": 0.00022972494618512317,
      "loss": 0.0514,
      "step": 5402
    },
    {
      "epoch": 0.7572529782761037,
      "grad_norm": 0.5572202801704407,
      "learning_rate": 0.0002297105955513035,
      "loss": 0.0708,
      "step": 5403
    },
    {
      "epoch": 0.7573931324456903,
      "grad_norm": 0.18339067697525024,
      "learning_rate": 0.00022969624491748383,
      "loss": 0.0291,
      "step": 5404
    },
    {
      "epoch": 0.7575332866152769,
      "grad_norm": 0.3518141508102417,
      "learning_rate": 0.00022968189428366416,
      "loss": 0.067,
      "step": 5405
    },
    {
      "epoch": 0.7576734407848633,
      "grad_norm": 0.34705042839050293,
      "learning_rate": 0.0002296675436498445,
      "loss": 0.0908,
      "step": 5406
    },
    {
      "epoch": 0.7578135949544499,
      "grad_norm": 0.45325136184692383,
      "learning_rate": 0.00022965319301602484,
      "loss": 0.1322,
      "step": 5407
    },
    {
      "epoch": 0.7579537491240365,
      "grad_norm": 0.7047853469848633,
      "learning_rate": 0.00022963884238220517,
      "loss": 0.1176,
      "step": 5408
    },
    {
      "epoch": 0.758093903293623,
      "grad_norm": 0.33671072125434875,
      "learning_rate": 0.00022962449174838555,
      "loss": 0.1151,
      "step": 5409
    },
    {
      "epoch": 0.7582340574632095,
      "grad_norm": 0.42787134647369385,
      "learning_rate": 0.00022961014111456588,
      "loss": 0.0534,
      "step": 5410
    },
    {
      "epoch": 0.7583742116327961,
      "grad_norm": 0.44981077313423157,
      "learning_rate": 0.0002295957904807462,
      "loss": 0.0874,
      "step": 5411
    },
    {
      "epoch": 0.7585143658023826,
      "grad_norm": 0.41092604398727417,
      "learning_rate": 0.00022958143984692656,
      "loss": 0.0944,
      "step": 5412
    },
    {
      "epoch": 0.7586545199719692,
      "grad_norm": 0.2224346250295639,
      "learning_rate": 0.0002295670892131069,
      "loss": 0.0549,
      "step": 5413
    },
    {
      "epoch": 0.7587946741415557,
      "grad_norm": 0.37301105260849,
      "learning_rate": 0.00022955273857928722,
      "loss": 0.0744,
      "step": 5414
    },
    {
      "epoch": 0.7589348283111422,
      "grad_norm": 0.36424317955970764,
      "learning_rate": 0.00022953838794546758,
      "loss": 0.087,
      "step": 5415
    },
    {
      "epoch": 0.7590749824807288,
      "grad_norm": 0.45953139662742615,
      "learning_rate": 0.0002295240373116479,
      "loss": 0.118,
      "step": 5416
    },
    {
      "epoch": 0.7592151366503154,
      "grad_norm": 0.19470801949501038,
      "learning_rate": 0.00022950968667782823,
      "loss": 0.041,
      "step": 5417
    },
    {
      "epoch": 0.7593552908199019,
      "grad_norm": 0.2255251556634903,
      "learning_rate": 0.00022949533604400862,
      "loss": 0.0513,
      "step": 5418
    },
    {
      "epoch": 0.7594954449894884,
      "grad_norm": 0.3890061676502228,
      "learning_rate": 0.00022948098541018895,
      "loss": 0.1079,
      "step": 5419
    },
    {
      "epoch": 0.759635599159075,
      "grad_norm": 0.6219236850738525,
      "learning_rate": 0.00022946663477636927,
      "loss": 0.0885,
      "step": 5420
    },
    {
      "epoch": 0.7597757533286615,
      "grad_norm": 0.5062904953956604,
      "learning_rate": 0.00022945228414254963,
      "loss": 0.0696,
      "step": 5421
    },
    {
      "epoch": 0.7599159074982481,
      "grad_norm": 0.5156422853469849,
      "learning_rate": 0.00022943793350872996,
      "loss": 0.0962,
      "step": 5422
    },
    {
      "epoch": 0.7600560616678346,
      "grad_norm": 0.2950113117694855,
      "learning_rate": 0.00022942358287491029,
      "loss": 0.028,
      "step": 5423
    },
    {
      "epoch": 0.7601962158374211,
      "grad_norm": 0.503536581993103,
      "learning_rate": 0.00022940923224109061,
      "loss": 0.1453,
      "step": 5424
    },
    {
      "epoch": 0.7603363700070077,
      "grad_norm": 0.3930160105228424,
      "learning_rate": 0.00022939488160727097,
      "loss": 0.0938,
      "step": 5425
    },
    {
      "epoch": 0.7604765241765943,
      "grad_norm": 0.33009910583496094,
      "learning_rate": 0.0002293805309734513,
      "loss": 0.0731,
      "step": 5426
    },
    {
      "epoch": 0.7606166783461809,
      "grad_norm": 0.34729063510894775,
      "learning_rate": 0.00022936618033963163,
      "loss": 0.0412,
      "step": 5427
    },
    {
      "epoch": 0.7607568325157673,
      "grad_norm": 0.6240994334220886,
      "learning_rate": 0.00022935182970581198,
      "loss": 0.0817,
      "step": 5428
    },
    {
      "epoch": 0.7608969866853539,
      "grad_norm": 0.25242751836776733,
      "learning_rate": 0.0002293374790719923,
      "loss": 0.0788,
      "step": 5429
    },
    {
      "epoch": 0.7610371408549405,
      "grad_norm": 0.26445019245147705,
      "learning_rate": 0.00022932312843817267,
      "loss": 0.0651,
      "step": 5430
    },
    {
      "epoch": 0.761177295024527,
      "grad_norm": 0.6258429884910583,
      "learning_rate": 0.00022930877780435302,
      "loss": 0.0684,
      "step": 5431
    },
    {
      "epoch": 0.7613174491941135,
      "grad_norm": 0.24849849939346313,
      "learning_rate": 0.00022929442717053335,
      "loss": 0.0277,
      "step": 5432
    },
    {
      "epoch": 0.7614576033637,
      "grad_norm": 0.7237057685852051,
      "learning_rate": 0.00022928007653671368,
      "loss": 0.0614,
      "step": 5433
    },
    {
      "epoch": 0.7615977575332866,
      "grad_norm": 0.7451324462890625,
      "learning_rate": 0.00022926572590289404,
      "loss": 0.1271,
      "step": 5434
    },
    {
      "epoch": 0.7617379117028732,
      "grad_norm": 0.7456595301628113,
      "learning_rate": 0.00022925137526907436,
      "loss": 0.1131,
      "step": 5435
    },
    {
      "epoch": 0.7618780658724597,
      "grad_norm": 0.844342827796936,
      "learning_rate": 0.0002292370246352547,
      "loss": 0.0393,
      "step": 5436
    },
    {
      "epoch": 0.7620182200420462,
      "grad_norm": 0.3638335168361664,
      "learning_rate": 0.00022922267400143505,
      "loss": 0.0494,
      "step": 5437
    },
    {
      "epoch": 0.7621583742116328,
      "grad_norm": 0.944242000579834,
      "learning_rate": 0.00022920832336761538,
      "loss": 0.1483,
      "step": 5438
    },
    {
      "epoch": 0.7622985283812194,
      "grad_norm": 0.68157958984375,
      "learning_rate": 0.0002291939727337957,
      "loss": 0.0743,
      "step": 5439
    },
    {
      "epoch": 0.7624386825508059,
      "grad_norm": 3.959632635116577,
      "learning_rate": 0.00022917962209997603,
      "loss": 0.1273,
      "step": 5440
    },
    {
      "epoch": 0.7625788367203924,
      "grad_norm": 0.7042611241340637,
      "learning_rate": 0.00022916527146615642,
      "loss": 0.0701,
      "step": 5441
    },
    {
      "epoch": 0.762718990889979,
      "grad_norm": 0.45904669165611267,
      "learning_rate": 0.00022915092083233675,
      "loss": 0.1252,
      "step": 5442
    },
    {
      "epoch": 0.7628591450595655,
      "grad_norm": 0.4239557683467865,
      "learning_rate": 0.00022913657019851707,
      "loss": 0.0293,
      "step": 5443
    },
    {
      "epoch": 0.7629992992291521,
      "grad_norm": 0.7637688517570496,
      "learning_rate": 0.00022912221956469743,
      "loss": 0.0882,
      "step": 5444
    },
    {
      "epoch": 0.7631394533987386,
      "grad_norm": 0.35936346650123596,
      "learning_rate": 0.00022910786893087776,
      "loss": 0.0491,
      "step": 5445
    },
    {
      "epoch": 0.7632796075683251,
      "grad_norm": 0.38990676403045654,
      "learning_rate": 0.0002290935182970581,
      "loss": 0.0825,
      "step": 5446
    },
    {
      "epoch": 0.7634197617379117,
      "grad_norm": 0.8177030682563782,
      "learning_rate": 0.00022907916766323844,
      "loss": 0.323,
      "step": 5447
    },
    {
      "epoch": 0.7635599159074983,
      "grad_norm": 0.8256582617759705,
      "learning_rate": 0.00022906481702941877,
      "loss": 0.0806,
      "step": 5448
    },
    {
      "epoch": 0.7637000700770848,
      "grad_norm": 0.9747647047042847,
      "learning_rate": 0.0002290504663955991,
      "loss": 0.2827,
      "step": 5449
    },
    {
      "epoch": 0.7638402242466713,
      "grad_norm": 0.8744803667068481,
      "learning_rate": 0.00022903611576177948,
      "loss": 0.0365,
      "step": 5450
    },
    {
      "epoch": 0.7639803784162579,
      "grad_norm": 0.41220322251319885,
      "learning_rate": 0.0002290217651279598,
      "loss": 0.164,
      "step": 5451
    },
    {
      "epoch": 0.7641205325858444,
      "grad_norm": 0.4641821086406708,
      "learning_rate": 0.00022900741449414014,
      "loss": 0.0726,
      "step": 5452
    },
    {
      "epoch": 0.764260686755431,
      "grad_norm": 0.24509745836257935,
      "learning_rate": 0.0002289930638603205,
      "loss": 0.0591,
      "step": 5453
    },
    {
      "epoch": 0.7644008409250175,
      "grad_norm": 0.7141879200935364,
      "learning_rate": 0.00022897871322650082,
      "loss": 0.2115,
      "step": 5454
    },
    {
      "epoch": 0.764540995094604,
      "grad_norm": 0.22111722826957703,
      "learning_rate": 0.00022896436259268115,
      "loss": 0.0713,
      "step": 5455
    },
    {
      "epoch": 0.7646811492641906,
      "grad_norm": 0.488768607378006,
      "learning_rate": 0.0002289500119588615,
      "loss": 0.1028,
      "step": 5456
    },
    {
      "epoch": 0.7648213034337772,
      "grad_norm": 0.4556877017021179,
      "learning_rate": 0.00022893566132504184,
      "loss": 0.1679,
      "step": 5457
    },
    {
      "epoch": 0.7649614576033636,
      "grad_norm": 0.37825000286102295,
      "learning_rate": 0.00022892131069122217,
      "loss": 0.0938,
      "step": 5458
    },
    {
      "epoch": 0.7651016117729502,
      "grad_norm": 0.3884667456150055,
      "learning_rate": 0.0002289069600574025,
      "loss": 0.1208,
      "step": 5459
    },
    {
      "epoch": 0.7652417659425368,
      "grad_norm": 0.30482277274131775,
      "learning_rate": 0.00022889260942358285,
      "loss": 0.1089,
      "step": 5460
    },
    {
      "epoch": 0.7653819201121234,
      "grad_norm": 0.27011653780937195,
      "learning_rate": 0.00022887825878976318,
      "loss": 0.0664,
      "step": 5461
    },
    {
      "epoch": 0.7655220742817099,
      "grad_norm": 0.30397647619247437,
      "learning_rate": 0.00022886390815594353,
      "loss": 0.1149,
      "step": 5462
    },
    {
      "epoch": 0.7656622284512964,
      "grad_norm": 0.26909390091896057,
      "learning_rate": 0.0002288495575221239,
      "loss": 0.0796,
      "step": 5463
    },
    {
      "epoch": 0.765802382620883,
      "grad_norm": 0.4857081174850464,
      "learning_rate": 0.00022883520688830422,
      "loss": 0.1152,
      "step": 5464
    },
    {
      "epoch": 0.7659425367904695,
      "grad_norm": 0.2117944061756134,
      "learning_rate": 0.00022882085625448455,
      "loss": 0.0783,
      "step": 5465
    },
    {
      "epoch": 0.7660826909600561,
      "grad_norm": 0.47209733724594116,
      "learning_rate": 0.0002288065056206649,
      "loss": 0.1347,
      "step": 5466
    },
    {
      "epoch": 0.7662228451296426,
      "grad_norm": 0.350393146276474,
      "learning_rate": 0.00022879215498684523,
      "loss": 0.1044,
      "step": 5467
    },
    {
      "epoch": 0.7663629992992291,
      "grad_norm": 0.42804351449012756,
      "learning_rate": 0.00022877780435302556,
      "loss": 0.1,
      "step": 5468
    },
    {
      "epoch": 0.7665031534688157,
      "grad_norm": 0.6100720763206482,
      "learning_rate": 0.00022876345371920592,
      "loss": 0.144,
      "step": 5469
    },
    {
      "epoch": 0.7666433076384023,
      "grad_norm": 0.3709159791469574,
      "learning_rate": 0.00022874910308538624,
      "loss": 0.0533,
      "step": 5470
    },
    {
      "epoch": 0.7667834618079888,
      "grad_norm": 0.22639378905296326,
      "learning_rate": 0.00022873475245156657,
      "loss": 0.0427,
      "step": 5471
    },
    {
      "epoch": 0.7669236159775753,
      "grad_norm": 0.733083188533783,
      "learning_rate": 0.00022872040181774696,
      "loss": 0.0775,
      "step": 5472
    },
    {
      "epoch": 0.7670637701471619,
      "grad_norm": 0.40559929609298706,
      "learning_rate": 0.00022870605118392728,
      "loss": 0.056,
      "step": 5473
    },
    {
      "epoch": 0.7672039243167484,
      "grad_norm": 0.2793610990047455,
      "learning_rate": 0.0002286917005501076,
      "loss": 0.0879,
      "step": 5474
    },
    {
      "epoch": 0.767344078486335,
      "grad_norm": 0.42612722516059875,
      "learning_rate": 0.00022867734991628794,
      "loss": 0.0729,
      "step": 5475
    },
    {
      "epoch": 0.7674842326559215,
      "grad_norm": 0.3715907633304596,
      "learning_rate": 0.0002286629992824683,
      "loss": 0.0926,
      "step": 5476
    },
    {
      "epoch": 0.767624386825508,
      "grad_norm": 0.2092791050672531,
      "learning_rate": 0.00022864864864864862,
      "loss": 0.0438,
      "step": 5477
    },
    {
      "epoch": 0.7677645409950946,
      "grad_norm": 0.7470300197601318,
      "learning_rate": 0.00022863429801482895,
      "loss": 0.0986,
      "step": 5478
    },
    {
      "epoch": 0.7679046951646812,
      "grad_norm": 0.3440161347389221,
      "learning_rate": 0.0002286199473810093,
      "loss": 0.0833,
      "step": 5479
    },
    {
      "epoch": 0.7680448493342676,
      "grad_norm": 0.40457969903945923,
      "learning_rate": 0.00022860559674718964,
      "loss": 0.0823,
      "step": 5480
    },
    {
      "epoch": 0.7681850035038542,
      "grad_norm": 0.5405162572860718,
      "learning_rate": 0.00022859124611336997,
      "loss": 0.1184,
      "step": 5481
    },
    {
      "epoch": 0.7683251576734408,
      "grad_norm": 0.5273644328117371,
      "learning_rate": 0.00022857689547955035,
      "loss": 0.0734,
      "step": 5482
    },
    {
      "epoch": 0.7684653118430274,
      "grad_norm": 0.20312072336673737,
      "learning_rate": 0.00022856254484573068,
      "loss": 0.0216,
      "step": 5483
    },
    {
      "epoch": 0.7686054660126139,
      "grad_norm": 0.3540211319923401,
      "learning_rate": 0.000228548194211911,
      "loss": 0.0378,
      "step": 5484
    },
    {
      "epoch": 0.7687456201822004,
      "grad_norm": 0.18328943848609924,
      "learning_rate": 0.00022853384357809136,
      "loss": 0.0338,
      "step": 5485
    },
    {
      "epoch": 0.768885774351787,
      "grad_norm": 0.47405415773391724,
      "learning_rate": 0.0002285194929442717,
      "loss": 0.0738,
      "step": 5486
    },
    {
      "epoch": 0.7690259285213735,
      "grad_norm": 0.23524805903434753,
      "learning_rate": 0.00022850514231045202,
      "loss": 0.0356,
      "step": 5487
    },
    {
      "epoch": 0.7691660826909601,
      "grad_norm": 1.4024724960327148,
      "learning_rate": 0.00022849079167663237,
      "loss": 0.1725,
      "step": 5488
    },
    {
      "epoch": 0.7693062368605466,
      "grad_norm": 0.9750211834907532,
      "learning_rate": 0.0002284764410428127,
      "loss": 0.1854,
      "step": 5489
    },
    {
      "epoch": 0.7694463910301331,
      "grad_norm": 0.40581846237182617,
      "learning_rate": 0.00022846209040899303,
      "loss": 0.0844,
      "step": 5490
    },
    {
      "epoch": 0.7695865451997197,
      "grad_norm": 0.8945640921592712,
      "learning_rate": 0.0002284477397751734,
      "loss": 0.0576,
      "step": 5491
    },
    {
      "epoch": 0.7697266993693063,
      "grad_norm": 0.4466705024242401,
      "learning_rate": 0.00022843338914135372,
      "loss": 0.0445,
      "step": 5492
    },
    {
      "epoch": 0.7698668535388928,
      "grad_norm": 0.17964980006217957,
      "learning_rate": 0.00022841903850753404,
      "loss": 0.0286,
      "step": 5493
    },
    {
      "epoch": 0.7700070077084793,
      "grad_norm": 0.6608170866966248,
      "learning_rate": 0.0002284046878737144,
      "loss": 0.0531,
      "step": 5494
    },
    {
      "epoch": 0.7701471618780659,
      "grad_norm": 0.42862468957901,
      "learning_rate": 0.00022839033723989476,
      "loss": 0.064,
      "step": 5495
    },
    {
      "epoch": 0.7702873160476524,
      "grad_norm": 1.3754823207855225,
      "learning_rate": 0.00022837598660607508,
      "loss": 0.2525,
      "step": 5496
    },
    {
      "epoch": 0.770427470217239,
      "grad_norm": 0.5401486754417419,
      "learning_rate": 0.0002283616359722554,
      "loss": 0.087,
      "step": 5497
    },
    {
      "epoch": 0.7705676243868255,
      "grad_norm": 1.3065012693405151,
      "learning_rate": 0.00022834728533843577,
      "loss": 0.192,
      "step": 5498
    },
    {
      "epoch": 0.770707778556412,
      "grad_norm": 1.142110824584961,
      "learning_rate": 0.0002283329347046161,
      "loss": 0.1133,
      "step": 5499
    },
    {
      "epoch": 0.7708479327259986,
      "grad_norm": 0.9058482050895691,
      "learning_rate": 0.00022831858407079643,
      "loss": 0.1584,
      "step": 5500
    },
    {
      "epoch": 0.7709880868955852,
      "grad_norm": 0.6781784892082214,
      "learning_rate": 0.00022830423343697678,
      "loss": 0.1607,
      "step": 5501
    },
    {
      "epoch": 0.7711282410651716,
      "grad_norm": 0.3875426650047302,
      "learning_rate": 0.0002282898828031571,
      "loss": 0.0908,
      "step": 5502
    },
    {
      "epoch": 0.7712683952347582,
      "grad_norm": 0.24912330508232117,
      "learning_rate": 0.00022827553216933744,
      "loss": 0.0497,
      "step": 5503
    },
    {
      "epoch": 0.7714085494043448,
      "grad_norm": 0.44237616658210754,
      "learning_rate": 0.00022826118153551782,
      "loss": 0.0606,
      "step": 5504
    },
    {
      "epoch": 0.7715487035739313,
      "grad_norm": 0.36543911695480347,
      "learning_rate": 0.00022824683090169815,
      "loss": 0.0768,
      "step": 5505
    },
    {
      "epoch": 0.7716888577435179,
      "grad_norm": 0.5312711596488953,
      "learning_rate": 0.00022823248026787848,
      "loss": 0.0506,
      "step": 5506
    },
    {
      "epoch": 0.7718290119131044,
      "grad_norm": 0.28648728132247925,
      "learning_rate": 0.00022821812963405883,
      "loss": 0.0689,
      "step": 5507
    },
    {
      "epoch": 0.771969166082691,
      "grad_norm": 0.6324368715286255,
      "learning_rate": 0.00022820377900023916,
      "loss": 0.1072,
      "step": 5508
    },
    {
      "epoch": 0.7721093202522775,
      "grad_norm": 0.5460099577903748,
      "learning_rate": 0.0002281894283664195,
      "loss": 0.0702,
      "step": 5509
    },
    {
      "epoch": 0.7722494744218641,
      "grad_norm": 0.9754613041877747,
      "learning_rate": 0.00022817507773259982,
      "loss": 0.0719,
      "step": 5510
    },
    {
      "epoch": 0.7723896285914505,
      "grad_norm": 0.4745437800884247,
      "learning_rate": 0.00022816072709878018,
      "loss": 0.0959,
      "step": 5511
    },
    {
      "epoch": 0.7725297827610371,
      "grad_norm": 0.39778056740760803,
      "learning_rate": 0.0002281463764649605,
      "loss": 0.055,
      "step": 5512
    },
    {
      "epoch": 0.7726699369306237,
      "grad_norm": 0.4247722029685974,
      "learning_rate": 0.00022813202583114083,
      "loss": 0.1013,
      "step": 5513
    },
    {
      "epoch": 0.7728100911002103,
      "grad_norm": 0.6100055575370789,
      "learning_rate": 0.00022811767519732122,
      "loss": 0.0998,
      "step": 5514
    },
    {
      "epoch": 0.7729502452697968,
      "grad_norm": 0.3268587291240692,
      "learning_rate": 0.00022810332456350154,
      "loss": 0.1111,
      "step": 5515
    },
    {
      "epoch": 0.7730903994393833,
      "grad_norm": 0.33326730132102966,
      "learning_rate": 0.00022808897392968187,
      "loss": 0.0685,
      "step": 5516
    },
    {
      "epoch": 0.7732305536089699,
      "grad_norm": 0.648952305316925,
      "learning_rate": 0.00022807462329586223,
      "loss": 0.1035,
      "step": 5517
    },
    {
      "epoch": 0.7733707077785564,
      "grad_norm": 0.3474395275115967,
      "learning_rate": 0.00022806027266204256,
      "loss": 0.1074,
      "step": 5518
    },
    {
      "epoch": 0.773510861948143,
      "grad_norm": 0.35976314544677734,
      "learning_rate": 0.00022804592202822289,
      "loss": 0.1051,
      "step": 5519
    },
    {
      "epoch": 0.7736510161177295,
      "grad_norm": 0.5398932099342346,
      "learning_rate": 0.00022803157139440324,
      "loss": 0.074,
      "step": 5520
    },
    {
      "epoch": 0.773791170287316,
      "grad_norm": 0.6685708165168762,
      "learning_rate": 0.00022801722076058357,
      "loss": 0.125,
      "step": 5521
    },
    {
      "epoch": 0.7739313244569026,
      "grad_norm": 0.7419278621673584,
      "learning_rate": 0.0002280028701267639,
      "loss": 0.1251,
      "step": 5522
    },
    {
      "epoch": 0.7740714786264892,
      "grad_norm": 0.20199072360992432,
      "learning_rate": 0.00022798851949294425,
      "loss": 0.0231,
      "step": 5523
    },
    {
      "epoch": 0.7742116327960756,
      "grad_norm": 0.5185670852661133,
      "learning_rate": 0.00022797416885912458,
      "loss": 0.1321,
      "step": 5524
    },
    {
      "epoch": 0.7743517869656622,
      "grad_norm": 0.17992335557937622,
      "learning_rate": 0.0002279598182253049,
      "loss": 0.0463,
      "step": 5525
    },
    {
      "epoch": 0.7744919411352488,
      "grad_norm": 0.705573558807373,
      "learning_rate": 0.0002279454675914853,
      "loss": 0.0426,
      "step": 5526
    },
    {
      "epoch": 0.7746320953048353,
      "grad_norm": 0.7857452630996704,
      "learning_rate": 0.00022793111695766562,
      "loss": 0.0827,
      "step": 5527
    },
    {
      "epoch": 0.7747722494744219,
      "grad_norm": 0.5969287157058716,
      "learning_rate": 0.00022791676632384595,
      "loss": 0.0575,
      "step": 5528
    },
    {
      "epoch": 0.7749124036440084,
      "grad_norm": 0.18924486637115479,
      "learning_rate": 0.00022790241569002628,
      "loss": 0.038,
      "step": 5529
    },
    {
      "epoch": 0.775052557813595,
      "grad_norm": 0.4760206937789917,
      "learning_rate": 0.00022788806505620663,
      "loss": 0.1088,
      "step": 5530
    },
    {
      "epoch": 0.7751927119831815,
      "grad_norm": 0.518638551235199,
      "learning_rate": 0.00022787371442238696,
      "loss": 0.0353,
      "step": 5531
    },
    {
      "epoch": 0.7753328661527681,
      "grad_norm": 0.3066500425338745,
      "learning_rate": 0.0002278593637885673,
      "loss": 0.0466,
      "step": 5532
    },
    {
      "epoch": 0.7754730203223545,
      "grad_norm": 0.2970122992992401,
      "learning_rate": 0.00022784501315474765,
      "loss": 0.0913,
      "step": 5533
    },
    {
      "epoch": 0.7756131744919411,
      "grad_norm": 0.4344521760940552,
      "learning_rate": 0.00022783066252092798,
      "loss": 0.1538,
      "step": 5534
    },
    {
      "epoch": 0.7757533286615277,
      "grad_norm": 0.4884020686149597,
      "learning_rate": 0.0002278163118871083,
      "loss": 0.1396,
      "step": 5535
    },
    {
      "epoch": 0.7758934828311143,
      "grad_norm": 0.38114333152770996,
      "learning_rate": 0.0002278019612532887,
      "loss": 0.1576,
      "step": 5536
    },
    {
      "epoch": 0.7760336370007007,
      "grad_norm": 0.49452364444732666,
      "learning_rate": 0.00022778761061946902,
      "loss": 0.1974,
      "step": 5537
    },
    {
      "epoch": 0.7761737911702873,
      "grad_norm": 0.4518037438392639,
      "learning_rate": 0.00022777325998564934,
      "loss": 0.0757,
      "step": 5538
    },
    {
      "epoch": 0.7763139453398739,
      "grad_norm": 0.4614986181259155,
      "learning_rate": 0.0002277589093518297,
      "loss": 0.0693,
      "step": 5539
    },
    {
      "epoch": 0.7764540995094604,
      "grad_norm": 0.5054397583007812,
      "learning_rate": 0.00022774455871801003,
      "loss": 0.2056,
      "step": 5540
    },
    {
      "epoch": 0.776594253679047,
      "grad_norm": 0.4532138705253601,
      "learning_rate": 0.00022773020808419036,
      "loss": 0.164,
      "step": 5541
    },
    {
      "epoch": 0.7767344078486335,
      "grad_norm": 0.34185174107551575,
      "learning_rate": 0.0002277158574503707,
      "loss": 0.0416,
      "step": 5542
    },
    {
      "epoch": 0.77687456201822,
      "grad_norm": 0.4061909317970276,
      "learning_rate": 0.00022770150681655104,
      "loss": 0.0812,
      "step": 5543
    },
    {
      "epoch": 0.7770147161878066,
      "grad_norm": 1.1434621810913086,
      "learning_rate": 0.00022768715618273137,
      "loss": 0.0949,
      "step": 5544
    },
    {
      "epoch": 0.7771548703573932,
      "grad_norm": 0.2393876314163208,
      "learning_rate": 0.0002276728055489117,
      "loss": 0.0683,
      "step": 5545
    },
    {
      "epoch": 0.7772950245269796,
      "grad_norm": 0.5975365042686462,
      "learning_rate": 0.00022765845491509208,
      "loss": 0.205,
      "step": 5546
    },
    {
      "epoch": 0.7774351786965662,
      "grad_norm": 0.5446653366088867,
      "learning_rate": 0.0002276441042812724,
      "loss": 0.0645,
      "step": 5547
    },
    {
      "epoch": 0.7775753328661528,
      "grad_norm": 4.504129886627197,
      "learning_rate": 0.00022762975364745274,
      "loss": 0.3726,
      "step": 5548
    },
    {
      "epoch": 0.7777154870357393,
      "grad_norm": 3.6292026042938232,
      "learning_rate": 0.0002276154030136331,
      "loss": 0.7406,
      "step": 5549
    },
    {
      "epoch": 0.7778556412053259,
      "grad_norm": 1.1210706233978271,
      "learning_rate": 0.00022760105237981342,
      "loss": 0.1189,
      "step": 5550
    },
    {
      "epoch": 0.7779957953749124,
      "grad_norm": 0.18803133070468903,
      "learning_rate": 0.00022758670174599375,
      "loss": 0.0493,
      "step": 5551
    },
    {
      "epoch": 0.7781359495444989,
      "grad_norm": 0.4657420217990875,
      "learning_rate": 0.0002275723511121741,
      "loss": 0.1338,
      "step": 5552
    },
    {
      "epoch": 0.7782761037140855,
      "grad_norm": 0.2918739318847656,
      "learning_rate": 0.00022755800047835444,
      "loss": 0.0375,
      "step": 5553
    },
    {
      "epoch": 0.7784162578836721,
      "grad_norm": 0.49253174662590027,
      "learning_rate": 0.00022754364984453476,
      "loss": 0.0814,
      "step": 5554
    },
    {
      "epoch": 0.7785564120532585,
      "grad_norm": 0.5060675740242004,
      "learning_rate": 0.00022752929921071512,
      "loss": 0.0947,
      "step": 5555
    },
    {
      "epoch": 0.7786965662228451,
      "grad_norm": 0.3380374014377594,
      "learning_rate": 0.00022751494857689545,
      "loss": 0.0475,
      "step": 5556
    },
    {
      "epoch": 0.7788367203924317,
      "grad_norm": 0.35931336879730225,
      "learning_rate": 0.0002275005979430758,
      "loss": 0.1079,
      "step": 5557
    },
    {
      "epoch": 0.7789768745620183,
      "grad_norm": 0.3204613924026489,
      "learning_rate": 0.00022748624730925616,
      "loss": 0.1118,
      "step": 5558
    },
    {
      "epoch": 0.7791170287316047,
      "grad_norm": 0.42459946870803833,
      "learning_rate": 0.0002274718966754365,
      "loss": 0.0752,
      "step": 5559
    },
    {
      "epoch": 0.7792571829011913,
      "grad_norm": 0.5983796119689941,
      "learning_rate": 0.00022745754604161682,
      "loss": 0.0808,
      "step": 5560
    },
    {
      "epoch": 0.7793973370707779,
      "grad_norm": 0.6460114121437073,
      "learning_rate": 0.00022744319540779717,
      "loss": 0.1252,
      "step": 5561
    },
    {
      "epoch": 0.7795374912403644,
      "grad_norm": 0.39227792620658875,
      "learning_rate": 0.0002274288447739775,
      "loss": 0.1058,
      "step": 5562
    },
    {
      "epoch": 0.779677645409951,
      "grad_norm": 0.2218901365995407,
      "learning_rate": 0.00022741449414015783,
      "loss": 0.0308,
      "step": 5563
    },
    {
      "epoch": 0.7798177995795375,
      "grad_norm": 0.27046000957489014,
      "learning_rate": 0.00022740014350633816,
      "loss": 0.0665,
      "step": 5564
    },
    {
      "epoch": 0.779957953749124,
      "grad_norm": 0.41917768120765686,
      "learning_rate": 0.00022738579287251851,
      "loss": 0.1094,
      "step": 5565
    },
    {
      "epoch": 0.7800981079187106,
      "grad_norm": 0.3697170317173004,
      "learning_rate": 0.00022737144223869884,
      "loss": 0.0818,
      "step": 5566
    },
    {
      "epoch": 0.7802382620882972,
      "grad_norm": 0.37233036756515503,
      "learning_rate": 0.00022735709160487917,
      "loss": 0.02,
      "step": 5567
    },
    {
      "epoch": 0.7803784162578836,
      "grad_norm": 0.1780182421207428,
      "learning_rate": 0.00022734274097105955,
      "loss": 0.0445,
      "step": 5568
    },
    {
      "epoch": 0.7805185704274702,
      "grad_norm": 0.35164687037467957,
      "learning_rate": 0.00022732839033723988,
      "loss": 0.0635,
      "step": 5569
    },
    {
      "epoch": 0.7806587245970568,
      "grad_norm": 0.24476350843906403,
      "learning_rate": 0.0002273140397034202,
      "loss": 0.065,
      "step": 5570
    },
    {
      "epoch": 0.7807988787666433,
      "grad_norm": 0.39904242753982544,
      "learning_rate": 0.00022729968906960057,
      "loss": 0.0792,
      "step": 5571
    },
    {
      "epoch": 0.7809390329362299,
      "grad_norm": 0.41253888607025146,
      "learning_rate": 0.0002272853384357809,
      "loss": 0.1132,
      "step": 5572
    },
    {
      "epoch": 0.7810791871058164,
      "grad_norm": 0.3747071921825409,
      "learning_rate": 0.00022727098780196122,
      "loss": 0.0721,
      "step": 5573
    },
    {
      "epoch": 0.7812193412754029,
      "grad_norm": 0.3455543518066406,
      "learning_rate": 0.00022725663716814158,
      "loss": 0.1271,
      "step": 5574
    },
    {
      "epoch": 0.7813594954449895,
      "grad_norm": 0.449190229177475,
      "learning_rate": 0.0002272422865343219,
      "loss": 0.0698,
      "step": 5575
    },
    {
      "epoch": 0.7814996496145761,
      "grad_norm": 0.3513028919696808,
      "learning_rate": 0.00022722793590050224,
      "loss": 0.1134,
      "step": 5576
    },
    {
      "epoch": 0.7816398037841625,
      "grad_norm": 0.3159481883049011,
      "learning_rate": 0.00022721358526668262,
      "loss": 0.0691,
      "step": 5577
    },
    {
      "epoch": 0.7817799579537491,
      "grad_norm": 0.49728885293006897,
      "learning_rate": 0.00022719923463286295,
      "loss": 0.2723,
      "step": 5578
    },
    {
      "epoch": 0.7819201121233357,
      "grad_norm": 0.570715069770813,
      "learning_rate": 0.00022718488399904328,
      "loss": 0.1158,
      "step": 5579
    },
    {
      "epoch": 0.7820602662929222,
      "grad_norm": 0.36925241351127625,
      "learning_rate": 0.0002271705333652236,
      "loss": 0.1393,
      "step": 5580
    },
    {
      "epoch": 0.7822004204625087,
      "grad_norm": 0.42141807079315186,
      "learning_rate": 0.00022715618273140396,
      "loss": 0.1219,
      "step": 5581
    },
    {
      "epoch": 0.7823405746320953,
      "grad_norm": 0.40359222888946533,
      "learning_rate": 0.0002271418320975843,
      "loss": 0.1132,
      "step": 5582
    },
    {
      "epoch": 0.7824807288016818,
      "grad_norm": 0.21122585237026215,
      "learning_rate": 0.00022712748146376462,
      "loss": 0.0418,
      "step": 5583
    },
    {
      "epoch": 0.7826208829712684,
      "grad_norm": 0.4212229549884796,
      "learning_rate": 0.00022711313082994497,
      "loss": 0.0647,
      "step": 5584
    },
    {
      "epoch": 0.782761037140855,
      "grad_norm": 0.20950578153133392,
      "learning_rate": 0.0002270987801961253,
      "loss": 0.0249,
      "step": 5585
    },
    {
      "epoch": 0.7829011913104414,
      "grad_norm": 0.23099462687969208,
      "learning_rate": 0.00022708442956230563,
      "loss": 0.0597,
      "step": 5586
    },
    {
      "epoch": 0.783041345480028,
      "grad_norm": 0.4331852197647095,
      "learning_rate": 0.00022707007892848599,
      "loss": 0.1412,
      "step": 5587
    },
    {
      "epoch": 0.7831814996496146,
      "grad_norm": 0.6406581401824951,
      "learning_rate": 0.00022705572829466631,
      "loss": 0.1319,
      "step": 5588
    },
    {
      "epoch": 0.7833216538192012,
      "grad_norm": 0.6134215593338013,
      "learning_rate": 0.00022704137766084667,
      "loss": 0.1797,
      "step": 5589
    },
    {
      "epoch": 0.7834618079887876,
      "grad_norm": 0.3799291253089905,
      "learning_rate": 0.00022702702702702703,
      "loss": 0.0395,
      "step": 5590
    },
    {
      "epoch": 0.7836019621583742,
      "grad_norm": 0.3801416754722595,
      "learning_rate": 0.00022701267639320735,
      "loss": 0.0646,
      "step": 5591
    },
    {
      "epoch": 0.7837421163279608,
      "grad_norm": 0.8118061423301697,
      "learning_rate": 0.00022699832575938768,
      "loss": 0.317,
      "step": 5592
    },
    {
      "epoch": 0.7838822704975473,
      "grad_norm": 0.6364262700080872,
      "learning_rate": 0.00022698397512556804,
      "loss": 0.0468,
      "step": 5593
    },
    {
      "epoch": 0.7840224246671339,
      "grad_norm": 0.4522698223590851,
      "learning_rate": 0.00022696962449174837,
      "loss": 0.0702,
      "step": 5594
    },
    {
      "epoch": 0.7841625788367204,
      "grad_norm": 0.7264848351478577,
      "learning_rate": 0.0002269552738579287,
      "loss": 0.2152,
      "step": 5595
    },
    {
      "epoch": 0.7843027330063069,
      "grad_norm": 0.39555561542510986,
      "learning_rate": 0.00022694092322410905,
      "loss": 0.0501,
      "step": 5596
    },
    {
      "epoch": 0.7844428871758935,
      "grad_norm": 0.8773069381713867,
      "learning_rate": 0.00022692657259028938,
      "loss": 0.1856,
      "step": 5597
    },
    {
      "epoch": 0.7845830413454801,
      "grad_norm": 1.3970768451690674,
      "learning_rate": 0.0002269122219564697,
      "loss": 0.0983,
      "step": 5598
    },
    {
      "epoch": 0.7847231955150665,
      "grad_norm": 1.2268863916397095,
      "learning_rate": 0.00022689787132265004,
      "loss": 0.1516,
      "step": 5599
    },
    {
      "epoch": 0.7848633496846531,
      "grad_norm": 0.813997745513916,
      "learning_rate": 0.00022688352068883042,
      "loss": 0.0945,
      "step": 5600
    },
    {
      "epoch": 0.7850035038542397,
      "grad_norm": 0.19086599349975586,
      "learning_rate": 0.00022686917005501075,
      "loss": 0.0811,
      "step": 5601
    },
    {
      "epoch": 0.7851436580238262,
      "grad_norm": 0.13190264999866486,
      "learning_rate": 0.00022685481942119108,
      "loss": 0.0437,
      "step": 5602
    },
    {
      "epoch": 0.7852838121934127,
      "grad_norm": 0.22725702822208405,
      "learning_rate": 0.00022684046878737143,
      "loss": 0.1023,
      "step": 5603
    },
    {
      "epoch": 0.7854239663629993,
      "grad_norm": 0.4065977931022644,
      "learning_rate": 0.00022682611815355176,
      "loss": 0.0552,
      "step": 5604
    },
    {
      "epoch": 0.7855641205325858,
      "grad_norm": 0.5338484048843384,
      "learning_rate": 0.0002268117675197321,
      "loss": 0.0617,
      "step": 5605
    },
    {
      "epoch": 0.7857042747021724,
      "grad_norm": 0.39808306097984314,
      "learning_rate": 0.00022679741688591245,
      "loss": 0.0964,
      "step": 5606
    },
    {
      "epoch": 0.785844428871759,
      "grad_norm": 0.21572478115558624,
      "learning_rate": 0.00022678306625209277,
      "loss": 0.034,
      "step": 5607
    },
    {
      "epoch": 0.7859845830413454,
      "grad_norm": 0.19689863920211792,
      "learning_rate": 0.0002267687156182731,
      "loss": 0.0424,
      "step": 5608
    },
    {
      "epoch": 0.786124737210932,
      "grad_norm": 0.2556608319282532,
      "learning_rate": 0.00022675436498445349,
      "loss": 0.0455,
      "step": 5609
    },
    {
      "epoch": 0.7862648913805186,
      "grad_norm": 0.46096962690353394,
      "learning_rate": 0.00022674001435063381,
      "loss": 0.1173,
      "step": 5610
    },
    {
      "epoch": 0.7864050455501052,
      "grad_norm": 0.5848700404167175,
      "learning_rate": 0.00022672566371681414,
      "loss": 0.086,
      "step": 5611
    },
    {
      "epoch": 0.7865451997196916,
      "grad_norm": 0.256839394569397,
      "learning_rate": 0.0002267113130829945,
      "loss": 0.1097,
      "step": 5612
    },
    {
      "epoch": 0.7866853538892782,
      "grad_norm": 0.5051348209381104,
      "learning_rate": 0.00022669696244917483,
      "loss": 0.1139,
      "step": 5613
    },
    {
      "epoch": 0.7868255080588648,
      "grad_norm": 0.23546195030212402,
      "learning_rate": 0.00022668261181535516,
      "loss": 0.06,
      "step": 5614
    },
    {
      "epoch": 0.7869656622284513,
      "grad_norm": 0.3419448435306549,
      "learning_rate": 0.00022666826118153548,
      "loss": 0.0564,
      "step": 5615
    },
    {
      "epoch": 0.7871058163980379,
      "grad_norm": 0.32146215438842773,
      "learning_rate": 0.00022665391054771584,
      "loss": 0.0697,
      "step": 5616
    },
    {
      "epoch": 0.7872459705676244,
      "grad_norm": 0.6872513890266418,
      "learning_rate": 0.00022663955991389617,
      "loss": 0.1559,
      "step": 5617
    },
    {
      "epoch": 0.7873861247372109,
      "grad_norm": 0.4116996228694916,
      "learning_rate": 0.0002266252092800765,
      "loss": 0.0482,
      "step": 5618
    },
    {
      "epoch": 0.7875262789067975,
      "grad_norm": 0.27954569458961487,
      "learning_rate": 0.00022661085864625685,
      "loss": 0.0606,
      "step": 5619
    },
    {
      "epoch": 0.7876664330763841,
      "grad_norm": 0.46233686804771423,
      "learning_rate": 0.00022659650801243718,
      "loss": 0.0344,
      "step": 5620
    },
    {
      "epoch": 0.7878065872459705,
      "grad_norm": 0.267116516828537,
      "learning_rate": 0.00022658215737861754,
      "loss": 0.0802,
      "step": 5621
    },
    {
      "epoch": 0.7879467414155571,
      "grad_norm": 0.659602701663971,
      "learning_rate": 0.0002265678067447979,
      "loss": 0.0883,
      "step": 5622
    },
    {
      "epoch": 0.7880868955851437,
      "grad_norm": 0.4538685381412506,
      "learning_rate": 0.00022655345611097822,
      "loss": 0.0907,
      "step": 5623
    },
    {
      "epoch": 0.7882270497547302,
      "grad_norm": 1.5112347602844238,
      "learning_rate": 0.00022653910547715855,
      "loss": 0.1227,
      "step": 5624
    },
    {
      "epoch": 0.7883672039243167,
      "grad_norm": 0.16230528056621552,
      "learning_rate": 0.0002265247548433389,
      "loss": 0.0209,
      "step": 5625
    },
    {
      "epoch": 0.7885073580939033,
      "grad_norm": 0.6399327516555786,
      "learning_rate": 0.00022651040420951923,
      "loss": 0.0835,
      "step": 5626
    },
    {
      "epoch": 0.7886475122634898,
      "grad_norm": 0.4576042592525482,
      "learning_rate": 0.00022649605357569956,
      "loss": 0.0631,
      "step": 5627
    },
    {
      "epoch": 0.7887876664330764,
      "grad_norm": 0.8167616724967957,
      "learning_rate": 0.00022648170294187992,
      "loss": 0.0718,
      "step": 5628
    },
    {
      "epoch": 0.788927820602663,
      "grad_norm": 0.9858236908912659,
      "learning_rate": 0.00022646735230806025,
      "loss": 0.1232,
      "step": 5629
    },
    {
      "epoch": 0.7890679747722494,
      "grad_norm": 0.3055446147918701,
      "learning_rate": 0.00022645300167424058,
      "loss": 0.0735,
      "step": 5630
    },
    {
      "epoch": 0.789208128941836,
      "grad_norm": 0.3733578622341156,
      "learning_rate": 0.00022643865104042096,
      "loss": 0.0351,
      "step": 5631
    },
    {
      "epoch": 0.7893482831114226,
      "grad_norm": 0.6542314887046814,
      "learning_rate": 0.0002264243004066013,
      "loss": 0.1698,
      "step": 5632
    },
    {
      "epoch": 0.7894884372810091,
      "grad_norm": 0.2872672379016876,
      "learning_rate": 0.00022640994977278162,
      "loss": 0.0698,
      "step": 5633
    },
    {
      "epoch": 0.7896285914505956,
      "grad_norm": 0.5845891237258911,
      "learning_rate": 0.00022639559913896194,
      "loss": 0.1136,
      "step": 5634
    },
    {
      "epoch": 0.7897687456201822,
      "grad_norm": 0.6551676988601685,
      "learning_rate": 0.0002263812485051423,
      "loss": 0.0572,
      "step": 5635
    },
    {
      "epoch": 0.7899088997897687,
      "grad_norm": 0.3795095980167389,
      "learning_rate": 0.00022636689787132263,
      "loss": 0.1343,
      "step": 5636
    },
    {
      "epoch": 0.7900490539593553,
      "grad_norm": 0.7125422358512878,
      "learning_rate": 0.00022635254723750296,
      "loss": 0.0723,
      "step": 5637
    },
    {
      "epoch": 0.7901892081289419,
      "grad_norm": 0.6917170882225037,
      "learning_rate": 0.0002263381966036833,
      "loss": 0.138,
      "step": 5638
    },
    {
      "epoch": 0.7903293622985283,
      "grad_norm": 0.3661969304084778,
      "learning_rate": 0.00022632384596986364,
      "loss": 0.1047,
      "step": 5639
    },
    {
      "epoch": 0.7904695164681149,
      "grad_norm": 0.749763548374176,
      "learning_rate": 0.00022630949533604397,
      "loss": 0.1705,
      "step": 5640
    },
    {
      "epoch": 0.7906096706377015,
      "grad_norm": 0.519035816192627,
      "learning_rate": 0.00022629514470222435,
      "loss": 0.0955,
      "step": 5641
    },
    {
      "epoch": 0.7907498248072881,
      "grad_norm": 0.27287420630455017,
      "learning_rate": 0.00022628079406840468,
      "loss": 0.0674,
      "step": 5642
    },
    {
      "epoch": 0.7908899789768745,
      "grad_norm": 0.9136955738067627,
      "learning_rate": 0.000226266443434585,
      "loss": 0.0895,
      "step": 5643
    },
    {
      "epoch": 0.7910301331464611,
      "grad_norm": 0.44580692052841187,
      "learning_rate": 0.00022625209280076536,
      "loss": 0.0711,
      "step": 5644
    },
    {
      "epoch": 0.7911702873160477,
      "grad_norm": 0.9792532324790955,
      "learning_rate": 0.0002262377421669457,
      "loss": 0.0996,
      "step": 5645
    },
    {
      "epoch": 0.7913104414856342,
      "grad_norm": 0.29885461926460266,
      "learning_rate": 0.00022622339153312602,
      "loss": 0.0475,
      "step": 5646
    },
    {
      "epoch": 0.7914505956552207,
      "grad_norm": 0.30109041929244995,
      "learning_rate": 0.00022620904089930638,
      "loss": 0.0692,
      "step": 5647
    },
    {
      "epoch": 0.7915907498248073,
      "grad_norm": 1.1716101169586182,
      "learning_rate": 0.0002261946902654867,
      "loss": 0.3932,
      "step": 5648
    },
    {
      "epoch": 0.7917309039943938,
      "grad_norm": 1.0599429607391357,
      "learning_rate": 0.00022618033963166703,
      "loss": 0.2659,
      "step": 5649
    },
    {
      "epoch": 0.7918710581639804,
      "grad_norm": 1.750972867012024,
      "learning_rate": 0.00022616598899784736,
      "loss": 0.1428,
      "step": 5650
    },
    {
      "epoch": 0.792011212333567,
      "grad_norm": 0.5109270811080933,
      "learning_rate": 0.00022615163836402772,
      "loss": 0.0581,
      "step": 5651
    },
    {
      "epoch": 0.7921513665031534,
      "grad_norm": 0.22958607971668243,
      "learning_rate": 0.00022613728773020807,
      "loss": 0.0488,
      "step": 5652
    },
    {
      "epoch": 0.79229152067274,
      "grad_norm": 0.29007983207702637,
      "learning_rate": 0.0002261229370963884,
      "loss": 0.0509,
      "step": 5653
    },
    {
      "epoch": 0.7924316748423266,
      "grad_norm": 0.3183847963809967,
      "learning_rate": 0.00022610858646256876,
      "loss": 0.0606,
      "step": 5654
    },
    {
      "epoch": 0.7925718290119131,
      "grad_norm": 0.2664157450199127,
      "learning_rate": 0.0002260942358287491,
      "loss": 0.0413,
      "step": 5655
    },
    {
      "epoch": 0.7927119831814996,
      "grad_norm": 0.5876112580299377,
      "learning_rate": 0.00022607988519492942,
      "loss": 0.1074,
      "step": 5656
    },
    {
      "epoch": 0.7928521373510862,
      "grad_norm": 0.49838733673095703,
      "learning_rate": 0.00022606553456110977,
      "loss": 0.1187,
      "step": 5657
    },
    {
      "epoch": 0.7929922915206727,
      "grad_norm": 0.46227696537971497,
      "learning_rate": 0.0002260511839272901,
      "loss": 0.1082,
      "step": 5658
    },
    {
      "epoch": 0.7931324456902593,
      "grad_norm": 0.5369463562965393,
      "learning_rate": 0.00022603683329347043,
      "loss": 0.119,
      "step": 5659
    },
    {
      "epoch": 0.7932725998598459,
      "grad_norm": 0.4320513606071472,
      "learning_rate": 0.00022602248265965078,
      "loss": 0.0811,
      "step": 5660
    },
    {
      "epoch": 0.7934127540294323,
      "grad_norm": 0.43080368638038635,
      "learning_rate": 0.0002260081320258311,
      "loss": 0.0968,
      "step": 5661
    },
    {
      "epoch": 0.7935529081990189,
      "grad_norm": 0.3401932120323181,
      "learning_rate": 0.00022599378139201144,
      "loss": 0.1087,
      "step": 5662
    },
    {
      "epoch": 0.7936930623686055,
      "grad_norm": 0.5740426778793335,
      "learning_rate": 0.00022597943075819182,
      "loss": 0.0921,
      "step": 5663
    },
    {
      "epoch": 0.793833216538192,
      "grad_norm": 0.3936419188976288,
      "learning_rate": 0.00022596508012437215,
      "loss": 0.1118,
      "step": 5664
    },
    {
      "epoch": 0.7939733707077785,
      "grad_norm": 1.1023430824279785,
      "learning_rate": 0.00022595072949055248,
      "loss": 0.0962,
      "step": 5665
    },
    {
      "epoch": 0.7941135248773651,
      "grad_norm": 0.2322964370250702,
      "learning_rate": 0.0002259363788567328,
      "loss": 0.0769,
      "step": 5666
    },
    {
      "epoch": 0.7942536790469517,
      "grad_norm": 0.33208635449409485,
      "learning_rate": 0.00022592202822291317,
      "loss": 0.0521,
      "step": 5667
    },
    {
      "epoch": 0.7943938332165382,
      "grad_norm": 0.7058702111244202,
      "learning_rate": 0.0002259076775890935,
      "loss": 0.1431,
      "step": 5668
    },
    {
      "epoch": 0.7945339873861247,
      "grad_norm": 0.4282469153404236,
      "learning_rate": 0.00022589332695527382,
      "loss": 0.0603,
      "step": 5669
    },
    {
      "epoch": 0.7946741415557113,
      "grad_norm": 0.7234341502189636,
      "learning_rate": 0.00022587897632145418,
      "loss": 0.0782,
      "step": 5670
    },
    {
      "epoch": 0.7948142957252978,
      "grad_norm": 0.3753330111503601,
      "learning_rate": 0.0002258646256876345,
      "loss": 0.109,
      "step": 5671
    },
    {
      "epoch": 0.7949544498948844,
      "grad_norm": 0.6509559750556946,
      "learning_rate": 0.00022585027505381484,
      "loss": 0.0855,
      "step": 5672
    },
    {
      "epoch": 0.795094604064471,
      "grad_norm": 0.11885274201631546,
      "learning_rate": 0.00022583592441999522,
      "loss": 0.0194,
      "step": 5673
    },
    {
      "epoch": 0.7952347582340574,
      "grad_norm": 0.32345446944236755,
      "learning_rate": 0.00022582157378617555,
      "loss": 0.071,
      "step": 5674
    },
    {
      "epoch": 0.795374912403644,
      "grad_norm": 0.440663605928421,
      "learning_rate": 0.00022580722315235588,
      "loss": 0.0324,
      "step": 5675
    },
    {
      "epoch": 0.7955150665732306,
      "grad_norm": 0.4897230863571167,
      "learning_rate": 0.00022579287251853623,
      "loss": 0.0775,
      "step": 5676
    },
    {
      "epoch": 0.7956552207428171,
      "grad_norm": 0.24911180138587952,
      "learning_rate": 0.00022577852188471656,
      "loss": 0.0419,
      "step": 5677
    },
    {
      "epoch": 0.7957953749124036,
      "grad_norm": 0.3806205093860626,
      "learning_rate": 0.0002257641712508969,
      "loss": 0.111,
      "step": 5678
    },
    {
      "epoch": 0.7959355290819902,
      "grad_norm": 0.8356031179428101,
      "learning_rate": 0.00022574982061707724,
      "loss": 0.0537,
      "step": 5679
    },
    {
      "epoch": 0.7960756832515767,
      "grad_norm": 0.544108510017395,
      "learning_rate": 0.00022573546998325757,
      "loss": 0.0807,
      "step": 5680
    },
    {
      "epoch": 0.7962158374211633,
      "grad_norm": 0.32353276014328003,
      "learning_rate": 0.0002257211193494379,
      "loss": 0.0638,
      "step": 5681
    },
    {
      "epoch": 0.7963559915907499,
      "grad_norm": 0.6253033876419067,
      "learning_rate": 0.00022570676871561826,
      "loss": 0.0785,
      "step": 5682
    },
    {
      "epoch": 0.7964961457603363,
      "grad_norm": 0.6292871236801147,
      "learning_rate": 0.00022569241808179859,
      "loss": 0.1029,
      "step": 5683
    },
    {
      "epoch": 0.7966362999299229,
      "grad_norm": 0.670504629611969,
      "learning_rate": 0.00022567806744797894,
      "loss": 0.0557,
      "step": 5684
    },
    {
      "epoch": 0.7967764540995095,
      "grad_norm": 0.2368209958076477,
      "learning_rate": 0.00022566371681415927,
      "loss": 0.051,
      "step": 5685
    },
    {
      "epoch": 0.796916608269096,
      "grad_norm": 0.23010286688804626,
      "learning_rate": 0.00022564936618033963,
      "loss": 0.0573,
      "step": 5686
    },
    {
      "epoch": 0.7970567624386825,
      "grad_norm": 0.37756916880607605,
      "learning_rate": 0.00022563501554651995,
      "loss": 0.0656,
      "step": 5687
    },
    {
      "epoch": 0.7971969166082691,
      "grad_norm": 0.26231664419174194,
      "learning_rate": 0.00022562066491270028,
      "loss": 0.0291,
      "step": 5688
    },
    {
      "epoch": 0.7973370707778556,
      "grad_norm": 0.6982463002204895,
      "learning_rate": 0.00022560631427888064,
      "loss": 0.1424,
      "step": 5689
    },
    {
      "epoch": 0.7974772249474422,
      "grad_norm": 0.4191700518131256,
      "learning_rate": 0.00022559196364506097,
      "loss": 0.0466,
      "step": 5690
    },
    {
      "epoch": 0.7976173791170287,
      "grad_norm": 0.928926408290863,
      "learning_rate": 0.0002255776130112413,
      "loss": 0.2481,
      "step": 5691
    },
    {
      "epoch": 0.7977575332866153,
      "grad_norm": 0.4801519811153412,
      "learning_rate": 0.00022556326237742165,
      "loss": 0.1646,
      "step": 5692
    },
    {
      "epoch": 0.7978976874562018,
      "grad_norm": 0.4181602895259857,
      "learning_rate": 0.00022554891174360198,
      "loss": 0.0868,
      "step": 5693
    },
    {
      "epoch": 0.7980378416257884,
      "grad_norm": 0.3504292666912079,
      "learning_rate": 0.0002255345611097823,
      "loss": 0.1661,
      "step": 5694
    },
    {
      "epoch": 0.798177995795375,
      "grad_norm": 1.0696107149124146,
      "learning_rate": 0.0002255202104759627,
      "loss": 0.1435,
      "step": 5695
    },
    {
      "epoch": 0.7983181499649614,
      "grad_norm": 0.4525793790817261,
      "learning_rate": 0.00022550585984214302,
      "loss": 0.0449,
      "step": 5696
    },
    {
      "epoch": 0.798458304134548,
      "grad_norm": 0.483450323343277,
      "learning_rate": 0.00022549150920832335,
      "loss": 0.0923,
      "step": 5697
    },
    {
      "epoch": 0.7985984583041346,
      "grad_norm": 1.3730570077896118,
      "learning_rate": 0.0002254771585745037,
      "loss": 0.2055,
      "step": 5698
    },
    {
      "epoch": 0.7987386124737211,
      "grad_norm": 2.2538154125213623,
      "learning_rate": 0.00022546280794068403,
      "loss": 0.1648,
      "step": 5699
    },
    {
      "epoch": 0.7988787666433076,
      "grad_norm": 2.04182505607605,
      "learning_rate": 0.00022544845730686436,
      "loss": 0.0512,
      "step": 5700
    },
    {
      "epoch": 0.7990189208128942,
      "grad_norm": 0.33751675486564636,
      "learning_rate": 0.0002254341066730447,
      "loss": 0.0894,
      "step": 5701
    },
    {
      "epoch": 0.7991590749824807,
      "grad_norm": 0.3281555771827698,
      "learning_rate": 0.00022541975603922504,
      "loss": 0.0623,
      "step": 5702
    },
    {
      "epoch": 0.7992992291520673,
      "grad_norm": 0.5163365006446838,
      "learning_rate": 0.00022540540540540537,
      "loss": 0.0995,
      "step": 5703
    },
    {
      "epoch": 0.7994393833216539,
      "grad_norm": 0.2273314893245697,
      "learning_rate": 0.0002253910547715857,
      "loss": 0.0475,
      "step": 5704
    },
    {
      "epoch": 0.7995795374912403,
      "grad_norm": 0.20162153244018555,
      "learning_rate": 0.00022537670413776608,
      "loss": 0.0315,
      "step": 5705
    },
    {
      "epoch": 0.7997196916608269,
      "grad_norm": 0.23051901161670685,
      "learning_rate": 0.0002253623535039464,
      "loss": 0.0342,
      "step": 5706
    },
    {
      "epoch": 0.7998598458304135,
      "grad_norm": 0.3236226737499237,
      "learning_rate": 0.00022534800287012674,
      "loss": 0.1278,
      "step": 5707
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.33776968717575073,
      "learning_rate": 0.0002253336522363071,
      "loss": 0.1005,
      "step": 5708
    },
    {
      "epoch": 0.8001401541695865,
      "grad_norm": 0.7175341844558716,
      "learning_rate": 0.00022531930160248743,
      "loss": 0.1846,
      "step": 5709
    },
    {
      "epoch": 0.8002803083391731,
      "grad_norm": 0.4873778223991394,
      "learning_rate": 0.00022530495096866775,
      "loss": 0.0326,
      "step": 5710
    },
    {
      "epoch": 0.8004204625087596,
      "grad_norm": 0.39647483825683594,
      "learning_rate": 0.0002252906003348481,
      "loss": 0.0768,
      "step": 5711
    },
    {
      "epoch": 0.8005606166783462,
      "grad_norm": 0.6795065402984619,
      "learning_rate": 0.00022527624970102844,
      "loss": 0.0995,
      "step": 5712
    },
    {
      "epoch": 0.8007007708479327,
      "grad_norm": 0.265045702457428,
      "learning_rate": 0.00022526189906720877,
      "loss": 0.0606,
      "step": 5713
    },
    {
      "epoch": 0.8008409250175192,
      "grad_norm": 0.33189502358436584,
      "learning_rate": 0.00022524754843338912,
      "loss": 0.0733,
      "step": 5714
    },
    {
      "epoch": 0.8009810791871058,
      "grad_norm": 0.19205160439014435,
      "learning_rate": 0.00022523319779956945,
      "loss": 0.0416,
      "step": 5715
    },
    {
      "epoch": 0.8011212333566924,
      "grad_norm": 0.45440179109573364,
      "learning_rate": 0.0002252188471657498,
      "loss": 0.0671,
      "step": 5716
    },
    {
      "epoch": 0.801261387526279,
      "grad_norm": 0.3498120605945587,
      "learning_rate": 0.00022520449653193016,
      "loss": 0.0775,
      "step": 5717
    },
    {
      "epoch": 0.8014015416958654,
      "grad_norm": 0.4218611419200897,
      "learning_rate": 0.0002251901458981105,
      "loss": 0.1143,
      "step": 5718
    },
    {
      "epoch": 0.801541695865452,
      "grad_norm": 0.6124241948127747,
      "learning_rate": 0.00022517579526429082,
      "loss": 0.0991,
      "step": 5719
    },
    {
      "epoch": 0.8016818500350386,
      "grad_norm": 0.6980561017990112,
      "learning_rate": 0.00022516144463047115,
      "loss": 0.0781,
      "step": 5720
    },
    {
      "epoch": 0.8018220042046251,
      "grad_norm": 0.5807704925537109,
      "learning_rate": 0.0002251470939966515,
      "loss": 0.0438,
      "step": 5721
    },
    {
      "epoch": 0.8019621583742116,
      "grad_norm": 0.7196307182312012,
      "learning_rate": 0.00022513274336283183,
      "loss": 0.1886,
      "step": 5722
    },
    {
      "epoch": 0.8021023125437982,
      "grad_norm": 0.23914958536624908,
      "learning_rate": 0.00022511839272901216,
      "loss": 0.0531,
      "step": 5723
    },
    {
      "epoch": 0.8022424667133847,
      "grad_norm": 0.32259708642959595,
      "learning_rate": 0.00022510404209519252,
      "loss": 0.0908,
      "step": 5724
    },
    {
      "epoch": 0.8023826208829713,
      "grad_norm": 0.5749737024307251,
      "learning_rate": 0.00022508969146137285,
      "loss": 0.0836,
      "step": 5725
    },
    {
      "epoch": 0.8025227750525579,
      "grad_norm": 0.19791029393672943,
      "learning_rate": 0.00022507534082755317,
      "loss": 0.0341,
      "step": 5726
    },
    {
      "epoch": 0.8026629292221443,
      "grad_norm": 0.3118803799152374,
      "learning_rate": 0.00022506099019373356,
      "loss": 0.0748,
      "step": 5727
    },
    {
      "epoch": 0.8028030833917309,
      "grad_norm": 0.40618568658828735,
      "learning_rate": 0.00022504663955991389,
      "loss": 0.1146,
      "step": 5728
    },
    {
      "epoch": 0.8029432375613175,
      "grad_norm": 0.47597694396972656,
      "learning_rate": 0.00022503228892609421,
      "loss": 0.1232,
      "step": 5729
    },
    {
      "epoch": 0.803083391730904,
      "grad_norm": 1.004313588142395,
      "learning_rate": 0.00022501793829227457,
      "loss": 0.1368,
      "step": 5730
    },
    {
      "epoch": 0.8032235459004905,
      "grad_norm": 0.4950355291366577,
      "learning_rate": 0.0002250035876584549,
      "loss": 0.0774,
      "step": 5731
    },
    {
      "epoch": 0.8033637000700771,
      "grad_norm": 0.3069462478160858,
      "learning_rate": 0.00022498923702463523,
      "loss": 0.0592,
      "step": 5732
    },
    {
      "epoch": 0.8035038542396636,
      "grad_norm": 0.8063193559646606,
      "learning_rate": 0.00022497488639081558,
      "loss": 0.1375,
      "step": 5733
    },
    {
      "epoch": 0.8036440084092502,
      "grad_norm": 0.34387367963790894,
      "learning_rate": 0.0002249605357569959,
      "loss": 0.0559,
      "step": 5734
    },
    {
      "epoch": 0.8037841625788367,
      "grad_norm": 0.1765451431274414,
      "learning_rate": 0.00022494618512317624,
      "loss": 0.0188,
      "step": 5735
    },
    {
      "epoch": 0.8039243167484232,
      "grad_norm": 0.5219098329544067,
      "learning_rate": 0.00022493183448935657,
      "loss": 0.062,
      "step": 5736
    },
    {
      "epoch": 0.8040644709180098,
      "grad_norm": 0.7034408450126648,
      "learning_rate": 0.00022491748385553695,
      "loss": 0.1295,
      "step": 5737
    },
    {
      "epoch": 0.8042046250875964,
      "grad_norm": 0.24827638268470764,
      "learning_rate": 0.00022490313322171728,
      "loss": 0.0754,
      "step": 5738
    },
    {
      "epoch": 0.804344779257183,
      "grad_norm": 0.5348634123802185,
      "learning_rate": 0.0002248887825878976,
      "loss": 0.1076,
      "step": 5739
    },
    {
      "epoch": 0.8044849334267694,
      "grad_norm": 0.61797696352005,
      "learning_rate": 0.00022487443195407796,
      "loss": 0.1031,
      "step": 5740
    },
    {
      "epoch": 0.804625087596356,
      "grad_norm": 0.9629631042480469,
      "learning_rate": 0.0002248600813202583,
      "loss": 0.1487,
      "step": 5741
    },
    {
      "epoch": 0.8047652417659426,
      "grad_norm": 0.8176407217979431,
      "learning_rate": 0.00022484573068643862,
      "loss": 0.0783,
      "step": 5742
    },
    {
      "epoch": 0.8049053959355291,
      "grad_norm": 0.38148871064186096,
      "learning_rate": 0.00022483138005261898,
      "loss": 0.0533,
      "step": 5743
    },
    {
      "epoch": 0.8050455501051156,
      "grad_norm": 1.0174052715301514,
      "learning_rate": 0.0002248170294187993,
      "loss": 0.2,
      "step": 5744
    },
    {
      "epoch": 0.8051857042747022,
      "grad_norm": 0.8366091847419739,
      "learning_rate": 0.00022480267878497963,
      "loss": 0.2349,
      "step": 5745
    },
    {
      "epoch": 0.8053258584442887,
      "grad_norm": 0.2741507589817047,
      "learning_rate": 0.00022478832815116,
      "loss": 0.0494,
      "step": 5746
    },
    {
      "epoch": 0.8054660126138753,
      "grad_norm": 0.4730146825313568,
      "learning_rate": 0.00022477397751734032,
      "loss": 0.1758,
      "step": 5747
    },
    {
      "epoch": 0.8056061667834619,
      "grad_norm": 1.077645182609558,
      "learning_rate": 0.00022475962688352067,
      "loss": 0.1563,
      "step": 5748
    },
    {
      "epoch": 0.8057463209530483,
      "grad_norm": 1.776120662689209,
      "learning_rate": 0.00022474527624970103,
      "loss": 0.1765,
      "step": 5749
    },
    {
      "epoch": 0.8058864751226349,
      "grad_norm": 1.5276013612747192,
      "learning_rate": 0.00022473092561588136,
      "loss": 0.1371,
      "step": 5750
    },
    {
      "epoch": 0.8060266292922215,
      "grad_norm": 0.2624564468860626,
      "learning_rate": 0.00022471657498206169,
      "loss": 0.0297,
      "step": 5751
    },
    {
      "epoch": 0.806166783461808,
      "grad_norm": 0.5053154826164246,
      "learning_rate": 0.00022470222434824204,
      "loss": 0.0731,
      "step": 5752
    },
    {
      "epoch": 0.8063069376313945,
      "grad_norm": 0.4104936122894287,
      "learning_rate": 0.00022468787371442237,
      "loss": 0.0983,
      "step": 5753
    },
    {
      "epoch": 0.8064470918009811,
      "grad_norm": 0.30682364106178284,
      "learning_rate": 0.0002246735230806027,
      "loss": 0.0673,
      "step": 5754
    },
    {
      "epoch": 0.8065872459705676,
      "grad_norm": 0.2573780119419098,
      "learning_rate": 0.00022465917244678303,
      "loss": 0.0321,
      "step": 5755
    },
    {
      "epoch": 0.8067274001401542,
      "grad_norm": 0.20758210122585297,
      "learning_rate": 0.00022464482181296338,
      "loss": 0.0426,
      "step": 5756
    },
    {
      "epoch": 0.8068675543097407,
      "grad_norm": 0.7352272272109985,
      "learning_rate": 0.0002246304711791437,
      "loss": 0.1429,
      "step": 5757
    },
    {
      "epoch": 0.8070077084793272,
      "grad_norm": 0.5787365436553955,
      "learning_rate": 0.00022461612054532404,
      "loss": 0.0959,
      "step": 5758
    },
    {
      "epoch": 0.8071478626489138,
      "grad_norm": 0.20439884066581726,
      "learning_rate": 0.00022460176991150442,
      "loss": 0.0349,
      "step": 5759
    },
    {
      "epoch": 0.8072880168185004,
      "grad_norm": 0.18559855222702026,
      "learning_rate": 0.00022458741927768475,
      "loss": 0.0401,
      "step": 5760
    },
    {
      "epoch": 0.807428170988087,
      "grad_norm": 0.3092149794101715,
      "learning_rate": 0.00022457306864386508,
      "loss": 0.0593,
      "step": 5761
    },
    {
      "epoch": 0.8075683251576734,
      "grad_norm": 1.3547148704528809,
      "learning_rate": 0.00022455871801004544,
      "loss": 0.1303,
      "step": 5762
    },
    {
      "epoch": 0.80770847932726,
      "grad_norm": 0.2493240386247635,
      "learning_rate": 0.00022454436737622576,
      "loss": 0.0503,
      "step": 5763
    },
    {
      "epoch": 0.8078486334968465,
      "grad_norm": 0.2818904519081116,
      "learning_rate": 0.0002245300167424061,
      "loss": 0.0516,
      "step": 5764
    },
    {
      "epoch": 0.8079887876664331,
      "grad_norm": 0.38934430480003357,
      "learning_rate": 0.00022451566610858645,
      "loss": 0.1413,
      "step": 5765
    },
    {
      "epoch": 0.8081289418360196,
      "grad_norm": 0.48351791501045227,
      "learning_rate": 0.00022450131547476678,
      "loss": 0.0753,
      "step": 5766
    },
    {
      "epoch": 0.8082690960056061,
      "grad_norm": 0.3817238211631775,
      "learning_rate": 0.0002244869648409471,
      "loss": 0.0713,
      "step": 5767
    },
    {
      "epoch": 0.8084092501751927,
      "grad_norm": 0.3684089481830597,
      "learning_rate": 0.0002244726142071275,
      "loss": 0.1094,
      "step": 5768
    },
    {
      "epoch": 0.8085494043447793,
      "grad_norm": 0.9726632237434387,
      "learning_rate": 0.00022445826357330782,
      "loss": 0.1708,
      "step": 5769
    },
    {
      "epoch": 0.8086895585143657,
      "grad_norm": 1.1153467893600464,
      "learning_rate": 0.00022444391293948815,
      "loss": 0.2096,
      "step": 5770
    },
    {
      "epoch": 0.8088297126839523,
      "grad_norm": 0.49349766969680786,
      "learning_rate": 0.00022442956230566847,
      "loss": 0.1011,
      "step": 5771
    },
    {
      "epoch": 0.8089698668535389,
      "grad_norm": 0.5160754323005676,
      "learning_rate": 0.00022441521167184883,
      "loss": 0.1151,
      "step": 5772
    },
    {
      "epoch": 0.8091100210231255,
      "grad_norm": 0.2126186043024063,
      "learning_rate": 0.00022440086103802916,
      "loss": 0.0674,
      "step": 5773
    },
    {
      "epoch": 0.809250175192712,
      "grad_norm": 0.22338438034057617,
      "learning_rate": 0.0002243865104042095,
      "loss": 0.0767,
      "step": 5774
    },
    {
      "epoch": 0.8093903293622985,
      "grad_norm": 0.4069470763206482,
      "learning_rate": 0.00022437215977038984,
      "loss": 0.071,
      "step": 5775
    },
    {
      "epoch": 0.8095304835318851,
      "grad_norm": 0.5842556357383728,
      "learning_rate": 0.00022435780913657017,
      "loss": 0.1704,
      "step": 5776
    },
    {
      "epoch": 0.8096706377014716,
      "grad_norm": 1.6323877573013306,
      "learning_rate": 0.0002243434585027505,
      "loss": 0.0505,
      "step": 5777
    },
    {
      "epoch": 0.8098107918710582,
      "grad_norm": 0.4551006555557251,
      "learning_rate": 0.00022432910786893086,
      "loss": 0.1373,
      "step": 5778
    },
    {
      "epoch": 0.8099509460406447,
      "grad_norm": 0.5307178497314453,
      "learning_rate": 0.0002243147572351112,
      "loss": 0.0807,
      "step": 5779
    },
    {
      "epoch": 0.8100911002102312,
      "grad_norm": 0.5500494837760925,
      "learning_rate": 0.00022430040660129154,
      "loss": 0.0925,
      "step": 5780
    },
    {
      "epoch": 0.8102312543798178,
      "grad_norm": 0.47352516651153564,
      "learning_rate": 0.0002242860559674719,
      "loss": 0.1001,
      "step": 5781
    },
    {
      "epoch": 0.8103714085494044,
      "grad_norm": 0.3682084083557129,
      "learning_rate": 0.00022427170533365222,
      "loss": 0.0381,
      "step": 5782
    },
    {
      "epoch": 0.8105115627189909,
      "grad_norm": 0.7366754412651062,
      "learning_rate": 0.00022425735469983255,
      "loss": 0.1299,
      "step": 5783
    },
    {
      "epoch": 0.8106517168885774,
      "grad_norm": 0.4509728252887726,
      "learning_rate": 0.0002242430040660129,
      "loss": 0.0657,
      "step": 5784
    },
    {
      "epoch": 0.810791871058164,
      "grad_norm": 0.3227069079875946,
      "learning_rate": 0.00022422865343219324,
      "loss": 0.1501,
      "step": 5785
    },
    {
      "epoch": 0.8109320252277505,
      "grad_norm": 0.6010557413101196,
      "learning_rate": 0.00022421430279837357,
      "loss": 0.2439,
      "step": 5786
    },
    {
      "epoch": 0.8110721793973371,
      "grad_norm": 0.5877246260643005,
      "learning_rate": 0.00022419995216455392,
      "loss": 0.1577,
      "step": 5787
    },
    {
      "epoch": 0.8112123335669236,
      "grad_norm": 0.6178323030471802,
      "learning_rate": 0.00022418560153073425,
      "loss": 0.1568,
      "step": 5788
    },
    {
      "epoch": 0.8113524877365101,
      "grad_norm": 1.3562370538711548,
      "learning_rate": 0.00022417125089691458,
      "loss": 0.0626,
      "step": 5789
    },
    {
      "epoch": 0.8114926419060967,
      "grad_norm": 0.33482205867767334,
      "learning_rate": 0.0002241569002630949,
      "loss": 0.0561,
      "step": 5790
    },
    {
      "epoch": 0.8116327960756833,
      "grad_norm": 0.6903762221336365,
      "learning_rate": 0.0002241425496292753,
      "loss": 0.103,
      "step": 5791
    },
    {
      "epoch": 0.8117729502452697,
      "grad_norm": 0.5988749265670776,
      "learning_rate": 0.00022412819899545562,
      "loss": 0.1001,
      "step": 5792
    },
    {
      "epoch": 0.8119131044148563,
      "grad_norm": 1.2338522672653198,
      "learning_rate": 0.00022411384836163595,
      "loss": 0.1074,
      "step": 5793
    },
    {
      "epoch": 0.8120532585844429,
      "grad_norm": 0.9725277423858643,
      "learning_rate": 0.0002240994977278163,
      "loss": 0.1568,
      "step": 5794
    },
    {
      "epoch": 0.8121934127540295,
      "grad_norm": 2.4673266410827637,
      "learning_rate": 0.00022408514709399663,
      "loss": 0.1338,
      "step": 5795
    },
    {
      "epoch": 0.812333566923616,
      "grad_norm": 1.4851458072662354,
      "learning_rate": 0.00022407079646017696,
      "loss": 0.0907,
      "step": 5796
    },
    {
      "epoch": 0.8124737210932025,
      "grad_norm": 0.7200006246566772,
      "learning_rate": 0.00022405644582635731,
      "loss": 0.0541,
      "step": 5797
    },
    {
      "epoch": 0.812613875262789,
      "grad_norm": 1.2098779678344727,
      "learning_rate": 0.00022404209519253764,
      "loss": 0.1066,
      "step": 5798
    },
    {
      "epoch": 0.8127540294323756,
      "grad_norm": 1.121396541595459,
      "learning_rate": 0.00022402774455871797,
      "loss": 0.0875,
      "step": 5799
    },
    {
      "epoch": 0.8128941836019622,
      "grad_norm": 6.2350687980651855,
      "learning_rate": 0.00022401339392489835,
      "loss": 0.8495,
      "step": 5800
    },
    {
      "epoch": 0.8130343377715487,
      "grad_norm": 0.1380949169397354,
      "learning_rate": 0.00022399904329107868,
      "loss": 0.0294,
      "step": 5801
    },
    {
      "epoch": 0.8131744919411352,
      "grad_norm": 0.5304329991340637,
      "learning_rate": 0.000223984692657259,
      "loss": 0.133,
      "step": 5802
    },
    {
      "epoch": 0.8133146461107218,
      "grad_norm": 0.31048327684402466,
      "learning_rate": 0.00022397034202343937,
      "loss": 0.0642,
      "step": 5803
    },
    {
      "epoch": 0.8134548002803084,
      "grad_norm": 0.4840710759162903,
      "learning_rate": 0.0002239559913896197,
      "loss": 0.057,
      "step": 5804
    },
    {
      "epoch": 0.8135949544498949,
      "grad_norm": 0.37240302562713623,
      "learning_rate": 0.00022394164075580002,
      "loss": 0.1105,
      "step": 5805
    },
    {
      "epoch": 0.8137351086194814,
      "grad_norm": 0.1933498978614807,
      "learning_rate": 0.00022392729012198035,
      "loss": 0.0291,
      "step": 5806
    },
    {
      "epoch": 0.813875262789068,
      "grad_norm": 0.20238856971263885,
      "learning_rate": 0.0002239129394881607,
      "loss": 0.0394,
      "step": 5807
    },
    {
      "epoch": 0.8140154169586545,
      "grad_norm": 0.4747373163700104,
      "learning_rate": 0.00022389858885434104,
      "loss": 0.111,
      "step": 5808
    },
    {
      "epoch": 0.8141555711282411,
      "grad_norm": 0.3709792196750641,
      "learning_rate": 0.00022388423822052137,
      "loss": 0.1024,
      "step": 5809
    },
    {
      "epoch": 0.8142957252978276,
      "grad_norm": 0.48583874106407166,
      "learning_rate": 0.00022386988758670172,
      "loss": 0.1087,
      "step": 5810
    },
    {
      "epoch": 0.8144358794674141,
      "grad_norm": 0.14270281791687012,
      "learning_rate": 0.00022385553695288208,
      "loss": 0.0385,
      "step": 5811
    },
    {
      "epoch": 0.8145760336370007,
      "grad_norm": 0.6404475569725037,
      "learning_rate": 0.0002238411863190624,
      "loss": 0.1617,
      "step": 5812
    },
    {
      "epoch": 0.8147161878065873,
      "grad_norm": 0.2255086600780487,
      "learning_rate": 0.00022382683568524276,
      "loss": 0.0283,
      "step": 5813
    },
    {
      "epoch": 0.8148563419761737,
      "grad_norm": 0.27812886238098145,
      "learning_rate": 0.0002238124850514231,
      "loss": 0.0932,
      "step": 5814
    },
    {
      "epoch": 0.8149964961457603,
      "grad_norm": 0.28141188621520996,
      "learning_rate": 0.00022379813441760342,
      "loss": 0.0936,
      "step": 5815
    },
    {
      "epoch": 0.8151366503153469,
      "grad_norm": 0.2614344656467438,
      "learning_rate": 0.00022378378378378377,
      "loss": 0.0626,
      "step": 5816
    },
    {
      "epoch": 0.8152768044849334,
      "grad_norm": 0.3129107356071472,
      "learning_rate": 0.0002237694331499641,
      "loss": 0.0641,
      "step": 5817
    },
    {
      "epoch": 0.81541695865452,
      "grad_norm": 0.5294141173362732,
      "learning_rate": 0.00022375508251614443,
      "loss": 0.0578,
      "step": 5818
    },
    {
      "epoch": 0.8155571128241065,
      "grad_norm": 0.26989293098449707,
      "learning_rate": 0.0002237407318823248,
      "loss": 0.1115,
      "step": 5819
    },
    {
      "epoch": 0.815697266993693,
      "grad_norm": 0.3717700242996216,
      "learning_rate": 0.00022372638124850512,
      "loss": 0.0803,
      "step": 5820
    },
    {
      "epoch": 0.8158374211632796,
      "grad_norm": 0.4185395836830139,
      "learning_rate": 0.00022371203061468544,
      "loss": 0.1166,
      "step": 5821
    },
    {
      "epoch": 0.8159775753328662,
      "grad_norm": 0.427578866481781,
      "learning_rate": 0.00022369767998086583,
      "loss": 0.0472,
      "step": 5822
    },
    {
      "epoch": 0.8161177295024526,
      "grad_norm": 0.2768086791038513,
      "learning_rate": 0.00022368332934704616,
      "loss": 0.0486,
      "step": 5823
    },
    {
      "epoch": 0.8162578836720392,
      "grad_norm": 0.16227155923843384,
      "learning_rate": 0.00022366897871322648,
      "loss": 0.0364,
      "step": 5824
    },
    {
      "epoch": 0.8163980378416258,
      "grad_norm": 0.6284443736076355,
      "learning_rate": 0.0002236546280794068,
      "loss": 0.0852,
      "step": 5825
    },
    {
      "epoch": 0.8165381920112124,
      "grad_norm": 0.46867087483406067,
      "learning_rate": 0.00022364027744558717,
      "loss": 0.0786,
      "step": 5826
    },
    {
      "epoch": 0.8166783461807989,
      "grad_norm": 0.2443040907382965,
      "learning_rate": 0.0002236259268117675,
      "loss": 0.0617,
      "step": 5827
    },
    {
      "epoch": 0.8168185003503854,
      "grad_norm": 0.3987634479999542,
      "learning_rate": 0.00022361157617794783,
      "loss": 0.1299,
      "step": 5828
    },
    {
      "epoch": 0.816958654519972,
      "grad_norm": 0.2775398790836334,
      "learning_rate": 0.00022359722554412818,
      "loss": 0.1027,
      "step": 5829
    },
    {
      "epoch": 0.8170988086895585,
      "grad_norm": 0.3505871295928955,
      "learning_rate": 0.0002235828749103085,
      "loss": 0.0662,
      "step": 5830
    },
    {
      "epoch": 0.8172389628591451,
      "grad_norm": 0.3641906678676605,
      "learning_rate": 0.00022356852427648884,
      "loss": 0.1043,
      "step": 5831
    },
    {
      "epoch": 0.8173791170287316,
      "grad_norm": 0.9073621034622192,
      "learning_rate": 0.00022355417364266922,
      "loss": 0.1224,
      "step": 5832
    },
    {
      "epoch": 0.8175192711983181,
      "grad_norm": 0.20240426063537598,
      "learning_rate": 0.00022353982300884955,
      "loss": 0.0455,
      "step": 5833
    },
    {
      "epoch": 0.8176594253679047,
      "grad_norm": 0.4446265697479248,
      "learning_rate": 0.00022352547237502988,
      "loss": 0.0858,
      "step": 5834
    },
    {
      "epoch": 0.8177995795374913,
      "grad_norm": 0.23124608397483826,
      "learning_rate": 0.00022351112174121023,
      "loss": 0.109,
      "step": 5835
    },
    {
      "epoch": 0.8179397337070777,
      "grad_norm": 0.7142744660377502,
      "learning_rate": 0.00022349677110739056,
      "loss": 0.1285,
      "step": 5836
    },
    {
      "epoch": 0.8180798878766643,
      "grad_norm": 0.24515023827552795,
      "learning_rate": 0.0002234824204735709,
      "loss": 0.0459,
      "step": 5837
    },
    {
      "epoch": 0.8182200420462509,
      "grad_norm": 0.3710487484931946,
      "learning_rate": 0.00022346806983975125,
      "loss": 0.0945,
      "step": 5838
    },
    {
      "epoch": 0.8183601962158374,
      "grad_norm": 0.4483098089694977,
      "learning_rate": 0.00022345371920593158,
      "loss": 0.0514,
      "step": 5839
    },
    {
      "epoch": 0.818500350385424,
      "grad_norm": 0.2247656136751175,
      "learning_rate": 0.0002234393685721119,
      "loss": 0.0368,
      "step": 5840
    },
    {
      "epoch": 0.8186405045550105,
      "grad_norm": 0.5277408957481384,
      "learning_rate": 0.00022342501793829223,
      "loss": 0.0903,
      "step": 5841
    },
    {
      "epoch": 0.818780658724597,
      "grad_norm": 0.2889882028102875,
      "learning_rate": 0.0002234106673044726,
      "loss": 0.0817,
      "step": 5842
    },
    {
      "epoch": 0.8189208128941836,
      "grad_norm": 0.4435153305530548,
      "learning_rate": 0.00022339631667065294,
      "loss": 0.0444,
      "step": 5843
    },
    {
      "epoch": 0.8190609670637702,
      "grad_norm": 0.530709981918335,
      "learning_rate": 0.00022338196603683327,
      "loss": 0.0748,
      "step": 5844
    },
    {
      "epoch": 0.8192011212333566,
      "grad_norm": 0.3209708034992218,
      "learning_rate": 0.00022336761540301363,
      "loss": 0.0424,
      "step": 5845
    },
    {
      "epoch": 0.8193412754029432,
      "grad_norm": 1.7540130615234375,
      "learning_rate": 0.00022335326476919396,
      "loss": 0.1205,
      "step": 5846
    },
    {
      "epoch": 0.8194814295725298,
      "grad_norm": 1.0781437158584595,
      "learning_rate": 0.00022333891413537429,
      "loss": 0.0966,
      "step": 5847
    },
    {
      "epoch": 0.8196215837421164,
      "grad_norm": 0.09919463843107224,
      "learning_rate": 0.00022332456350155464,
      "loss": 0.0111,
      "step": 5848
    },
    {
      "epoch": 0.8197617379117029,
      "grad_norm": 2.7272090911865234,
      "learning_rate": 0.00022331021286773497,
      "loss": 0.351,
      "step": 5849
    },
    {
      "epoch": 0.8199018920812894,
      "grad_norm": 0.835983157157898,
      "learning_rate": 0.0002232958622339153,
      "loss": 0.0842,
      "step": 5850
    },
    {
      "epoch": 0.820042046250876,
      "grad_norm": 0.2904302775859833,
      "learning_rate": 0.00022328151160009565,
      "loss": 0.0398,
      "step": 5851
    },
    {
      "epoch": 0.8201822004204625,
      "grad_norm": 0.6469781994819641,
      "learning_rate": 0.00022326716096627598,
      "loss": 0.0915,
      "step": 5852
    },
    {
      "epoch": 0.8203223545900491,
      "grad_norm": 0.9117149114608765,
      "learning_rate": 0.0002232528103324563,
      "loss": 0.0832,
      "step": 5853
    },
    {
      "epoch": 0.8204625087596356,
      "grad_norm": 0.243239164352417,
      "learning_rate": 0.0002232384596986367,
      "loss": 0.0696,
      "step": 5854
    },
    {
      "epoch": 0.8206026629292221,
      "grad_norm": 0.3760521113872528,
      "learning_rate": 0.00022322410906481702,
      "loss": 0.0883,
      "step": 5855
    },
    {
      "epoch": 0.8207428170988087,
      "grad_norm": 0.4392783045768738,
      "learning_rate": 0.00022320975843099735,
      "loss": 0.0422,
      "step": 5856
    },
    {
      "epoch": 0.8208829712683953,
      "grad_norm": 0.3056473135948181,
      "learning_rate": 0.0002231954077971777,
      "loss": 0.0887,
      "step": 5857
    },
    {
      "epoch": 0.8210231254379817,
      "grad_norm": 0.5537986159324646,
      "learning_rate": 0.00022318105716335803,
      "loss": 0.0625,
      "step": 5858
    },
    {
      "epoch": 0.8211632796075683,
      "grad_norm": 0.672101616859436,
      "learning_rate": 0.00022316670652953836,
      "loss": 0.1328,
      "step": 5859
    },
    {
      "epoch": 0.8213034337771549,
      "grad_norm": 0.45674440264701843,
      "learning_rate": 0.0002231523558957187,
      "loss": 0.0582,
      "step": 5860
    },
    {
      "epoch": 0.8214435879467414,
      "grad_norm": 0.5701112747192383,
      "learning_rate": 0.00022313800526189905,
      "loss": 0.0796,
      "step": 5861
    },
    {
      "epoch": 0.821583742116328,
      "grad_norm": 0.38268429040908813,
      "learning_rate": 0.00022312365462807938,
      "loss": 0.0419,
      "step": 5862
    },
    {
      "epoch": 0.8217238962859145,
      "grad_norm": 0.31397050619125366,
      "learning_rate": 0.0002231093039942597,
      "loss": 0.0295,
      "step": 5863
    },
    {
      "epoch": 0.821864050455501,
      "grad_norm": 0.3330279290676117,
      "learning_rate": 0.0002230949533604401,
      "loss": 0.0268,
      "step": 5864
    },
    {
      "epoch": 0.8220042046250876,
      "grad_norm": 0.23322667181491852,
      "learning_rate": 0.00022308060272662042,
      "loss": 0.0697,
      "step": 5865
    },
    {
      "epoch": 0.8221443587946742,
      "grad_norm": 0.2703094482421875,
      "learning_rate": 0.00022306625209280074,
      "loss": 0.0683,
      "step": 5866
    },
    {
      "epoch": 0.8222845129642606,
      "grad_norm": 0.1767052859067917,
      "learning_rate": 0.0002230519014589811,
      "loss": 0.0383,
      "step": 5867
    },
    {
      "epoch": 0.8224246671338472,
      "grad_norm": 0.5386223793029785,
      "learning_rate": 0.00022303755082516143,
      "loss": 0.1271,
      "step": 5868
    },
    {
      "epoch": 0.8225648213034338,
      "grad_norm": 0.36625948548316956,
      "learning_rate": 0.00022302320019134176,
      "loss": 0.1122,
      "step": 5869
    },
    {
      "epoch": 0.8227049754730204,
      "grad_norm": 0.28537946939468384,
      "learning_rate": 0.0002230088495575221,
      "loss": 0.0513,
      "step": 5870
    },
    {
      "epoch": 0.8228451296426069,
      "grad_norm": 0.19931276142597198,
      "learning_rate": 0.00022299449892370244,
      "loss": 0.0289,
      "step": 5871
    },
    {
      "epoch": 0.8229852838121934,
      "grad_norm": 0.6263478398323059,
      "learning_rate": 0.00022298014828988277,
      "loss": 0.1233,
      "step": 5872
    },
    {
      "epoch": 0.82312543798178,
      "grad_norm": 0.29002660512924194,
      "learning_rate": 0.00022296579765606313,
      "loss": 0.0967,
      "step": 5873
    },
    {
      "epoch": 0.8232655921513665,
      "grad_norm": 0.31298041343688965,
      "learning_rate": 0.00022295144702224348,
      "loss": 0.058,
      "step": 5874
    },
    {
      "epoch": 0.8234057463209531,
      "grad_norm": 0.21126209199428558,
      "learning_rate": 0.0002229370963884238,
      "loss": 0.0486,
      "step": 5875
    },
    {
      "epoch": 0.8235459004905396,
      "grad_norm": 0.7859790921211243,
      "learning_rate": 0.00022292274575460414,
      "loss": 0.059,
      "step": 5876
    },
    {
      "epoch": 0.8236860546601261,
      "grad_norm": 0.33230945467948914,
      "learning_rate": 0.0002229083951207845,
      "loss": 0.0784,
      "step": 5877
    },
    {
      "epoch": 0.8238262088297127,
      "grad_norm": 0.3842691481113434,
      "learning_rate": 0.00022289404448696482,
      "loss": 0.0637,
      "step": 5878
    },
    {
      "epoch": 0.8239663629992993,
      "grad_norm": 0.41438397765159607,
      "learning_rate": 0.00022287969385314515,
      "loss": 0.1201,
      "step": 5879
    },
    {
      "epoch": 0.8241065171688857,
      "grad_norm": 0.45761585235595703,
      "learning_rate": 0.0002228653432193255,
      "loss": 0.1326,
      "step": 5880
    },
    {
      "epoch": 0.8242466713384723,
      "grad_norm": 0.32369065284729004,
      "learning_rate": 0.00022285099258550584,
      "loss": 0.062,
      "step": 5881
    },
    {
      "epoch": 0.8243868255080589,
      "grad_norm": 0.6559203863143921,
      "learning_rate": 0.00022283664195168616,
      "loss": 0.0779,
      "step": 5882
    },
    {
      "epoch": 0.8245269796776454,
      "grad_norm": 0.8068817257881165,
      "learning_rate": 0.00022282229131786652,
      "loss": 0.0845,
      "step": 5883
    },
    {
      "epoch": 0.824667133847232,
      "grad_norm": 0.2590385675430298,
      "learning_rate": 0.00022280794068404685,
      "loss": 0.0255,
      "step": 5884
    },
    {
      "epoch": 0.8248072880168185,
      "grad_norm": 0.5571833252906799,
      "learning_rate": 0.00022279359005022718,
      "loss": 0.0443,
      "step": 5885
    },
    {
      "epoch": 0.824947442186405,
      "grad_norm": 0.39487841725349426,
      "learning_rate": 0.00022277923941640756,
      "loss": 0.0643,
      "step": 5886
    },
    {
      "epoch": 0.8250875963559916,
      "grad_norm": 0.8468427062034607,
      "learning_rate": 0.0002227648887825879,
      "loss": 0.1607,
      "step": 5887
    },
    {
      "epoch": 0.8252277505255782,
      "grad_norm": 0.49531927704811096,
      "learning_rate": 0.00022275053814876822,
      "loss": 0.1023,
      "step": 5888
    },
    {
      "epoch": 0.8253679046951646,
      "grad_norm": 0.2164953500032425,
      "learning_rate": 0.00022273618751494857,
      "loss": 0.0302,
      "step": 5889
    },
    {
      "epoch": 0.8255080588647512,
      "grad_norm": 0.4535398781299591,
      "learning_rate": 0.0002227218368811289,
      "loss": 0.1252,
      "step": 5890
    },
    {
      "epoch": 0.8256482130343378,
      "grad_norm": 0.38749533891677856,
      "learning_rate": 0.00022270748624730923,
      "loss": 0.0642,
      "step": 5891
    },
    {
      "epoch": 0.8257883672039243,
      "grad_norm": 0.5470385551452637,
      "learning_rate": 0.00022269313561348959,
      "loss": 0.0984,
      "step": 5892
    },
    {
      "epoch": 0.8259285213735109,
      "grad_norm": 0.5258098244667053,
      "learning_rate": 0.00022267878497966991,
      "loss": 0.0558,
      "step": 5893
    },
    {
      "epoch": 0.8260686755430974,
      "grad_norm": 0.7119280695915222,
      "learning_rate": 0.00022266443434585024,
      "loss": 0.0993,
      "step": 5894
    },
    {
      "epoch": 0.826208829712684,
      "grad_norm": 1.5012911558151245,
      "learning_rate": 0.00022265008371203057,
      "loss": 0.0783,
      "step": 5895
    },
    {
      "epoch": 0.8263489838822705,
      "grad_norm": 0.911300539970398,
      "learning_rate": 0.00022263573307821095,
      "loss": 0.1116,
      "step": 5896
    },
    {
      "epoch": 0.8264891380518571,
      "grad_norm": 0.43797796964645386,
      "learning_rate": 0.00022262138244439128,
      "loss": 0.0453,
      "step": 5897
    },
    {
      "epoch": 0.8266292922214435,
      "grad_norm": 2.994680881500244,
      "learning_rate": 0.0002226070318105716,
      "loss": 0.2262,
      "step": 5898
    },
    {
      "epoch": 0.8267694463910301,
      "grad_norm": 1.300034761428833,
      "learning_rate": 0.00022259268117675197,
      "loss": 0.1085,
      "step": 5899
    },
    {
      "epoch": 0.8269096005606167,
      "grad_norm": 2.6300525665283203,
      "learning_rate": 0.0002225783305429323,
      "loss": 0.2636,
      "step": 5900
    },
    {
      "epoch": 0.8270497547302033,
      "grad_norm": 0.8500336408615112,
      "learning_rate": 0.00022256397990911262,
      "loss": 0.234,
      "step": 5901
    },
    {
      "epoch": 0.8271899088997897,
      "grad_norm": 0.2362862080335617,
      "learning_rate": 0.00022254962927529298,
      "loss": 0.0762,
      "step": 5902
    },
    {
      "epoch": 0.8273300630693763,
      "grad_norm": 0.2184583991765976,
      "learning_rate": 0.0002225352786414733,
      "loss": 0.0409,
      "step": 5903
    },
    {
      "epoch": 0.8274702172389629,
      "grad_norm": 0.33143872022628784,
      "learning_rate": 0.00022252092800765364,
      "loss": 0.1045,
      "step": 5904
    },
    {
      "epoch": 0.8276103714085494,
      "grad_norm": 0.580802857875824,
      "learning_rate": 0.000222506577373834,
      "loss": 0.2047,
      "step": 5905
    },
    {
      "epoch": 0.827750525578136,
      "grad_norm": 0.3154565095901489,
      "learning_rate": 0.00022249222674001435,
      "loss": 0.0866,
      "step": 5906
    },
    {
      "epoch": 0.8278906797477225,
      "grad_norm": 0.6071909070014954,
      "learning_rate": 0.00022247787610619468,
      "loss": 0.122,
      "step": 5907
    },
    {
      "epoch": 0.828030833917309,
      "grad_norm": 0.41419368982315063,
      "learning_rate": 0.00022246352547237503,
      "loss": 0.0845,
      "step": 5908
    },
    {
      "epoch": 0.8281709880868956,
      "grad_norm": 0.45883846282958984,
      "learning_rate": 0.00022244917483855536,
      "loss": 0.0807,
      "step": 5909
    },
    {
      "epoch": 0.8283111422564822,
      "grad_norm": 0.454587459564209,
      "learning_rate": 0.0002224348242047357,
      "loss": 0.09,
      "step": 5910
    },
    {
      "epoch": 0.8284512964260686,
      "grad_norm": 0.489064484834671,
      "learning_rate": 0.00022242047357091602,
      "loss": 0.0775,
      "step": 5911
    },
    {
      "epoch": 0.8285914505956552,
      "grad_norm": 0.4669618308544159,
      "learning_rate": 0.00022240612293709637,
      "loss": 0.0519,
      "step": 5912
    },
    {
      "epoch": 0.8287316047652418,
      "grad_norm": 0.4833787977695465,
      "learning_rate": 0.0002223917723032767,
      "loss": 0.1109,
      "step": 5913
    },
    {
      "epoch": 0.8288717589348283,
      "grad_norm": 0.26459962129592896,
      "learning_rate": 0.00022237742166945703,
      "loss": 0.0363,
      "step": 5914
    },
    {
      "epoch": 0.8290119131044149,
      "grad_norm": 0.42832279205322266,
      "learning_rate": 0.00022236307103563739,
      "loss": 0.0988,
      "step": 5915
    },
    {
      "epoch": 0.8291520672740014,
      "grad_norm": 0.33144134283065796,
      "learning_rate": 0.00022234872040181771,
      "loss": 0.0597,
      "step": 5916
    },
    {
      "epoch": 0.8292922214435879,
      "grad_norm": 0.25004151463508606,
      "learning_rate": 0.00022233436976799804,
      "loss": 0.0602,
      "step": 5917
    },
    {
      "epoch": 0.8294323756131745,
      "grad_norm": 0.687878429889679,
      "learning_rate": 0.00022232001913417843,
      "loss": 0.0955,
      "step": 5918
    },
    {
      "epoch": 0.8295725297827611,
      "grad_norm": 0.13679423928260803,
      "learning_rate": 0.00022230566850035875,
      "loss": 0.0233,
      "step": 5919
    },
    {
      "epoch": 0.8297126839523475,
      "grad_norm": 0.42223963141441345,
      "learning_rate": 0.00022229131786653908,
      "loss": 0.1175,
      "step": 5920
    },
    {
      "epoch": 0.8298528381219341,
      "grad_norm": 0.3276427984237671,
      "learning_rate": 0.00022227696723271944,
      "loss": 0.0648,
      "step": 5921
    },
    {
      "epoch": 0.8299929922915207,
      "grad_norm": 0.5942558646202087,
      "learning_rate": 0.00022226261659889977,
      "loss": 0.0724,
      "step": 5922
    },
    {
      "epoch": 0.8301331464611073,
      "grad_norm": 0.2597961723804474,
      "learning_rate": 0.0002222482659650801,
      "loss": 0.0618,
      "step": 5923
    },
    {
      "epoch": 0.8302733006306937,
      "grad_norm": 0.3672686815261841,
      "learning_rate": 0.00022223391533126045,
      "loss": 0.075,
      "step": 5924
    },
    {
      "epoch": 0.8304134548002803,
      "grad_norm": 0.3052956163883209,
      "learning_rate": 0.00022221956469744078,
      "loss": 0.0785,
      "step": 5925
    },
    {
      "epoch": 0.8305536089698669,
      "grad_norm": 0.29609790444374084,
      "learning_rate": 0.0002222052140636211,
      "loss": 0.0401,
      "step": 5926
    },
    {
      "epoch": 0.8306937631394534,
      "grad_norm": 0.48648586869239807,
      "learning_rate": 0.0002221908634298015,
      "loss": 0.076,
      "step": 5927
    },
    {
      "epoch": 0.83083391730904,
      "grad_norm": 0.6608290672302246,
      "learning_rate": 0.00022217651279598182,
      "loss": 0.1657,
      "step": 5928
    },
    {
      "epoch": 0.8309740714786265,
      "grad_norm": 0.24940335750579834,
      "learning_rate": 0.00022216216216216215,
      "loss": 0.048,
      "step": 5929
    },
    {
      "epoch": 0.831114225648213,
      "grad_norm": 0.4135406017303467,
      "learning_rate": 0.00022214781152834248,
      "loss": 0.0443,
      "step": 5930
    },
    {
      "epoch": 0.8312543798177996,
      "grad_norm": 0.5187567472457886,
      "learning_rate": 0.00022213346089452283,
      "loss": 0.0642,
      "step": 5931
    },
    {
      "epoch": 0.8313945339873862,
      "grad_norm": 0.25692304968833923,
      "learning_rate": 0.00022211911026070316,
      "loss": 0.0451,
      "step": 5932
    },
    {
      "epoch": 0.8315346881569726,
      "grad_norm": 0.3205924332141876,
      "learning_rate": 0.0002221047596268835,
      "loss": 0.0696,
      "step": 5933
    },
    {
      "epoch": 0.8316748423265592,
      "grad_norm": 0.33431941270828247,
      "learning_rate": 0.00022209040899306385,
      "loss": 0.0709,
      "step": 5934
    },
    {
      "epoch": 0.8318149964961458,
      "grad_norm": 0.43129605054855347,
      "learning_rate": 0.00022207605835924417,
      "loss": 0.0859,
      "step": 5935
    },
    {
      "epoch": 0.8319551506657323,
      "grad_norm": 0.9323621988296509,
      "learning_rate": 0.0002220617077254245,
      "loss": 0.0196,
      "step": 5936
    },
    {
      "epoch": 0.8320953048353189,
      "grad_norm": 0.24619603157043457,
      "learning_rate": 0.00022204735709160486,
      "loss": 0.0929,
      "step": 5937
    },
    {
      "epoch": 0.8322354590049054,
      "grad_norm": 0.5866087675094604,
      "learning_rate": 0.00022203300645778521,
      "loss": 0.0252,
      "step": 5938
    },
    {
      "epoch": 0.8323756131744919,
      "grad_norm": 0.7721034288406372,
      "learning_rate": 0.00022201865582396554,
      "loss": 0.1732,
      "step": 5939
    },
    {
      "epoch": 0.8325157673440785,
      "grad_norm": 0.5209160447120667,
      "learning_rate": 0.0002220043051901459,
      "loss": 0.1002,
      "step": 5940
    },
    {
      "epoch": 0.8326559215136651,
      "grad_norm": 0.3384513556957245,
      "learning_rate": 0.00022198995455632623,
      "loss": 0.0477,
      "step": 5941
    },
    {
      "epoch": 0.8327960756832515,
      "grad_norm": 0.32500627636909485,
      "learning_rate": 0.00022197560392250656,
      "loss": 0.0853,
      "step": 5942
    },
    {
      "epoch": 0.8329362298528381,
      "grad_norm": 1.1235716342926025,
      "learning_rate": 0.0002219612532886869,
      "loss": 0.1431,
      "step": 5943
    },
    {
      "epoch": 0.8330763840224247,
      "grad_norm": 0.2482185959815979,
      "learning_rate": 0.00022194690265486724,
      "loss": 0.0261,
      "step": 5944
    },
    {
      "epoch": 0.8332165381920112,
      "grad_norm": 0.773568868637085,
      "learning_rate": 0.00022193255202104757,
      "loss": 0.1627,
      "step": 5945
    },
    {
      "epoch": 0.8333566923615977,
      "grad_norm": 0.737103283405304,
      "learning_rate": 0.0002219182013872279,
      "loss": 0.0852,
      "step": 5946
    },
    {
      "epoch": 0.8334968465311843,
      "grad_norm": 1.8099327087402344,
      "learning_rate": 0.00022190385075340825,
      "loss": 0.2878,
      "step": 5947
    },
    {
      "epoch": 0.8336370007007708,
      "grad_norm": 0.8018434047698975,
      "learning_rate": 0.00022188950011958858,
      "loss": 0.0873,
      "step": 5948
    },
    {
      "epoch": 0.8337771548703574,
      "grad_norm": 2.499140501022339,
      "learning_rate": 0.0002218751494857689,
      "loss": 0.3278,
      "step": 5949
    },
    {
      "epoch": 0.833917309039944,
      "grad_norm": 1.1449474096298218,
      "learning_rate": 0.0002218607988519493,
      "loss": 0.3077,
      "step": 5950
    },
    {
      "epoch": 0.8340574632095304,
      "grad_norm": 0.5713278651237488,
      "learning_rate": 0.00022184644821812962,
      "loss": 0.0957,
      "step": 5951
    },
    {
      "epoch": 0.834197617379117,
      "grad_norm": 0.47468826174736023,
      "learning_rate": 0.00022183209758430995,
      "loss": 0.0752,
      "step": 5952
    },
    {
      "epoch": 0.8343377715487036,
      "grad_norm": 0.1592639684677124,
      "learning_rate": 0.0002218177469504903,
      "loss": 0.0306,
      "step": 5953
    },
    {
      "epoch": 0.8344779257182902,
      "grad_norm": 0.3566886782646179,
      "learning_rate": 0.00022180339631667063,
      "loss": 0.0929,
      "step": 5954
    },
    {
      "epoch": 0.8346180798878766,
      "grad_norm": 0.21627689898014069,
      "learning_rate": 0.00022178904568285096,
      "loss": 0.0506,
      "step": 5955
    },
    {
      "epoch": 0.8347582340574632,
      "grad_norm": 0.6209261417388916,
      "learning_rate": 0.00022177469504903132,
      "loss": 0.0825,
      "step": 5956
    },
    {
      "epoch": 0.8348983882270498,
      "grad_norm": 0.5948876738548279,
      "learning_rate": 0.00022176034441521165,
      "loss": 0.1176,
      "step": 5957
    },
    {
      "epoch": 0.8350385423966363,
      "grad_norm": 0.20633652806282043,
      "learning_rate": 0.00022174599378139197,
      "loss": 0.0372,
      "step": 5958
    },
    {
      "epoch": 0.8351786965662229,
      "grad_norm": 0.6971942186355591,
      "learning_rate": 0.00022173164314757236,
      "loss": 0.1589,
      "step": 5959
    },
    {
      "epoch": 0.8353188507358094,
      "grad_norm": 0.22771628201007843,
      "learning_rate": 0.00022171729251375269,
      "loss": 0.0281,
      "step": 5960
    },
    {
      "epoch": 0.8354590049053959,
      "grad_norm": 1.0242925882339478,
      "learning_rate": 0.00022170294187993301,
      "loss": 0.1849,
      "step": 5961
    },
    {
      "epoch": 0.8355991590749825,
      "grad_norm": 0.5470095872879028,
      "learning_rate": 0.00022168859124611334,
      "loss": 0.0822,
      "step": 5962
    },
    {
      "epoch": 0.8357393132445691,
      "grad_norm": 0.8643908500671387,
      "learning_rate": 0.0002216742406122937,
      "loss": 0.0873,
      "step": 5963
    },
    {
      "epoch": 0.8358794674141555,
      "grad_norm": 0.21783781051635742,
      "learning_rate": 0.00022165988997847403,
      "loss": 0.0338,
      "step": 5964
    },
    {
      "epoch": 0.8360196215837421,
      "grad_norm": 0.4660545587539673,
      "learning_rate": 0.00022164553934465436,
      "loss": 0.0826,
      "step": 5965
    },
    {
      "epoch": 0.8361597757533287,
      "grad_norm": 0.2196199744939804,
      "learning_rate": 0.0002216311887108347,
      "loss": 0.0519,
      "step": 5966
    },
    {
      "epoch": 0.8362999299229152,
      "grad_norm": 0.16608960926532745,
      "learning_rate": 0.00022161683807701504,
      "loss": 0.0237,
      "step": 5967
    },
    {
      "epoch": 0.8364400840925017,
      "grad_norm": 0.2225629985332489,
      "learning_rate": 0.00022160248744319537,
      "loss": 0.0227,
      "step": 5968
    },
    {
      "epoch": 0.8365802382620883,
      "grad_norm": 0.4260781705379486,
      "learning_rate": 0.00022158813680937572,
      "loss": 0.1013,
      "step": 5969
    },
    {
      "epoch": 0.8367203924316748,
      "grad_norm": 0.48793455958366394,
      "learning_rate": 0.00022157378617555608,
      "loss": 0.0772,
      "step": 5970
    },
    {
      "epoch": 0.8368605466012614,
      "grad_norm": 0.3152455985546112,
      "learning_rate": 0.0002215594355417364,
      "loss": 0.0512,
      "step": 5971
    },
    {
      "epoch": 0.837000700770848,
      "grad_norm": 0.254903644323349,
      "learning_rate": 0.00022154508490791676,
      "loss": 0.11,
      "step": 5972
    },
    {
      "epoch": 0.8371408549404344,
      "grad_norm": 0.6641345024108887,
      "learning_rate": 0.0002215307342740971,
      "loss": 0.1697,
      "step": 5973
    },
    {
      "epoch": 0.837281009110021,
      "grad_norm": 0.31637808680534363,
      "learning_rate": 0.00022151638364027742,
      "loss": 0.0296,
      "step": 5974
    },
    {
      "epoch": 0.8374211632796076,
      "grad_norm": 0.7511847615242004,
      "learning_rate": 0.00022150203300645778,
      "loss": 0.151,
      "step": 5975
    },
    {
      "epoch": 0.8375613174491942,
      "grad_norm": 0.3790588974952698,
      "learning_rate": 0.0002214876823726381,
      "loss": 0.0578,
      "step": 5976
    },
    {
      "epoch": 0.8377014716187806,
      "grad_norm": 0.24922585487365723,
      "learning_rate": 0.00022147333173881843,
      "loss": 0.0792,
      "step": 5977
    },
    {
      "epoch": 0.8378416257883672,
      "grad_norm": 0.1837388575077057,
      "learning_rate": 0.0002214589811049988,
      "loss": 0.0597,
      "step": 5978
    },
    {
      "epoch": 0.8379817799579538,
      "grad_norm": 0.4960850477218628,
      "learning_rate": 0.00022144463047117912,
      "loss": 0.1342,
      "step": 5979
    },
    {
      "epoch": 0.8381219341275403,
      "grad_norm": 0.28974518179893494,
      "learning_rate": 0.00022143027983735945,
      "loss": 0.1104,
      "step": 5980
    },
    {
      "epoch": 0.8382620882971269,
      "grad_norm": 0.8710290193557739,
      "learning_rate": 0.00022141592920353978,
      "loss": 0.0755,
      "step": 5981
    },
    {
      "epoch": 0.8384022424667134,
      "grad_norm": 0.3801742494106293,
      "learning_rate": 0.00022140157856972016,
      "loss": 0.057,
      "step": 5982
    },
    {
      "epoch": 0.8385423966362999,
      "grad_norm": 0.1014970988035202,
      "learning_rate": 0.0002213872279359005,
      "loss": 0.0114,
      "step": 5983
    },
    {
      "epoch": 0.8386825508058865,
      "grad_norm": 0.3004118502140045,
      "learning_rate": 0.00022137287730208082,
      "loss": 0.0459,
      "step": 5984
    },
    {
      "epoch": 0.8388227049754731,
      "grad_norm": 0.29902586340904236,
      "learning_rate": 0.00022135852666826117,
      "loss": 0.0579,
      "step": 5985
    },
    {
      "epoch": 0.8389628591450595,
      "grad_norm": 0.5611525177955627,
      "learning_rate": 0.0002213441760344415,
      "loss": 0.0704,
      "step": 5986
    },
    {
      "epoch": 0.8391030133146461,
      "grad_norm": 0.8856817483901978,
      "learning_rate": 0.00022132982540062183,
      "loss": 0.1368,
      "step": 5987
    },
    {
      "epoch": 0.8392431674842327,
      "grad_norm": 0.4001542627811432,
      "learning_rate": 0.00022131547476680218,
      "loss": 0.0501,
      "step": 5988
    },
    {
      "epoch": 0.8393833216538192,
      "grad_norm": 0.3253493905067444,
      "learning_rate": 0.0002213011241329825,
      "loss": 0.0343,
      "step": 5989
    },
    {
      "epoch": 0.8395234758234057,
      "grad_norm": 0.45205754041671753,
      "learning_rate": 0.00022128677349916284,
      "loss": 0.0853,
      "step": 5990
    },
    {
      "epoch": 0.8396636299929923,
      "grad_norm": 0.41299745440483093,
      "learning_rate": 0.00022127242286534322,
      "loss": 0.1086,
      "step": 5991
    },
    {
      "epoch": 0.8398037841625788,
      "grad_norm": 0.2580900192260742,
      "learning_rate": 0.00022125807223152355,
      "loss": 0.0811,
      "step": 5992
    },
    {
      "epoch": 0.8399439383321654,
      "grad_norm": 0.13802768290042877,
      "learning_rate": 0.00022124372159770388,
      "loss": 0.0367,
      "step": 5993
    },
    {
      "epoch": 0.840084092501752,
      "grad_norm": 0.414334774017334,
      "learning_rate": 0.00022122937096388424,
      "loss": 0.0713,
      "step": 5994
    },
    {
      "epoch": 0.8402242466713384,
      "grad_norm": 1.27986478805542,
      "learning_rate": 0.00022121502033006457,
      "loss": 0.2287,
      "step": 5995
    },
    {
      "epoch": 0.840364400840925,
      "grad_norm": 0.42905113101005554,
      "learning_rate": 0.0002212006696962449,
      "loss": 0.0413,
      "step": 5996
    },
    {
      "epoch": 0.8405045550105116,
      "grad_norm": 0.44794151186943054,
      "learning_rate": 0.00022118631906242522,
      "loss": 0.0375,
      "step": 5997
    },
    {
      "epoch": 0.8406447091800981,
      "grad_norm": 0.883091390132904,
      "learning_rate": 0.00022117196842860558,
      "loss": 0.2403,
      "step": 5998
    },
    {
      "epoch": 0.8407848633496846,
      "grad_norm": 0.8485433459281921,
      "learning_rate": 0.0002211576177947859,
      "loss": 0.0415,
      "step": 5999
    },
    {
      "epoch": 0.8409250175192712,
      "grad_norm": 2.7033872604370117,
      "learning_rate": 0.00022114326716096624,
      "loss": 0.3868,
      "step": 6000
    },
    {
      "epoch": 0.8410651716888577,
      "grad_norm": 0.42621660232543945,
      "learning_rate": 0.00022112891652714662,
      "loss": 0.0569,
      "step": 6001
    },
    {
      "epoch": 0.8412053258584443,
      "grad_norm": 0.40759584307670593,
      "learning_rate": 0.00022111456589332695,
      "loss": 0.0953,
      "step": 6002
    },
    {
      "epoch": 0.8413454800280308,
      "grad_norm": 0.16722503304481506,
      "learning_rate": 0.00022110021525950728,
      "loss": 0.0444,
      "step": 6003
    },
    {
      "epoch": 0.8414856341976173,
      "grad_norm": 0.2938117980957031,
      "learning_rate": 0.00022108586462568763,
      "loss": 0.0992,
      "step": 6004
    },
    {
      "epoch": 0.8416257883672039,
      "grad_norm": 0.3903235197067261,
      "learning_rate": 0.00022107151399186796,
      "loss": 0.1352,
      "step": 6005
    },
    {
      "epoch": 0.8417659425367905,
      "grad_norm": 0.4225024878978729,
      "learning_rate": 0.0002210571633580483,
      "loss": 0.0702,
      "step": 6006
    },
    {
      "epoch": 0.8419060967063771,
      "grad_norm": 0.4744155704975128,
      "learning_rate": 0.00022104281272422864,
      "loss": 0.1148,
      "step": 6007
    },
    {
      "epoch": 0.8420462508759635,
      "grad_norm": 0.229496031999588,
      "learning_rate": 0.00022102846209040897,
      "loss": 0.0342,
      "step": 6008
    },
    {
      "epoch": 0.8421864050455501,
      "grad_norm": 0.24346406757831573,
      "learning_rate": 0.0002210141114565893,
      "loss": 0.0371,
      "step": 6009
    },
    {
      "epoch": 0.8423265592151367,
      "grad_norm": 0.5424032211303711,
      "learning_rate": 0.00022099976082276966,
      "loss": 0.1306,
      "step": 6010
    },
    {
      "epoch": 0.8424667133847232,
      "grad_norm": 0.4252154529094696,
      "learning_rate": 0.00022098541018894998,
      "loss": 0.0905,
      "step": 6011
    },
    {
      "epoch": 0.8426068675543097,
      "grad_norm": 0.2243867814540863,
      "learning_rate": 0.0002209710595551303,
      "loss": 0.0709,
      "step": 6012
    },
    {
      "epoch": 0.8427470217238963,
      "grad_norm": 0.5185843706130981,
      "learning_rate": 0.0002209567089213107,
      "loss": 0.0984,
      "step": 6013
    },
    {
      "epoch": 0.8428871758934828,
      "grad_norm": 0.27996328473091125,
      "learning_rate": 0.00022094235828749102,
      "loss": 0.0425,
      "step": 6014
    },
    {
      "epoch": 0.8430273300630694,
      "grad_norm": 0.4408407509326935,
      "learning_rate": 0.00022092800765367135,
      "loss": 0.1021,
      "step": 6015
    },
    {
      "epoch": 0.843167484232656,
      "grad_norm": 0.5334112644195557,
      "learning_rate": 0.00022091365701985168,
      "loss": 0.101,
      "step": 6016
    },
    {
      "epoch": 0.8433076384022424,
      "grad_norm": 0.3409167230129242,
      "learning_rate": 0.00022089930638603204,
      "loss": 0.0573,
      "step": 6017
    },
    {
      "epoch": 0.843447792571829,
      "grad_norm": 0.21492350101470947,
      "learning_rate": 0.00022088495575221237,
      "loss": 0.0426,
      "step": 6018
    },
    {
      "epoch": 0.8435879467414156,
      "grad_norm": 0.18287070095539093,
      "learning_rate": 0.0002208706051183927,
      "loss": 0.0541,
      "step": 6019
    },
    {
      "epoch": 0.8437281009110021,
      "grad_norm": 0.6449204683303833,
      "learning_rate": 0.00022085625448457305,
      "loss": 0.0998,
      "step": 6020
    },
    {
      "epoch": 0.8438682550805886,
      "grad_norm": 0.8856091499328613,
      "learning_rate": 0.00022084190385075338,
      "loss": 0.0529,
      "step": 6021
    },
    {
      "epoch": 0.8440084092501752,
      "grad_norm": 0.4567941427230835,
      "learning_rate": 0.0002208275532169337,
      "loss": 0.0735,
      "step": 6022
    },
    {
      "epoch": 0.8441485634197617,
      "grad_norm": 0.5005852580070496,
      "learning_rate": 0.0002208132025831141,
      "loss": 0.0553,
      "step": 6023
    },
    {
      "epoch": 0.8442887175893483,
      "grad_norm": 0.31614986062049866,
      "learning_rate": 0.00022079885194929442,
      "loss": 0.0892,
      "step": 6024
    },
    {
      "epoch": 0.8444288717589348,
      "grad_norm": 0.4281679093837738,
      "learning_rate": 0.00022078450131547475,
      "loss": 0.0733,
      "step": 6025
    },
    {
      "epoch": 0.8445690259285213,
      "grad_norm": 0.18619506061077118,
      "learning_rate": 0.0002207701506816551,
      "loss": 0.0363,
      "step": 6026
    },
    {
      "epoch": 0.8447091800981079,
      "grad_norm": 0.29114648699760437,
      "learning_rate": 0.00022075580004783543,
      "loss": 0.0735,
      "step": 6027
    },
    {
      "epoch": 0.8448493342676945,
      "grad_norm": 0.29656782746315,
      "learning_rate": 0.00022074144941401576,
      "loss": 0.0536,
      "step": 6028
    },
    {
      "epoch": 0.844989488437281,
      "grad_norm": 0.3082650601863861,
      "learning_rate": 0.00022072709878019612,
      "loss": 0.065,
      "step": 6029
    },
    {
      "epoch": 0.8451296426068675,
      "grad_norm": 0.3323041796684265,
      "learning_rate": 0.00022071274814637644,
      "loss": 0.0422,
      "step": 6030
    },
    {
      "epoch": 0.8452697967764541,
      "grad_norm": 0.4899621903896332,
      "learning_rate": 0.00022069839751255677,
      "loss": 0.102,
      "step": 6031
    },
    {
      "epoch": 0.8454099509460407,
      "grad_norm": 0.6736263036727905,
      "learning_rate": 0.0002206840468787371,
      "loss": 0.1277,
      "step": 6032
    },
    {
      "epoch": 0.8455501051156272,
      "grad_norm": 0.18957358598709106,
      "learning_rate": 0.00022066969624491748,
      "loss": 0.0512,
      "step": 6033
    },
    {
      "epoch": 0.8456902592852137,
      "grad_norm": 0.32455477118492126,
      "learning_rate": 0.0002206553456110978,
      "loss": 0.0611,
      "step": 6034
    },
    {
      "epoch": 0.8458304134548003,
      "grad_norm": 0.7106560468673706,
      "learning_rate": 0.00022064099497727814,
      "loss": 0.0795,
      "step": 6035
    },
    {
      "epoch": 0.8459705676243868,
      "grad_norm": 0.36234986782073975,
      "learning_rate": 0.0002206266443434585,
      "loss": 0.0746,
      "step": 6036
    },
    {
      "epoch": 0.8461107217939734,
      "grad_norm": 0.4556647539138794,
      "learning_rate": 0.00022061229370963883,
      "loss": 0.0827,
      "step": 6037
    },
    {
      "epoch": 0.84625087596356,
      "grad_norm": 0.5617167353630066,
      "learning_rate": 0.00022059794307581915,
      "loss": 0.0394,
      "step": 6038
    },
    {
      "epoch": 0.8463910301331464,
      "grad_norm": 0.5563653111457825,
      "learning_rate": 0.0002205835924419995,
      "loss": 0.1189,
      "step": 6039
    },
    {
      "epoch": 0.846531184302733,
      "grad_norm": 0.31766507029533386,
      "learning_rate": 0.00022056924180817984,
      "loss": 0.0726,
      "step": 6040
    },
    {
      "epoch": 0.8466713384723196,
      "grad_norm": 0.7167912125587463,
      "learning_rate": 0.00022055489117436017,
      "loss": 0.0684,
      "step": 6041
    },
    {
      "epoch": 0.8468114926419061,
      "grad_norm": 0.4309585690498352,
      "learning_rate": 0.00022054054054054052,
      "loss": 0.1414,
      "step": 6042
    },
    {
      "epoch": 0.8469516468114926,
      "grad_norm": 0.3490201532840729,
      "learning_rate": 0.00022052618990672085,
      "loss": 0.047,
      "step": 6043
    },
    {
      "epoch": 0.8470918009810792,
      "grad_norm": 0.30682438611984253,
      "learning_rate": 0.00022051183927290118,
      "loss": 0.0248,
      "step": 6044
    },
    {
      "epoch": 0.8472319551506657,
      "grad_norm": 0.2593974173069,
      "learning_rate": 0.00022049748863908156,
      "loss": 0.0635,
      "step": 6045
    },
    {
      "epoch": 0.8473721093202523,
      "grad_norm": 0.7133006453514099,
      "learning_rate": 0.0002204831380052619,
      "loss": 0.0758,
      "step": 6046
    },
    {
      "epoch": 0.8475122634898388,
      "grad_norm": 2.085472345352173,
      "learning_rate": 0.00022046878737144222,
      "loss": 0.1418,
      "step": 6047
    },
    {
      "epoch": 0.8476524176594253,
      "grad_norm": 0.7213760614395142,
      "learning_rate": 0.00022045443673762258,
      "loss": 0.0489,
      "step": 6048
    },
    {
      "epoch": 0.8477925718290119,
      "grad_norm": 1.4674125909805298,
      "learning_rate": 0.0002204400861038029,
      "loss": 0.0746,
      "step": 6049
    },
    {
      "epoch": 0.8479327259985985,
      "grad_norm": 4.065183639526367,
      "learning_rate": 0.00022042573546998323,
      "loss": 0.2873,
      "step": 6050
    },
    {
      "epoch": 0.848072880168185,
      "grad_norm": 0.3302542567253113,
      "learning_rate": 0.00022041138483616356,
      "loss": 0.0644,
      "step": 6051
    },
    {
      "epoch": 0.8482130343377715,
      "grad_norm": 0.3476284146308899,
      "learning_rate": 0.00022039703420234392,
      "loss": 0.0699,
      "step": 6052
    },
    {
      "epoch": 0.8483531885073581,
      "grad_norm": 0.6748327612876892,
      "learning_rate": 0.00022038268356852425,
      "loss": 0.0505,
      "step": 6053
    },
    {
      "epoch": 0.8484933426769447,
      "grad_norm": 0.32755163311958313,
      "learning_rate": 0.00022036833293470457,
      "loss": 0.0212,
      "step": 6054
    },
    {
      "epoch": 0.8486334968465312,
      "grad_norm": 0.4533473551273346,
      "learning_rate": 0.00022035398230088496,
      "loss": 0.0838,
      "step": 6055
    },
    {
      "epoch": 0.8487736510161177,
      "grad_norm": 0.9042192697525024,
      "learning_rate": 0.00022033963166706529,
      "loss": 0.0677,
      "step": 6056
    },
    {
      "epoch": 0.8489138051857043,
      "grad_norm": 0.5788269639015198,
      "learning_rate": 0.00022032528103324561,
      "loss": 0.055,
      "step": 6057
    },
    {
      "epoch": 0.8490539593552908,
      "grad_norm": 0.6356707811355591,
      "learning_rate": 0.00022031093039942597,
      "loss": 0.0863,
      "step": 6058
    },
    {
      "epoch": 0.8491941135248774,
      "grad_norm": 0.2752299904823303,
      "learning_rate": 0.0002202965797656063,
      "loss": 0.0533,
      "step": 6059
    },
    {
      "epoch": 0.849334267694464,
      "grad_norm": 0.701941728591919,
      "learning_rate": 0.00022028222913178663,
      "loss": 0.0734,
      "step": 6060
    },
    {
      "epoch": 0.8494744218640504,
      "grad_norm": 0.8902832269668579,
      "learning_rate": 0.00022026787849796698,
      "loss": 0.2098,
      "step": 6061
    },
    {
      "epoch": 0.849614576033637,
      "grad_norm": 0.6169689297676086,
      "learning_rate": 0.0002202535278641473,
      "loss": 0.0677,
      "step": 6062
    },
    {
      "epoch": 0.8497547302032236,
      "grad_norm": 0.5556852221488953,
      "learning_rate": 0.00022023917723032764,
      "loss": 0.103,
      "step": 6063
    },
    {
      "epoch": 0.8498948843728101,
      "grad_norm": 0.862409234046936,
      "learning_rate": 0.000220224826596508,
      "loss": 0.1048,
      "step": 6064
    },
    {
      "epoch": 0.8500350385423966,
      "grad_norm": 0.39001962542533875,
      "learning_rate": 0.00022021047596268835,
      "loss": 0.0542,
      "step": 6065
    },
    {
      "epoch": 0.8501751927119832,
      "grad_norm": 0.5409131050109863,
      "learning_rate": 0.00022019612532886868,
      "loss": 0.1172,
      "step": 6066
    },
    {
      "epoch": 0.8503153468815697,
      "grad_norm": 0.35064318776130676,
      "learning_rate": 0.000220181774695049,
      "loss": 0.0713,
      "step": 6067
    },
    {
      "epoch": 0.8504555010511563,
      "grad_norm": 0.43928372859954834,
      "learning_rate": 0.00022016742406122936,
      "loss": 0.0826,
      "step": 6068
    },
    {
      "epoch": 0.8505956552207428,
      "grad_norm": 0.38914406299591064,
      "learning_rate": 0.0002201530734274097,
      "loss": 0.0881,
      "step": 6069
    },
    {
      "epoch": 0.8507358093903293,
      "grad_norm": 0.5665842890739441,
      "learning_rate": 0.00022013872279359002,
      "loss": 0.081,
      "step": 6070
    },
    {
      "epoch": 0.8508759635599159,
      "grad_norm": 0.3907313048839569,
      "learning_rate": 0.00022012437215977038,
      "loss": 0.1094,
      "step": 6071
    },
    {
      "epoch": 0.8510161177295025,
      "grad_norm": 0.573666512966156,
      "learning_rate": 0.0002201100215259507,
      "loss": 0.0917,
      "step": 6072
    },
    {
      "epoch": 0.851156271899089,
      "grad_norm": 0.3204288184642792,
      "learning_rate": 0.00022009567089213103,
      "loss": 0.0816,
      "step": 6073
    },
    {
      "epoch": 0.8512964260686755,
      "grad_norm": 0.3869044780731201,
      "learning_rate": 0.0002200813202583114,
      "loss": 0.1327,
      "step": 6074
    },
    {
      "epoch": 0.8514365802382621,
      "grad_norm": 0.20069493353366852,
      "learning_rate": 0.00022006696962449172,
      "loss": 0.0352,
      "step": 6075
    },
    {
      "epoch": 0.8515767344078486,
      "grad_norm": 0.4491002857685089,
      "learning_rate": 0.00022005261899067205,
      "loss": 0.0912,
      "step": 6076
    },
    {
      "epoch": 0.8517168885774352,
      "grad_norm": 0.599244236946106,
      "learning_rate": 0.00022003826835685243,
      "loss": 0.0785,
      "step": 6077
    },
    {
      "epoch": 0.8518570427470217,
      "grad_norm": 0.47733041644096375,
      "learning_rate": 0.00022002391772303276,
      "loss": 0.0843,
      "step": 6078
    },
    {
      "epoch": 0.8519971969166082,
      "grad_norm": 0.2956453561782837,
      "learning_rate": 0.00022000956708921309,
      "loss": 0.0459,
      "step": 6079
    },
    {
      "epoch": 0.8521373510861948,
      "grad_norm": 0.5410683155059814,
      "learning_rate": 0.00021999521645539344,
      "loss": 0.1125,
      "step": 6080
    },
    {
      "epoch": 0.8522775052557814,
      "grad_norm": 0.5709959864616394,
      "learning_rate": 0.00021998086582157377,
      "loss": 0.0658,
      "step": 6081
    },
    {
      "epoch": 0.852417659425368,
      "grad_norm": 0.4829818904399872,
      "learning_rate": 0.0002199665151877541,
      "loss": 0.1498,
      "step": 6082
    },
    {
      "epoch": 0.8525578135949544,
      "grad_norm": 0.17665866017341614,
      "learning_rate": 0.00021995216455393445,
      "loss": 0.0187,
      "step": 6083
    },
    {
      "epoch": 0.852697967764541,
      "grad_norm": 0.37586143612861633,
      "learning_rate": 0.00021993781392011478,
      "loss": 0.0456,
      "step": 6084
    },
    {
      "epoch": 0.8528381219341276,
      "grad_norm": 0.7087547183036804,
      "learning_rate": 0.0002199234632862951,
      "loss": 0.0668,
      "step": 6085
    },
    {
      "epoch": 0.8529782761037141,
      "grad_norm": 0.6223840117454529,
      "learning_rate": 0.00021990911265247544,
      "loss": 0.0613,
      "step": 6086
    },
    {
      "epoch": 0.8531184302733006,
      "grad_norm": 0.26645681262016296,
      "learning_rate": 0.00021989476201865582,
      "loss": 0.0354,
      "step": 6087
    },
    {
      "epoch": 0.8532585844428872,
      "grad_norm": 0.35039687156677246,
      "learning_rate": 0.00021988041138483615,
      "loss": 0.0446,
      "step": 6088
    },
    {
      "epoch": 0.8533987386124737,
      "grad_norm": 0.19350412487983704,
      "learning_rate": 0.00021986606075101648,
      "loss": 0.026,
      "step": 6089
    },
    {
      "epoch": 0.8535388927820603,
      "grad_norm": 0.7810064554214478,
      "learning_rate": 0.00021985171011719684,
      "loss": 0.1607,
      "step": 6090
    },
    {
      "epoch": 0.8536790469516468,
      "grad_norm": 0.22369100153446198,
      "learning_rate": 0.00021983735948337716,
      "loss": 0.0633,
      "step": 6091
    },
    {
      "epoch": 0.8538192011212333,
      "grad_norm": 0.8078294396400452,
      "learning_rate": 0.0002198230088495575,
      "loss": 0.0968,
      "step": 6092
    },
    {
      "epoch": 0.8539593552908199,
      "grad_norm": 0.44355660676956177,
      "learning_rate": 0.00021980865821573785,
      "loss": 0.0661,
      "step": 6093
    },
    {
      "epoch": 0.8540995094604065,
      "grad_norm": 1.3358955383300781,
      "learning_rate": 0.00021979430758191818,
      "loss": 0.134,
      "step": 6094
    },
    {
      "epoch": 0.854239663629993,
      "grad_norm": 2.5596325397491455,
      "learning_rate": 0.0002197799569480985,
      "loss": 0.1636,
      "step": 6095
    },
    {
      "epoch": 0.8543798177995795,
      "grad_norm": 1.5910701751708984,
      "learning_rate": 0.00021976560631427886,
      "loss": 0.1689,
      "step": 6096
    },
    {
      "epoch": 0.8545199719691661,
      "grad_norm": 4.034012317657471,
      "learning_rate": 0.00021975125568045922,
      "loss": 0.1846,
      "step": 6097
    },
    {
      "epoch": 0.8546601261387526,
      "grad_norm": 0.786284863948822,
      "learning_rate": 0.00021973690504663955,
      "loss": 0.0635,
      "step": 6098
    },
    {
      "epoch": 0.8548002803083392,
      "grad_norm": 2.224099636077881,
      "learning_rate": 0.0002197225544128199,
      "loss": 0.2622,
      "step": 6099
    },
    {
      "epoch": 0.8549404344779257,
      "grad_norm": 2.533465623855591,
      "learning_rate": 0.00021970820377900023,
      "loss": 0.4339,
      "step": 6100
    },
    {
      "epoch": 0.8550805886475122,
      "grad_norm": 0.2941974699497223,
      "learning_rate": 0.00021969385314518056,
      "loss": 0.0471,
      "step": 6101
    },
    {
      "epoch": 0.8552207428170988,
      "grad_norm": 0.47508567571640015,
      "learning_rate": 0.0002196795025113609,
      "loss": 0.0734,
      "step": 6102
    },
    {
      "epoch": 0.8553608969866854,
      "grad_norm": 0.4837630093097687,
      "learning_rate": 0.00021966515187754124,
      "loss": 0.122,
      "step": 6103
    },
    {
      "epoch": 0.855501051156272,
      "grad_norm": 0.4980991780757904,
      "learning_rate": 0.00021965080124372157,
      "loss": 0.0603,
      "step": 6104
    },
    {
      "epoch": 0.8556412053258584,
      "grad_norm": 0.38687628507614136,
      "learning_rate": 0.0002196364506099019,
      "loss": 0.0572,
      "step": 6105
    },
    {
      "epoch": 0.855781359495445,
      "grad_norm": 0.26383933424949646,
      "learning_rate": 0.00021962209997608226,
      "loss": 0.0326,
      "step": 6106
    },
    {
      "epoch": 0.8559215136650316,
      "grad_norm": 0.5728917717933655,
      "learning_rate": 0.00021960774934226258,
      "loss": 0.0654,
      "step": 6107
    },
    {
      "epoch": 0.8560616678346181,
      "grad_norm": 0.2652728259563446,
      "learning_rate": 0.0002195933987084429,
      "loss": 0.0438,
      "step": 6108
    },
    {
      "epoch": 0.8562018220042046,
      "grad_norm": 0.1694273203611374,
      "learning_rate": 0.0002195790480746233,
      "loss": 0.0324,
      "step": 6109
    },
    {
      "epoch": 0.8563419761737912,
      "grad_norm": 0.3238029479980469,
      "learning_rate": 0.00021956469744080362,
      "loss": 0.0713,
      "step": 6110
    },
    {
      "epoch": 0.8564821303433777,
      "grad_norm": 0.16493548452854156,
      "learning_rate": 0.00021955034680698395,
      "loss": 0.0187,
      "step": 6111
    },
    {
      "epoch": 0.8566222845129643,
      "grad_norm": 0.4229266941547394,
      "learning_rate": 0.0002195359961731643,
      "loss": 0.0629,
      "step": 6112
    },
    {
      "epoch": 0.8567624386825508,
      "grad_norm": 0.7142085433006287,
      "learning_rate": 0.00021952164553934464,
      "loss": 0.1472,
      "step": 6113
    },
    {
      "epoch": 0.8569025928521373,
      "grad_norm": 0.17299939692020416,
      "learning_rate": 0.00021950729490552497,
      "loss": 0.0693,
      "step": 6114
    },
    {
      "epoch": 0.8570427470217239,
      "grad_norm": 0.5140867829322815,
      "learning_rate": 0.00021949294427170532,
      "loss": 0.1833,
      "step": 6115
    },
    {
      "epoch": 0.8571829011913105,
      "grad_norm": 0.33215197920799255,
      "learning_rate": 0.00021947859363788565,
      "loss": 0.033,
      "step": 6116
    },
    {
      "epoch": 0.857323055360897,
      "grad_norm": 0.31363826990127563,
      "learning_rate": 0.00021946424300406598,
      "loss": 0.08,
      "step": 6117
    },
    {
      "epoch": 0.8574632095304835,
      "grad_norm": 0.4356216788291931,
      "learning_rate": 0.00021944989237024636,
      "loss": 0.1139,
      "step": 6118
    },
    {
      "epoch": 0.8576033637000701,
      "grad_norm": 0.3813140392303467,
      "learning_rate": 0.0002194355417364267,
      "loss": 0.0852,
      "step": 6119
    },
    {
      "epoch": 0.8577435178696566,
      "grad_norm": 1.0481899976730347,
      "learning_rate": 0.00021942119110260702,
      "loss": 0.1134,
      "step": 6120
    },
    {
      "epoch": 0.8578836720392432,
      "grad_norm": 0.5337585210800171,
      "learning_rate": 0.00021940684046878735,
      "loss": 0.0556,
      "step": 6121
    },
    {
      "epoch": 0.8580238262088297,
      "grad_norm": 0.674825131893158,
      "learning_rate": 0.0002193924898349677,
      "loss": 0.073,
      "step": 6122
    },
    {
      "epoch": 0.8581639803784162,
      "grad_norm": 0.1671558916568756,
      "learning_rate": 0.00021937813920114803,
      "loss": 0.0301,
      "step": 6123
    },
    {
      "epoch": 0.8583041345480028,
      "grad_norm": 0.46822091937065125,
      "learning_rate": 0.00021936378856732836,
      "loss": 0.0696,
      "step": 6124
    },
    {
      "epoch": 0.8584442887175894,
      "grad_norm": 0.45676594972610474,
      "learning_rate": 0.00021934943793350871,
      "loss": 0.1004,
      "step": 6125
    },
    {
      "epoch": 0.858584442887176,
      "grad_norm": 0.24710460007190704,
      "learning_rate": 0.00021933508729968904,
      "loss": 0.0553,
      "step": 6126
    },
    {
      "epoch": 0.8587245970567624,
      "grad_norm": 0.20501458644866943,
      "learning_rate": 0.00021932073666586937,
      "loss": 0.0247,
      "step": 6127
    },
    {
      "epoch": 0.858864751226349,
      "grad_norm": 0.6338192820549011,
      "learning_rate": 0.00021930638603204975,
      "loss": 0.1415,
      "step": 6128
    },
    {
      "epoch": 0.8590049053959355,
      "grad_norm": 0.3492337167263031,
      "learning_rate": 0.00021929203539823008,
      "loss": 0.1034,
      "step": 6129
    },
    {
      "epoch": 0.8591450595655221,
      "grad_norm": 0.46829754114151,
      "learning_rate": 0.0002192776847644104,
      "loss": 0.0914,
      "step": 6130
    },
    {
      "epoch": 0.8592852137351086,
      "grad_norm": 0.6821593642234802,
      "learning_rate": 0.00021926333413059077,
      "loss": 0.1096,
      "step": 6131
    },
    {
      "epoch": 0.8594253679046951,
      "grad_norm": 0.5158564448356628,
      "learning_rate": 0.0002192489834967711,
      "loss": 0.1258,
      "step": 6132
    },
    {
      "epoch": 0.8595655220742817,
      "grad_norm": 0.4335736930370331,
      "learning_rate": 0.00021923463286295142,
      "loss": 0.066,
      "step": 6133
    },
    {
      "epoch": 0.8597056762438683,
      "grad_norm": 0.21926026046276093,
      "learning_rate": 0.00021922028222913178,
      "loss": 0.0374,
      "step": 6134
    },
    {
      "epoch": 0.8598458304134547,
      "grad_norm": 0.764644980430603,
      "learning_rate": 0.0002192059315953121,
      "loss": 0.0572,
      "step": 6135
    },
    {
      "epoch": 0.8599859845830413,
      "grad_norm": 0.21171337366104126,
      "learning_rate": 0.00021919158096149244,
      "loss": 0.0577,
      "step": 6136
    },
    {
      "epoch": 0.8601261387526279,
      "grad_norm": 0.2932329475879669,
      "learning_rate": 0.00021917723032767277,
      "loss": 0.0548,
      "step": 6137
    },
    {
      "epoch": 0.8602662929222145,
      "grad_norm": 0.4444162845611572,
      "learning_rate": 0.00021916287969385312,
      "loss": 0.0578,
      "step": 6138
    },
    {
      "epoch": 0.860406447091801,
      "grad_norm": 0.9112905263900757,
      "learning_rate": 0.00021914852906003345,
      "loss": 0.0954,
      "step": 6139
    },
    {
      "epoch": 0.8605466012613875,
      "grad_norm": 0.05830321088433266,
      "learning_rate": 0.00021913417842621378,
      "loss": 0.0078,
      "step": 6140
    },
    {
      "epoch": 0.8606867554309741,
      "grad_norm": 0.5870422720909119,
      "learning_rate": 0.00021911982779239416,
      "loss": 0.1373,
      "step": 6141
    },
    {
      "epoch": 0.8608269096005606,
      "grad_norm": 0.4302772879600525,
      "learning_rate": 0.0002191054771585745,
      "loss": 0.0466,
      "step": 6142
    },
    {
      "epoch": 0.8609670637701472,
      "grad_norm": 0.456406831741333,
      "learning_rate": 0.00021909112652475482,
      "loss": 0.1077,
      "step": 6143
    },
    {
      "epoch": 0.8611072179397337,
      "grad_norm": 0.5110020041465759,
      "learning_rate": 0.00021907677589093517,
      "loss": 0.121,
      "step": 6144
    },
    {
      "epoch": 0.8612473721093202,
      "grad_norm": 0.4321301579475403,
      "learning_rate": 0.0002190624252571155,
      "loss": 0.0874,
      "step": 6145
    },
    {
      "epoch": 0.8613875262789068,
      "grad_norm": 1.0213720798492432,
      "learning_rate": 0.00021904807462329583,
      "loss": 0.072,
      "step": 6146
    },
    {
      "epoch": 0.8615276804484934,
      "grad_norm": 0.35551443696022034,
      "learning_rate": 0.0002190337239894762,
      "loss": 0.0339,
      "step": 6147
    },
    {
      "epoch": 0.8616678346180799,
      "grad_norm": 0.24760396778583527,
      "learning_rate": 0.00021901937335565652,
      "loss": 0.0618,
      "step": 6148
    },
    {
      "epoch": 0.8618079887876664,
      "grad_norm": 0.8129737973213196,
      "learning_rate": 0.00021900502272183684,
      "loss": 0.1025,
      "step": 6149
    },
    {
      "epoch": 0.861948142957253,
      "grad_norm": 1.3780536651611328,
      "learning_rate": 0.00021899067208801723,
      "loss": 0.1216,
      "step": 6150
    },
    {
      "epoch": 0.8620882971268395,
      "grad_norm": 0.3356468975543976,
      "learning_rate": 0.00021897632145419756,
      "loss": 0.0794,
      "step": 6151
    },
    {
      "epoch": 0.8622284512964261,
      "grad_norm": 0.40836986899375916,
      "learning_rate": 0.00021896197082037788,
      "loss": 0.0582,
      "step": 6152
    },
    {
      "epoch": 0.8623686054660126,
      "grad_norm": 0.2814682126045227,
      "learning_rate": 0.00021894762018655824,
      "loss": 0.0612,
      "step": 6153
    },
    {
      "epoch": 0.8625087596355991,
      "grad_norm": 0.46097177267074585,
      "learning_rate": 0.00021893326955273857,
      "loss": 0.1014,
      "step": 6154
    },
    {
      "epoch": 0.8626489138051857,
      "grad_norm": 0.31306591629981995,
      "learning_rate": 0.0002189189189189189,
      "loss": 0.0446,
      "step": 6155
    },
    {
      "epoch": 0.8627890679747723,
      "grad_norm": 0.38307279348373413,
      "learning_rate": 0.00021890456828509923,
      "loss": 0.0931,
      "step": 6156
    },
    {
      "epoch": 0.8629292221443587,
      "grad_norm": 0.5275318622589111,
      "learning_rate": 0.00021889021765127958,
      "loss": 0.058,
      "step": 6157
    },
    {
      "epoch": 0.8630693763139453,
      "grad_norm": 0.27580344676971436,
      "learning_rate": 0.0002188758670174599,
      "loss": 0.0564,
      "step": 6158
    },
    {
      "epoch": 0.8632095304835319,
      "grad_norm": 0.27724042534828186,
      "learning_rate": 0.00021886151638364024,
      "loss": 0.0384,
      "step": 6159
    },
    {
      "epoch": 0.8633496846531185,
      "grad_norm": 0.28988340497016907,
      "learning_rate": 0.00021884716574982062,
      "loss": 0.0564,
      "step": 6160
    },
    {
      "epoch": 0.863489838822705,
      "grad_norm": 0.8585079908370972,
      "learning_rate": 0.00021883281511600095,
      "loss": 0.0551,
      "step": 6161
    },
    {
      "epoch": 0.8636299929922915,
      "grad_norm": 0.1465761661529541,
      "learning_rate": 0.00021881846448218128,
      "loss": 0.0558,
      "step": 6162
    },
    {
      "epoch": 0.863770147161878,
      "grad_norm": 0.42087769508361816,
      "learning_rate": 0.00021880411384836163,
      "loss": 0.085,
      "step": 6163
    },
    {
      "epoch": 0.8639103013314646,
      "grad_norm": 0.3326455354690552,
      "learning_rate": 0.00021878976321454196,
      "loss": 0.0417,
      "step": 6164
    },
    {
      "epoch": 0.8640504555010512,
      "grad_norm": 0.46806639432907104,
      "learning_rate": 0.0002187754125807223,
      "loss": 0.0458,
      "step": 6165
    },
    {
      "epoch": 0.8641906096706377,
      "grad_norm": 0.43334513902664185,
      "learning_rate": 0.00021876106194690265,
      "loss": 0.1083,
      "step": 6166
    },
    {
      "epoch": 0.8643307638402242,
      "grad_norm": 0.4545452892780304,
      "learning_rate": 0.00021874671131308298,
      "loss": 0.1221,
      "step": 6167
    },
    {
      "epoch": 0.8644709180098108,
      "grad_norm": 0.21493001282215118,
      "learning_rate": 0.0002187323606792633,
      "loss": 0.0409,
      "step": 6168
    },
    {
      "epoch": 0.8646110721793974,
      "grad_norm": 0.1310833841562271,
      "learning_rate": 0.00021871801004544366,
      "loss": 0.0314,
      "step": 6169
    },
    {
      "epoch": 0.8647512263489839,
      "grad_norm": 0.47558698058128357,
      "learning_rate": 0.000218703659411624,
      "loss": 0.0632,
      "step": 6170
    },
    {
      "epoch": 0.8648913805185704,
      "grad_norm": 0.12817169725894928,
      "learning_rate": 0.00021868930877780432,
      "loss": 0.0199,
      "step": 6171
    },
    {
      "epoch": 0.865031534688157,
      "grad_norm": 0.49068155884742737,
      "learning_rate": 0.00021867495814398464,
      "loss": 0.1007,
      "step": 6172
    },
    {
      "epoch": 0.8651716888577435,
      "grad_norm": 0.3276272416114807,
      "learning_rate": 0.00021866060751016503,
      "loss": 0.0747,
      "step": 6173
    },
    {
      "epoch": 0.8653118430273301,
      "grad_norm": 0.4701068699359894,
      "learning_rate": 0.00021864625687634536,
      "loss": 0.0868,
      "step": 6174
    },
    {
      "epoch": 0.8654519971969166,
      "grad_norm": 0.5828079581260681,
      "learning_rate": 0.00021863190624252568,
      "loss": 0.0779,
      "step": 6175
    },
    {
      "epoch": 0.8655921513665031,
      "grad_norm": 0.30745652318000793,
      "learning_rate": 0.00021861755560870604,
      "loss": 0.0384,
      "step": 6176
    },
    {
      "epoch": 0.8657323055360897,
      "grad_norm": 0.48013949394226074,
      "learning_rate": 0.00021860320497488637,
      "loss": 0.0954,
      "step": 6177
    },
    {
      "epoch": 0.8658724597056763,
      "grad_norm": 0.32466575503349304,
      "learning_rate": 0.0002185888543410667,
      "loss": 0.0511,
      "step": 6178
    },
    {
      "epoch": 0.8660126138752627,
      "grad_norm": 0.41292816400527954,
      "learning_rate": 0.00021857450370724705,
      "loss": 0.0662,
      "step": 6179
    },
    {
      "epoch": 0.8661527680448493,
      "grad_norm": 0.741558849811554,
      "learning_rate": 0.00021856015307342738,
      "loss": 0.0439,
      "step": 6180
    },
    {
      "epoch": 0.8662929222144359,
      "grad_norm": 0.5732241272926331,
      "learning_rate": 0.0002185458024396077,
      "loss": 0.1156,
      "step": 6181
    },
    {
      "epoch": 0.8664330763840224,
      "grad_norm": 0.7777132987976074,
      "learning_rate": 0.0002185314518057881,
      "loss": 0.1861,
      "step": 6182
    },
    {
      "epoch": 0.866573230553609,
      "grad_norm": 0.24981337785720825,
      "learning_rate": 0.00021851710117196842,
      "loss": 0.0569,
      "step": 6183
    },
    {
      "epoch": 0.8667133847231955,
      "grad_norm": 0.2616785168647766,
      "learning_rate": 0.00021850275053814875,
      "loss": 0.0694,
      "step": 6184
    },
    {
      "epoch": 0.866853538892782,
      "grad_norm": 0.4602219760417938,
      "learning_rate": 0.0002184883999043291,
      "loss": 0.1396,
      "step": 6185
    },
    {
      "epoch": 0.8669936930623686,
      "grad_norm": 0.4059341549873352,
      "learning_rate": 0.00021847404927050943,
      "loss": 0.0954,
      "step": 6186
    },
    {
      "epoch": 0.8671338472319552,
      "grad_norm": 0.4319641888141632,
      "learning_rate": 0.00021845969863668976,
      "loss": 0.1626,
      "step": 6187
    },
    {
      "epoch": 0.8672740014015416,
      "grad_norm": 0.3092079162597656,
      "learning_rate": 0.00021844534800287012,
      "loss": 0.0861,
      "step": 6188
    },
    {
      "epoch": 0.8674141555711282,
      "grad_norm": 0.46257224678993225,
      "learning_rate": 0.00021843099736905045,
      "loss": 0.1117,
      "step": 6189
    },
    {
      "epoch": 0.8675543097407148,
      "grad_norm": 1.5813649892807007,
      "learning_rate": 0.00021841664673523078,
      "loss": 0.0719,
      "step": 6190
    },
    {
      "epoch": 0.8676944639103014,
      "grad_norm": 0.5169780254364014,
      "learning_rate": 0.0002184022961014111,
      "loss": 0.0668,
      "step": 6191
    },
    {
      "epoch": 0.8678346180798879,
      "grad_norm": 0.357990026473999,
      "learning_rate": 0.0002183879454675915,
      "loss": 0.0846,
      "step": 6192
    },
    {
      "epoch": 0.8679747722494744,
      "grad_norm": 0.227934330701828,
      "learning_rate": 0.00021837359483377182,
      "loss": 0.0458,
      "step": 6193
    },
    {
      "epoch": 0.868114926419061,
      "grad_norm": 0.8706230521202087,
      "learning_rate": 0.00021835924419995214,
      "loss": 0.1192,
      "step": 6194
    },
    {
      "epoch": 0.8682550805886475,
      "grad_norm": 0.7386736869812012,
      "learning_rate": 0.0002183448935661325,
      "loss": 0.0663,
      "step": 6195
    },
    {
      "epoch": 0.8683952347582341,
      "grad_norm": 0.4754023849964142,
      "learning_rate": 0.00021833054293231283,
      "loss": 0.1005,
      "step": 6196
    },
    {
      "epoch": 0.8685353889278206,
      "grad_norm": 0.2744595408439636,
      "learning_rate": 0.00021831619229849316,
      "loss": 0.0286,
      "step": 6197
    },
    {
      "epoch": 0.8686755430974071,
      "grad_norm": 0.8672484755516052,
      "learning_rate": 0.0002183018416646735,
      "loss": 0.1787,
      "step": 6198
    },
    {
      "epoch": 0.8688156972669937,
      "grad_norm": 4.3144097328186035,
      "learning_rate": 0.00021828749103085384,
      "loss": 0.1939,
      "step": 6199
    },
    {
      "epoch": 0.8689558514365803,
      "grad_norm": 3.1200180053710938,
      "learning_rate": 0.00021827314039703417,
      "loss": 0.0956,
      "step": 6200
    },
    {
      "epoch": 0.8690960056061667,
      "grad_norm": 0.2564171850681305,
      "learning_rate": 0.00021825878976321453,
      "loss": 0.0764,
      "step": 6201
    },
    {
      "epoch": 0.8692361597757533,
      "grad_norm": 0.5688282251358032,
      "learning_rate": 0.00021824443912939485,
      "loss": 0.1047,
      "step": 6202
    },
    {
      "epoch": 0.8693763139453399,
      "grad_norm": 0.33884042501449585,
      "learning_rate": 0.00021823008849557518,
      "loss": 0.0617,
      "step": 6203
    },
    {
      "epoch": 0.8695164681149264,
      "grad_norm": 0.5766940116882324,
      "learning_rate": 0.00021821573786175557,
      "loss": 0.1023,
      "step": 6204
    },
    {
      "epoch": 0.869656622284513,
      "grad_norm": 0.3584175109863281,
      "learning_rate": 0.0002182013872279359,
      "loss": 0.1451,
      "step": 6205
    },
    {
      "epoch": 0.8697967764540995,
      "grad_norm": 0.12489283829927444,
      "learning_rate": 0.00021818703659411622,
      "loss": 0.0218,
      "step": 6206
    },
    {
      "epoch": 0.869936930623686,
      "grad_norm": 0.539472758769989,
      "learning_rate": 0.00021817268596029655,
      "loss": 0.1171,
      "step": 6207
    },
    {
      "epoch": 0.8700770847932726,
      "grad_norm": 0.2593488097190857,
      "learning_rate": 0.0002181583353264769,
      "loss": 0.0935,
      "step": 6208
    },
    {
      "epoch": 0.8702172389628592,
      "grad_norm": 0.683388352394104,
      "learning_rate": 0.00021814398469265724,
      "loss": 0.0976,
      "step": 6209
    },
    {
      "epoch": 0.8703573931324456,
      "grad_norm": 0.19402046501636505,
      "learning_rate": 0.00021812963405883756,
      "loss": 0.0381,
      "step": 6210
    },
    {
      "epoch": 0.8704975473020322,
      "grad_norm": 0.2791079580783844,
      "learning_rate": 0.00021811528342501792,
      "loss": 0.054,
      "step": 6211
    },
    {
      "epoch": 0.8706377014716188,
      "grad_norm": 0.5183849334716797,
      "learning_rate": 0.00021810093279119825,
      "loss": 0.0672,
      "step": 6212
    },
    {
      "epoch": 0.8707778556412054,
      "grad_norm": 0.579476535320282,
      "learning_rate": 0.00021808658215737858,
      "loss": 0.1057,
      "step": 6213
    },
    {
      "epoch": 0.8709180098107919,
      "grad_norm": 0.3718048334121704,
      "learning_rate": 0.00021807223152355896,
      "loss": 0.0475,
      "step": 6214
    },
    {
      "epoch": 0.8710581639803784,
      "grad_norm": 0.25893765687942505,
      "learning_rate": 0.0002180578808897393,
      "loss": 0.0448,
      "step": 6215
    },
    {
      "epoch": 0.871198318149965,
      "grad_norm": 0.2760825753211975,
      "learning_rate": 0.00021804353025591962,
      "loss": 0.0716,
      "step": 6216
    },
    {
      "epoch": 0.8713384723195515,
      "grad_norm": 0.47999587655067444,
      "learning_rate": 0.00021802917962209997,
      "loss": 0.1643,
      "step": 6217
    },
    {
      "epoch": 0.8714786264891381,
      "grad_norm": 0.5208139419555664,
      "learning_rate": 0.0002180148289882803,
      "loss": 0.0716,
      "step": 6218
    },
    {
      "epoch": 0.8716187806587246,
      "grad_norm": 0.4334179162979126,
      "learning_rate": 0.00021800047835446063,
      "loss": 0.0715,
      "step": 6219
    },
    {
      "epoch": 0.8717589348283111,
      "grad_norm": 0.18401111662387848,
      "learning_rate": 0.00021798612772064099,
      "loss": 0.0283,
      "step": 6220
    },
    {
      "epoch": 0.8718990889978977,
      "grad_norm": 0.6516224145889282,
      "learning_rate": 0.00021797177708682131,
      "loss": 0.1209,
      "step": 6221
    },
    {
      "epoch": 0.8720392431674843,
      "grad_norm": 0.29093581438064575,
      "learning_rate": 0.00021795742645300164,
      "loss": 0.0601,
      "step": 6222
    },
    {
      "epoch": 0.8721793973370707,
      "grad_norm": 0.6586615443229675,
      "learning_rate": 0.00021794307581918197,
      "loss": 0.1371,
      "step": 6223
    },
    {
      "epoch": 0.8723195515066573,
      "grad_norm": 0.20381219685077667,
      "learning_rate": 0.00021792872518536235,
      "loss": 0.0353,
      "step": 6224
    },
    {
      "epoch": 0.8724597056762439,
      "grad_norm": 0.2556091248989105,
      "learning_rate": 0.00021791437455154268,
      "loss": 0.0302,
      "step": 6225
    },
    {
      "epoch": 0.8725998598458304,
      "grad_norm": 0.320342093706131,
      "learning_rate": 0.000217900023917723,
      "loss": 0.0525,
      "step": 6226
    },
    {
      "epoch": 0.872740014015417,
      "grad_norm": 0.4603051543235779,
      "learning_rate": 0.00021788567328390337,
      "loss": 0.0499,
      "step": 6227
    },
    {
      "epoch": 0.8728801681850035,
      "grad_norm": 0.4017927646636963,
      "learning_rate": 0.0002178713226500837,
      "loss": 0.0705,
      "step": 6228
    },
    {
      "epoch": 0.87302032235459,
      "grad_norm": 0.3869698941707611,
      "learning_rate": 0.00021785697201626402,
      "loss": 0.0737,
      "step": 6229
    },
    {
      "epoch": 0.8731604765241766,
      "grad_norm": 0.6222916841506958,
      "learning_rate": 0.00021784262138244438,
      "loss": 0.0762,
      "step": 6230
    },
    {
      "epoch": 0.8733006306937632,
      "grad_norm": 0.6190270185470581,
      "learning_rate": 0.0002178282707486247,
      "loss": 0.1041,
      "step": 6231
    },
    {
      "epoch": 0.8734407848633496,
      "grad_norm": 0.4681747853755951,
      "learning_rate": 0.00021781392011480504,
      "loss": 0.0545,
      "step": 6232
    },
    {
      "epoch": 0.8735809390329362,
      "grad_norm": 0.24020327627658844,
      "learning_rate": 0.0002177995694809854,
      "loss": 0.0371,
      "step": 6233
    },
    {
      "epoch": 0.8737210932025228,
      "grad_norm": 0.46187299489974976,
      "learning_rate": 0.00021778521884716572,
      "loss": 0.1199,
      "step": 6234
    },
    {
      "epoch": 0.8738612473721094,
      "grad_norm": 0.1926620900630951,
      "learning_rate": 0.00021777086821334605,
      "loss": 0.0254,
      "step": 6235
    },
    {
      "epoch": 0.8740014015416958,
      "grad_norm": 0.5401365160942078,
      "learning_rate": 0.00021775651757952643,
      "loss": 0.102,
      "step": 6236
    },
    {
      "epoch": 0.8741415557112824,
      "grad_norm": 0.676092267036438,
      "learning_rate": 0.00021774216694570676,
      "loss": 0.0826,
      "step": 6237
    },
    {
      "epoch": 0.874281709880869,
      "grad_norm": 0.34450727701187134,
      "learning_rate": 0.0002177278163118871,
      "loss": 0.0746,
      "step": 6238
    },
    {
      "epoch": 0.8744218640504555,
      "grad_norm": 0.37508848309516907,
      "learning_rate": 0.00021771346567806744,
      "loss": 0.0597,
      "step": 6239
    },
    {
      "epoch": 0.8745620182200421,
      "grad_norm": 0.4466211795806885,
      "learning_rate": 0.00021769911504424777,
      "loss": 0.068,
      "step": 6240
    },
    {
      "epoch": 0.8747021723896286,
      "grad_norm": 0.512169361114502,
      "learning_rate": 0.0002176847644104281,
      "loss": 0.0848,
      "step": 6241
    },
    {
      "epoch": 0.8748423265592151,
      "grad_norm": 0.18908928334712982,
      "learning_rate": 0.00021767041377660843,
      "loss": 0.0126,
      "step": 6242
    },
    {
      "epoch": 0.8749824807288017,
      "grad_norm": 0.22945968806743622,
      "learning_rate": 0.00021765606314278879,
      "loss": 0.0202,
      "step": 6243
    },
    {
      "epoch": 0.8751226348983883,
      "grad_norm": 0.7049269676208496,
      "learning_rate": 0.00021764171250896911,
      "loss": 0.1779,
      "step": 6244
    },
    {
      "epoch": 0.8752627890679747,
      "grad_norm": 0.34868231415748596,
      "learning_rate": 0.00021762736187514944,
      "loss": 0.0205,
      "step": 6245
    },
    {
      "epoch": 0.8754029432375613,
      "grad_norm": 0.4349097013473511,
      "learning_rate": 0.00021761301124132983,
      "loss": 0.0432,
      "step": 6246
    },
    {
      "epoch": 0.8755430974071479,
      "grad_norm": 1.3755170106887817,
      "learning_rate": 0.00021759866060751015,
      "loss": 0.0657,
      "step": 6247
    },
    {
      "epoch": 0.8756832515767344,
      "grad_norm": 1.2230827808380127,
      "learning_rate": 0.00021758430997369048,
      "loss": 0.1141,
      "step": 6248
    },
    {
      "epoch": 0.875823405746321,
      "grad_norm": 0.8426931500434875,
      "learning_rate": 0.00021756995933987084,
      "loss": 0.1539,
      "step": 6249
    },
    {
      "epoch": 0.8759635599159075,
      "grad_norm": 1.4148598909378052,
      "learning_rate": 0.00021755560870605117,
      "loss": 0.2293,
      "step": 6250
    },
    {
      "epoch": 0.876103714085494,
      "grad_norm": 0.32737067341804504,
      "learning_rate": 0.0002175412580722315,
      "loss": 0.0927,
      "step": 6251
    },
    {
      "epoch": 0.8762438682550806,
      "grad_norm": 0.6828939318656921,
      "learning_rate": 0.00021752690743841185,
      "loss": 0.1564,
      "step": 6252
    },
    {
      "epoch": 0.8763840224246672,
      "grad_norm": 0.5200299620628357,
      "learning_rate": 0.00021751255680459218,
      "loss": 0.1048,
      "step": 6253
    },
    {
      "epoch": 0.8765241765942536,
      "grad_norm": 0.6152176856994629,
      "learning_rate": 0.0002174982061707725,
      "loss": 0.146,
      "step": 6254
    },
    {
      "epoch": 0.8766643307638402,
      "grad_norm": 0.5028912425041199,
      "learning_rate": 0.0002174838555369529,
      "loss": 0.0702,
      "step": 6255
    },
    {
      "epoch": 0.8768044849334268,
      "grad_norm": 0.260227769613266,
      "learning_rate": 0.00021746950490313322,
      "loss": 0.0394,
      "step": 6256
    },
    {
      "epoch": 0.8769446391030133,
      "grad_norm": 0.5616887807846069,
      "learning_rate": 0.00021745515426931355,
      "loss": 0.1816,
      "step": 6257
    },
    {
      "epoch": 0.8770847932725998,
      "grad_norm": 0.20666776597499847,
      "learning_rate": 0.00021744080363549388,
      "loss": 0.0458,
      "step": 6258
    },
    {
      "epoch": 0.8772249474421864,
      "grad_norm": 0.2023920863866806,
      "learning_rate": 0.00021742645300167423,
      "loss": 0.0339,
      "step": 6259
    },
    {
      "epoch": 0.877365101611773,
      "grad_norm": 0.6767168045043945,
      "learning_rate": 0.00021741210236785456,
      "loss": 0.1386,
      "step": 6260
    },
    {
      "epoch": 0.8775052557813595,
      "grad_norm": 0.3841035068035126,
      "learning_rate": 0.0002173977517340349,
      "loss": 0.0836,
      "step": 6261
    },
    {
      "epoch": 0.8776454099509461,
      "grad_norm": 0.2366967350244522,
      "learning_rate": 0.00021738340110021525,
      "loss": 0.0452,
      "step": 6262
    },
    {
      "epoch": 0.8777855641205325,
      "grad_norm": 0.6210862994194031,
      "learning_rate": 0.00021736905046639557,
      "loss": 0.143,
      "step": 6263
    },
    {
      "epoch": 0.8779257182901191,
      "grad_norm": 0.5211275219917297,
      "learning_rate": 0.0002173546998325759,
      "loss": 0.159,
      "step": 6264
    },
    {
      "epoch": 0.8780658724597057,
      "grad_norm": 0.2777138352394104,
      "learning_rate": 0.00021734034919875626,
      "loss": 0.0962,
      "step": 6265
    },
    {
      "epoch": 0.8782060266292923,
      "grad_norm": 0.3432758152484894,
      "learning_rate": 0.0002173259985649366,
      "loss": 0.1401,
      "step": 6266
    },
    {
      "epoch": 0.8783461807988787,
      "grad_norm": 0.2281433790922165,
      "learning_rate": 0.00021731164793111692,
      "loss": 0.0452,
      "step": 6267
    },
    {
      "epoch": 0.8784863349684653,
      "grad_norm": 0.12181345373392105,
      "learning_rate": 0.0002172972972972973,
      "loss": 0.0238,
      "step": 6268
    },
    {
      "epoch": 0.8786264891380519,
      "grad_norm": 0.18437130749225616,
      "learning_rate": 0.00021728294666347763,
      "loss": 0.0315,
      "step": 6269
    },
    {
      "epoch": 0.8787666433076384,
      "grad_norm": 0.40676167607307434,
      "learning_rate": 0.00021726859602965796,
      "loss": 0.0694,
      "step": 6270
    },
    {
      "epoch": 0.878906797477225,
      "grad_norm": 0.6529611349105835,
      "learning_rate": 0.0002172542453958383,
      "loss": 0.0652,
      "step": 6271
    },
    {
      "epoch": 0.8790469516468115,
      "grad_norm": 0.3763105869293213,
      "learning_rate": 0.00021723989476201864,
      "loss": 0.0785,
      "step": 6272
    },
    {
      "epoch": 0.879187105816398,
      "grad_norm": 0.5702271461486816,
      "learning_rate": 0.00021722554412819897,
      "loss": 0.1388,
      "step": 6273
    },
    {
      "epoch": 0.8793272599859846,
      "grad_norm": 0.39969533681869507,
      "learning_rate": 0.00021721119349437932,
      "loss": 0.084,
      "step": 6274
    },
    {
      "epoch": 0.8794674141555712,
      "grad_norm": 0.3073981702327728,
      "learning_rate": 0.00021719684286055965,
      "loss": 0.045,
      "step": 6275
    },
    {
      "epoch": 0.8796075683251576,
      "grad_norm": 0.49991291761398315,
      "learning_rate": 0.00021718249222673998,
      "loss": 0.0778,
      "step": 6276
    },
    {
      "epoch": 0.8797477224947442,
      "grad_norm": 0.2920743525028229,
      "learning_rate": 0.0002171681415929203,
      "loss": 0.0828,
      "step": 6277
    },
    {
      "epoch": 0.8798878766643308,
      "grad_norm": 0.33542996644973755,
      "learning_rate": 0.0002171537909591007,
      "loss": 0.047,
      "step": 6278
    },
    {
      "epoch": 0.8800280308339173,
      "grad_norm": 0.4488348364830017,
      "learning_rate": 0.00021713944032528102,
      "loss": 0.0976,
      "step": 6279
    },
    {
      "epoch": 0.8801681850035038,
      "grad_norm": 0.2969159185886383,
      "learning_rate": 0.00021712508969146135,
      "loss": 0.0358,
      "step": 6280
    },
    {
      "epoch": 0.8803083391730904,
      "grad_norm": 0.3600703775882721,
      "learning_rate": 0.0002171107390576417,
      "loss": 0.1016,
      "step": 6281
    },
    {
      "epoch": 0.8804484933426769,
      "grad_norm": 0.6032580733299255,
      "learning_rate": 0.00021709638842382203,
      "loss": 0.0852,
      "step": 6282
    },
    {
      "epoch": 0.8805886475122635,
      "grad_norm": 0.4375476539134979,
      "learning_rate": 0.00021708203779000236,
      "loss": 0.0629,
      "step": 6283
    },
    {
      "epoch": 0.8807288016818501,
      "grad_norm": 0.5672846436500549,
      "learning_rate": 0.00021706768715618272,
      "loss": 0.0732,
      "step": 6284
    },
    {
      "epoch": 0.8808689558514365,
      "grad_norm": 0.32331809401512146,
      "learning_rate": 0.00021705333652236305,
      "loss": 0.0801,
      "step": 6285
    },
    {
      "epoch": 0.8810091100210231,
      "grad_norm": 0.335155189037323,
      "learning_rate": 0.00021703898588854337,
      "loss": 0.0378,
      "step": 6286
    },
    {
      "epoch": 0.8811492641906097,
      "grad_norm": 0.29419201612472534,
      "learning_rate": 0.00021702463525472376,
      "loss": 0.0489,
      "step": 6287
    },
    {
      "epoch": 0.8812894183601963,
      "grad_norm": 0.6636062860488892,
      "learning_rate": 0.00021701028462090409,
      "loss": 0.1538,
      "step": 6288
    },
    {
      "epoch": 0.8814295725297827,
      "grad_norm": 0.45243704319000244,
      "learning_rate": 0.00021699593398708441,
      "loss": 0.0546,
      "step": 6289
    },
    {
      "epoch": 0.8815697266993693,
      "grad_norm": 0.7146409153938293,
      "learning_rate": 0.00021698158335326477,
      "loss": 0.0798,
      "step": 6290
    },
    {
      "epoch": 0.8817098808689559,
      "grad_norm": 0.5848540663719177,
      "learning_rate": 0.0002169672327194451,
      "loss": 0.0477,
      "step": 6291
    },
    {
      "epoch": 0.8818500350385424,
      "grad_norm": 0.32826846837997437,
      "learning_rate": 0.00021695288208562543,
      "loss": 0.0209,
      "step": 6292
    },
    {
      "epoch": 0.881990189208129,
      "grad_norm": 0.650705873966217,
      "learning_rate": 0.00021693853145180576,
      "loss": 0.0637,
      "step": 6293
    },
    {
      "epoch": 0.8821303433777155,
      "grad_norm": 0.804492712020874,
      "learning_rate": 0.0002169241808179861,
      "loss": 0.1327,
      "step": 6294
    },
    {
      "epoch": 0.882270497547302,
      "grad_norm": 0.38751909136772156,
      "learning_rate": 0.00021690983018416644,
      "loss": 0.0385,
      "step": 6295
    },
    {
      "epoch": 0.8824106517168886,
      "grad_norm": 0.6429628133773804,
      "learning_rate": 0.00021689547955034677,
      "loss": 0.0569,
      "step": 6296
    },
    {
      "epoch": 0.8825508058864752,
      "grad_norm": 0.49627959728240967,
      "learning_rate": 0.00021688112891652712,
      "loss": 0.0756,
      "step": 6297
    },
    {
      "epoch": 0.8826909600560616,
      "grad_norm": 2.548335552215576,
      "learning_rate": 0.00021686677828270745,
      "loss": 0.2245,
      "step": 6298
    },
    {
      "epoch": 0.8828311142256482,
      "grad_norm": 0.9784413576126099,
      "learning_rate": 0.00021685242764888778,
      "loss": 0.1249,
      "step": 6299
    },
    {
      "epoch": 0.8829712683952348,
      "grad_norm": 1.6628530025482178,
      "learning_rate": 0.00021683807701506816,
      "loss": 0.1621,
      "step": 6300
    },
    {
      "epoch": 0.8831114225648213,
      "grad_norm": 0.5426815748214722,
      "learning_rate": 0.0002168237263812485,
      "loss": 0.1478,
      "step": 6301
    },
    {
      "epoch": 0.8832515767344078,
      "grad_norm": 0.31795185804367065,
      "learning_rate": 0.00021680937574742882,
      "loss": 0.0959,
      "step": 6302
    },
    {
      "epoch": 0.8833917309039944,
      "grad_norm": 0.23255884647369385,
      "learning_rate": 0.00021679502511360918,
      "loss": 0.0343,
      "step": 6303
    },
    {
      "epoch": 0.8835318850735809,
      "grad_norm": 0.37613925337791443,
      "learning_rate": 0.0002167806744797895,
      "loss": 0.1066,
      "step": 6304
    },
    {
      "epoch": 0.8836720392431675,
      "grad_norm": 0.4515974819660187,
      "learning_rate": 0.00021676632384596983,
      "loss": 0.0811,
      "step": 6305
    },
    {
      "epoch": 0.8838121934127541,
      "grad_norm": 0.24465088546276093,
      "learning_rate": 0.0002167519732121502,
      "loss": 0.0663,
      "step": 6306
    },
    {
      "epoch": 0.8839523475823405,
      "grad_norm": 0.3040670156478882,
      "learning_rate": 0.00021673762257833052,
      "loss": 0.0625,
      "step": 6307
    },
    {
      "epoch": 0.8840925017519271,
      "grad_norm": 0.4737105369567871,
      "learning_rate": 0.00021672327194451085,
      "loss": 0.1029,
      "step": 6308
    },
    {
      "epoch": 0.8842326559215137,
      "grad_norm": 0.6553725004196167,
      "learning_rate": 0.00021670892131069123,
      "loss": 0.1536,
      "step": 6309
    },
    {
      "epoch": 0.8843728100911002,
      "grad_norm": 0.4673299789428711,
      "learning_rate": 0.00021669457067687156,
      "loss": 0.0418,
      "step": 6310
    },
    {
      "epoch": 0.8845129642606867,
      "grad_norm": 0.3112906217575073,
      "learning_rate": 0.0002166802200430519,
      "loss": 0.0972,
      "step": 6311
    },
    {
      "epoch": 0.8846531184302733,
      "grad_norm": 0.3027682602405548,
      "learning_rate": 0.00021666586940923222,
      "loss": 0.1196,
      "step": 6312
    },
    {
      "epoch": 0.8847932725998598,
      "grad_norm": 0.48045089840888977,
      "learning_rate": 0.00021665151877541257,
      "loss": 0.059,
      "step": 6313
    },
    {
      "epoch": 0.8849334267694464,
      "grad_norm": 0.45420539379119873,
      "learning_rate": 0.0002166371681415929,
      "loss": 0.0886,
      "step": 6314
    },
    {
      "epoch": 0.885073580939033,
      "grad_norm": 0.2544019818305969,
      "learning_rate": 0.00021662281750777323,
      "loss": 0.0671,
      "step": 6315
    },
    {
      "epoch": 0.8852137351086194,
      "grad_norm": 0.2321062833070755,
      "learning_rate": 0.00021660846687395358,
      "loss": 0.0608,
      "step": 6316
    },
    {
      "epoch": 0.885353889278206,
      "grad_norm": 0.5656206011772156,
      "learning_rate": 0.0002165941162401339,
      "loss": 0.0795,
      "step": 6317
    },
    {
      "epoch": 0.8854940434477926,
      "grad_norm": 0.19876818358898163,
      "learning_rate": 0.00021657976560631424,
      "loss": 0.0219,
      "step": 6318
    },
    {
      "epoch": 0.8856341976173792,
      "grad_norm": 0.1730661392211914,
      "learning_rate": 0.00021656541497249462,
      "loss": 0.0294,
      "step": 6319
    },
    {
      "epoch": 0.8857743517869656,
      "grad_norm": 0.1284327358007431,
      "learning_rate": 0.00021655106433867495,
      "loss": 0.0249,
      "step": 6320
    },
    {
      "epoch": 0.8859145059565522,
      "grad_norm": 0.34040510654449463,
      "learning_rate": 0.00021653671370485528,
      "loss": 0.0564,
      "step": 6321
    },
    {
      "epoch": 0.8860546601261388,
      "grad_norm": 0.372646689414978,
      "learning_rate": 0.00021652236307103564,
      "loss": 0.0531,
      "step": 6322
    },
    {
      "epoch": 0.8861948142957253,
      "grad_norm": 0.41042637825012207,
      "learning_rate": 0.00021650801243721597,
      "loss": 0.0355,
      "step": 6323
    },
    {
      "epoch": 0.8863349684653118,
      "grad_norm": 0.3855037987232208,
      "learning_rate": 0.0002164936618033963,
      "loss": 0.071,
      "step": 6324
    },
    {
      "epoch": 0.8864751226348984,
      "grad_norm": 0.37636715173721313,
      "learning_rate": 0.00021647931116957665,
      "loss": 0.0653,
      "step": 6325
    },
    {
      "epoch": 0.8866152768044849,
      "grad_norm": 0.46927669644355774,
      "learning_rate": 0.00021646496053575698,
      "loss": 0.1323,
      "step": 6326
    },
    {
      "epoch": 0.8867554309740715,
      "grad_norm": 0.37175631523132324,
      "learning_rate": 0.0002164506099019373,
      "loss": 0.0587,
      "step": 6327
    },
    {
      "epoch": 0.8868955851436581,
      "grad_norm": 0.21671995520591736,
      "learning_rate": 0.00021643625926811764,
      "loss": 0.0607,
      "step": 6328
    },
    {
      "epoch": 0.8870357393132445,
      "grad_norm": 0.4263468384742737,
      "learning_rate": 0.000216421908634298,
      "loss": 0.0625,
      "step": 6329
    },
    {
      "epoch": 0.8871758934828311,
      "grad_norm": 0.5397272706031799,
      "learning_rate": 0.00021640755800047832,
      "loss": 0.0685,
      "step": 6330
    },
    {
      "epoch": 0.8873160476524177,
      "grad_norm": 0.35666197538375854,
      "learning_rate": 0.00021639320736665865,
      "loss": 0.0717,
      "step": 6331
    },
    {
      "epoch": 0.8874562018220042,
      "grad_norm": 0.5617634654045105,
      "learning_rate": 0.00021637885673283903,
      "loss": 0.1761,
      "step": 6332
    },
    {
      "epoch": 0.8875963559915907,
      "grad_norm": 0.21159803867340088,
      "learning_rate": 0.00021636450609901936,
      "loss": 0.0305,
      "step": 6333
    },
    {
      "epoch": 0.8877365101611773,
      "grad_norm": 0.40066707134246826,
      "learning_rate": 0.0002163501554651997,
      "loss": 0.0715,
      "step": 6334
    },
    {
      "epoch": 0.8878766643307638,
      "grad_norm": 0.6241999864578247,
      "learning_rate": 0.00021633580483138004,
      "loss": 0.0892,
      "step": 6335
    },
    {
      "epoch": 0.8880168185003504,
      "grad_norm": 0.37987348437309265,
      "learning_rate": 0.00021632145419756037,
      "loss": 0.0755,
      "step": 6336
    },
    {
      "epoch": 0.888156972669937,
      "grad_norm": 0.4611404836177826,
      "learning_rate": 0.0002163071035637407,
      "loss": 0.067,
      "step": 6337
    },
    {
      "epoch": 0.8882971268395234,
      "grad_norm": 0.33832624554634094,
      "learning_rate": 0.00021629275292992106,
      "loss": 0.0771,
      "step": 6338
    },
    {
      "epoch": 0.88843728100911,
      "grad_norm": 0.4897250235080719,
      "learning_rate": 0.00021627840229610138,
      "loss": 0.1297,
      "step": 6339
    },
    {
      "epoch": 0.8885774351786966,
      "grad_norm": 0.5794617533683777,
      "learning_rate": 0.0002162640516622817,
      "loss": 0.0511,
      "step": 6340
    },
    {
      "epoch": 0.8887175893482832,
      "grad_norm": 1.030187964439392,
      "learning_rate": 0.0002162497010284621,
      "loss": 0.1939,
      "step": 6341
    },
    {
      "epoch": 0.8888577435178696,
      "grad_norm": 0.46357640624046326,
      "learning_rate": 0.00021623535039464242,
      "loss": 0.0546,
      "step": 6342
    },
    {
      "epoch": 0.8889978976874562,
      "grad_norm": 0.49379438161849976,
      "learning_rate": 0.00021622099976082275,
      "loss": 0.0248,
      "step": 6343
    },
    {
      "epoch": 0.8891380518570428,
      "grad_norm": 0.34207683801651,
      "learning_rate": 0.0002162066491270031,
      "loss": 0.0262,
      "step": 6344
    },
    {
      "epoch": 0.8892782060266293,
      "grad_norm": 0.42017680406570435,
      "learning_rate": 0.00021619229849318344,
      "loss": 0.0757,
      "step": 6345
    },
    {
      "epoch": 0.8894183601962158,
      "grad_norm": 0.8866990804672241,
      "learning_rate": 0.00021617794785936377,
      "loss": 0.1058,
      "step": 6346
    },
    {
      "epoch": 0.8895585143658024,
      "grad_norm": 0.2930103838443756,
      "learning_rate": 0.0002161635972255441,
      "loss": 0.0172,
      "step": 6347
    },
    {
      "epoch": 0.8896986685353889,
      "grad_norm": 0.7346456050872803,
      "learning_rate": 0.00021614924659172445,
      "loss": 0.1232,
      "step": 6348
    },
    {
      "epoch": 0.8898388227049755,
      "grad_norm": 1.468702793121338,
      "learning_rate": 0.00021613489595790478,
      "loss": 0.1692,
      "step": 6349
    },
    {
      "epoch": 0.8899789768745621,
      "grad_norm": 1.1133503913879395,
      "learning_rate": 0.0002161205453240851,
      "loss": 0.4245,
      "step": 6350
    },
    {
      "epoch": 0.8901191310441485,
      "grad_norm": 0.47393229603767395,
      "learning_rate": 0.0002161061946902655,
      "loss": 0.0605,
      "step": 6351
    },
    {
      "epoch": 0.8902592852137351,
      "grad_norm": 0.35352733731269836,
      "learning_rate": 0.00021609184405644582,
      "loss": 0.0644,
      "step": 6352
    },
    {
      "epoch": 0.8903994393833217,
      "grad_norm": 0.29689136147499084,
      "learning_rate": 0.00021607749342262615,
      "loss": 0.0453,
      "step": 6353
    },
    {
      "epoch": 0.8905395935529082,
      "grad_norm": 0.3777115046977997,
      "learning_rate": 0.0002160631427888065,
      "loss": 0.0884,
      "step": 6354
    },
    {
      "epoch": 0.8906797477224947,
      "grad_norm": 0.281775563955307,
      "learning_rate": 0.00021604879215498683,
      "loss": 0.0236,
      "step": 6355
    },
    {
      "epoch": 0.8908199018920813,
      "grad_norm": 0.3897337019443512,
      "learning_rate": 0.00021603444152116716,
      "loss": 0.0967,
      "step": 6356
    },
    {
      "epoch": 0.8909600560616678,
      "grad_norm": 0.4497261643409729,
      "learning_rate": 0.00021602009088734752,
      "loss": 0.0868,
      "step": 6357
    },
    {
      "epoch": 0.8911002102312544,
      "grad_norm": 0.31632840633392334,
      "learning_rate": 0.00021600574025352784,
      "loss": 0.0594,
      "step": 6358
    },
    {
      "epoch": 0.891240364400841,
      "grad_norm": 0.5239577889442444,
      "learning_rate": 0.00021599138961970817,
      "loss": 0.079,
      "step": 6359
    },
    {
      "epoch": 0.8913805185704274,
      "grad_norm": 0.45975571870803833,
      "learning_rate": 0.00021597703898588853,
      "loss": 0.09,
      "step": 6360
    },
    {
      "epoch": 0.891520672740014,
      "grad_norm": 1.152085781097412,
      "learning_rate": 0.00021596268835206886,
      "loss": 0.1501,
      "step": 6361
    },
    {
      "epoch": 0.8916608269096006,
      "grad_norm": 0.2481631636619568,
      "learning_rate": 0.00021594833771824919,
      "loss": 0.0623,
      "step": 6362
    },
    {
      "epoch": 0.8918009810791871,
      "grad_norm": 0.33675843477249146,
      "learning_rate": 0.00021593398708442951,
      "loss": 0.0868,
      "step": 6363
    },
    {
      "epoch": 0.8919411352487736,
      "grad_norm": 0.6384389400482178,
      "learning_rate": 0.0002159196364506099,
      "loss": 0.1124,
      "step": 6364
    },
    {
      "epoch": 0.8920812894183602,
      "grad_norm": 1.1308022737503052,
      "learning_rate": 0.00021590528581679023,
      "loss": 0.0278,
      "step": 6365
    },
    {
      "epoch": 0.8922214435879467,
      "grad_norm": 0.460615873336792,
      "learning_rate": 0.00021589093518297055,
      "loss": 0.1178,
      "step": 6366
    },
    {
      "epoch": 0.8923615977575333,
      "grad_norm": 0.18188369274139404,
      "learning_rate": 0.0002158765845491509,
      "loss": 0.0795,
      "step": 6367
    },
    {
      "epoch": 0.8925017519271198,
      "grad_norm": 0.29433348774909973,
      "learning_rate": 0.00021586223391533124,
      "loss": 0.0578,
      "step": 6368
    },
    {
      "epoch": 0.8926419060967064,
      "grad_norm": 0.1542975902557373,
      "learning_rate": 0.00021584788328151157,
      "loss": 0.0268,
      "step": 6369
    },
    {
      "epoch": 0.8927820602662929,
      "grad_norm": 0.37385544180870056,
      "learning_rate": 0.00021583353264769192,
      "loss": 0.0967,
      "step": 6370
    },
    {
      "epoch": 0.8929222144358795,
      "grad_norm": 0.1319681704044342,
      "learning_rate": 0.00021581918201387225,
      "loss": 0.0155,
      "step": 6371
    },
    {
      "epoch": 0.8930623686054661,
      "grad_norm": 0.48037686944007874,
      "learning_rate": 0.00021580483138005258,
      "loss": 0.0875,
      "step": 6372
    },
    {
      "epoch": 0.8932025227750525,
      "grad_norm": 0.18546746671199799,
      "learning_rate": 0.00021579048074623296,
      "loss": 0.0361,
      "step": 6373
    },
    {
      "epoch": 0.8933426769446391,
      "grad_norm": 0.2430107295513153,
      "learning_rate": 0.0002157761301124133,
      "loss": 0.0502,
      "step": 6374
    },
    {
      "epoch": 0.8934828311142257,
      "grad_norm": 0.41412192583084106,
      "learning_rate": 0.00021576177947859362,
      "loss": 0.1078,
      "step": 6375
    },
    {
      "epoch": 0.8936229852838122,
      "grad_norm": 0.5208156108856201,
      "learning_rate": 0.00021574742884477398,
      "loss": 0.0711,
      "step": 6376
    },
    {
      "epoch": 0.8937631394533987,
      "grad_norm": 0.30979371070861816,
      "learning_rate": 0.0002157330782109543,
      "loss": 0.0524,
      "step": 6377
    },
    {
      "epoch": 0.8939032936229853,
      "grad_norm": 0.7480751872062683,
      "learning_rate": 0.00021571872757713463,
      "loss": 0.1444,
      "step": 6378
    },
    {
      "epoch": 0.8940434477925718,
      "grad_norm": 0.5892516374588013,
      "learning_rate": 0.000215704376943315,
      "loss": 0.1176,
      "step": 6379
    },
    {
      "epoch": 0.8941836019621584,
      "grad_norm": 0.3465835154056549,
      "learning_rate": 0.00021569002630949532,
      "loss": 0.0706,
      "step": 6380
    },
    {
      "epoch": 0.894323756131745,
      "grad_norm": 0.469944030046463,
      "learning_rate": 0.00021567567567567565,
      "loss": 0.1279,
      "step": 6381
    },
    {
      "epoch": 0.8944639103013314,
      "grad_norm": 0.6023955345153809,
      "learning_rate": 0.00021566132504185597,
      "loss": 0.0629,
      "step": 6382
    },
    {
      "epoch": 0.894604064470918,
      "grad_norm": 1.1703921556472778,
      "learning_rate": 0.00021564697440803636,
      "loss": 0.1073,
      "step": 6383
    },
    {
      "epoch": 0.8947442186405046,
      "grad_norm": 0.4299857020378113,
      "learning_rate": 0.00021563262377421668,
      "loss": 0.0607,
      "step": 6384
    },
    {
      "epoch": 0.8948843728100911,
      "grad_norm": 0.18168188631534576,
      "learning_rate": 0.000215618273140397,
      "loss": 0.0464,
      "step": 6385
    },
    {
      "epoch": 0.8950245269796776,
      "grad_norm": 0.26161521673202515,
      "learning_rate": 0.00021560392250657737,
      "loss": 0.0708,
      "step": 6386
    },
    {
      "epoch": 0.8951646811492642,
      "grad_norm": 0.37883812189102173,
      "learning_rate": 0.0002155895718727577,
      "loss": 0.0906,
      "step": 6387
    },
    {
      "epoch": 0.8953048353188507,
      "grad_norm": 0.2373247742652893,
      "learning_rate": 0.00021557522123893803,
      "loss": 0.0216,
      "step": 6388
    },
    {
      "epoch": 0.8954449894884373,
      "grad_norm": 0.5080769658088684,
      "learning_rate": 0.00021556087060511838,
      "loss": 0.0593,
      "step": 6389
    },
    {
      "epoch": 0.8955851436580238,
      "grad_norm": 0.5784989595413208,
      "learning_rate": 0.0002155465199712987,
      "loss": 0.084,
      "step": 6390
    },
    {
      "epoch": 0.8957252978276103,
      "grad_norm": 0.8219420909881592,
      "learning_rate": 0.00021553216933747904,
      "loss": 0.0639,
      "step": 6391
    },
    {
      "epoch": 0.8958654519971969,
      "grad_norm": 0.28659573197364807,
      "learning_rate": 0.0002155178187036594,
      "loss": 0.0403,
      "step": 6392
    },
    {
      "epoch": 0.8960056061667835,
      "grad_norm": 0.20385098457336426,
      "learning_rate": 0.00021550346806983972,
      "loss": 0.0229,
      "step": 6393
    },
    {
      "epoch": 0.8961457603363701,
      "grad_norm": 0.3711751103401184,
      "learning_rate": 0.00021548911743602005,
      "loss": 0.0559,
      "step": 6394
    },
    {
      "epoch": 0.8962859145059565,
      "grad_norm": 0.9714674949645996,
      "learning_rate": 0.00021547476680220043,
      "loss": 0.3136,
      "step": 6395
    },
    {
      "epoch": 0.8964260686755431,
      "grad_norm": 1.0008267164230347,
      "learning_rate": 0.00021546041616838076,
      "loss": 0.076,
      "step": 6396
    },
    {
      "epoch": 0.8965662228451297,
      "grad_norm": 1.8442028760910034,
      "learning_rate": 0.0002154460655345611,
      "loss": 0.1535,
      "step": 6397
    },
    {
      "epoch": 0.8967063770147162,
      "grad_norm": 1.917199969291687,
      "learning_rate": 0.00021543171490074142,
      "loss": 0.2855,
      "step": 6398
    },
    {
      "epoch": 0.8968465311843027,
      "grad_norm": 1.8050737380981445,
      "learning_rate": 0.00021541736426692178,
      "loss": 0.2695,
      "step": 6399
    },
    {
      "epoch": 0.8969866853538893,
      "grad_norm": 1.1386626958847046,
      "learning_rate": 0.0002154030136331021,
      "loss": 0.1617,
      "step": 6400
    },
    {
      "epoch": 0.8971268395234758,
      "grad_norm": 0.4858371317386627,
      "learning_rate": 0.00021538866299928243,
      "loss": 0.0469,
      "step": 6401
    },
    {
      "epoch": 0.8972669936930624,
      "grad_norm": 0.2051311731338501,
      "learning_rate": 0.0002153743123654628,
      "loss": 0.0257,
      "step": 6402
    },
    {
      "epoch": 0.897407147862649,
      "grad_norm": 0.6071633696556091,
      "learning_rate": 0.00021535996173164312,
      "loss": 0.0948,
      "step": 6403
    },
    {
      "epoch": 0.8975473020322354,
      "grad_norm": 0.4932337701320648,
      "learning_rate": 0.00021534561109782345,
      "loss": 0.0854,
      "step": 6404
    },
    {
      "epoch": 0.897687456201822,
      "grad_norm": 0.5392564535140991,
      "learning_rate": 0.00021533126046400383,
      "loss": 0.0818,
      "step": 6405
    },
    {
      "epoch": 0.8978276103714086,
      "grad_norm": 0.2504193186759949,
      "learning_rate": 0.00021531690983018416,
      "loss": 0.0906,
      "step": 6406
    },
    {
      "epoch": 0.8979677645409951,
      "grad_norm": 0.20426024496555328,
      "learning_rate": 0.00021530255919636449,
      "loss": 0.0632,
      "step": 6407
    },
    {
      "epoch": 0.8981079187105816,
      "grad_norm": 0.24049171805381775,
      "learning_rate": 0.00021528820856254484,
      "loss": 0.0541,
      "step": 6408
    },
    {
      "epoch": 0.8982480728801682,
      "grad_norm": 0.3818652033805847,
      "learning_rate": 0.00021527385792872517,
      "loss": 0.082,
      "step": 6409
    },
    {
      "epoch": 0.8983882270497547,
      "grad_norm": 0.41560080647468567,
      "learning_rate": 0.0002152595072949055,
      "loss": 0.063,
      "step": 6410
    },
    {
      "epoch": 0.8985283812193413,
      "grad_norm": 0.8477592468261719,
      "learning_rate": 0.00021524515666108585,
      "loss": 0.1278,
      "step": 6411
    },
    {
      "epoch": 0.8986685353889278,
      "grad_norm": 0.6891153454780579,
      "learning_rate": 0.00021523080602726618,
      "loss": 0.0605,
      "step": 6412
    },
    {
      "epoch": 0.8988086895585143,
      "grad_norm": 0.38674548268318176,
      "learning_rate": 0.0002152164553934465,
      "loss": 0.0761,
      "step": 6413
    },
    {
      "epoch": 0.8989488437281009,
      "grad_norm": 0.3405720591545105,
      "learning_rate": 0.0002152021047596269,
      "loss": 0.094,
      "step": 6414
    },
    {
      "epoch": 0.8990889978976875,
      "grad_norm": 0.4232446551322937,
      "learning_rate": 0.00021518775412580722,
      "loss": 0.0762,
      "step": 6415
    },
    {
      "epoch": 0.899229152067274,
      "grad_norm": 0.49405723810195923,
      "learning_rate": 0.00021517340349198755,
      "loss": 0.1253,
      "step": 6416
    },
    {
      "epoch": 0.8993693062368605,
      "grad_norm": 0.6300173997879028,
      "learning_rate": 0.00021515905285816788,
      "loss": 0.1427,
      "step": 6417
    },
    {
      "epoch": 0.8995094604064471,
      "grad_norm": 0.36156022548675537,
      "learning_rate": 0.00021514470222434824,
      "loss": 0.0829,
      "step": 6418
    },
    {
      "epoch": 0.8996496145760337,
      "grad_norm": 0.3899405002593994,
      "learning_rate": 0.00021513035159052856,
      "loss": 0.063,
      "step": 6419
    },
    {
      "epoch": 0.8997897687456202,
      "grad_norm": 0.4408261775970459,
      "learning_rate": 0.0002151160009567089,
      "loss": 0.1094,
      "step": 6420
    },
    {
      "epoch": 0.8999299229152067,
      "grad_norm": 0.3543868362903595,
      "learning_rate": 0.00021510165032288925,
      "loss": 0.0594,
      "step": 6421
    },
    {
      "epoch": 0.9000700770847933,
      "grad_norm": 0.4986155331134796,
      "learning_rate": 0.00021508729968906958,
      "loss": 0.0548,
      "step": 6422
    },
    {
      "epoch": 0.9002102312543798,
      "grad_norm": 0.26845064759254456,
      "learning_rate": 0.0002150729490552499,
      "loss": 0.05,
      "step": 6423
    },
    {
      "epoch": 0.9003503854239664,
      "grad_norm": 0.20656314492225647,
      "learning_rate": 0.00021505859842143026,
      "loss": 0.0239,
      "step": 6424
    },
    {
      "epoch": 0.900490539593553,
      "grad_norm": 0.4421077370643616,
      "learning_rate": 0.0002150442477876106,
      "loss": 0.0364,
      "step": 6425
    },
    {
      "epoch": 0.9006306937631394,
      "grad_norm": 0.5424543023109436,
      "learning_rate": 0.00021502989715379092,
      "loss": 0.0692,
      "step": 6426
    },
    {
      "epoch": 0.900770847932726,
      "grad_norm": 0.4379134178161621,
      "learning_rate": 0.0002150155465199713,
      "loss": 0.1015,
      "step": 6427
    },
    {
      "epoch": 0.9009110021023126,
      "grad_norm": 0.2991301417350769,
      "learning_rate": 0.00021500119588615163,
      "loss": 0.0404,
      "step": 6428
    },
    {
      "epoch": 0.9010511562718991,
      "grad_norm": 0.5305184721946716,
      "learning_rate": 0.00021498684525233196,
      "loss": 0.0713,
      "step": 6429
    },
    {
      "epoch": 0.9011913104414856,
      "grad_norm": 0.286765456199646,
      "learning_rate": 0.00021497249461851231,
      "loss": 0.0351,
      "step": 6430
    },
    {
      "epoch": 0.9013314646110722,
      "grad_norm": 0.2445044219493866,
      "learning_rate": 0.00021495814398469264,
      "loss": 0.0795,
      "step": 6431
    },
    {
      "epoch": 0.9014716187806587,
      "grad_norm": 0.35519471764564514,
      "learning_rate": 0.00021494379335087297,
      "loss": 0.0819,
      "step": 6432
    },
    {
      "epoch": 0.9016117729502453,
      "grad_norm": 0.2523954212665558,
      "learning_rate": 0.0002149294427170533,
      "loss": 0.0937,
      "step": 6433
    },
    {
      "epoch": 0.9017519271198318,
      "grad_norm": 0.1853904128074646,
      "learning_rate": 0.00021491509208323366,
      "loss": 0.025,
      "step": 6434
    },
    {
      "epoch": 0.9018920812894183,
      "grad_norm": 0.6573611497879028,
      "learning_rate": 0.00021490074144941398,
      "loss": 0.1004,
      "step": 6435
    },
    {
      "epoch": 0.9020322354590049,
      "grad_norm": 0.37836775183677673,
      "learning_rate": 0.0002148863908155943,
      "loss": 0.053,
      "step": 6436
    },
    {
      "epoch": 0.9021723896285915,
      "grad_norm": 0.31566399335861206,
      "learning_rate": 0.0002148720401817747,
      "loss": 0.049,
      "step": 6437
    },
    {
      "epoch": 0.902312543798178,
      "grad_norm": 0.8381720185279846,
      "learning_rate": 0.00021485768954795502,
      "loss": 0.1271,
      "step": 6438
    },
    {
      "epoch": 0.9024526979677645,
      "grad_norm": 0.15564432740211487,
      "learning_rate": 0.00021484333891413535,
      "loss": 0.0135,
      "step": 6439
    },
    {
      "epoch": 0.9025928521373511,
      "grad_norm": 0.6811659336090088,
      "learning_rate": 0.0002148289882803157,
      "loss": 0.1118,
      "step": 6440
    },
    {
      "epoch": 0.9027330063069376,
      "grad_norm": 0.4137628376483917,
      "learning_rate": 0.00021481463764649604,
      "loss": 0.0277,
      "step": 6441
    },
    {
      "epoch": 0.9028731604765242,
      "grad_norm": 0.40326789021492004,
      "learning_rate": 0.00021480028701267636,
      "loss": 0.0826,
      "step": 6442
    },
    {
      "epoch": 0.9030133146461107,
      "grad_norm": 0.3607103228569031,
      "learning_rate": 0.00021478593637885672,
      "loss": 0.071,
      "step": 6443
    },
    {
      "epoch": 0.9031534688156972,
      "grad_norm": 0.5023744106292725,
      "learning_rate": 0.00021477158574503705,
      "loss": 0.0638,
      "step": 6444
    },
    {
      "epoch": 0.9032936229852838,
      "grad_norm": 0.6061773300170898,
      "learning_rate": 0.00021475723511121738,
      "loss": 0.046,
      "step": 6445
    },
    {
      "epoch": 0.9034337771548704,
      "grad_norm": 0.23244105279445648,
      "learning_rate": 0.00021474288447739776,
      "loss": 0.0192,
      "step": 6446
    },
    {
      "epoch": 0.9035739313244568,
      "grad_norm": 1.5404292345046997,
      "learning_rate": 0.0002147285338435781,
      "loss": 0.0591,
      "step": 6447
    },
    {
      "epoch": 0.9037140854940434,
      "grad_norm": 0.9891332983970642,
      "learning_rate": 0.00021471418320975842,
      "loss": 0.0598,
      "step": 6448
    },
    {
      "epoch": 0.90385423966363,
      "grad_norm": 1.560807228088379,
      "learning_rate": 0.00021469983257593877,
      "loss": 0.1885,
      "step": 6449
    },
    {
      "epoch": 0.9039943938332166,
      "grad_norm": 0.8083940744400024,
      "learning_rate": 0.0002146854819421191,
      "loss": 0.1647,
      "step": 6450
    },
    {
      "epoch": 0.9041345480028031,
      "grad_norm": 0.27475714683532715,
      "learning_rate": 0.00021467113130829943,
      "loss": 0.045,
      "step": 6451
    },
    {
      "epoch": 0.9042747021723896,
      "grad_norm": 0.5286942720413208,
      "learning_rate": 0.00021465678067447976,
      "loss": 0.0777,
      "step": 6452
    },
    {
      "epoch": 0.9044148563419762,
      "grad_norm": 0.3517037630081177,
      "learning_rate": 0.00021464243004066011,
      "loss": 0.0309,
      "step": 6453
    },
    {
      "epoch": 0.9045550105115627,
      "grad_norm": 0.38065117597579956,
      "learning_rate": 0.00021462807940684044,
      "loss": 0.053,
      "step": 6454
    },
    {
      "epoch": 0.9046951646811493,
      "grad_norm": 0.5434138178825378,
      "learning_rate": 0.00021461372877302077,
      "loss": 0.0563,
      "step": 6455
    },
    {
      "epoch": 0.9048353188507358,
      "grad_norm": 0.7512357234954834,
      "learning_rate": 0.00021459937813920113,
      "loss": 0.0961,
      "step": 6456
    },
    {
      "epoch": 0.9049754730203223,
      "grad_norm": 0.4269784390926361,
      "learning_rate": 0.00021458502750538146,
      "loss": 0.0501,
      "step": 6457
    },
    {
      "epoch": 0.9051156271899089,
      "grad_norm": 0.23223674297332764,
      "learning_rate": 0.00021457067687156178,
      "loss": 0.0357,
      "step": 6458
    },
    {
      "epoch": 0.9052557813594955,
      "grad_norm": 0.2480769008398056,
      "learning_rate": 0.00021455632623774217,
      "loss": 0.0333,
      "step": 6459
    },
    {
      "epoch": 0.905395935529082,
      "grad_norm": 0.18888044357299805,
      "learning_rate": 0.0002145419756039225,
      "loss": 0.0504,
      "step": 6460
    },
    {
      "epoch": 0.9055360896986685,
      "grad_norm": 0.493127703666687,
      "learning_rate": 0.00021452762497010282,
      "loss": 0.1079,
      "step": 6461
    },
    {
      "epoch": 0.9056762438682551,
      "grad_norm": 0.32565072178840637,
      "learning_rate": 0.00021451327433628318,
      "loss": 0.0711,
      "step": 6462
    },
    {
      "epoch": 0.9058163980378416,
      "grad_norm": 0.48333826661109924,
      "learning_rate": 0.0002144989237024635,
      "loss": 0.0641,
      "step": 6463
    },
    {
      "epoch": 0.9059565522074282,
      "grad_norm": 0.40877336263656616,
      "learning_rate": 0.00021448457306864384,
      "loss": 0.0314,
      "step": 6464
    },
    {
      "epoch": 0.9060967063770147,
      "grad_norm": 0.2629973292350769,
      "learning_rate": 0.0002144702224348242,
      "loss": 0.0586,
      "step": 6465
    },
    {
      "epoch": 0.9062368605466012,
      "grad_norm": 0.37494751811027527,
      "learning_rate": 0.00021445587180100452,
      "loss": 0.0958,
      "step": 6466
    },
    {
      "epoch": 0.9063770147161878,
      "grad_norm": 0.7391054034233093,
      "learning_rate": 0.00021444152116718485,
      "loss": 0.121,
      "step": 6467
    },
    {
      "epoch": 0.9065171688857744,
      "grad_norm": 0.281841903924942,
      "learning_rate": 0.00021442717053336518,
      "loss": 0.04,
      "step": 6468
    },
    {
      "epoch": 0.9066573230553608,
      "grad_norm": 0.3329413831233978,
      "learning_rate": 0.00021441281989954556,
      "loss": 0.0551,
      "step": 6469
    },
    {
      "epoch": 0.9067974772249474,
      "grad_norm": 0.3830040991306305,
      "learning_rate": 0.0002143984692657259,
      "loss": 0.0683,
      "step": 6470
    },
    {
      "epoch": 0.906937631394534,
      "grad_norm": 0.6144102811813354,
      "learning_rate": 0.00021438411863190622,
      "loss": 0.1029,
      "step": 6471
    },
    {
      "epoch": 0.9070777855641206,
      "grad_norm": 0.6852854490280151,
      "learning_rate": 0.00021436976799808657,
      "loss": 0.1522,
      "step": 6472
    },
    {
      "epoch": 0.9072179397337071,
      "grad_norm": 0.3438425064086914,
      "learning_rate": 0.0002143554173642669,
      "loss": 0.0777,
      "step": 6473
    },
    {
      "epoch": 0.9073580939032936,
      "grad_norm": 0.17858552932739258,
      "learning_rate": 0.00021434106673044723,
      "loss": 0.0314,
      "step": 6474
    },
    {
      "epoch": 0.9074982480728802,
      "grad_norm": 0.5745185017585754,
      "learning_rate": 0.0002143267160966276,
      "loss": 0.0891,
      "step": 6475
    },
    {
      "epoch": 0.9076384022424667,
      "grad_norm": 0.3353009819984436,
      "learning_rate": 0.00021431236546280792,
      "loss": 0.0706,
      "step": 6476
    },
    {
      "epoch": 0.9077785564120533,
      "grad_norm": 0.9571337699890137,
      "learning_rate": 0.00021429801482898824,
      "loss": 0.0873,
      "step": 6477
    },
    {
      "epoch": 0.9079187105816398,
      "grad_norm": 0.31498396396636963,
      "learning_rate": 0.00021428366419516863,
      "loss": 0.0399,
      "step": 6478
    },
    {
      "epoch": 0.9080588647512263,
      "grad_norm": 0.8018702268600464,
      "learning_rate": 0.00021426931356134896,
      "loss": 0.127,
      "step": 6479
    },
    {
      "epoch": 0.9081990189208129,
      "grad_norm": 0.46543970704078674,
      "learning_rate": 0.00021425496292752928,
      "loss": 0.0792,
      "step": 6480
    },
    {
      "epoch": 0.9083391730903995,
      "grad_norm": 0.2422894537448883,
      "learning_rate": 0.00021424061229370964,
      "loss": 0.0352,
      "step": 6481
    },
    {
      "epoch": 0.908479327259986,
      "grad_norm": 0.33804574608802795,
      "learning_rate": 0.00021422626165988997,
      "loss": 0.0426,
      "step": 6482
    },
    {
      "epoch": 0.9086194814295725,
      "grad_norm": 0.22067680954933167,
      "learning_rate": 0.0002142119110260703,
      "loss": 0.0618,
      "step": 6483
    },
    {
      "epoch": 0.9087596355991591,
      "grad_norm": 0.2905866801738739,
      "learning_rate": 0.00021419756039225065,
      "loss": 0.0563,
      "step": 6484
    },
    {
      "epoch": 0.9088997897687456,
      "grad_norm": 0.1399889439344406,
      "learning_rate": 0.00021418320975843098,
      "loss": 0.0185,
      "step": 6485
    },
    {
      "epoch": 0.9090399439383322,
      "grad_norm": 0.6650295853614807,
      "learning_rate": 0.0002141688591246113,
      "loss": 0.0672,
      "step": 6486
    },
    {
      "epoch": 0.9091800981079187,
      "grad_norm": 0.4174114465713501,
      "learning_rate": 0.00021415450849079164,
      "loss": 0.0686,
      "step": 6487
    },
    {
      "epoch": 0.9093202522775052,
      "grad_norm": 0.5293183326721191,
      "learning_rate": 0.000214140157856972,
      "loss": 0.0523,
      "step": 6488
    },
    {
      "epoch": 0.9094604064470918,
      "grad_norm": 0.9272147417068481,
      "learning_rate": 0.00021412580722315232,
      "loss": 0.1536,
      "step": 6489
    },
    {
      "epoch": 0.9096005606166784,
      "grad_norm": 0.19976091384887695,
      "learning_rate": 0.00021411145658933265,
      "loss": 0.0331,
      "step": 6490
    },
    {
      "epoch": 0.9097407147862648,
      "grad_norm": 0.2879762649536133,
      "learning_rate": 0.00021409710595551303,
      "loss": 0.0335,
      "step": 6491
    },
    {
      "epoch": 0.9098808689558514,
      "grad_norm": 1.6049736738204956,
      "learning_rate": 0.00021408275532169336,
      "loss": 0.2334,
      "step": 6492
    },
    {
      "epoch": 0.910021023125438,
      "grad_norm": 0.4523888826370239,
      "learning_rate": 0.0002140684046878737,
      "loss": 0.1455,
      "step": 6493
    },
    {
      "epoch": 0.9101611772950245,
      "grad_norm": 0.29907873272895813,
      "learning_rate": 0.00021405405405405405,
      "loss": 0.0357,
      "step": 6494
    },
    {
      "epoch": 0.9103013314646111,
      "grad_norm": 0.7394307255744934,
      "learning_rate": 0.00021403970342023437,
      "loss": 0.0545,
      "step": 6495
    },
    {
      "epoch": 0.9104414856341976,
      "grad_norm": 0.5330753922462463,
      "learning_rate": 0.0002140253527864147,
      "loss": 0.2582,
      "step": 6496
    },
    {
      "epoch": 0.9105816398037841,
      "grad_norm": 1.8949971199035645,
      "learning_rate": 0.00021401100215259506,
      "loss": 0.1642,
      "step": 6497
    },
    {
      "epoch": 0.9107217939733707,
      "grad_norm": 0.9914364218711853,
      "learning_rate": 0.0002139966515187754,
      "loss": 0.0872,
      "step": 6498
    },
    {
      "epoch": 0.9108619481429573,
      "grad_norm": 0.4456580877304077,
      "learning_rate": 0.00021398230088495572,
      "loss": 0.041,
      "step": 6499
    },
    {
      "epoch": 0.9110021023125437,
      "grad_norm": 2.9075493812561035,
      "learning_rate": 0.0002139679502511361,
      "loss": 0.1986,
      "step": 6500
    },
    {
      "epoch": 0.9111422564821303,
      "grad_norm": 0.4145531952381134,
      "learning_rate": 0.00021395359961731643,
      "loss": 0.1075,
      "step": 6501
    },
    {
      "epoch": 0.9112824106517169,
      "grad_norm": 0.8073126673698425,
      "learning_rate": 0.00021393924898349676,
      "loss": 0.1935,
      "step": 6502
    },
    {
      "epoch": 0.9114225648213035,
      "grad_norm": 0.218635693192482,
      "learning_rate": 0.00021392489834967708,
      "loss": 0.0496,
      "step": 6503
    },
    {
      "epoch": 0.91156271899089,
      "grad_norm": 0.538770318031311,
      "learning_rate": 0.00021391054771585744,
      "loss": 0.0942,
      "step": 6504
    },
    {
      "epoch": 0.9117028731604765,
      "grad_norm": 0.38798198103904724,
      "learning_rate": 0.00021389619708203777,
      "loss": 0.0702,
      "step": 6505
    },
    {
      "epoch": 0.9118430273300631,
      "grad_norm": 0.30979520082473755,
      "learning_rate": 0.0002138818464482181,
      "loss": 0.082,
      "step": 6506
    },
    {
      "epoch": 0.9119831814996496,
      "grad_norm": 0.3060492277145386,
      "learning_rate": 0.00021386749581439845,
      "loss": 0.0514,
      "step": 6507
    },
    {
      "epoch": 0.9121233356692362,
      "grad_norm": 0.48615023493766785,
      "learning_rate": 0.00021385314518057878,
      "loss": 0.0858,
      "step": 6508
    },
    {
      "epoch": 0.9122634898388227,
      "grad_norm": 0.5969953536987305,
      "learning_rate": 0.0002138387945467591,
      "loss": 0.1424,
      "step": 6509
    },
    {
      "epoch": 0.9124036440084092,
      "grad_norm": 0.28630274534225464,
      "learning_rate": 0.0002138244439129395,
      "loss": 0.0789,
      "step": 6510
    },
    {
      "epoch": 0.9125437981779958,
      "grad_norm": 0.6329449415206909,
      "learning_rate": 0.00021381009327911982,
      "loss": 0.0966,
      "step": 6511
    },
    {
      "epoch": 0.9126839523475824,
      "grad_norm": 0.3260738253593445,
      "learning_rate": 0.00021379574264530015,
      "loss": 0.0361,
      "step": 6512
    },
    {
      "epoch": 0.9128241065171688,
      "grad_norm": 0.31602317094802856,
      "learning_rate": 0.0002137813920114805,
      "loss": 0.0509,
      "step": 6513
    },
    {
      "epoch": 0.9129642606867554,
      "grad_norm": 0.3736482262611389,
      "learning_rate": 0.00021376704137766083,
      "loss": 0.0621,
      "step": 6514
    },
    {
      "epoch": 0.913104414856342,
      "grad_norm": 0.2639308273792267,
      "learning_rate": 0.00021375269074384116,
      "loss": 0.0571,
      "step": 6515
    },
    {
      "epoch": 0.9132445690259285,
      "grad_norm": 0.4399544894695282,
      "learning_rate": 0.00021373834011002152,
      "loss": 0.1004,
      "step": 6516
    },
    {
      "epoch": 0.9133847231955151,
      "grad_norm": 0.6382057070732117,
      "learning_rate": 0.00021372398947620185,
      "loss": 0.0881,
      "step": 6517
    },
    {
      "epoch": 0.9135248773651016,
      "grad_norm": 0.3910166919231415,
      "learning_rate": 0.00021370963884238218,
      "loss": 0.0522,
      "step": 6518
    },
    {
      "epoch": 0.9136650315346881,
      "grad_norm": 0.1977645456790924,
      "learning_rate": 0.0002136952882085625,
      "loss": 0.0637,
      "step": 6519
    },
    {
      "epoch": 0.9138051857042747,
      "grad_norm": 0.44781142473220825,
      "learning_rate": 0.00021368093757474286,
      "loss": 0.1074,
      "step": 6520
    },
    {
      "epoch": 0.9139453398738613,
      "grad_norm": 0.7284477353096008,
      "learning_rate": 0.0002136665869409232,
      "loss": 0.0971,
      "step": 6521
    },
    {
      "epoch": 0.9140854940434477,
      "grad_norm": 0.2804543972015381,
      "learning_rate": 0.00021365223630710352,
      "loss": 0.024,
      "step": 6522
    },
    {
      "epoch": 0.9142256482130343,
      "grad_norm": 0.1625552773475647,
      "learning_rate": 0.0002136378856732839,
      "loss": 0.0401,
      "step": 6523
    },
    {
      "epoch": 0.9143658023826209,
      "grad_norm": 0.4540673494338989,
      "learning_rate": 0.00021362353503946423,
      "loss": 0.1122,
      "step": 6524
    },
    {
      "epoch": 0.9145059565522075,
      "grad_norm": 0.4620015025138855,
      "learning_rate": 0.00021360918440564456,
      "loss": 0.1412,
      "step": 6525
    },
    {
      "epoch": 0.914646110721794,
      "grad_norm": 0.25984036922454834,
      "learning_rate": 0.0002135948337718249,
      "loss": 0.0607,
      "step": 6526
    },
    {
      "epoch": 0.9147862648913805,
      "grad_norm": 0.11482558399438858,
      "learning_rate": 0.00021358048313800524,
      "loss": 0.0398,
      "step": 6527
    },
    {
      "epoch": 0.914926419060967,
      "grad_norm": 0.7788105010986328,
      "learning_rate": 0.00021356613250418557,
      "loss": 0.1029,
      "step": 6528
    },
    {
      "epoch": 0.9150665732305536,
      "grad_norm": 0.5086371302604675,
      "learning_rate": 0.00021355178187036593,
      "loss": 0.161,
      "step": 6529
    },
    {
      "epoch": 0.9152067274001402,
      "grad_norm": 0.4723289906978607,
      "learning_rate": 0.00021353743123654625,
      "loss": 0.1568,
      "step": 6530
    },
    {
      "epoch": 0.9153468815697267,
      "grad_norm": 0.4866516590118408,
      "learning_rate": 0.00021352308060272658,
      "loss": 0.0446,
      "step": 6531
    },
    {
      "epoch": 0.9154870357393132,
      "grad_norm": 0.48494449257850647,
      "learning_rate": 0.00021350872996890697,
      "loss": 0.0985,
      "step": 6532
    },
    {
      "epoch": 0.9156271899088998,
      "grad_norm": 0.13357658684253693,
      "learning_rate": 0.0002134943793350873,
      "loss": 0.0123,
      "step": 6533
    },
    {
      "epoch": 0.9157673440784864,
      "grad_norm": 0.4283812940120697,
      "learning_rate": 0.00021348002870126762,
      "loss": 0.1017,
      "step": 6534
    },
    {
      "epoch": 0.9159074982480728,
      "grad_norm": 0.6210263967514038,
      "learning_rate": 0.00021346567806744798,
      "loss": 0.0429,
      "step": 6535
    },
    {
      "epoch": 0.9160476524176594,
      "grad_norm": 0.5816081762313843,
      "learning_rate": 0.0002134513274336283,
      "loss": 0.1058,
      "step": 6536
    },
    {
      "epoch": 0.916187806587246,
      "grad_norm": 0.1629910171031952,
      "learning_rate": 0.00021343697679980864,
      "loss": 0.0339,
      "step": 6537
    },
    {
      "epoch": 0.9163279607568325,
      "grad_norm": 0.5754914879798889,
      "learning_rate": 0.00021342262616598896,
      "loss": 0.1801,
      "step": 6538
    },
    {
      "epoch": 0.9164681149264191,
      "grad_norm": 0.570254921913147,
      "learning_rate": 0.00021340827553216932,
      "loss": 0.0449,
      "step": 6539
    },
    {
      "epoch": 0.9166082690960056,
      "grad_norm": 0.9843643307685852,
      "learning_rate": 0.00021339392489834965,
      "loss": 0.221,
      "step": 6540
    },
    {
      "epoch": 0.9167484232655921,
      "grad_norm": 0.17277079820632935,
      "learning_rate": 0.00021337957426452998,
      "loss": 0.0217,
      "step": 6541
    },
    {
      "epoch": 0.9168885774351787,
      "grad_norm": 0.7701627612113953,
      "learning_rate": 0.00021336522363071036,
      "loss": 0.0457,
      "step": 6542
    },
    {
      "epoch": 0.9170287316047653,
      "grad_norm": 1.2003687620162964,
      "learning_rate": 0.0002133508729968907,
      "loss": 0.078,
      "step": 6543
    },
    {
      "epoch": 0.9171688857743517,
      "grad_norm": 3.0578112602233887,
      "learning_rate": 0.00021333652236307102,
      "loss": 0.2153,
      "step": 6544
    },
    {
      "epoch": 0.9173090399439383,
      "grad_norm": 2.0003116130828857,
      "learning_rate": 0.00021332217172925137,
      "loss": 0.1878,
      "step": 6545
    },
    {
      "epoch": 0.9174491941135249,
      "grad_norm": 0.39463701844215393,
      "learning_rate": 0.0002133078210954317,
      "loss": 0.1965,
      "step": 6546
    },
    {
      "epoch": 0.9175893482831115,
      "grad_norm": 4.3486104011535645,
      "learning_rate": 0.00021329347046161203,
      "loss": 0.0856,
      "step": 6547
    },
    {
      "epoch": 0.917729502452698,
      "grad_norm": 0.974236249923706,
      "learning_rate": 0.00021327911982779238,
      "loss": 0.2314,
      "step": 6548
    },
    {
      "epoch": 0.9178696566222845,
      "grad_norm": 1.3744697570800781,
      "learning_rate": 0.0002132647691939727,
      "loss": 0.2313,
      "step": 6549
    },
    {
      "epoch": 0.918009810791871,
      "grad_norm": 0.6843933463096619,
      "learning_rate": 0.00021325041856015304,
      "loss": 0.0861,
      "step": 6550
    },
    {
      "epoch": 0.9181499649614576,
      "grad_norm": 0.4104370176792145,
      "learning_rate": 0.0002132360679263334,
      "loss": 0.1074,
      "step": 6551
    },
    {
      "epoch": 0.9182901191310442,
      "grad_norm": 0.2638469636440277,
      "learning_rate": 0.00021322171729251373,
      "loss": 0.0888,
      "step": 6552
    },
    {
      "epoch": 0.9184302733006307,
      "grad_norm": 0.3258189260959625,
      "learning_rate": 0.00021320736665869405,
      "loss": 0.0733,
      "step": 6553
    },
    {
      "epoch": 0.9185704274702172,
      "grad_norm": 0.2016063779592514,
      "learning_rate": 0.00021319301602487438,
      "loss": 0.0333,
      "step": 6554
    },
    {
      "epoch": 0.9187105816398038,
      "grad_norm": 0.276944100856781,
      "learning_rate": 0.00021317866539105477,
      "loss": 0.0857,
      "step": 6555
    },
    {
      "epoch": 0.9188507358093904,
      "grad_norm": 0.4742622673511505,
      "learning_rate": 0.0002131643147572351,
      "loss": 0.1611,
      "step": 6556
    },
    {
      "epoch": 0.9189908899789768,
      "grad_norm": 0.433238685131073,
      "learning_rate": 0.00021314996412341542,
      "loss": 0.1017,
      "step": 6557
    },
    {
      "epoch": 0.9191310441485634,
      "grad_norm": 0.362604558467865,
      "learning_rate": 0.00021313561348959578,
      "loss": 0.057,
      "step": 6558
    },
    {
      "epoch": 0.91927119831815,
      "grad_norm": 0.7808148264884949,
      "learning_rate": 0.0002131212628557761,
      "loss": 0.1487,
      "step": 6559
    },
    {
      "epoch": 0.9194113524877365,
      "grad_norm": 0.33569496870040894,
      "learning_rate": 0.00021310691222195644,
      "loss": 0.0945,
      "step": 6560
    },
    {
      "epoch": 0.9195515066573231,
      "grad_norm": 0.6150591969490051,
      "learning_rate": 0.0002130925615881368,
      "loss": 0.1404,
      "step": 6561
    },
    {
      "epoch": 0.9196916608269096,
      "grad_norm": 0.36744436621665955,
      "learning_rate": 0.00021307821095431712,
      "loss": 0.0664,
      "step": 6562
    },
    {
      "epoch": 0.9198318149964961,
      "grad_norm": 0.7724167704582214,
      "learning_rate": 0.00021306386032049745,
      "loss": 0.137,
      "step": 6563
    },
    {
      "epoch": 0.9199719691660827,
      "grad_norm": 0.4401775598526001,
      "learning_rate": 0.00021304950968667783,
      "loss": 0.0615,
      "step": 6564
    },
    {
      "epoch": 0.9201121233356693,
      "grad_norm": 0.3308345675468445,
      "learning_rate": 0.00021303515905285816,
      "loss": 0.0381,
      "step": 6565
    },
    {
      "epoch": 0.9202522775052557,
      "grad_norm": 0.6252098679542542,
      "learning_rate": 0.0002130208084190385,
      "loss": 0.0447,
      "step": 6566
    },
    {
      "epoch": 0.9203924316748423,
      "grad_norm": 0.22807785868644714,
      "learning_rate": 0.00021300645778521884,
      "loss": 0.0379,
      "step": 6567
    },
    {
      "epoch": 0.9205325858444289,
      "grad_norm": 0.8972699642181396,
      "learning_rate": 0.00021299210715139917,
      "loss": 0.0766,
      "step": 6568
    },
    {
      "epoch": 0.9206727400140154,
      "grad_norm": 0.26445111632347107,
      "learning_rate": 0.0002129777565175795,
      "loss": 0.0681,
      "step": 6569
    },
    {
      "epoch": 0.920812894183602,
      "grad_norm": 0.28286445140838623,
      "learning_rate": 0.00021296340588375986,
      "loss": 0.0585,
      "step": 6570
    },
    {
      "epoch": 0.9209530483531885,
      "grad_norm": 0.583223283290863,
      "learning_rate": 0.00021294905524994019,
      "loss": 0.0862,
      "step": 6571
    },
    {
      "epoch": 0.921093202522775,
      "grad_norm": 0.14293135702610016,
      "learning_rate": 0.00021293470461612051,
      "loss": 0.0293,
      "step": 6572
    },
    {
      "epoch": 0.9212333566923616,
      "grad_norm": 0.4340853989124298,
      "learning_rate": 0.00021292035398230084,
      "loss": 0.0415,
      "step": 6573
    },
    {
      "epoch": 0.9213735108619482,
      "grad_norm": 0.6256099343299866,
      "learning_rate": 0.00021290600334848123,
      "loss": 0.0629,
      "step": 6574
    },
    {
      "epoch": 0.9215136650315346,
      "grad_norm": 0.321030855178833,
      "learning_rate": 0.00021289165271466155,
      "loss": 0.0604,
      "step": 6575
    },
    {
      "epoch": 0.9216538192011212,
      "grad_norm": 0.34738689661026,
      "learning_rate": 0.00021287730208084188,
      "loss": 0.052,
      "step": 6576
    },
    {
      "epoch": 0.9217939733707078,
      "grad_norm": 0.3290344774723053,
      "learning_rate": 0.00021286295144702224,
      "loss": 0.0591,
      "step": 6577
    },
    {
      "epoch": 0.9219341275402944,
      "grad_norm": 0.18225766718387604,
      "learning_rate": 0.00021284860081320257,
      "loss": 0.0362,
      "step": 6578
    },
    {
      "epoch": 0.9220742817098808,
      "grad_norm": 0.2788483500480652,
      "learning_rate": 0.0002128342501793829,
      "loss": 0.0467,
      "step": 6579
    },
    {
      "epoch": 0.9222144358794674,
      "grad_norm": 0.6952465772628784,
      "learning_rate": 0.00021281989954556325,
      "loss": 0.1207,
      "step": 6580
    },
    {
      "epoch": 0.922354590049054,
      "grad_norm": 0.6459895968437195,
      "learning_rate": 0.00021280554891174358,
      "loss": 0.1193,
      "step": 6581
    },
    {
      "epoch": 0.9224947442186405,
      "grad_norm": 0.3681592345237732,
      "learning_rate": 0.0002127911982779239,
      "loss": 0.0461,
      "step": 6582
    },
    {
      "epoch": 0.9226348983882271,
      "grad_norm": 0.5143775939941406,
      "learning_rate": 0.00021277684764410426,
      "loss": 0.1356,
      "step": 6583
    },
    {
      "epoch": 0.9227750525578136,
      "grad_norm": 0.4886704683303833,
      "learning_rate": 0.0002127624970102846,
      "loss": 0.0705,
      "step": 6584
    },
    {
      "epoch": 0.9229152067274001,
      "grad_norm": 0.4671679139137268,
      "learning_rate": 0.00021274814637646492,
      "loss": 0.1072,
      "step": 6585
    },
    {
      "epoch": 0.9230553608969867,
      "grad_norm": 0.2895188331604004,
      "learning_rate": 0.0002127337957426453,
      "loss": 0.0279,
      "step": 6586
    },
    {
      "epoch": 0.9231955150665733,
      "grad_norm": 0.548511803150177,
      "learning_rate": 0.00021271944510882563,
      "loss": 0.0762,
      "step": 6587
    },
    {
      "epoch": 0.9233356692361597,
      "grad_norm": 0.20739451050758362,
      "learning_rate": 0.00021270509447500596,
      "loss": 0.0207,
      "step": 6588
    },
    {
      "epoch": 0.9234758234057463,
      "grad_norm": 0.6376757025718689,
      "learning_rate": 0.0002126907438411863,
      "loss": 0.1053,
      "step": 6589
    },
    {
      "epoch": 0.9236159775753329,
      "grad_norm": 0.6974294185638428,
      "learning_rate": 0.00021267639320736665,
      "loss": 0.1186,
      "step": 6590
    },
    {
      "epoch": 0.9237561317449194,
      "grad_norm": 0.18099217116832733,
      "learning_rate": 0.00021266204257354697,
      "loss": 0.0186,
      "step": 6591
    },
    {
      "epoch": 0.923896285914506,
      "grad_norm": 0.6088647842407227,
      "learning_rate": 0.0002126476919397273,
      "loss": 0.1091,
      "step": 6592
    },
    {
      "epoch": 0.9240364400840925,
      "grad_norm": 0.3992898762226105,
      "learning_rate": 0.00021263334130590766,
      "loss": 0.1839,
      "step": 6593
    },
    {
      "epoch": 0.924176594253679,
      "grad_norm": 0.34131160378456116,
      "learning_rate": 0.000212618990672088,
      "loss": 0.0599,
      "step": 6594
    },
    {
      "epoch": 0.9243167484232656,
      "grad_norm": 0.38721826672554016,
      "learning_rate": 0.00021260464003826832,
      "loss": 0.0714,
      "step": 6595
    },
    {
      "epoch": 0.9244569025928522,
      "grad_norm": 0.04819757118821144,
      "learning_rate": 0.0002125902894044487,
      "loss": 0.0065,
      "step": 6596
    },
    {
      "epoch": 0.9245970567624386,
      "grad_norm": 0.573275625705719,
      "learning_rate": 0.00021257593877062903,
      "loss": 0.0939,
      "step": 6597
    },
    {
      "epoch": 0.9247372109320252,
      "grad_norm": 1.7822608947753906,
      "learning_rate": 0.00021256158813680935,
      "loss": 0.4323,
      "step": 6598
    },
    {
      "epoch": 0.9248773651016118,
      "grad_norm": 1.5164729356765747,
      "learning_rate": 0.0002125472375029897,
      "loss": 0.153,
      "step": 6599
    },
    {
      "epoch": 0.9250175192711984,
      "grad_norm": 2.6761419773101807,
      "learning_rate": 0.00021253288686917004,
      "loss": 0.3312,
      "step": 6600
    },
    {
      "epoch": 0.9251576734407848,
      "grad_norm": 0.35326775908470154,
      "learning_rate": 0.00021251853623535037,
      "loss": 0.1139,
      "step": 6601
    },
    {
      "epoch": 0.9252978276103714,
      "grad_norm": 0.23541373014450073,
      "learning_rate": 0.00021250418560153072,
      "loss": 0.041,
      "step": 6602
    },
    {
      "epoch": 0.925437981779958,
      "grad_norm": 0.15127332508563995,
      "learning_rate": 0.00021248983496771105,
      "loss": 0.0242,
      "step": 6603
    },
    {
      "epoch": 0.9255781359495445,
      "grad_norm": 0.8747516870498657,
      "learning_rate": 0.00021247548433389138,
      "loss": 0.0661,
      "step": 6604
    },
    {
      "epoch": 0.9257182901191311,
      "grad_norm": 0.9469834566116333,
      "learning_rate": 0.00021246113370007176,
      "loss": 0.1381,
      "step": 6605
    },
    {
      "epoch": 0.9258584442887176,
      "grad_norm": 0.37461382150650024,
      "learning_rate": 0.0002124467830662521,
      "loss": 0.0961,
      "step": 6606
    },
    {
      "epoch": 0.9259985984583041,
      "grad_norm": 0.337449848651886,
      "learning_rate": 0.00021243243243243242,
      "loss": 0.0723,
      "step": 6607
    },
    {
      "epoch": 0.9261387526278907,
      "grad_norm": 0.6606259346008301,
      "learning_rate": 0.00021241808179861275,
      "loss": 0.1,
      "step": 6608
    },
    {
      "epoch": 0.9262789067974773,
      "grad_norm": 0.47763779759407043,
      "learning_rate": 0.0002124037311647931,
      "loss": 0.0817,
      "step": 6609
    },
    {
      "epoch": 0.9264190609670637,
      "grad_norm": 0.15813365578651428,
      "learning_rate": 0.00021238938053097343,
      "loss": 0.0211,
      "step": 6610
    },
    {
      "epoch": 0.9265592151366503,
      "grad_norm": 0.2945040464401245,
      "learning_rate": 0.00021237502989715376,
      "loss": 0.1142,
      "step": 6611
    },
    {
      "epoch": 0.9266993693062369,
      "grad_norm": 0.3160688579082489,
      "learning_rate": 0.00021236067926333412,
      "loss": 0.0596,
      "step": 6612
    },
    {
      "epoch": 0.9268395234758234,
      "grad_norm": 0.4523538649082184,
      "learning_rate": 0.00021234632862951445,
      "loss": 0.0591,
      "step": 6613
    },
    {
      "epoch": 0.92697967764541,
      "grad_norm": 0.3221744894981384,
      "learning_rate": 0.00021233197799569477,
      "loss": 0.0918,
      "step": 6614
    },
    {
      "epoch": 0.9271198318149965,
      "grad_norm": 0.8121134638786316,
      "learning_rate": 0.00021231762736187513,
      "loss": 0.1557,
      "step": 6615
    },
    {
      "epoch": 0.927259985984583,
      "grad_norm": 0.7170042395591736,
      "learning_rate": 0.00021230327672805546,
      "loss": 0.0939,
      "step": 6616
    },
    {
      "epoch": 0.9274001401541696,
      "grad_norm": 0.4265516698360443,
      "learning_rate": 0.0002122889260942358,
      "loss": 0.0566,
      "step": 6617
    },
    {
      "epoch": 0.9275402943237562,
      "grad_norm": 0.5969727039337158,
      "learning_rate": 0.00021227457546041617,
      "loss": 0.0828,
      "step": 6618
    },
    {
      "epoch": 0.9276804484933426,
      "grad_norm": 0.4594188928604126,
      "learning_rate": 0.0002122602248265965,
      "loss": 0.0586,
      "step": 6619
    },
    {
      "epoch": 0.9278206026629292,
      "grad_norm": 0.6752777099609375,
      "learning_rate": 0.00021224587419277683,
      "loss": 0.1821,
      "step": 6620
    },
    {
      "epoch": 0.9279607568325158,
      "grad_norm": 0.6409759521484375,
      "learning_rate": 0.00021223152355895718,
      "loss": 0.0458,
      "step": 6621
    },
    {
      "epoch": 0.9281009110021023,
      "grad_norm": 0.48027411103248596,
      "learning_rate": 0.0002122171729251375,
      "loss": 0.0774,
      "step": 6622
    },
    {
      "epoch": 0.9282410651716888,
      "grad_norm": 0.6678305864334106,
      "learning_rate": 0.00021220282229131784,
      "loss": 0.0797,
      "step": 6623
    },
    {
      "epoch": 0.9283812193412754,
      "grad_norm": 0.35657790303230286,
      "learning_rate": 0.00021218847165749817,
      "loss": 0.0676,
      "step": 6624
    },
    {
      "epoch": 0.928521373510862,
      "grad_norm": 0.49849340319633484,
      "learning_rate": 0.00021217412102367852,
      "loss": 0.0825,
      "step": 6625
    },
    {
      "epoch": 0.9286615276804485,
      "grad_norm": 0.48096442222595215,
      "learning_rate": 0.00021215977038985885,
      "loss": 0.0528,
      "step": 6626
    },
    {
      "epoch": 0.9288016818500351,
      "grad_norm": 0.31301823258399963,
      "learning_rate": 0.00021214541975603918,
      "loss": 0.0558,
      "step": 6627
    },
    {
      "epoch": 0.9289418360196215,
      "grad_norm": 0.5326740741729736,
      "learning_rate": 0.00021213106912221956,
      "loss": 0.027,
      "step": 6628
    },
    {
      "epoch": 0.9290819901892081,
      "grad_norm": 0.3684687912464142,
      "learning_rate": 0.0002121167184883999,
      "loss": 0.0578,
      "step": 6629
    },
    {
      "epoch": 0.9292221443587947,
      "grad_norm": 0.7051708102226257,
      "learning_rate": 0.00021210236785458022,
      "loss": 0.1289,
      "step": 6630
    },
    {
      "epoch": 0.9293622985283813,
      "grad_norm": 0.717178225517273,
      "learning_rate": 0.00021208801722076058,
      "loss": 0.1313,
      "step": 6631
    },
    {
      "epoch": 0.9295024526979677,
      "grad_norm": 0.6747783422470093,
      "learning_rate": 0.0002120736665869409,
      "loss": 0.2192,
      "step": 6632
    },
    {
      "epoch": 0.9296426068675543,
      "grad_norm": 0.5534868836402893,
      "learning_rate": 0.00021205931595312123,
      "loss": 0.1306,
      "step": 6633
    },
    {
      "epoch": 0.9297827610371409,
      "grad_norm": 0.6477826833724976,
      "learning_rate": 0.0002120449653193016,
      "loss": 0.1114,
      "step": 6634
    },
    {
      "epoch": 0.9299229152067274,
      "grad_norm": 0.7142448425292969,
      "learning_rate": 0.00021203061468548192,
      "loss": 0.1409,
      "step": 6635
    },
    {
      "epoch": 0.930063069376314,
      "grad_norm": 0.8179980516433716,
      "learning_rate": 0.00021201626405166225,
      "loss": 0.0778,
      "step": 6636
    },
    {
      "epoch": 0.9302032235459005,
      "grad_norm": 0.4205307960510254,
      "learning_rate": 0.00021200191341784263,
      "loss": 0.0485,
      "step": 6637
    },
    {
      "epoch": 0.930343377715487,
      "grad_norm": 0.21393099427223206,
      "learning_rate": 0.00021198756278402296,
      "loss": 0.054,
      "step": 6638
    },
    {
      "epoch": 0.9304835318850736,
      "grad_norm": 0.4255819022655487,
      "learning_rate": 0.0002119732121502033,
      "loss": 0.0441,
      "step": 6639
    },
    {
      "epoch": 0.9306236860546602,
      "grad_norm": 0.4818372130393982,
      "learning_rate": 0.00021195886151638364,
      "loss": 0.058,
      "step": 6640
    },
    {
      "epoch": 0.9307638402242466,
      "grad_norm": 1.068794846534729,
      "learning_rate": 0.00021194451088256397,
      "loss": 0.0808,
      "step": 6641
    },
    {
      "epoch": 0.9309039943938332,
      "grad_norm": 0.49423131346702576,
      "learning_rate": 0.0002119301602487443,
      "loss": 0.2034,
      "step": 6642
    },
    {
      "epoch": 0.9310441485634198,
      "grad_norm": 0.7991312742233276,
      "learning_rate": 0.00021191580961492463,
      "loss": 0.2301,
      "step": 6643
    },
    {
      "epoch": 0.9311843027330063,
      "grad_norm": 0.8272519707679749,
      "learning_rate": 0.00021190145898110498,
      "loss": 0.0626,
      "step": 6644
    },
    {
      "epoch": 0.9313244569025928,
      "grad_norm": 0.7400698661804199,
      "learning_rate": 0.0002118871083472853,
      "loss": 0.1193,
      "step": 6645
    },
    {
      "epoch": 0.9314646110721794,
      "grad_norm": 1.1985710859298706,
      "learning_rate": 0.00021187275771346564,
      "loss": 0.1766,
      "step": 6646
    },
    {
      "epoch": 0.9316047652417659,
      "grad_norm": 0.9758482575416565,
      "learning_rate": 0.000211858407079646,
      "loss": 0.0627,
      "step": 6647
    },
    {
      "epoch": 0.9317449194113525,
      "grad_norm": 1.2363991737365723,
      "learning_rate": 0.00021184405644582633,
      "loss": 0.2157,
      "step": 6648
    },
    {
      "epoch": 0.9318850735809391,
      "grad_norm": 2.4562606811523438,
      "learning_rate": 0.00021182970581200665,
      "loss": 0.0921,
      "step": 6649
    },
    {
      "epoch": 0.9320252277505255,
      "grad_norm": 1.0566496849060059,
      "learning_rate": 0.00021181535517818704,
      "loss": 0.0508,
      "step": 6650
    },
    {
      "epoch": 0.9321653819201121,
      "grad_norm": 0.31610044836997986,
      "learning_rate": 0.00021180100454436736,
      "loss": 0.0617,
      "step": 6651
    },
    {
      "epoch": 0.9323055360896987,
      "grad_norm": 0.5460981726646423,
      "learning_rate": 0.0002117866539105477,
      "loss": 0.0563,
      "step": 6652
    },
    {
      "epoch": 0.9324456902592853,
      "grad_norm": 0.4073355197906494,
      "learning_rate": 0.00021177230327672805,
      "loss": 0.1002,
      "step": 6653
    },
    {
      "epoch": 0.9325858444288717,
      "grad_norm": 0.26478004455566406,
      "learning_rate": 0.00021175795264290838,
      "loss": 0.072,
      "step": 6654
    },
    {
      "epoch": 0.9327259985984583,
      "grad_norm": 0.24905385076999664,
      "learning_rate": 0.0002117436020090887,
      "loss": 0.0332,
      "step": 6655
    },
    {
      "epoch": 0.9328661527680449,
      "grad_norm": 0.5655146837234497,
      "learning_rate": 0.00021172925137526906,
      "loss": 0.0858,
      "step": 6656
    },
    {
      "epoch": 0.9330063069376314,
      "grad_norm": 0.2765238583087921,
      "learning_rate": 0.0002117149007414494,
      "loss": 0.0658,
      "step": 6657
    },
    {
      "epoch": 0.933146461107218,
      "grad_norm": 0.4552586078643799,
      "learning_rate": 0.00021170055010762972,
      "loss": 0.1167,
      "step": 6658
    },
    {
      "epoch": 0.9332866152768045,
      "grad_norm": 0.26092761754989624,
      "learning_rate": 0.00021168619947381005,
      "loss": 0.0586,
      "step": 6659
    },
    {
      "epoch": 0.933426769446391,
      "grad_norm": 0.15380969643592834,
      "learning_rate": 0.00021167184883999043,
      "loss": 0.0486,
      "step": 6660
    },
    {
      "epoch": 0.9335669236159776,
      "grad_norm": 0.5019110441207886,
      "learning_rate": 0.00021165749820617076,
      "loss": 0.0773,
      "step": 6661
    },
    {
      "epoch": 0.9337070777855642,
      "grad_norm": 0.36999794840812683,
      "learning_rate": 0.0002116431475723511,
      "loss": 0.0727,
      "step": 6662
    },
    {
      "epoch": 0.9338472319551506,
      "grad_norm": 0.3883402943611145,
      "learning_rate": 0.00021162879693853144,
      "loss": 0.1147,
      "step": 6663
    },
    {
      "epoch": 0.9339873861247372,
      "grad_norm": 0.3678326904773712,
      "learning_rate": 0.00021161444630471177,
      "loss": 0.0507,
      "step": 6664
    },
    {
      "epoch": 0.9341275402943238,
      "grad_norm": 0.2711755633354187,
      "learning_rate": 0.0002116000956708921,
      "loss": 0.0363,
      "step": 6665
    },
    {
      "epoch": 0.9342676944639103,
      "grad_norm": 0.32458898425102234,
      "learning_rate": 0.00021158574503707246,
      "loss": 0.0271,
      "step": 6666
    },
    {
      "epoch": 0.9344078486334968,
      "grad_norm": 0.11644943803548813,
      "learning_rate": 0.00021157139440325278,
      "loss": 0.0227,
      "step": 6667
    },
    {
      "epoch": 0.9345480028030834,
      "grad_norm": 0.5873533487319946,
      "learning_rate": 0.0002115570437694331,
      "loss": 0.0803,
      "step": 6668
    },
    {
      "epoch": 0.9346881569726699,
      "grad_norm": 0.3252265751361847,
      "learning_rate": 0.0002115426931356135,
      "loss": 0.0817,
      "step": 6669
    },
    {
      "epoch": 0.9348283111422565,
      "grad_norm": 0.8226193785667419,
      "learning_rate": 0.00021152834250179382,
      "loss": 0.1717,
      "step": 6670
    },
    {
      "epoch": 0.9349684653118431,
      "grad_norm": 0.15587258338928223,
      "learning_rate": 0.00021151399186797415,
      "loss": 0.0202,
      "step": 6671
    },
    {
      "epoch": 0.9351086194814295,
      "grad_norm": 0.46227923035621643,
      "learning_rate": 0.0002114996412341545,
      "loss": 0.1086,
      "step": 6672
    },
    {
      "epoch": 0.9352487736510161,
      "grad_norm": 0.43766385316848755,
      "learning_rate": 0.00021148529060033484,
      "loss": 0.0928,
      "step": 6673
    },
    {
      "epoch": 0.9353889278206027,
      "grad_norm": 0.5527503490447998,
      "learning_rate": 0.00021147093996651517,
      "loss": 0.124,
      "step": 6674
    },
    {
      "epoch": 0.9355290819901892,
      "grad_norm": 0.3553295135498047,
      "learning_rate": 0.00021145658933269552,
      "loss": 0.0416,
      "step": 6675
    },
    {
      "epoch": 0.9356692361597757,
      "grad_norm": 0.4060499966144562,
      "learning_rate": 0.00021144223869887585,
      "loss": 0.0853,
      "step": 6676
    },
    {
      "epoch": 0.9358093903293623,
      "grad_norm": 0.5635578036308289,
      "learning_rate": 0.00021142788806505618,
      "loss": 0.1293,
      "step": 6677
    },
    {
      "epoch": 0.9359495444989488,
      "grad_norm": 0.4267987310886383,
      "learning_rate": 0.0002114135374312365,
      "loss": 0.0862,
      "step": 6678
    },
    {
      "epoch": 0.9360896986685354,
      "grad_norm": 0.26879268884658813,
      "learning_rate": 0.00021139918679741686,
      "loss": 0.0299,
      "step": 6679
    },
    {
      "epoch": 0.9362298528381219,
      "grad_norm": 0.13448967039585114,
      "learning_rate": 0.0002113848361635972,
      "loss": 0.0116,
      "step": 6680
    },
    {
      "epoch": 0.9363700070077084,
      "grad_norm": 1.026279091835022,
      "learning_rate": 0.00021137048552977755,
      "loss": 0.1186,
      "step": 6681
    },
    {
      "epoch": 0.936510161177295,
      "grad_norm": 0.5111008286476135,
      "learning_rate": 0.0002113561348959579,
      "loss": 0.0694,
      "step": 6682
    },
    {
      "epoch": 0.9366503153468816,
      "grad_norm": 0.4464820921421051,
      "learning_rate": 0.00021134178426213823,
      "loss": 0.0767,
      "step": 6683
    },
    {
      "epoch": 0.9367904695164682,
      "grad_norm": 1.2222315073013306,
      "learning_rate": 0.00021132743362831856,
      "loss": 0.0854,
      "step": 6684
    },
    {
      "epoch": 0.9369306236860546,
      "grad_norm": 0.3423914313316345,
      "learning_rate": 0.00021131308299449892,
      "loss": 0.0517,
      "step": 6685
    },
    {
      "epoch": 0.9370707778556412,
      "grad_norm": 0.16706036031246185,
      "learning_rate": 0.00021129873236067924,
      "loss": 0.0438,
      "step": 6686
    },
    {
      "epoch": 0.9372109320252278,
      "grad_norm": 0.23028790950775146,
      "learning_rate": 0.00021128438172685957,
      "loss": 0.0841,
      "step": 6687
    },
    {
      "epoch": 0.9373510861948143,
      "grad_norm": 0.225498229265213,
      "learning_rate": 0.00021127003109303993,
      "loss": 0.0372,
      "step": 6688
    },
    {
      "epoch": 0.9374912403644008,
      "grad_norm": 0.2766595184803009,
      "learning_rate": 0.00021125568045922026,
      "loss": 0.0226,
      "step": 6689
    },
    {
      "epoch": 0.9376313945339874,
      "grad_norm": 0.3983423709869385,
      "learning_rate": 0.00021124132982540059,
      "loss": 0.0321,
      "step": 6690
    },
    {
      "epoch": 0.9377715487035739,
      "grad_norm": 0.5107288956642151,
      "learning_rate": 0.00021122697919158097,
      "loss": 0.1206,
      "step": 6691
    },
    {
      "epoch": 0.9379117028731605,
      "grad_norm": 0.28537335991859436,
      "learning_rate": 0.0002112126285577613,
      "loss": 0.0406,
      "step": 6692
    },
    {
      "epoch": 0.9380518570427471,
      "grad_norm": 1.2690504789352417,
      "learning_rate": 0.00021119827792394163,
      "loss": 0.2139,
      "step": 6693
    },
    {
      "epoch": 0.9381920112123335,
      "grad_norm": 0.6160872578620911,
      "learning_rate": 0.00021118392729012195,
      "loss": 0.0974,
      "step": 6694
    },
    {
      "epoch": 0.9383321653819201,
      "grad_norm": 0.330723375082016,
      "learning_rate": 0.0002111695766563023,
      "loss": 0.0855,
      "step": 6695
    },
    {
      "epoch": 0.9384723195515067,
      "grad_norm": 0.7720799446105957,
      "learning_rate": 0.00021115522602248264,
      "loss": 0.0961,
      "step": 6696
    },
    {
      "epoch": 0.9386124737210932,
      "grad_norm": 0.29477834701538086,
      "learning_rate": 0.00021114087538866297,
      "loss": 0.0714,
      "step": 6697
    },
    {
      "epoch": 0.9387526278906797,
      "grad_norm": 0.24162638187408447,
      "learning_rate": 0.00021112652475484332,
      "loss": 0.0244,
      "step": 6698
    },
    {
      "epoch": 0.9388927820602663,
      "grad_norm": 2.0211286544799805,
      "learning_rate": 0.00021111217412102365,
      "loss": 0.0739,
      "step": 6699
    },
    {
      "epoch": 0.9390329362298528,
      "grad_norm": 1.0248448848724365,
      "learning_rate": 0.00021109782348720398,
      "loss": 0.1163,
      "step": 6700
    },
    {
      "epoch": 0.9391730903994394,
      "grad_norm": 1.0221667289733887,
      "learning_rate": 0.00021108347285338436,
      "loss": 0.1436,
      "step": 6701
    },
    {
      "epoch": 0.9393132445690259,
      "grad_norm": 0.3628167510032654,
      "learning_rate": 0.0002110691222195647,
      "loss": 0.1063,
      "step": 6702
    },
    {
      "epoch": 0.9394533987386124,
      "grad_norm": 0.33413881063461304,
      "learning_rate": 0.00021105477158574502,
      "loss": 0.0538,
      "step": 6703
    },
    {
      "epoch": 0.939593552908199,
      "grad_norm": 0.48904046416282654,
      "learning_rate": 0.00021104042095192537,
      "loss": 0.1099,
      "step": 6704
    },
    {
      "epoch": 0.9397337070777856,
      "grad_norm": 0.4129713773727417,
      "learning_rate": 0.0002110260703181057,
      "loss": 0.0506,
      "step": 6705
    },
    {
      "epoch": 0.9398738612473722,
      "grad_norm": 0.48088040947914124,
      "learning_rate": 0.00021101171968428603,
      "loss": 0.0319,
      "step": 6706
    },
    {
      "epoch": 0.9400140154169586,
      "grad_norm": 0.195185124874115,
      "learning_rate": 0.0002109973690504664,
      "loss": 0.0484,
      "step": 6707
    },
    {
      "epoch": 0.9401541695865452,
      "grad_norm": 0.25378546118736267,
      "learning_rate": 0.00021098301841664672,
      "loss": 0.0933,
      "step": 6708
    },
    {
      "epoch": 0.9402943237561318,
      "grad_norm": 0.48153361678123474,
      "learning_rate": 0.00021096866778282704,
      "loss": 0.0718,
      "step": 6709
    },
    {
      "epoch": 0.9404344779257183,
      "grad_norm": 0.6445961594581604,
      "learning_rate": 0.0002109543171490074,
      "loss": 0.0807,
      "step": 6710
    },
    {
      "epoch": 0.9405746320953048,
      "grad_norm": 0.2878335416316986,
      "learning_rate": 0.00021093996651518773,
      "loss": 0.0604,
      "step": 6711
    },
    {
      "epoch": 0.9407147862648914,
      "grad_norm": 0.344350129365921,
      "learning_rate": 0.00021092561588136806,
      "loss": 0.1058,
      "step": 6712
    },
    {
      "epoch": 0.9408549404344779,
      "grad_norm": 0.30809956789016724,
      "learning_rate": 0.0002109112652475484,
      "loss": 0.0373,
      "step": 6713
    },
    {
      "epoch": 0.9409950946040645,
      "grad_norm": 0.5173668265342712,
      "learning_rate": 0.00021089691461372877,
      "loss": 0.0896,
      "step": 6714
    },
    {
      "epoch": 0.9411352487736511,
      "grad_norm": 0.22702008485794067,
      "learning_rate": 0.0002108825639799091,
      "loss": 0.0514,
      "step": 6715
    },
    {
      "epoch": 0.9412754029432375,
      "grad_norm": 0.20474450290203094,
      "learning_rate": 0.00021086821334608943,
      "loss": 0.0271,
      "step": 6716
    },
    {
      "epoch": 0.9414155571128241,
      "grad_norm": 0.21513846516609192,
      "learning_rate": 0.00021085386271226978,
      "loss": 0.0414,
      "step": 6717
    },
    {
      "epoch": 0.9415557112824107,
      "grad_norm": 0.23136316239833832,
      "learning_rate": 0.0002108395120784501,
      "loss": 0.0633,
      "step": 6718
    },
    {
      "epoch": 0.9416958654519972,
      "grad_norm": 0.5780016779899597,
      "learning_rate": 0.00021082516144463044,
      "loss": 0.1144,
      "step": 6719
    },
    {
      "epoch": 0.9418360196215837,
      "grad_norm": 0.407420814037323,
      "learning_rate": 0.0002108108108108108,
      "loss": 0.0837,
      "step": 6720
    },
    {
      "epoch": 0.9419761737911703,
      "grad_norm": 0.3637130558490753,
      "learning_rate": 0.00021079646017699112,
      "loss": 0.0695,
      "step": 6721
    },
    {
      "epoch": 0.9421163279607568,
      "grad_norm": 0.3148209750652313,
      "learning_rate": 0.00021078210954317145,
      "loss": 0.0671,
      "step": 6722
    },
    {
      "epoch": 0.9422564821303434,
      "grad_norm": 0.4187985062599182,
      "learning_rate": 0.00021076775890935183,
      "loss": 0.0698,
      "step": 6723
    },
    {
      "epoch": 0.9423966362999299,
      "grad_norm": 0.3156029284000397,
      "learning_rate": 0.00021075340827553216,
      "loss": 0.0243,
      "step": 6724
    },
    {
      "epoch": 0.9425367904695164,
      "grad_norm": 0.32550543546676636,
      "learning_rate": 0.0002107390576417125,
      "loss": 0.087,
      "step": 6725
    },
    {
      "epoch": 0.942676944639103,
      "grad_norm": 0.4009532034397125,
      "learning_rate": 0.00021072470700789285,
      "loss": 0.0842,
      "step": 6726
    },
    {
      "epoch": 0.9428170988086896,
      "grad_norm": 0.28254517912864685,
      "learning_rate": 0.00021071035637407318,
      "loss": 0.0762,
      "step": 6727
    },
    {
      "epoch": 0.9429572529782762,
      "grad_norm": 0.4632325768470764,
      "learning_rate": 0.0002106960057402535,
      "loss": 0.1208,
      "step": 6728
    },
    {
      "epoch": 0.9430974071478626,
      "grad_norm": 0.21587690711021423,
      "learning_rate": 0.00021068165510643383,
      "loss": 0.0327,
      "step": 6729
    },
    {
      "epoch": 0.9432375613174492,
      "grad_norm": 0.23405754566192627,
      "learning_rate": 0.0002106673044726142,
      "loss": 0.0556,
      "step": 6730
    },
    {
      "epoch": 0.9433777154870358,
      "grad_norm": 0.5658929944038391,
      "learning_rate": 0.00021065295383879452,
      "loss": 0.1037,
      "step": 6731
    },
    {
      "epoch": 0.9435178696566223,
      "grad_norm": 0.3672569692134857,
      "learning_rate": 0.00021063860320497485,
      "loss": 0.0583,
      "step": 6732
    },
    {
      "epoch": 0.9436580238262088,
      "grad_norm": 0.7151066064834595,
      "learning_rate": 0.00021062425257115523,
      "loss": 0.0719,
      "step": 6733
    },
    {
      "epoch": 0.9437981779957954,
      "grad_norm": 0.3362157940864563,
      "learning_rate": 0.00021060990193733556,
      "loss": 0.0322,
      "step": 6734
    },
    {
      "epoch": 0.9439383321653819,
      "grad_norm": 1.0419331789016724,
      "learning_rate": 0.00021059555130351589,
      "loss": 0.1414,
      "step": 6735
    },
    {
      "epoch": 0.9440784863349685,
      "grad_norm": 0.9612257480621338,
      "learning_rate": 0.00021058120066969624,
      "loss": 0.0391,
      "step": 6736
    },
    {
      "epoch": 0.9442186405045551,
      "grad_norm": 0.4866899847984314,
      "learning_rate": 0.00021056685003587657,
      "loss": 0.157,
      "step": 6737
    },
    {
      "epoch": 0.9443587946741415,
      "grad_norm": 0.3909037411212921,
      "learning_rate": 0.0002105524994020569,
      "loss": 0.0657,
      "step": 6738
    },
    {
      "epoch": 0.9444989488437281,
      "grad_norm": 0.4714002311229706,
      "learning_rate": 0.00021053814876823725,
      "loss": 0.0389,
      "step": 6739
    },
    {
      "epoch": 0.9446391030133147,
      "grad_norm": 0.6339796185493469,
      "learning_rate": 0.00021052379813441758,
      "loss": 0.0971,
      "step": 6740
    },
    {
      "epoch": 0.9447792571829012,
      "grad_norm": 0.7312930226325989,
      "learning_rate": 0.0002105094475005979,
      "loss": 0.0693,
      "step": 6741
    },
    {
      "epoch": 0.9449194113524877,
      "grad_norm": 0.5390293598175049,
      "learning_rate": 0.00021049509686677827,
      "loss": 0.0382,
      "step": 6742
    },
    {
      "epoch": 0.9450595655220743,
      "grad_norm": 0.2725268602371216,
      "learning_rate": 0.0002104807462329586,
      "loss": 0.0415,
      "step": 6743
    },
    {
      "epoch": 0.9451997196916608,
      "grad_norm": 0.3685208261013031,
      "learning_rate": 0.00021046639559913892,
      "loss": 0.0488,
      "step": 6744
    },
    {
      "epoch": 0.9453398738612474,
      "grad_norm": 0.45445573329925537,
      "learning_rate": 0.0002104520449653193,
      "loss": 0.1105,
      "step": 6745
    },
    {
      "epoch": 0.9454800280308339,
      "grad_norm": 0.541092038154602,
      "learning_rate": 0.00021043769433149964,
      "loss": 0.0806,
      "step": 6746
    },
    {
      "epoch": 0.9456201822004204,
      "grad_norm": 0.6137456893920898,
      "learning_rate": 0.00021042334369767996,
      "loss": 0.1166,
      "step": 6747
    },
    {
      "epoch": 0.945760336370007,
      "grad_norm": 0.811437726020813,
      "learning_rate": 0.0002104089930638603,
      "loss": 0.1063,
      "step": 6748
    },
    {
      "epoch": 0.9459004905395936,
      "grad_norm": 1.289063572883606,
      "learning_rate": 0.00021039464243004065,
      "loss": 0.252,
      "step": 6749
    },
    {
      "epoch": 0.9460406447091801,
      "grad_norm": 1.4864723682403564,
      "learning_rate": 0.00021038029179622098,
      "loss": 0.2891,
      "step": 6750
    },
    {
      "epoch": 0.9461807988787666,
      "grad_norm": 0.19582898914813995,
      "learning_rate": 0.0002103659411624013,
      "loss": 0.0377,
      "step": 6751
    },
    {
      "epoch": 0.9463209530483532,
      "grad_norm": 0.4129214882850647,
      "learning_rate": 0.00021035159052858166,
      "loss": 0.0297,
      "step": 6752
    },
    {
      "epoch": 0.9464611072179397,
      "grad_norm": 0.39841458201408386,
      "learning_rate": 0.000210337239894762,
      "loss": 0.104,
      "step": 6753
    },
    {
      "epoch": 0.9466012613875263,
      "grad_norm": 0.5760337114334106,
      "learning_rate": 0.00021032288926094232,
      "loss": 0.1442,
      "step": 6754
    },
    {
      "epoch": 0.9467414155571128,
      "grad_norm": 0.2270519733428955,
      "learning_rate": 0.0002103085386271227,
      "loss": 0.0937,
      "step": 6755
    },
    {
      "epoch": 0.9468815697266993,
      "grad_norm": 0.4276168942451477,
      "learning_rate": 0.00021029418799330303,
      "loss": 0.0626,
      "step": 6756
    },
    {
      "epoch": 0.9470217238962859,
      "grad_norm": 0.44169506430625916,
      "learning_rate": 0.00021027983735948336,
      "loss": 0.1065,
      "step": 6757
    },
    {
      "epoch": 0.9471618780658725,
      "grad_norm": 0.4154358208179474,
      "learning_rate": 0.0002102654867256637,
      "loss": 0.0761,
      "step": 6758
    },
    {
      "epoch": 0.9473020322354591,
      "grad_norm": 0.3619321584701538,
      "learning_rate": 0.00021025113609184404,
      "loss": 0.0821,
      "step": 6759
    },
    {
      "epoch": 0.9474421864050455,
      "grad_norm": 0.36187854409217834,
      "learning_rate": 0.00021023678545802437,
      "loss": 0.0576,
      "step": 6760
    },
    {
      "epoch": 0.9475823405746321,
      "grad_norm": 0.1873820424079895,
      "learning_rate": 0.00021022243482420473,
      "loss": 0.0428,
      "step": 6761
    },
    {
      "epoch": 0.9477224947442187,
      "grad_norm": 0.3125673234462738,
      "learning_rate": 0.00021020808419038505,
      "loss": 0.0613,
      "step": 6762
    },
    {
      "epoch": 0.9478626489138052,
      "grad_norm": 0.32027217745780945,
      "learning_rate": 0.00021019373355656538,
      "loss": 0.0556,
      "step": 6763
    },
    {
      "epoch": 0.9480028030833917,
      "grad_norm": 0.20015032589435577,
      "learning_rate": 0.0002101793829227457,
      "loss": 0.0572,
      "step": 6764
    },
    {
      "epoch": 0.9481429572529783,
      "grad_norm": 0.19977861642837524,
      "learning_rate": 0.0002101650322889261,
      "loss": 0.0312,
      "step": 6765
    },
    {
      "epoch": 0.9482831114225648,
      "grad_norm": 0.4260909855365753,
      "learning_rate": 0.00021015068165510642,
      "loss": 0.1323,
      "step": 6766
    },
    {
      "epoch": 0.9484232655921514,
      "grad_norm": 0.2937268316745758,
      "learning_rate": 0.00021013633102128675,
      "loss": 0.0237,
      "step": 6767
    },
    {
      "epoch": 0.9485634197617379,
      "grad_norm": 0.34284707903862,
      "learning_rate": 0.0002101219803874671,
      "loss": 0.0858,
      "step": 6768
    },
    {
      "epoch": 0.9487035739313244,
      "grad_norm": 0.29405221343040466,
      "learning_rate": 0.00021010762975364744,
      "loss": 0.0968,
      "step": 6769
    },
    {
      "epoch": 0.948843728100911,
      "grad_norm": 0.343694806098938,
      "learning_rate": 0.00021009327911982776,
      "loss": 0.0822,
      "step": 6770
    },
    {
      "epoch": 0.9489838822704976,
      "grad_norm": 0.5976319909095764,
      "learning_rate": 0.00021007892848600812,
      "loss": 0.1108,
      "step": 6771
    },
    {
      "epoch": 0.9491240364400841,
      "grad_norm": 0.6340566873550415,
      "learning_rate": 0.00021006457785218845,
      "loss": 0.1635,
      "step": 6772
    },
    {
      "epoch": 0.9492641906096706,
      "grad_norm": 0.3809947669506073,
      "learning_rate": 0.00021005022721836878,
      "loss": 0.0672,
      "step": 6773
    },
    {
      "epoch": 0.9494043447792572,
      "grad_norm": 0.315130352973938,
      "learning_rate": 0.00021003587658454913,
      "loss": 0.0563,
      "step": 6774
    },
    {
      "epoch": 0.9495444989488437,
      "grad_norm": 0.19299724698066711,
      "learning_rate": 0.00021002152595072946,
      "loss": 0.0453,
      "step": 6775
    },
    {
      "epoch": 0.9496846531184303,
      "grad_norm": 0.536948025226593,
      "learning_rate": 0.0002100071753169098,
      "loss": 0.1265,
      "step": 6776
    },
    {
      "epoch": 0.9498248072880168,
      "grad_norm": 0.17534567415714264,
      "learning_rate": 0.00020999282468309017,
      "loss": 0.0404,
      "step": 6777
    },
    {
      "epoch": 0.9499649614576033,
      "grad_norm": 0.23549401760101318,
      "learning_rate": 0.0002099784740492705,
      "loss": 0.0315,
      "step": 6778
    },
    {
      "epoch": 0.9501051156271899,
      "grad_norm": 0.3954560160636902,
      "learning_rate": 0.00020996412341545083,
      "loss": 0.0896,
      "step": 6779
    },
    {
      "epoch": 0.9502452697967765,
      "grad_norm": 0.5285711884498596,
      "learning_rate": 0.00020994977278163116,
      "loss": 0.0474,
      "step": 6780
    },
    {
      "epoch": 0.950385423966363,
      "grad_norm": 0.245276540517807,
      "learning_rate": 0.00020993542214781151,
      "loss": 0.0411,
      "step": 6781
    },
    {
      "epoch": 0.9505255781359495,
      "grad_norm": 0.3574492335319519,
      "learning_rate": 0.00020992107151399184,
      "loss": 0.0355,
      "step": 6782
    },
    {
      "epoch": 0.9506657323055361,
      "grad_norm": 0.3773401975631714,
      "learning_rate": 0.00020990672088017217,
      "loss": 0.0869,
      "step": 6783
    },
    {
      "epoch": 0.9508058864751227,
      "grad_norm": 0.3458418548107147,
      "learning_rate": 0.00020989237024635253,
      "loss": 0.0535,
      "step": 6784
    },
    {
      "epoch": 0.9509460406447092,
      "grad_norm": 0.40093284845352173,
      "learning_rate": 0.00020987801961253286,
      "loss": 0.1455,
      "step": 6785
    },
    {
      "epoch": 0.9510861948142957,
      "grad_norm": 0.3073892891407013,
      "learning_rate": 0.00020986366897871318,
      "loss": 0.0501,
      "step": 6786
    },
    {
      "epoch": 0.9512263489838823,
      "grad_norm": 0.2885949909687042,
      "learning_rate": 0.00020984931834489357,
      "loss": 0.0425,
      "step": 6787
    },
    {
      "epoch": 0.9513665031534688,
      "grad_norm": 0.2747536301612854,
      "learning_rate": 0.0002098349677110739,
      "loss": 0.0363,
      "step": 6788
    },
    {
      "epoch": 0.9515066573230554,
      "grad_norm": 0.5417380332946777,
      "learning_rate": 0.00020982061707725422,
      "loss": 0.0995,
      "step": 6789
    },
    {
      "epoch": 0.9516468114926419,
      "grad_norm": 0.3912383019924164,
      "learning_rate": 0.00020980626644343458,
      "loss": 0.0453,
      "step": 6790
    },
    {
      "epoch": 0.9517869656622284,
      "grad_norm": 0.4273107647895813,
      "learning_rate": 0.0002097919158096149,
      "loss": 0.1067,
      "step": 6791
    },
    {
      "epoch": 0.951927119831815,
      "grad_norm": 0.44154760241508484,
      "learning_rate": 0.00020977756517579524,
      "loss": 0.0452,
      "step": 6792
    },
    {
      "epoch": 0.9520672740014016,
      "grad_norm": 0.35051122307777405,
      "learning_rate": 0.0002097632145419756,
      "loss": 0.0438,
      "step": 6793
    },
    {
      "epoch": 0.9522074281709881,
      "grad_norm": 0.7021182775497437,
      "learning_rate": 0.00020974886390815592,
      "loss": 0.2359,
      "step": 6794
    },
    {
      "epoch": 0.9523475823405746,
      "grad_norm": 0.7743340730667114,
      "learning_rate": 0.00020973451327433625,
      "loss": 0.2383,
      "step": 6795
    },
    {
      "epoch": 0.9524877365101612,
      "grad_norm": 0.9278534054756165,
      "learning_rate": 0.00020972016264051663,
      "loss": 0.1161,
      "step": 6796
    },
    {
      "epoch": 0.9526278906797477,
      "grad_norm": 0.9077466130256653,
      "learning_rate": 0.00020970581200669696,
      "loss": 0.0422,
      "step": 6797
    },
    {
      "epoch": 0.9527680448493343,
      "grad_norm": 3.7644705772399902,
      "learning_rate": 0.0002096914613728773,
      "loss": 0.1579,
      "step": 6798
    },
    {
      "epoch": 0.9529081990189208,
      "grad_norm": 0.3081324100494385,
      "learning_rate": 0.00020967711073905762,
      "loss": 0.0181,
      "step": 6799
    },
    {
      "epoch": 0.9530483531885073,
      "grad_norm": 3.044713020324707,
      "learning_rate": 0.00020966276010523797,
      "loss": 0.1534,
      "step": 6800
    },
    {
      "epoch": 0.9531885073580939,
      "grad_norm": 0.3266991674900055,
      "learning_rate": 0.0002096484094714183,
      "loss": 0.057,
      "step": 6801
    },
    {
      "epoch": 0.9533286615276805,
      "grad_norm": 0.32300278544425964,
      "learning_rate": 0.00020963405883759863,
      "loss": 0.0708,
      "step": 6802
    },
    {
      "epoch": 0.953468815697267,
      "grad_norm": 0.26398855447769165,
      "learning_rate": 0.000209619708203779,
      "loss": 0.0441,
      "step": 6803
    },
    {
      "epoch": 0.9536089698668535,
      "grad_norm": 0.6699583530426025,
      "learning_rate": 0.00020960535756995932,
      "loss": 0.1139,
      "step": 6804
    },
    {
      "epoch": 0.9537491240364401,
      "grad_norm": 0.6843666434288025,
      "learning_rate": 0.00020959100693613964,
      "loss": 0.1313,
      "step": 6805
    },
    {
      "epoch": 0.9538892782060266,
      "grad_norm": 0.2818840742111206,
      "learning_rate": 0.00020957665630232,
      "loss": 0.0511,
      "step": 6806
    },
    {
      "epoch": 0.9540294323756132,
      "grad_norm": 0.3320895731449127,
      "learning_rate": 0.00020956230566850033,
      "loss": 0.0384,
      "step": 6807
    },
    {
      "epoch": 0.9541695865451997,
      "grad_norm": 0.413745641708374,
      "learning_rate": 0.00020954795503468068,
      "loss": 0.0849,
      "step": 6808
    },
    {
      "epoch": 0.9543097407147862,
      "grad_norm": 0.1976531744003296,
      "learning_rate": 0.00020953360440086104,
      "loss": 0.0297,
      "step": 6809
    },
    {
      "epoch": 0.9544498948843728,
      "grad_norm": 0.4792803227901459,
      "learning_rate": 0.00020951925376704137,
      "loss": 0.076,
      "step": 6810
    },
    {
      "epoch": 0.9545900490539594,
      "grad_norm": 0.13696984946727753,
      "learning_rate": 0.0002095049031332217,
      "loss": 0.0221,
      "step": 6811
    },
    {
      "epoch": 0.9547302032235458,
      "grad_norm": 0.19744239747524261,
      "learning_rate": 0.00020949055249940205,
      "loss": 0.0071,
      "step": 6812
    },
    {
      "epoch": 0.9548703573931324,
      "grad_norm": 0.2566864788532257,
      "learning_rate": 0.00020947620186558238,
      "loss": 0.0776,
      "step": 6813
    },
    {
      "epoch": 0.955010511562719,
      "grad_norm": 0.3636632561683655,
      "learning_rate": 0.0002094618512317627,
      "loss": 0.0733,
      "step": 6814
    },
    {
      "epoch": 0.9551506657323056,
      "grad_norm": 0.29802748560905457,
      "learning_rate": 0.00020944750059794304,
      "loss": 0.0434,
      "step": 6815
    },
    {
      "epoch": 0.9552908199018921,
      "grad_norm": 0.31809690594673157,
      "learning_rate": 0.0002094331499641234,
      "loss": 0.0355,
      "step": 6816
    },
    {
      "epoch": 0.9554309740714786,
      "grad_norm": 0.3092915415763855,
      "learning_rate": 0.00020941879933030372,
      "loss": 0.0675,
      "step": 6817
    },
    {
      "epoch": 0.9555711282410652,
      "grad_norm": 0.30978840589523315,
      "learning_rate": 0.00020940444869648405,
      "loss": 0.0416,
      "step": 6818
    },
    {
      "epoch": 0.9557112824106517,
      "grad_norm": 0.4784527122974396,
      "learning_rate": 0.00020939009806266443,
      "loss": 0.1491,
      "step": 6819
    },
    {
      "epoch": 0.9558514365802383,
      "grad_norm": 0.5266783237457275,
      "learning_rate": 0.00020937574742884476,
      "loss": 0.0664,
      "step": 6820
    },
    {
      "epoch": 0.9559915907498248,
      "grad_norm": 0.4265827238559723,
      "learning_rate": 0.0002093613967950251,
      "loss": 0.1297,
      "step": 6821
    },
    {
      "epoch": 0.9561317449194113,
      "grad_norm": 0.5543458461761475,
      "learning_rate": 0.00020934704616120545,
      "loss": 0.1197,
      "step": 6822
    },
    {
      "epoch": 0.9562718990889979,
      "grad_norm": 0.56550133228302,
      "learning_rate": 0.00020933269552738577,
      "loss": 0.0813,
      "step": 6823
    },
    {
      "epoch": 0.9564120532585845,
      "grad_norm": 0.3015633821487427,
      "learning_rate": 0.0002093183448935661,
      "loss": 0.0597,
      "step": 6824
    },
    {
      "epoch": 0.956552207428171,
      "grad_norm": 0.5544846057891846,
      "learning_rate": 0.00020930399425974646,
      "loss": 0.1174,
      "step": 6825
    },
    {
      "epoch": 0.9566923615977575,
      "grad_norm": 0.26918500661849976,
      "learning_rate": 0.0002092896436259268,
      "loss": 0.0497,
      "step": 6826
    },
    {
      "epoch": 0.9568325157673441,
      "grad_norm": 0.5477024912834167,
      "learning_rate": 0.00020927529299210712,
      "loss": 0.103,
      "step": 6827
    },
    {
      "epoch": 0.9569726699369306,
      "grad_norm": 0.5423645973205566,
      "learning_rate": 0.0002092609423582875,
      "loss": 0.0621,
      "step": 6828
    },
    {
      "epoch": 0.9571128241065172,
      "grad_norm": 0.2877272665500641,
      "learning_rate": 0.00020924659172446783,
      "loss": 0.0559,
      "step": 6829
    },
    {
      "epoch": 0.9572529782761037,
      "grad_norm": 0.16464126110076904,
      "learning_rate": 0.00020923224109064816,
      "loss": 0.0534,
      "step": 6830
    },
    {
      "epoch": 0.9573931324456902,
      "grad_norm": 0.296781063079834,
      "learning_rate": 0.0002092178904568285,
      "loss": 0.0739,
      "step": 6831
    },
    {
      "epoch": 0.9575332866152768,
      "grad_norm": 0.3618508279323578,
      "learning_rate": 0.00020920353982300884,
      "loss": 0.0714,
      "step": 6832
    },
    {
      "epoch": 0.9576734407848634,
      "grad_norm": 0.3697979152202606,
      "learning_rate": 0.00020918918918918917,
      "loss": 0.0571,
      "step": 6833
    },
    {
      "epoch": 0.9578135949544498,
      "grad_norm": 0.5758258700370789,
      "learning_rate": 0.0002091748385553695,
      "loss": 0.0572,
      "step": 6834
    },
    {
      "epoch": 0.9579537491240364,
      "grad_norm": 0.2583303451538086,
      "learning_rate": 0.00020916048792154985,
      "loss": 0.0311,
      "step": 6835
    },
    {
      "epoch": 0.958093903293623,
      "grad_norm": 0.24555346369743347,
      "learning_rate": 0.00020914613728773018,
      "loss": 0.0452,
      "step": 6836
    },
    {
      "epoch": 0.9582340574632096,
      "grad_norm": 0.3500046133995056,
      "learning_rate": 0.0002091317866539105,
      "loss": 0.1176,
      "step": 6837
    },
    {
      "epoch": 0.9583742116327961,
      "grad_norm": 0.4273541569709778,
      "learning_rate": 0.00020911743602009087,
      "loss": 0.0695,
      "step": 6838
    },
    {
      "epoch": 0.9585143658023826,
      "grad_norm": 1.1742777824401855,
      "learning_rate": 0.0002091030853862712,
      "loss": 0.0621,
      "step": 6839
    },
    {
      "epoch": 0.9586545199719692,
      "grad_norm": 0.23294374346733093,
      "learning_rate": 0.00020908873475245155,
      "loss": 0.0285,
      "step": 6840
    },
    {
      "epoch": 0.9587946741415557,
      "grad_norm": 0.29251956939697266,
      "learning_rate": 0.0002090743841186319,
      "loss": 0.0383,
      "step": 6841
    },
    {
      "epoch": 0.9589348283111423,
      "grad_norm": 0.44774389266967773,
      "learning_rate": 0.00020906003348481223,
      "loss": 0.1078,
      "step": 6842
    },
    {
      "epoch": 0.9590749824807288,
      "grad_norm": 0.3983364403247833,
      "learning_rate": 0.00020904568285099256,
      "loss": 0.0816,
      "step": 6843
    },
    {
      "epoch": 0.9592151366503153,
      "grad_norm": 0.01704859547317028,
      "learning_rate": 0.00020903133221717292,
      "loss": 0.0023,
      "step": 6844
    },
    {
      "epoch": 0.9593552908199019,
      "grad_norm": 1.1474777460098267,
      "learning_rate": 0.00020901698158335325,
      "loss": 0.141,
      "step": 6845
    },
    {
      "epoch": 0.9594954449894885,
      "grad_norm": 0.4115254282951355,
      "learning_rate": 0.00020900263094953358,
      "loss": 0.0157,
      "step": 6846
    },
    {
      "epoch": 0.959635599159075,
      "grad_norm": 1.123226284980774,
      "learning_rate": 0.00020898828031571393,
      "loss": 0.2343,
      "step": 6847
    },
    {
      "epoch": 0.9597757533286615,
      "grad_norm": 0.9458621144294739,
      "learning_rate": 0.00020897392968189426,
      "loss": 0.2189,
      "step": 6848
    },
    {
      "epoch": 0.9599159074982481,
      "grad_norm": 1.3432241678237915,
      "learning_rate": 0.0002089595790480746,
      "loss": 0.2261,
      "step": 6849
    },
    {
      "epoch": 0.9600560616678346,
      "grad_norm": 2.4930717945098877,
      "learning_rate": 0.00020894522841425492,
      "loss": 0.2259,
      "step": 6850
    },
    {
      "epoch": 0.9601962158374212,
      "grad_norm": 0.2192278802394867,
      "learning_rate": 0.0002089308777804353,
      "loss": 0.0405,
      "step": 6851
    },
    {
      "epoch": 0.9603363700070077,
      "grad_norm": 0.5236284136772156,
      "learning_rate": 0.00020891652714661563,
      "loss": 0.0782,
      "step": 6852
    },
    {
      "epoch": 0.9604765241765942,
      "grad_norm": 0.36357614398002625,
      "learning_rate": 0.00020890217651279596,
      "loss": 0.0332,
      "step": 6853
    },
    {
      "epoch": 0.9606166783461808,
      "grad_norm": 0.35396990180015564,
      "learning_rate": 0.0002088878258789763,
      "loss": 0.0362,
      "step": 6854
    },
    {
      "epoch": 0.9607568325157674,
      "grad_norm": 0.3969132602214813,
      "learning_rate": 0.00020887347524515664,
      "loss": 0.0615,
      "step": 6855
    },
    {
      "epoch": 0.9608969866853538,
      "grad_norm": 0.15618836879730225,
      "learning_rate": 0.00020885912461133697,
      "loss": 0.0484,
      "step": 6856
    },
    {
      "epoch": 0.9610371408549404,
      "grad_norm": 0.3552684783935547,
      "learning_rate": 0.00020884477397751733,
      "loss": 0.0893,
      "step": 6857
    },
    {
      "epoch": 0.961177295024527,
      "grad_norm": 0.26791292428970337,
      "learning_rate": 0.00020883042334369765,
      "loss": 0.0746,
      "step": 6858
    },
    {
      "epoch": 0.9613174491941135,
      "grad_norm": 0.8597043752670288,
      "learning_rate": 0.00020881607270987798,
      "loss": 0.0963,
      "step": 6859
    },
    {
      "epoch": 0.9614576033637001,
      "grad_norm": 0.4311549663543701,
      "learning_rate": 0.00020880172207605837,
      "loss": 0.1231,
      "step": 6860
    },
    {
      "epoch": 0.9615977575332866,
      "grad_norm": 0.3141905665397644,
      "learning_rate": 0.0002087873714422387,
      "loss": 0.0977,
      "step": 6861
    },
    {
      "epoch": 0.9617379117028731,
      "grad_norm": 0.3051708936691284,
      "learning_rate": 0.00020877302080841902,
      "loss": 0.1264,
      "step": 6862
    },
    {
      "epoch": 0.9618780658724597,
      "grad_norm": 0.32538965344429016,
      "learning_rate": 0.00020875867017459938,
      "loss": 0.062,
      "step": 6863
    },
    {
      "epoch": 0.9620182200420463,
      "grad_norm": 0.4162488877773285,
      "learning_rate": 0.0002087443195407797,
      "loss": 0.1089,
      "step": 6864
    },
    {
      "epoch": 0.9621583742116327,
      "grad_norm": 0.3207227885723114,
      "learning_rate": 0.00020872996890696003,
      "loss": 0.0537,
      "step": 6865
    },
    {
      "epoch": 0.9622985283812193,
      "grad_norm": 0.14528274536132812,
      "learning_rate": 0.0002087156182731404,
      "loss": 0.0201,
      "step": 6866
    },
    {
      "epoch": 0.9624386825508059,
      "grad_norm": 0.29829683899879456,
      "learning_rate": 0.00020870126763932072,
      "loss": 0.1148,
      "step": 6867
    },
    {
      "epoch": 0.9625788367203925,
      "grad_norm": 0.08368442207574844,
      "learning_rate": 0.00020868691700550105,
      "loss": 0.0174,
      "step": 6868
    },
    {
      "epoch": 0.962718990889979,
      "grad_norm": 0.14285236597061157,
      "learning_rate": 0.00020867256637168138,
      "loss": 0.0246,
      "step": 6869
    },
    {
      "epoch": 0.9628591450595655,
      "grad_norm": 3.1134514808654785,
      "learning_rate": 0.00020865821573786173,
      "loss": 0.0723,
      "step": 6870
    },
    {
      "epoch": 0.9629992992291521,
      "grad_norm": 0.38169342279434204,
      "learning_rate": 0.00020864386510404206,
      "loss": 0.0984,
      "step": 6871
    },
    {
      "epoch": 0.9631394533987386,
      "grad_norm": 0.4693698585033417,
      "learning_rate": 0.00020862951447022242,
      "loss": 0.1521,
      "step": 6872
    },
    {
      "epoch": 0.9632796075683252,
      "grad_norm": 0.14318130910396576,
      "learning_rate": 0.00020861516383640277,
      "loss": 0.0497,
      "step": 6873
    },
    {
      "epoch": 0.9634197617379117,
      "grad_norm": 0.21699407696723938,
      "learning_rate": 0.0002086008132025831,
      "loss": 0.0452,
      "step": 6874
    },
    {
      "epoch": 0.9635599159074982,
      "grad_norm": 0.36788904666900635,
      "learning_rate": 0.00020858646256876343,
      "loss": 0.0697,
      "step": 6875
    },
    {
      "epoch": 0.9637000700770848,
      "grad_norm": 0.8227933049201965,
      "learning_rate": 0.00020857211193494378,
      "loss": 0.1259,
      "step": 6876
    },
    {
      "epoch": 0.9638402242466714,
      "grad_norm": 0.4492322504520416,
      "learning_rate": 0.0002085577613011241,
      "loss": 0.1111,
      "step": 6877
    },
    {
      "epoch": 0.9639803784162578,
      "grad_norm": 1.0661616325378418,
      "learning_rate": 0.00020854341066730444,
      "loss": 0.0717,
      "step": 6878
    },
    {
      "epoch": 0.9641205325858444,
      "grad_norm": 0.3309056758880615,
      "learning_rate": 0.0002085290600334848,
      "loss": 0.0449,
      "step": 6879
    },
    {
      "epoch": 0.964260686755431,
      "grad_norm": 0.33337658643722534,
      "learning_rate": 0.00020851470939966513,
      "loss": 0.0858,
      "step": 6880
    },
    {
      "epoch": 0.9644008409250175,
      "grad_norm": 0.9851006269454956,
      "learning_rate": 0.00020850035876584545,
      "loss": 0.0643,
      "step": 6881
    },
    {
      "epoch": 0.9645409950946041,
      "grad_norm": 0.24895556271076202,
      "learning_rate": 0.00020848600813202584,
      "loss": 0.0363,
      "step": 6882
    },
    {
      "epoch": 0.9646811492641906,
      "grad_norm": 0.48956048488616943,
      "learning_rate": 0.00020847165749820617,
      "loss": 0.0932,
      "step": 6883
    },
    {
      "epoch": 0.9648213034337771,
      "grad_norm": 0.35453397035598755,
      "learning_rate": 0.0002084573068643865,
      "loss": 0.1305,
      "step": 6884
    },
    {
      "epoch": 0.9649614576033637,
      "grad_norm": 0.2098037451505661,
      "learning_rate": 0.00020844295623056682,
      "loss": 0.077,
      "step": 6885
    },
    {
      "epoch": 0.9651016117729503,
      "grad_norm": 0.23574477434158325,
      "learning_rate": 0.00020842860559674718,
      "loss": 0.0197,
      "step": 6886
    },
    {
      "epoch": 0.9652417659425367,
      "grad_norm": 0.36492350697517395,
      "learning_rate": 0.0002084142549629275,
      "loss": 0.0628,
      "step": 6887
    },
    {
      "epoch": 0.9653819201121233,
      "grad_norm": 0.7939193844795227,
      "learning_rate": 0.00020839990432910784,
      "loss": 0.059,
      "step": 6888
    },
    {
      "epoch": 0.9655220742817099,
      "grad_norm": 1.3341991901397705,
      "learning_rate": 0.0002083855536952882,
      "loss": 0.2687,
      "step": 6889
    },
    {
      "epoch": 0.9656622284512965,
      "grad_norm": 0.09365924447774887,
      "learning_rate": 0.00020837120306146852,
      "loss": 0.0157,
      "step": 6890
    },
    {
      "epoch": 0.965802382620883,
      "grad_norm": 0.4536382257938385,
      "learning_rate": 0.00020835685242764885,
      "loss": 0.0679,
      "step": 6891
    },
    {
      "epoch": 0.9659425367904695,
      "grad_norm": 0.6109830141067505,
      "learning_rate": 0.00020834250179382923,
      "loss": 0.088,
      "step": 6892
    },
    {
      "epoch": 0.9660826909600561,
      "grad_norm": 0.7137888669967651,
      "learning_rate": 0.00020832815116000956,
      "loss": 0.0421,
      "step": 6893
    },
    {
      "epoch": 0.9662228451296426,
      "grad_norm": 0.883314311504364,
      "learning_rate": 0.0002083138005261899,
      "loss": 0.0589,
      "step": 6894
    },
    {
      "epoch": 0.9663629992992292,
      "grad_norm": 0.6361221671104431,
      "learning_rate": 0.00020829944989237024,
      "loss": 0.062,
      "step": 6895
    },
    {
      "epoch": 0.9665031534688157,
      "grad_norm": 2.891608238220215,
      "learning_rate": 0.00020828509925855057,
      "loss": 0.4207,
      "step": 6896
    },
    {
      "epoch": 0.9666433076384022,
      "grad_norm": 0.7212381362915039,
      "learning_rate": 0.0002082707486247309,
      "loss": 0.0716,
      "step": 6897
    },
    {
      "epoch": 0.9667834618079888,
      "grad_norm": 1.5999410152435303,
      "learning_rate": 0.00020825639799091126,
      "loss": 0.2749,
      "step": 6898
    },
    {
      "epoch": 0.9669236159775754,
      "grad_norm": 1.2265528440475464,
      "learning_rate": 0.00020824204735709159,
      "loss": 0.2263,
      "step": 6899
    },
    {
      "epoch": 0.9670637701471618,
      "grad_norm": 0.8691014051437378,
      "learning_rate": 0.00020822769672327191,
      "loss": 0.0846,
      "step": 6900
    },
    {
      "epoch": 0.9672039243167484,
      "grad_norm": 0.22890891134738922,
      "learning_rate": 0.00020821334608945227,
      "loss": 0.0474,
      "step": 6901
    },
    {
      "epoch": 0.967344078486335,
      "grad_norm": 0.2901453375816345,
      "learning_rate": 0.0002081989954556326,
      "loss": 0.0563,
      "step": 6902
    },
    {
      "epoch": 0.9674842326559215,
      "grad_norm": 0.5743530988693237,
      "learning_rate": 0.00020818464482181293,
      "loss": 0.1326,
      "step": 6903
    },
    {
      "epoch": 0.9676243868255081,
      "grad_norm": 0.45388081669807434,
      "learning_rate": 0.00020817029418799328,
      "loss": 0.0881,
      "step": 6904
    },
    {
      "epoch": 0.9677645409950946,
      "grad_norm": 0.5331664085388184,
      "learning_rate": 0.00020815594355417364,
      "loss": 0.095,
      "step": 6905
    },
    {
      "epoch": 0.9679046951646811,
      "grad_norm": 0.17647810280323029,
      "learning_rate": 0.00020814159292035397,
      "loss": 0.0285,
      "step": 6906
    },
    {
      "epoch": 0.9680448493342677,
      "grad_norm": 0.41952958703041077,
      "learning_rate": 0.0002081272422865343,
      "loss": 0.0615,
      "step": 6907
    },
    {
      "epoch": 0.9681850035038543,
      "grad_norm": 0.4461071193218231,
      "learning_rate": 0.00020811289165271465,
      "loss": 0.1123,
      "step": 6908
    },
    {
      "epoch": 0.9683251576734407,
      "grad_norm": 0.6692301630973816,
      "learning_rate": 0.00020809854101889498,
      "loss": 0.0817,
      "step": 6909
    },
    {
      "epoch": 0.9684653118430273,
      "grad_norm": 0.36530736088752747,
      "learning_rate": 0.0002080841903850753,
      "loss": 0.0506,
      "step": 6910
    },
    {
      "epoch": 0.9686054660126139,
      "grad_norm": 0.4973869025707245,
      "learning_rate": 0.00020806983975125566,
      "loss": 0.0826,
      "step": 6911
    },
    {
      "epoch": 0.9687456201822005,
      "grad_norm": 0.5696591138839722,
      "learning_rate": 0.000208055489117436,
      "loss": 0.0816,
      "step": 6912
    },
    {
      "epoch": 0.9688857743517869,
      "grad_norm": 0.3018736243247986,
      "learning_rate": 0.00020804113848361632,
      "loss": 0.0721,
      "step": 6913
    },
    {
      "epoch": 0.9690259285213735,
      "grad_norm": 0.33308374881744385,
      "learning_rate": 0.0002080267878497967,
      "loss": 0.0417,
      "step": 6914
    },
    {
      "epoch": 0.96916608269096,
      "grad_norm": 0.3336891233921051,
      "learning_rate": 0.00020801243721597703,
      "loss": 0.0741,
      "step": 6915
    },
    {
      "epoch": 0.9693062368605466,
      "grad_norm": 0.41221198439598083,
      "learning_rate": 0.00020799808658215736,
      "loss": 0.0593,
      "step": 6916
    },
    {
      "epoch": 0.9694463910301332,
      "grad_norm": 0.47218501567840576,
      "learning_rate": 0.00020798373594833772,
      "loss": 0.1557,
      "step": 6917
    },
    {
      "epoch": 0.9695865451997197,
      "grad_norm": 0.26762154698371887,
      "learning_rate": 0.00020796938531451804,
      "loss": 0.0625,
      "step": 6918
    },
    {
      "epoch": 0.9697266993693062,
      "grad_norm": 0.44407474994659424,
      "learning_rate": 0.00020795503468069837,
      "loss": 0.0765,
      "step": 6919
    },
    {
      "epoch": 0.9698668535388928,
      "grad_norm": 0.30590900778770447,
      "learning_rate": 0.0002079406840468787,
      "loss": 0.03,
      "step": 6920
    },
    {
      "epoch": 0.9700070077084794,
      "grad_norm": 0.41272279620170593,
      "learning_rate": 0.00020792633341305906,
      "loss": 0.0692,
      "step": 6921
    },
    {
      "epoch": 0.9701471618780658,
      "grad_norm": 0.6494072079658508,
      "learning_rate": 0.00020791198277923939,
      "loss": 0.1739,
      "step": 6922
    },
    {
      "epoch": 0.9702873160476524,
      "grad_norm": 0.315132200717926,
      "learning_rate": 0.00020789763214541971,
      "loss": 0.0839,
      "step": 6923
    },
    {
      "epoch": 0.970427470217239,
      "grad_norm": 0.3787939250469208,
      "learning_rate": 0.0002078832815116001,
      "loss": 0.0286,
      "step": 6924
    },
    {
      "epoch": 0.9705676243868255,
      "grad_norm": 1.0519497394561768,
      "learning_rate": 0.00020786893087778043,
      "loss": 0.1371,
      "step": 6925
    },
    {
      "epoch": 0.9707077785564121,
      "grad_norm": 0.5610406994819641,
      "learning_rate": 0.00020785458024396075,
      "loss": 0.1178,
      "step": 6926
    },
    {
      "epoch": 0.9708479327259986,
      "grad_norm": 0.35256150364875793,
      "learning_rate": 0.0002078402296101411,
      "loss": 0.0652,
      "step": 6927
    },
    {
      "epoch": 0.9709880868955851,
      "grad_norm": 0.13512149453163147,
      "learning_rate": 0.00020782587897632144,
      "loss": 0.0286,
      "step": 6928
    },
    {
      "epoch": 0.9711282410651717,
      "grad_norm": 0.2617959678173065,
      "learning_rate": 0.00020781152834250177,
      "loss": 0.0413,
      "step": 6929
    },
    {
      "epoch": 0.9712683952347583,
      "grad_norm": 0.1958327442407608,
      "learning_rate": 0.00020779717770868212,
      "loss": 0.0375,
      "step": 6930
    },
    {
      "epoch": 0.9714085494043447,
      "grad_norm": 0.4769895374774933,
      "learning_rate": 0.00020778282707486245,
      "loss": 0.07,
      "step": 6931
    },
    {
      "epoch": 0.9715487035739313,
      "grad_norm": 0.4198013246059418,
      "learning_rate": 0.00020776847644104278,
      "loss": 0.0905,
      "step": 6932
    },
    {
      "epoch": 0.9716888577435179,
      "grad_norm": 0.5336285829544067,
      "learning_rate": 0.00020775412580722314,
      "loss": 0.0755,
      "step": 6933
    },
    {
      "epoch": 0.9718290119131044,
      "grad_norm": 0.323056697845459,
      "learning_rate": 0.00020773977517340346,
      "loss": 0.0963,
      "step": 6934
    },
    {
      "epoch": 0.9719691660826909,
      "grad_norm": 0.9526858329772949,
      "learning_rate": 0.00020772542453958382,
      "loss": 0.0577,
      "step": 6935
    },
    {
      "epoch": 0.9721093202522775,
      "grad_norm": 0.3790183961391449,
      "learning_rate": 0.00020771107390576418,
      "loss": 0.059,
      "step": 6936
    },
    {
      "epoch": 0.972249474421864,
      "grad_norm": 0.8899807333946228,
      "learning_rate": 0.0002076967232719445,
      "loss": 0.1189,
      "step": 6937
    },
    {
      "epoch": 0.9723896285914506,
      "grad_norm": 1.0275369882583618,
      "learning_rate": 0.00020768237263812483,
      "loss": 0.185,
      "step": 6938
    },
    {
      "epoch": 0.9725297827610372,
      "grad_norm": 0.2496979981660843,
      "learning_rate": 0.00020766802200430516,
      "loss": 0.0688,
      "step": 6939
    },
    {
      "epoch": 0.9726699369306236,
      "grad_norm": 0.3520813286304474,
      "learning_rate": 0.00020765367137048552,
      "loss": 0.0375,
      "step": 6940
    },
    {
      "epoch": 0.9728100911002102,
      "grad_norm": 0.6201832890510559,
      "learning_rate": 0.00020763932073666585,
      "loss": 0.0721,
      "step": 6941
    },
    {
      "epoch": 0.9729502452697968,
      "grad_norm": 0.6734105944633484,
      "learning_rate": 0.00020762497010284617,
      "loss": 0.0732,
      "step": 6942
    },
    {
      "epoch": 0.9730903994393834,
      "grad_norm": 0.5790233016014099,
      "learning_rate": 0.00020761061946902653,
      "loss": 0.0253,
      "step": 6943
    },
    {
      "epoch": 0.9732305536089698,
      "grad_norm": 0.7184246182441711,
      "learning_rate": 0.00020759626883520686,
      "loss": 0.0483,
      "step": 6944
    },
    {
      "epoch": 0.9733707077785564,
      "grad_norm": 0.6030838489532471,
      "learning_rate": 0.0002075819182013872,
      "loss": 0.0599,
      "step": 6945
    },
    {
      "epoch": 0.973510861948143,
      "grad_norm": 0.9266127347946167,
      "learning_rate": 0.00020756756756756757,
      "loss": 0.1178,
      "step": 6946
    },
    {
      "epoch": 0.9736510161177295,
      "grad_norm": 2.7538599967956543,
      "learning_rate": 0.0002075532169337479,
      "loss": 0.1661,
      "step": 6947
    },
    {
      "epoch": 0.9737911702873161,
      "grad_norm": 0.5794621109962463,
      "learning_rate": 0.00020753886629992823,
      "loss": 0.1154,
      "step": 6948
    },
    {
      "epoch": 0.9739313244569026,
      "grad_norm": 1.5444072484970093,
      "learning_rate": 0.00020752451566610858,
      "loss": 0.173,
      "step": 6949
    },
    {
      "epoch": 0.9740714786264891,
      "grad_norm": 1.1836755275726318,
      "learning_rate": 0.0002075101650322889,
      "loss": 0.1901,
      "step": 6950
    },
    {
      "epoch": 0.9742116327960757,
      "grad_norm": 0.27281269431114197,
      "learning_rate": 0.00020749581439846924,
      "loss": 0.0366,
      "step": 6951
    },
    {
      "epoch": 0.9743517869656623,
      "grad_norm": 0.2458818256855011,
      "learning_rate": 0.0002074814637646496,
      "loss": 0.0381,
      "step": 6952
    },
    {
      "epoch": 0.9744919411352487,
      "grad_norm": 0.4554118812084198,
      "learning_rate": 0.00020746711313082992,
      "loss": 0.102,
      "step": 6953
    },
    {
      "epoch": 0.9746320953048353,
      "grad_norm": 0.5654162764549255,
      "learning_rate": 0.00020745276249701025,
      "loss": 0.099,
      "step": 6954
    },
    {
      "epoch": 0.9747722494744219,
      "grad_norm": 0.31239786744117737,
      "learning_rate": 0.00020743841186319058,
      "loss": 0.0437,
      "step": 6955
    },
    {
      "epoch": 0.9749124036440084,
      "grad_norm": 0.2250964492559433,
      "learning_rate": 0.00020742406122937096,
      "loss": 0.0627,
      "step": 6956
    },
    {
      "epoch": 0.9750525578135949,
      "grad_norm": 0.4371689260005951,
      "learning_rate": 0.0002074097105955513,
      "loss": 0.1013,
      "step": 6957
    },
    {
      "epoch": 0.9751927119831815,
      "grad_norm": 0.27315187454223633,
      "learning_rate": 0.00020739535996173162,
      "loss": 0.044,
      "step": 6958
    },
    {
      "epoch": 0.975332866152768,
      "grad_norm": 0.5150197744369507,
      "learning_rate": 0.00020738100932791198,
      "loss": 0.0893,
      "step": 6959
    },
    {
      "epoch": 0.9754730203223546,
      "grad_norm": 0.4323754906654358,
      "learning_rate": 0.0002073666586940923,
      "loss": 0.1044,
      "step": 6960
    },
    {
      "epoch": 0.9756131744919412,
      "grad_norm": 0.22092288732528687,
      "learning_rate": 0.00020735230806027263,
      "loss": 0.0403,
      "step": 6961
    },
    {
      "epoch": 0.9757533286615276,
      "grad_norm": 0.20307494699954987,
      "learning_rate": 0.000207337957426453,
      "loss": 0.0573,
      "step": 6962
    },
    {
      "epoch": 0.9758934828311142,
      "grad_norm": 0.41007351875305176,
      "learning_rate": 0.00020732360679263332,
      "loss": 0.1052,
      "step": 6963
    },
    {
      "epoch": 0.9760336370007008,
      "grad_norm": 0.2851572036743164,
      "learning_rate": 0.00020730925615881365,
      "loss": 0.0552,
      "step": 6964
    },
    {
      "epoch": 0.9761737911702874,
      "grad_norm": 0.961644172668457,
      "learning_rate": 0.000207294905524994,
      "loss": 0.132,
      "step": 6965
    },
    {
      "epoch": 0.9763139453398738,
      "grad_norm": 0.21831761300563812,
      "learning_rate": 0.00020728055489117433,
      "loss": 0.031,
      "step": 6966
    },
    {
      "epoch": 0.9764540995094604,
      "grad_norm": 0.4800322353839874,
      "learning_rate": 0.0002072662042573547,
      "loss": 0.0603,
      "step": 6967
    },
    {
      "epoch": 0.976594253679047,
      "grad_norm": 0.56587815284729,
      "learning_rate": 0.00020725185362353504,
      "loss": 0.0589,
      "step": 6968
    },
    {
      "epoch": 0.9767344078486335,
      "grad_norm": 0.34528833627700806,
      "learning_rate": 0.00020723750298971537,
      "loss": 0.0437,
      "step": 6969
    },
    {
      "epoch": 0.9768745620182201,
      "grad_norm": 0.34439995884895325,
      "learning_rate": 0.0002072231523558957,
      "loss": 0.058,
      "step": 6970
    },
    {
      "epoch": 0.9770147161878066,
      "grad_norm": 0.1708739697933197,
      "learning_rate": 0.00020720880172207605,
      "loss": 0.0311,
      "step": 6971
    },
    {
      "epoch": 0.9771548703573931,
      "grad_norm": 0.7745507955551147,
      "learning_rate": 0.00020719445108825638,
      "loss": 0.1903,
      "step": 6972
    },
    {
      "epoch": 0.9772950245269797,
      "grad_norm": 0.19944646954536438,
      "learning_rate": 0.0002071801004544367,
      "loss": 0.0424,
      "step": 6973
    },
    {
      "epoch": 0.9774351786965663,
      "grad_norm": 0.29700997471809387,
      "learning_rate": 0.00020716574982061704,
      "loss": 0.0321,
      "step": 6974
    },
    {
      "epoch": 0.9775753328661527,
      "grad_norm": 0.30736637115478516,
      "learning_rate": 0.0002071513991867974,
      "loss": 0.1211,
      "step": 6975
    },
    {
      "epoch": 0.9777154870357393,
      "grad_norm": 0.1292431801557541,
      "learning_rate": 0.00020713704855297772,
      "loss": 0.0226,
      "step": 6976
    },
    {
      "epoch": 0.9778556412053259,
      "grad_norm": 0.4061063230037689,
      "learning_rate": 0.00020712269791915805,
      "loss": 0.0728,
      "step": 6977
    },
    {
      "epoch": 0.9779957953749124,
      "grad_norm": 0.5302044153213501,
      "learning_rate": 0.00020710834728533844,
      "loss": 0.0895,
      "step": 6978
    },
    {
      "epoch": 0.9781359495444989,
      "grad_norm": 0.2492242306470871,
      "learning_rate": 0.00020709399665151876,
      "loss": 0.0291,
      "step": 6979
    },
    {
      "epoch": 0.9782761037140855,
      "grad_norm": 0.5246913433074951,
      "learning_rate": 0.0002070796460176991,
      "loss": 0.1142,
      "step": 6980
    },
    {
      "epoch": 0.978416257883672,
      "grad_norm": 0.3783250153064728,
      "learning_rate": 0.00020706529538387945,
      "loss": 0.041,
      "step": 6981
    },
    {
      "epoch": 0.9785564120532586,
      "grad_norm": 0.9292603731155396,
      "learning_rate": 0.00020705094475005978,
      "loss": 0.1006,
      "step": 6982
    },
    {
      "epoch": 0.9786965662228452,
      "grad_norm": 0.16517627239227295,
      "learning_rate": 0.0002070365941162401,
      "loss": 0.0327,
      "step": 6983
    },
    {
      "epoch": 0.9788367203924316,
      "grad_norm": 0.40302616357803345,
      "learning_rate": 0.00020702224348242046,
      "loss": 0.0198,
      "step": 6984
    },
    {
      "epoch": 0.9789768745620182,
      "grad_norm": 0.282459020614624,
      "learning_rate": 0.0002070078928486008,
      "loss": 0.0566,
      "step": 6985
    },
    {
      "epoch": 0.9791170287316048,
      "grad_norm": 0.129927396774292,
      "learning_rate": 0.00020699354221478112,
      "loss": 0.0082,
      "step": 6986
    },
    {
      "epoch": 0.9792571829011913,
      "grad_norm": 0.33250609040260315,
      "learning_rate": 0.0002069791915809615,
      "loss": 0.049,
      "step": 6987
    },
    {
      "epoch": 0.9793973370707778,
      "grad_norm": 0.41602328419685364,
      "learning_rate": 0.00020696484094714183,
      "loss": 0.0519,
      "step": 6988
    },
    {
      "epoch": 0.9795374912403644,
      "grad_norm": 0.5409842729568481,
      "learning_rate": 0.00020695049031332216,
      "loss": 0.054,
      "step": 6989
    },
    {
      "epoch": 0.979677645409951,
      "grad_norm": 0.5139724612236023,
      "learning_rate": 0.0002069361396795025,
      "loss": 0.0821,
      "step": 6990
    },
    {
      "epoch": 0.9798177995795375,
      "grad_norm": 0.5147519111633301,
      "learning_rate": 0.00020692178904568284,
      "loss": 0.0543,
      "step": 6991
    },
    {
      "epoch": 0.9799579537491241,
      "grad_norm": 0.6792997717857361,
      "learning_rate": 0.00020690743841186317,
      "loss": 0.1143,
      "step": 6992
    },
    {
      "epoch": 0.9800981079187105,
      "grad_norm": 0.5357325673103333,
      "learning_rate": 0.0002068930877780435,
      "loss": 0.1133,
      "step": 6993
    },
    {
      "epoch": 0.9802382620882971,
      "grad_norm": 0.9429832100868225,
      "learning_rate": 0.00020687873714422386,
      "loss": 0.1229,
      "step": 6994
    },
    {
      "epoch": 0.9803784162578837,
      "grad_norm": 0.30730298161506653,
      "learning_rate": 0.00020686438651040418,
      "loss": 0.0658,
      "step": 6995
    },
    {
      "epoch": 0.9805185704274703,
      "grad_norm": 0.9290444254875183,
      "learning_rate": 0.0002068500358765845,
      "loss": 0.1383,
      "step": 6996
    },
    {
      "epoch": 0.9806587245970567,
      "grad_norm": 1.5623475313186646,
      "learning_rate": 0.00020683568524276487,
      "loss": 0.0622,
      "step": 6997
    },
    {
      "epoch": 0.9807988787666433,
      "grad_norm": 2.994185209274292,
      "learning_rate": 0.0002068213346089452,
      "loss": 0.1936,
      "step": 6998
    },
    {
      "epoch": 0.9809390329362299,
      "grad_norm": 0.5363599061965942,
      "learning_rate": 0.00020680698397512555,
      "loss": 0.053,
      "step": 6999
    },
    {
      "epoch": 0.9810791871058164,
      "grad_norm": 0.4846128225326538,
      "learning_rate": 0.0002067926333413059,
      "loss": 0.0322,
      "step": 7000
    },
    {
      "epoch": 0.9812193412754029,
      "grad_norm": 0.30691367387771606,
      "learning_rate": 0.00020677828270748624,
      "loss": 0.0575,
      "step": 7001
    },
    {
      "epoch": 0.9813594954449895,
      "grad_norm": 0.2341775894165039,
      "learning_rate": 0.00020676393207366657,
      "loss": 0.0369,
      "step": 7002
    },
    {
      "epoch": 0.981499649614576,
      "grad_norm": 0.45710182189941406,
      "learning_rate": 0.00020674958143984692,
      "loss": 0.1012,
      "step": 7003
    },
    {
      "epoch": 0.9816398037841626,
      "grad_norm": 0.36037296056747437,
      "learning_rate": 0.00020673523080602725,
      "loss": 0.0721,
      "step": 7004
    },
    {
      "epoch": 0.9817799579537492,
      "grad_norm": 0.3422500789165497,
      "learning_rate": 0.00020672088017220758,
      "loss": 0.0141,
      "step": 7005
    },
    {
      "epoch": 0.9819201121233356,
      "grad_norm": 0.31993263959884644,
      "learning_rate": 0.00020670652953838793,
      "loss": 0.0453,
      "step": 7006
    },
    {
      "epoch": 0.9820602662929222,
      "grad_norm": 0.2883129119873047,
      "learning_rate": 0.00020669217890456826,
      "loss": 0.0328,
      "step": 7007
    },
    {
      "epoch": 0.9822004204625088,
      "grad_norm": 0.26729217171669006,
      "learning_rate": 0.0002066778282707486,
      "loss": 0.0564,
      "step": 7008
    },
    {
      "epoch": 0.9823405746320953,
      "grad_norm": 0.3590710163116455,
      "learning_rate": 0.00020666347763692892,
      "loss": 0.0599,
      "step": 7009
    },
    {
      "epoch": 0.9824807288016818,
      "grad_norm": 0.47798070311546326,
      "learning_rate": 0.0002066491270031093,
      "loss": 0.0468,
      "step": 7010
    },
    {
      "epoch": 0.9826208829712684,
      "grad_norm": 0.46382319927215576,
      "learning_rate": 0.00020663477636928963,
      "loss": 0.1042,
      "step": 7011
    },
    {
      "epoch": 0.9827610371408549,
      "grad_norm": 0.37648844718933105,
      "learning_rate": 0.00020662042573546996,
      "loss": 0.0452,
      "step": 7012
    },
    {
      "epoch": 0.9829011913104415,
      "grad_norm": 0.5499049425125122,
      "learning_rate": 0.00020660607510165032,
      "loss": 0.1431,
      "step": 7013
    },
    {
      "epoch": 0.9830413454800281,
      "grad_norm": 0.7389718294143677,
      "learning_rate": 0.00020659172446783064,
      "loss": 0.0712,
      "step": 7014
    },
    {
      "epoch": 0.9831814996496145,
      "grad_norm": 0.1977791041135788,
      "learning_rate": 0.00020657737383401097,
      "loss": 0.0544,
      "step": 7015
    },
    {
      "epoch": 0.9833216538192011,
      "grad_norm": 0.5178593993186951,
      "learning_rate": 0.00020656302320019133,
      "loss": 0.073,
      "step": 7016
    },
    {
      "epoch": 0.9834618079887877,
      "grad_norm": 0.9876886010169983,
      "learning_rate": 0.00020654867256637166,
      "loss": 0.07,
      "step": 7017
    },
    {
      "epoch": 0.9836019621583743,
      "grad_norm": 0.3890713155269623,
      "learning_rate": 0.00020653432193255199,
      "loss": 0.0954,
      "step": 7018
    },
    {
      "epoch": 0.9837421163279607,
      "grad_norm": 0.5815483331680298,
      "learning_rate": 0.00020651997129873237,
      "loss": 0.0768,
      "step": 7019
    },
    {
      "epoch": 0.9838822704975473,
      "grad_norm": 0.4890800416469574,
      "learning_rate": 0.0002065056206649127,
      "loss": 0.1239,
      "step": 7020
    },
    {
      "epoch": 0.9840224246671339,
      "grad_norm": 0.5563451647758484,
      "learning_rate": 0.00020649127003109303,
      "loss": 0.0492,
      "step": 7021
    },
    {
      "epoch": 0.9841625788367204,
      "grad_norm": 0.5135653018951416,
      "learning_rate": 0.00020647691939727338,
      "loss": 0.1072,
      "step": 7022
    },
    {
      "epoch": 0.9843027330063069,
      "grad_norm": 0.8146799206733704,
      "learning_rate": 0.0002064625687634537,
      "loss": 0.0791,
      "step": 7023
    },
    {
      "epoch": 0.9844428871758935,
      "grad_norm": 0.3026643991470337,
      "learning_rate": 0.00020644821812963404,
      "loss": 0.0611,
      "step": 7024
    },
    {
      "epoch": 0.98458304134548,
      "grad_norm": 0.43137699365615845,
      "learning_rate": 0.00020643386749581437,
      "loss": 0.1023,
      "step": 7025
    },
    {
      "epoch": 0.9847231955150666,
      "grad_norm": 0.299809992313385,
      "learning_rate": 0.00020641951686199472,
      "loss": 0.0829,
      "step": 7026
    },
    {
      "epoch": 0.9848633496846532,
      "grad_norm": 0.5949607491493225,
      "learning_rate": 0.00020640516622817505,
      "loss": 0.0851,
      "step": 7027
    },
    {
      "epoch": 0.9850035038542396,
      "grad_norm": 0.38227760791778564,
      "learning_rate": 0.00020639081559435538,
      "loss": 0.0794,
      "step": 7028
    },
    {
      "epoch": 0.9851436580238262,
      "grad_norm": 0.540929913520813,
      "learning_rate": 0.00020637646496053573,
      "loss": 0.0371,
      "step": 7029
    },
    {
      "epoch": 0.9852838121934128,
      "grad_norm": 0.3091493844985962,
      "learning_rate": 0.0002063621143267161,
      "loss": 0.0339,
      "step": 7030
    },
    {
      "epoch": 0.9854239663629993,
      "grad_norm": 0.4328194856643677,
      "learning_rate": 0.00020634776369289642,
      "loss": 0.0772,
      "step": 7031
    },
    {
      "epoch": 0.9855641205325858,
      "grad_norm": 0.5255628824234009,
      "learning_rate": 0.00020633341305907677,
      "loss": 0.1025,
      "step": 7032
    },
    {
      "epoch": 0.9857042747021724,
      "grad_norm": 0.3415969908237457,
      "learning_rate": 0.0002063190624252571,
      "loss": 0.0847,
      "step": 7033
    },
    {
      "epoch": 0.9858444288717589,
      "grad_norm": 0.30125463008880615,
      "learning_rate": 0.00020630471179143743,
      "loss": 0.0998,
      "step": 7034
    },
    {
      "epoch": 0.9859845830413455,
      "grad_norm": 0.1967931091785431,
      "learning_rate": 0.0002062903611576178,
      "loss": 0.0239,
      "step": 7035
    },
    {
      "epoch": 0.9861247372109321,
      "grad_norm": 0.6195602416992188,
      "learning_rate": 0.00020627601052379812,
      "loss": 0.0901,
      "step": 7036
    },
    {
      "epoch": 0.9862648913805185,
      "grad_norm": 0.2731892168521881,
      "learning_rate": 0.00020626165988997844,
      "loss": 0.063,
      "step": 7037
    },
    {
      "epoch": 0.9864050455501051,
      "grad_norm": 0.3098612129688263,
      "learning_rate": 0.0002062473092561588,
      "loss": 0.0926,
      "step": 7038
    },
    {
      "epoch": 0.9865451997196917,
      "grad_norm": 0.2808132767677307,
      "learning_rate": 0.00020623295862233913,
      "loss": 0.0421,
      "step": 7039
    },
    {
      "epoch": 0.9866853538892782,
      "grad_norm": 0.34991103410720825,
      "learning_rate": 0.00020621860798851946,
      "loss": 0.1143,
      "step": 7040
    },
    {
      "epoch": 0.9868255080588647,
      "grad_norm": 1.4301457405090332,
      "learning_rate": 0.00020620425735469984,
      "loss": 0.0788,
      "step": 7041
    },
    {
      "epoch": 0.9869656622284513,
      "grad_norm": 0.26166096329689026,
      "learning_rate": 0.00020618990672088017,
      "loss": 0.0323,
      "step": 7042
    },
    {
      "epoch": 0.9871058163980378,
      "grad_norm": 0.8020908832550049,
      "learning_rate": 0.0002061755560870605,
      "loss": 0.162,
      "step": 7043
    },
    {
      "epoch": 0.9872459705676244,
      "grad_norm": 0.8709140419960022,
      "learning_rate": 0.00020616120545324083,
      "loss": 0.1758,
      "step": 7044
    },
    {
      "epoch": 0.9873861247372109,
      "grad_norm": 0.8525486588478088,
      "learning_rate": 0.00020614685481942118,
      "loss": 0.0917,
      "step": 7045
    },
    {
      "epoch": 0.9875262789067975,
      "grad_norm": 1.0763194561004639,
      "learning_rate": 0.0002061325041856015,
      "loss": 0.1291,
      "step": 7046
    },
    {
      "epoch": 0.987666433076384,
      "grad_norm": 0.530787467956543,
      "learning_rate": 0.00020611815355178184,
      "loss": 0.0735,
      "step": 7047
    },
    {
      "epoch": 0.9878065872459706,
      "grad_norm": 1.07344388961792,
      "learning_rate": 0.0002061038029179622,
      "loss": 0.1167,
      "step": 7048
    },
    {
      "epoch": 0.9879467414155572,
      "grad_norm": 1.048150658607483,
      "learning_rate": 0.00020608945228414252,
      "loss": 0.1266,
      "step": 7049
    },
    {
      "epoch": 0.9880868955851436,
      "grad_norm": 4.462587833404541,
      "learning_rate": 0.00020607510165032285,
      "loss": 0.3827,
      "step": 7050
    },
    {
      "epoch": 0.9882270497547302,
      "grad_norm": 0.37979450821876526,
      "learning_rate": 0.00020606075101650323,
      "loss": 0.12,
      "step": 7051
    },
    {
      "epoch": 0.9883672039243168,
      "grad_norm": 0.26517969369888306,
      "learning_rate": 0.00020604640038268356,
      "loss": 0.0923,
      "step": 7052
    },
    {
      "epoch": 0.9885073580939033,
      "grad_norm": 0.2702295780181885,
      "learning_rate": 0.0002060320497488639,
      "loss": 0.0633,
      "step": 7053
    },
    {
      "epoch": 0.9886475122634898,
      "grad_norm": 0.45654866099357605,
      "learning_rate": 0.00020601769911504425,
      "loss": 0.0955,
      "step": 7054
    },
    {
      "epoch": 0.9887876664330764,
      "grad_norm": 0.5021839737892151,
      "learning_rate": 0.00020600334848122458,
      "loss": 0.0971,
      "step": 7055
    },
    {
      "epoch": 0.9889278206026629,
      "grad_norm": 0.3004218637943268,
      "learning_rate": 0.0002059889978474049,
      "loss": 0.12,
      "step": 7056
    },
    {
      "epoch": 0.9890679747722495,
      "grad_norm": 0.33169451355934143,
      "learning_rate": 0.00020597464721358526,
      "loss": 0.0886,
      "step": 7057
    },
    {
      "epoch": 0.9892081289418361,
      "grad_norm": 0.2871354818344116,
      "learning_rate": 0.0002059602965797656,
      "loss": 0.0313,
      "step": 7058
    },
    {
      "epoch": 0.9893482831114225,
      "grad_norm": 0.9618695378303528,
      "learning_rate": 0.00020594594594594592,
      "loss": 0.0979,
      "step": 7059
    },
    {
      "epoch": 0.9894884372810091,
      "grad_norm": 0.25487926602363586,
      "learning_rate": 0.00020593159531212625,
      "loss": 0.0388,
      "step": 7060
    },
    {
      "epoch": 0.9896285914505957,
      "grad_norm": 0.2643638551235199,
      "learning_rate": 0.0002059172446783066,
      "loss": 0.0427,
      "step": 7061
    },
    {
      "epoch": 0.9897687456201822,
      "grad_norm": 0.22066166996955872,
      "learning_rate": 0.00020590289404448696,
      "loss": 0.0677,
      "step": 7062
    },
    {
      "epoch": 0.9899088997897687,
      "grad_norm": 0.10679704695940018,
      "learning_rate": 0.00020588854341066729,
      "loss": 0.0187,
      "step": 7063
    },
    {
      "epoch": 0.9900490539593553,
      "grad_norm": 0.2386174350976944,
      "learning_rate": 0.00020587419277684764,
      "loss": 0.0562,
      "step": 7064
    },
    {
      "epoch": 0.9901892081289418,
      "grad_norm": 0.44058892130851746,
      "learning_rate": 0.00020585984214302797,
      "loss": 0.0825,
      "step": 7065
    },
    {
      "epoch": 0.9903293622985284,
      "grad_norm": 0.7175392508506775,
      "learning_rate": 0.0002058454915092083,
      "loss": 0.0641,
      "step": 7066
    },
    {
      "epoch": 0.9904695164681149,
      "grad_norm": 0.5369006991386414,
      "learning_rate": 0.00020583114087538865,
      "loss": 0.1013,
      "step": 7067
    },
    {
      "epoch": 0.9906096706377014,
      "grad_norm": 0.14437179267406464,
      "learning_rate": 0.00020581679024156898,
      "loss": 0.0151,
      "step": 7068
    },
    {
      "epoch": 0.990749824807288,
      "grad_norm": 0.3580746352672577,
      "learning_rate": 0.0002058024396077493,
      "loss": 0.0425,
      "step": 7069
    },
    {
      "epoch": 0.9908899789768746,
      "grad_norm": 0.20528078079223633,
      "learning_rate": 0.00020578808897392967,
      "loss": 0.028,
      "step": 7070
    },
    {
      "epoch": 0.9910301331464612,
      "grad_norm": 0.27812063694000244,
      "learning_rate": 0.00020577373834011,
      "loss": 0.0691,
      "step": 7071
    },
    {
      "epoch": 0.9911702873160476,
      "grad_norm": 0.2630431652069092,
      "learning_rate": 0.00020575938770629032,
      "loss": 0.0486,
      "step": 7072
    },
    {
      "epoch": 0.9913104414856342,
      "grad_norm": 0.20029127597808838,
      "learning_rate": 0.0002057450370724707,
      "loss": 0.0328,
      "step": 7073
    },
    {
      "epoch": 0.9914505956552208,
      "grad_norm": 0.8080753087997437,
      "learning_rate": 0.00020573068643865104,
      "loss": 0.1783,
      "step": 7074
    },
    {
      "epoch": 0.9915907498248073,
      "grad_norm": 0.3186112940311432,
      "learning_rate": 0.00020571633580483136,
      "loss": 0.0359,
      "step": 7075
    },
    {
      "epoch": 0.9917309039943938,
      "grad_norm": 0.18976891040802002,
      "learning_rate": 0.0002057019851710117,
      "loss": 0.0334,
      "step": 7076
    },
    {
      "epoch": 0.9918710581639804,
      "grad_norm": 0.3050951659679413,
      "learning_rate": 0.00020568763453719205,
      "loss": 0.0769,
      "step": 7077
    },
    {
      "epoch": 0.9920112123335669,
      "grad_norm": 0.42824047803878784,
      "learning_rate": 0.00020567328390337238,
      "loss": 0.0829,
      "step": 7078
    },
    {
      "epoch": 0.9921513665031535,
      "grad_norm": 0.6803753972053528,
      "learning_rate": 0.0002056589332695527,
      "loss": 0.1113,
      "step": 7079
    },
    {
      "epoch": 0.9922915206727401,
      "grad_norm": 0.38260748982429504,
      "learning_rate": 0.00020564458263573306,
      "loss": 0.1025,
      "step": 7080
    },
    {
      "epoch": 0.9924316748423265,
      "grad_norm": 0.20176684856414795,
      "learning_rate": 0.0002056302320019134,
      "loss": 0.0237,
      "step": 7081
    },
    {
      "epoch": 0.9925718290119131,
      "grad_norm": 0.3585328757762909,
      "learning_rate": 0.00020561588136809372,
      "loss": 0.0389,
      "step": 7082
    },
    {
      "epoch": 0.9927119831814997,
      "grad_norm": 0.5209757089614868,
      "learning_rate": 0.0002056015307342741,
      "loss": 0.1041,
      "step": 7083
    },
    {
      "epoch": 0.9928521373510862,
      "grad_norm": 0.45334312319755554,
      "learning_rate": 0.00020558718010045443,
      "loss": 0.0695,
      "step": 7084
    },
    {
      "epoch": 0.9929922915206727,
      "grad_norm": 0.5760122537612915,
      "learning_rate": 0.00020557282946663476,
      "loss": 0.1058,
      "step": 7085
    },
    {
      "epoch": 0.9931324456902593,
      "grad_norm": 0.5242316722869873,
      "learning_rate": 0.0002055584788328151,
      "loss": 0.0604,
      "step": 7086
    },
    {
      "epoch": 0.9932725998598458,
      "grad_norm": 0.382184773683548,
      "learning_rate": 0.00020554412819899544,
      "loss": 0.103,
      "step": 7087
    },
    {
      "epoch": 0.9934127540294324,
      "grad_norm": 0.6734559535980225,
      "learning_rate": 0.00020552977756517577,
      "loss": 0.049,
      "step": 7088
    },
    {
      "epoch": 0.9935529081990189,
      "grad_norm": 0.2894585430622101,
      "learning_rate": 0.00020551542693135613,
      "loss": 0.0576,
      "step": 7089
    },
    {
      "epoch": 0.9936930623686054,
      "grad_norm": 0.6706838607788086,
      "learning_rate": 0.00020550107629753645,
      "loss": 0.1325,
      "step": 7090
    },
    {
      "epoch": 0.993833216538192,
      "grad_norm": 0.33583569526672363,
      "learning_rate": 0.00020548672566371678,
      "loss": 0.0576,
      "step": 7091
    },
    {
      "epoch": 0.9939733707077786,
      "grad_norm": 0.3077923357486725,
      "learning_rate": 0.00020547237502989714,
      "loss": 0.0593,
      "step": 7092
    },
    {
      "epoch": 0.9941135248773652,
      "grad_norm": 0.18344588577747345,
      "learning_rate": 0.00020545802439607747,
      "loss": 0.0261,
      "step": 7093
    },
    {
      "epoch": 0.9942536790469516,
      "grad_norm": 0.7783629894256592,
      "learning_rate": 0.00020544367376225782,
      "loss": 0.069,
      "step": 7094
    },
    {
      "epoch": 0.9943938332165382,
      "grad_norm": 0.23166191577911377,
      "learning_rate": 0.00020542932312843815,
      "loss": 0.0495,
      "step": 7095
    },
    {
      "epoch": 0.9945339873861248,
      "grad_norm": 0.4915904402732849,
      "learning_rate": 0.0002054149724946185,
      "loss": 0.1026,
      "step": 7096
    },
    {
      "epoch": 0.9946741415557113,
      "grad_norm": 0.2845335900783539,
      "learning_rate": 0.00020540062186079884,
      "loss": 0.032,
      "step": 7097
    },
    {
      "epoch": 0.9948142957252978,
      "grad_norm": 1.0460699796676636,
      "learning_rate": 0.00020538627122697916,
      "loss": 0.1769,
      "step": 7098
    },
    {
      "epoch": 0.9949544498948844,
      "grad_norm": 1.540544867515564,
      "learning_rate": 0.00020537192059315952,
      "loss": 0.1013,
      "step": 7099
    },
    {
      "epoch": 0.9950946040644709,
      "grad_norm": 2.2109038829803467,
      "learning_rate": 0.00020535756995933985,
      "loss": 0.0651,
      "step": 7100
    },
    {
      "epoch": 0.9952347582340575,
      "grad_norm": 0.474007785320282,
      "learning_rate": 0.00020534321932552018,
      "loss": 0.0673,
      "step": 7101
    },
    {
      "epoch": 0.9953749124036441,
      "grad_norm": 0.5390128493309021,
      "learning_rate": 0.00020532886869170053,
      "loss": 0.1015,
      "step": 7102
    },
    {
      "epoch": 0.9955150665732305,
      "grad_norm": 0.3510839343070984,
      "learning_rate": 0.00020531451805788086,
      "loss": 0.0503,
      "step": 7103
    },
    {
      "epoch": 0.9956552207428171,
      "grad_norm": 0.755711555480957,
      "learning_rate": 0.0002053001674240612,
      "loss": 0.0858,
      "step": 7104
    },
    {
      "epoch": 0.9957953749124037,
      "grad_norm": 0.8366712927818298,
      "learning_rate": 0.00020528581679024157,
      "loss": 0.1422,
      "step": 7105
    },
    {
      "epoch": 0.9959355290819902,
      "grad_norm": 0.5402758717536926,
      "learning_rate": 0.0002052714661564219,
      "loss": 0.0571,
      "step": 7106
    },
    {
      "epoch": 0.9960756832515767,
      "grad_norm": 0.2994421720504761,
      "learning_rate": 0.00020525711552260223,
      "loss": 0.0333,
      "step": 7107
    },
    {
      "epoch": 0.9962158374211633,
      "grad_norm": 0.31326714158058167,
      "learning_rate": 0.00020524276488878259,
      "loss": 0.0439,
      "step": 7108
    },
    {
      "epoch": 0.9963559915907498,
      "grad_norm": 0.6625999808311462,
      "learning_rate": 0.00020522841425496291,
      "loss": 0.0824,
      "step": 7109
    },
    {
      "epoch": 0.9964961457603364,
      "grad_norm": 0.3427512049674988,
      "learning_rate": 0.00020521406362114324,
      "loss": 0.0581,
      "step": 7110
    },
    {
      "epoch": 0.9966362999299229,
      "grad_norm": 0.48263347148895264,
      "learning_rate": 0.00020519971298732357,
      "loss": 0.0888,
      "step": 7111
    },
    {
      "epoch": 0.9967764540995094,
      "grad_norm": 0.2981701195240021,
      "learning_rate": 0.00020518536235350393,
      "loss": 0.099,
      "step": 7112
    },
    {
      "epoch": 0.996916608269096,
      "grad_norm": 0.4356768727302551,
      "learning_rate": 0.00020517101171968426,
      "loss": 0.0602,
      "step": 7113
    },
    {
      "epoch": 0.9970567624386826,
      "grad_norm": 0.46916916966438293,
      "learning_rate": 0.00020515666108586458,
      "loss": 0.0668,
      "step": 7114
    },
    {
      "epoch": 0.9971969166082691,
      "grad_norm": 0.36121895909309387,
      "learning_rate": 0.00020514231045204497,
      "loss": 0.0555,
      "step": 7115
    },
    {
      "epoch": 0.9973370707778556,
      "grad_norm": 0.38666602969169617,
      "learning_rate": 0.0002051279598182253,
      "loss": 0.0372,
      "step": 7116
    },
    {
      "epoch": 0.9974772249474422,
      "grad_norm": 0.7229727506637573,
      "learning_rate": 0.00020511360918440562,
      "loss": 0.1263,
      "step": 7117
    },
    {
      "epoch": 0.9976173791170287,
      "grad_norm": 0.4332045316696167,
      "learning_rate": 0.00020509925855058598,
      "loss": 0.0761,
      "step": 7118
    },
    {
      "epoch": 0.9977575332866153,
      "grad_norm": 0.3635750412940979,
      "learning_rate": 0.0002050849079167663,
      "loss": 0.0424,
      "step": 7119
    },
    {
      "epoch": 0.9978976874562018,
      "grad_norm": 0.36419880390167236,
      "learning_rate": 0.00020507055728294664,
      "loss": 0.0807,
      "step": 7120
    },
    {
      "epoch": 0.9980378416257883,
      "grad_norm": 0.3582659065723419,
      "learning_rate": 0.000205056206649127,
      "loss": 0.0689,
      "step": 7121
    },
    {
      "epoch": 0.9981779957953749,
      "grad_norm": 0.14960364997386932,
      "learning_rate": 0.00020504185601530732,
      "loss": 0.0314,
      "step": 7122
    },
    {
      "epoch": 0.9983181499649615,
      "grad_norm": 0.4342552721500397,
      "learning_rate": 0.00020502750538148765,
      "loss": 0.046,
      "step": 7123
    },
    {
      "epoch": 0.9984583041345481,
      "grad_norm": 0.37765437364578247,
      "learning_rate": 0.000205013154747668,
      "loss": 0.1352,
      "step": 7124
    },
    {
      "epoch": 0.9985984583041345,
      "grad_norm": 0.8114736676216125,
      "learning_rate": 0.00020499880411384833,
      "loss": 0.1503,
      "step": 7125
    },
    {
      "epoch": 0.9987386124737211,
      "grad_norm": 0.24331362545490265,
      "learning_rate": 0.0002049844534800287,
      "loss": 0.0537,
      "step": 7126
    },
    {
      "epoch": 0.9988787666433077,
      "grad_norm": 0.3869306147098541,
      "learning_rate": 0.00020497010284620905,
      "loss": 0.1128,
      "step": 7127
    },
    {
      "epoch": 0.9990189208128942,
      "grad_norm": 0.7088903784751892,
      "learning_rate": 0.00020495575221238937,
      "loss": 0.1281,
      "step": 7128
    },
    {
      "epoch": 0.9991590749824807,
      "grad_norm": 0.6007456183433533,
      "learning_rate": 0.0002049414015785697,
      "loss": 0.1023,
      "step": 7129
    },
    {
      "epoch": 0.9992992291520673,
      "grad_norm": 0.3305444121360779,
      "learning_rate": 0.00020492705094475003,
      "loss": 0.0294,
      "step": 7130
    },
    {
      "epoch": 0.9994393833216538,
      "grad_norm": 0.21787963807582855,
      "learning_rate": 0.00020491270031093039,
      "loss": 0.0501,
      "step": 7131
    },
    {
      "epoch": 0.9995795374912404,
      "grad_norm": 0.3425024151802063,
      "learning_rate": 0.00020489834967711071,
      "loss": 0.0938,
      "step": 7132
    },
    {
      "epoch": 0.9997196916608269,
      "grad_norm": 1.214779019355774,
      "learning_rate": 0.00020488399904329104,
      "loss": 0.1061,
      "step": 7133
    },
    {
      "epoch": 0.9998598458304134,
      "grad_norm": 0.9670930504798889,
      "learning_rate": 0.0002048696484094714,
      "loss": 0.121,
      "step": 7134
    },
    {
      "epoch": 1.0,
      "grad_norm": 2.462139368057251,
      "learning_rate": 0.00020485529777565173,
      "loss": 0.1329,
      "step": 7135
    },
    {
      "epoch": 1.0001401541695865,
      "grad_norm": 0.2860074043273926,
      "learning_rate": 0.00020484094714183206,
      "loss": 0.067,
      "step": 7136
    },
    {
      "epoch": 1.0002803083391731,
      "grad_norm": 0.5792491436004639,
      "learning_rate": 0.00020482659650801244,
      "loss": 0.1233,
      "step": 7137
    },
    {
      "epoch": 1.0004204625087596,
      "grad_norm": 0.23094545304775238,
      "learning_rate": 0.00020481224587419277,
      "loss": 0.0709,
      "step": 7138
    },
    {
      "epoch": 1.0005606166783463,
      "grad_norm": 0.14148405194282532,
      "learning_rate": 0.0002047978952403731,
      "loss": 0.0276,
      "step": 7139
    },
    {
      "epoch": 1.0007007708479327,
      "grad_norm": 0.20996734499931335,
      "learning_rate": 0.00020478354460655345,
      "loss": 0.0311,
      "step": 7140
    },
    {
      "epoch": 1.0008409250175192,
      "grad_norm": 0.24814346432685852,
      "learning_rate": 0.00020476919397273378,
      "loss": 0.0463,
      "step": 7141
    },
    {
      "epoch": 1.0009810791871059,
      "grad_norm": 0.27665844559669495,
      "learning_rate": 0.0002047548433389141,
      "loss": 0.0838,
      "step": 7142
    },
    {
      "epoch": 1.0011212333566923,
      "grad_norm": 0.3728223145008087,
      "learning_rate": 0.00020474049270509446,
      "loss": 0.0543,
      "step": 7143
    },
    {
      "epoch": 1.0012613875262788,
      "grad_norm": 0.29025015234947205,
      "learning_rate": 0.0002047261420712748,
      "loss": 0.0834,
      "step": 7144
    },
    {
      "epoch": 1.0014015416958655,
      "grad_norm": 0.26596614718437195,
      "learning_rate": 0.00020471179143745512,
      "loss": 0.0405,
      "step": 7145
    },
    {
      "epoch": 1.001541695865452,
      "grad_norm": 0.22693288326263428,
      "learning_rate": 0.00020469744080363545,
      "loss": 0.0321,
      "step": 7146
    },
    {
      "epoch": 1.0016818500350386,
      "grad_norm": 0.46044325828552246,
      "learning_rate": 0.00020468309016981583,
      "loss": 0.0746,
      "step": 7147
    },
    {
      "epoch": 1.001822004204625,
      "grad_norm": 0.28197091817855835,
      "learning_rate": 0.00020466873953599616,
      "loss": 0.0357,
      "step": 7148
    },
    {
      "epoch": 1.0019621583742115,
      "grad_norm": 0.13461196422576904,
      "learning_rate": 0.0002046543889021765,
      "loss": 0.0276,
      "step": 7149
    },
    {
      "epoch": 1.0021023125437982,
      "grad_norm": 0.33018240332603455,
      "learning_rate": 0.00020464003826835685,
      "loss": 0.0626,
      "step": 7150
    },
    {
      "epoch": 1.0022424667133847,
      "grad_norm": 1.112799882888794,
      "learning_rate": 0.00020462568763453717,
      "loss": 0.0505,
      "step": 7151
    },
    {
      "epoch": 1.0023826208829714,
      "grad_norm": 0.3565979301929474,
      "learning_rate": 0.0002046113370007175,
      "loss": 0.0413,
      "step": 7152
    },
    {
      "epoch": 1.0025227750525578,
      "grad_norm": 0.40753549337387085,
      "learning_rate": 0.00020459698636689786,
      "loss": 0.0382,
      "step": 7153
    },
    {
      "epoch": 1.0026629292221443,
      "grad_norm": 0.5862252116203308,
      "learning_rate": 0.0002045826357330782,
      "loss": 0.1047,
      "step": 7154
    },
    {
      "epoch": 1.002803083391731,
      "grad_norm": 0.2554514408111572,
      "learning_rate": 0.00020456828509925852,
      "loss": 0.051,
      "step": 7155
    },
    {
      "epoch": 1.0029432375613174,
      "grad_norm": 0.5357496738433838,
      "learning_rate": 0.00020455393446543887,
      "loss": 0.0814,
      "step": 7156
    },
    {
      "epoch": 1.0030833917309039,
      "grad_norm": 0.2656848728656769,
      "learning_rate": 0.00020453958383161923,
      "loss": 0.0194,
      "step": 7157
    },
    {
      "epoch": 1.0032235459004906,
      "grad_norm": 0.5183130502700806,
      "learning_rate": 0.00020452523319779956,
      "loss": 0.0368,
      "step": 7158
    },
    {
      "epoch": 1.003363700070077,
      "grad_norm": 0.3824182152748108,
      "learning_rate": 0.0002045108825639799,
      "loss": 0.1042,
      "step": 7159
    },
    {
      "epoch": 1.0035038542396637,
      "grad_norm": 0.5910611748695374,
      "learning_rate": 0.00020449653193016024,
      "loss": 0.0571,
      "step": 7160
    },
    {
      "epoch": 1.0036440084092502,
      "grad_norm": 0.3872907757759094,
      "learning_rate": 0.00020448218129634057,
      "loss": 0.1088,
      "step": 7161
    },
    {
      "epoch": 1.0037841625788366,
      "grad_norm": 0.6107721924781799,
      "learning_rate": 0.00020446783066252092,
      "loss": 0.0857,
      "step": 7162
    },
    {
      "epoch": 1.0039243167484233,
      "grad_norm": 0.42903822660446167,
      "learning_rate": 0.00020445348002870125,
      "loss": 0.074,
      "step": 7163
    },
    {
      "epoch": 1.0040644709180098,
      "grad_norm": 0.4483715891838074,
      "learning_rate": 0.00020443912939488158,
      "loss": 0.0566,
      "step": 7164
    },
    {
      "epoch": 1.0042046250875964,
      "grad_norm": 0.25098010897636414,
      "learning_rate": 0.0002044247787610619,
      "loss": 0.0272,
      "step": 7165
    },
    {
      "epoch": 1.004344779257183,
      "grad_norm": 1.1176609992980957,
      "learning_rate": 0.00020441042812724227,
      "loss": 0.0824,
      "step": 7166
    },
    {
      "epoch": 1.0044849334267694,
      "grad_norm": 0.49698013067245483,
      "learning_rate": 0.0002043960774934226,
      "loss": 0.0643,
      "step": 7167
    },
    {
      "epoch": 1.004625087596356,
      "grad_norm": 0.4594554901123047,
      "learning_rate": 0.00020438172685960292,
      "loss": 0.0838,
      "step": 7168
    },
    {
      "epoch": 1.0047652417659425,
      "grad_norm": 0.41447314620018005,
      "learning_rate": 0.0002043673762257833,
      "loss": 0.0737,
      "step": 7169
    },
    {
      "epoch": 1.0049053959355292,
      "grad_norm": 0.46276921033859253,
      "learning_rate": 0.00020435302559196363,
      "loss": 0.0711,
      "step": 7170
    },
    {
      "epoch": 1.0050455501051156,
      "grad_norm": 0.5921447277069092,
      "learning_rate": 0.00020433867495814396,
      "loss": 0.0717,
      "step": 7171
    },
    {
      "epoch": 1.005185704274702,
      "grad_norm": 0.3937142491340637,
      "learning_rate": 0.00020432432432432432,
      "loss": 0.0452,
      "step": 7172
    },
    {
      "epoch": 1.0053258584442888,
      "grad_norm": 0.11307530105113983,
      "learning_rate": 0.00020430997369050465,
      "loss": 0.0168,
      "step": 7173
    },
    {
      "epoch": 1.0054660126138752,
      "grad_norm": 0.7994346618652344,
      "learning_rate": 0.00020429562305668498,
      "loss": 0.0775,
      "step": 7174
    },
    {
      "epoch": 1.0056061667834617,
      "grad_norm": 0.4385305643081665,
      "learning_rate": 0.00020428127242286533,
      "loss": 0.115,
      "step": 7175
    },
    {
      "epoch": 1.0057463209530484,
      "grad_norm": 0.3255448043346405,
      "learning_rate": 0.00020426692178904566,
      "loss": 0.0223,
      "step": 7176
    },
    {
      "epoch": 1.0058864751226348,
      "grad_norm": 0.5701867341995239,
      "learning_rate": 0.000204252571155226,
      "loss": 0.0666,
      "step": 7177
    },
    {
      "epoch": 1.0060266292922215,
      "grad_norm": 0.6423361897468567,
      "learning_rate": 0.00020423822052140637,
      "loss": 0.1026,
      "step": 7178
    },
    {
      "epoch": 1.006166783461808,
      "grad_norm": 0.38494694232940674,
      "learning_rate": 0.0002042238698875867,
      "loss": 0.0334,
      "step": 7179
    },
    {
      "epoch": 1.0063069376313944,
      "grad_norm": 0.6565921902656555,
      "learning_rate": 0.00020420951925376703,
      "loss": 0.0849,
      "step": 7180
    },
    {
      "epoch": 1.0064470918009811,
      "grad_norm": 0.43672704696655273,
      "learning_rate": 0.00020419516861994736,
      "loss": 0.1401,
      "step": 7181
    },
    {
      "epoch": 1.0065872459705676,
      "grad_norm": 0.4526292681694031,
      "learning_rate": 0.0002041808179861277,
      "loss": 0.0385,
      "step": 7182
    },
    {
      "epoch": 1.0067274001401543,
      "grad_norm": 0.22365699708461761,
      "learning_rate": 0.00020416646735230804,
      "loss": 0.037,
      "step": 7183
    },
    {
      "epoch": 1.0068675543097407,
      "grad_norm": 1.9282917976379395,
      "learning_rate": 0.00020415211671848837,
      "loss": 0.1219,
      "step": 7184
    },
    {
      "epoch": 1.0070077084793272,
      "grad_norm": 2.3473336696624756,
      "learning_rate": 0.00020413776608466872,
      "loss": 0.3394,
      "step": 7185
    },
    {
      "epoch": 1.0071478626489139,
      "grad_norm": 0.1598985493183136,
      "learning_rate": 0.00020412341545084905,
      "loss": 0.0296,
      "step": 7186
    },
    {
      "epoch": 1.0072880168185003,
      "grad_norm": 0.2726248800754547,
      "learning_rate": 0.00020410906481702938,
      "loss": 0.0383,
      "step": 7187
    },
    {
      "epoch": 1.0074281709880868,
      "grad_norm": 0.18712691962718964,
      "learning_rate": 0.00020409471418320974,
      "loss": 0.0425,
      "step": 7188
    },
    {
      "epoch": 1.0075683251576735,
      "grad_norm": 0.4840964674949646,
      "learning_rate": 0.0002040803635493901,
      "loss": 0.0896,
      "step": 7189
    },
    {
      "epoch": 1.00770847932726,
      "grad_norm": 0.3585018217563629,
      "learning_rate": 0.00020406601291557042,
      "loss": 0.0662,
      "step": 7190
    },
    {
      "epoch": 1.0078486334968466,
      "grad_norm": 0.3046760559082031,
      "learning_rate": 0.00020405166228175078,
      "loss": 0.0494,
      "step": 7191
    },
    {
      "epoch": 1.007988787666433,
      "grad_norm": 0.2689339816570282,
      "learning_rate": 0.0002040373116479311,
      "loss": 0.0578,
      "step": 7192
    },
    {
      "epoch": 1.0081289418360195,
      "grad_norm": 0.47273269295692444,
      "learning_rate": 0.00020402296101411143,
      "loss": 0.124,
      "step": 7193
    },
    {
      "epoch": 1.0082690960056062,
      "grad_norm": 0.6697574853897095,
      "learning_rate": 0.0002040086103802918,
      "loss": 0.0386,
      "step": 7194
    },
    {
      "epoch": 1.0084092501751927,
      "grad_norm": 0.5874694585800171,
      "learning_rate": 0.00020399425974647212,
      "loss": 0.109,
      "step": 7195
    },
    {
      "epoch": 1.0085494043447794,
      "grad_norm": 0.2584046423435211,
      "learning_rate": 0.00020397990911265245,
      "loss": 0.0254,
      "step": 7196
    },
    {
      "epoch": 1.0086895585143658,
      "grad_norm": 0.2864406704902649,
      "learning_rate": 0.0002039655584788328,
      "loss": 0.0323,
      "step": 7197
    },
    {
      "epoch": 1.0088297126839523,
      "grad_norm": 0.20431135594844818,
      "learning_rate": 0.00020395120784501313,
      "loss": 0.0292,
      "step": 7198
    },
    {
      "epoch": 1.008969866853539,
      "grad_norm": 0.5439653992652893,
      "learning_rate": 0.00020393685721119346,
      "loss": 0.057,
      "step": 7199
    },
    {
      "epoch": 1.0091100210231254,
      "grad_norm": 0.29200273752212524,
      "learning_rate": 0.0002039225065773738,
      "loss": 0.0341,
      "step": 7200
    },
    {
      "epoch": 1.0092501751927119,
      "grad_norm": 0.31157270073890686,
      "learning_rate": 0.00020390815594355417,
      "loss": 0.0485,
      "step": 7201
    },
    {
      "epoch": 1.0093903293622986,
      "grad_norm": 0.17556287348270416,
      "learning_rate": 0.0002038938053097345,
      "loss": 0.0326,
      "step": 7202
    },
    {
      "epoch": 1.009530483531885,
      "grad_norm": 0.510563850402832,
      "learning_rate": 0.00020387945467591483,
      "loss": 0.039,
      "step": 7203
    },
    {
      "epoch": 1.0096706377014717,
      "grad_norm": 0.3272097706794739,
      "learning_rate": 0.00020386510404209518,
      "loss": 0.0279,
      "step": 7204
    },
    {
      "epoch": 1.0098107918710582,
      "grad_norm": 0.694811999797821,
      "learning_rate": 0.0002038507534082755,
      "loss": 0.1093,
      "step": 7205
    },
    {
      "epoch": 1.0099509460406446,
      "grad_norm": 0.16517947614192963,
      "learning_rate": 0.00020383640277445584,
      "loss": 0.0274,
      "step": 7206
    },
    {
      "epoch": 1.0100911002102313,
      "grad_norm": 0.24342204630374908,
      "learning_rate": 0.0002038220521406362,
      "loss": 0.0234,
      "step": 7207
    },
    {
      "epoch": 1.0102312543798178,
      "grad_norm": 0.2100415825843811,
      "learning_rate": 0.00020380770150681653,
      "loss": 0.0424,
      "step": 7208
    },
    {
      "epoch": 1.0103714085494044,
      "grad_norm": 0.2612631916999817,
      "learning_rate": 0.00020379335087299685,
      "loss": 0.0428,
      "step": 7209
    },
    {
      "epoch": 1.010511562718991,
      "grad_norm": 0.15079949796199799,
      "learning_rate": 0.00020377900023917724,
      "loss": 0.0111,
      "step": 7210
    },
    {
      "epoch": 1.0106517168885774,
      "grad_norm": 0.3396081030368805,
      "learning_rate": 0.00020376464960535757,
      "loss": 0.0989,
      "step": 7211
    },
    {
      "epoch": 1.010791871058164,
      "grad_norm": 0.4676767587661743,
      "learning_rate": 0.0002037502989715379,
      "loss": 0.0604,
      "step": 7212
    },
    {
      "epoch": 1.0109320252277505,
      "grad_norm": 0.24942511320114136,
      "learning_rate": 0.00020373594833771825,
      "loss": 0.056,
      "step": 7213
    },
    {
      "epoch": 1.0110721793973372,
      "grad_norm": 0.40445375442504883,
      "learning_rate": 0.00020372159770389858,
      "loss": 0.1008,
      "step": 7214
    },
    {
      "epoch": 1.0112123335669236,
      "grad_norm": 0.3608073890209198,
      "learning_rate": 0.0002037072470700789,
      "loss": 0.0372,
      "step": 7215
    },
    {
      "epoch": 1.01135248773651,
      "grad_norm": 0.15398520231246948,
      "learning_rate": 0.00020369289643625924,
      "loss": 0.0171,
      "step": 7216
    },
    {
      "epoch": 1.0114926419060968,
      "grad_norm": 0.45473670959472656,
      "learning_rate": 0.0002036785458024396,
      "loss": 0.0668,
      "step": 7217
    },
    {
      "epoch": 1.0116327960756832,
      "grad_norm": 0.5309837460517883,
      "learning_rate": 0.00020366419516861992,
      "loss": 0.0651,
      "step": 7218
    },
    {
      "epoch": 1.0117729502452697,
      "grad_norm": 0.7329338192939758,
      "learning_rate": 0.00020364984453480025,
      "loss": 0.0932,
      "step": 7219
    },
    {
      "epoch": 1.0119131044148564,
      "grad_norm": 0.36260029673576355,
      "learning_rate": 0.0002036354939009806,
      "loss": 0.0656,
      "step": 7220
    },
    {
      "epoch": 1.0120532585844428,
      "grad_norm": 0.35101327300071716,
      "learning_rate": 0.00020362114326716096,
      "loss": 0.0616,
      "step": 7221
    },
    {
      "epoch": 1.0121934127540295,
      "grad_norm": 0.43613189458847046,
      "learning_rate": 0.0002036067926333413,
      "loss": 0.0398,
      "step": 7222
    },
    {
      "epoch": 1.012333566923616,
      "grad_norm": 0.478753924369812,
      "learning_rate": 0.00020359244199952164,
      "loss": 0.0605,
      "step": 7223
    },
    {
      "epoch": 1.0124737210932024,
      "grad_norm": 0.07942412048578262,
      "learning_rate": 0.00020357809136570197,
      "loss": 0.0104,
      "step": 7224
    },
    {
      "epoch": 1.0126138752627891,
      "grad_norm": 0.5202214121818542,
      "learning_rate": 0.0002035637407318823,
      "loss": 0.0941,
      "step": 7225
    },
    {
      "epoch": 1.0127540294323756,
      "grad_norm": 0.27747607231140137,
      "learning_rate": 0.00020354939009806266,
      "loss": 0.0459,
      "step": 7226
    },
    {
      "epoch": 1.0128941836019623,
      "grad_norm": 0.27870842814445496,
      "learning_rate": 0.00020353503946424299,
      "loss": 0.0239,
      "step": 7227
    },
    {
      "epoch": 1.0130343377715487,
      "grad_norm": 0.30336645245552063,
      "learning_rate": 0.00020352068883042331,
      "loss": 0.01,
      "step": 7228
    },
    {
      "epoch": 1.0131744919411352,
      "grad_norm": 0.8620482683181763,
      "learning_rate": 0.00020350633819660367,
      "loss": 0.0779,
      "step": 7229
    },
    {
      "epoch": 1.0133146461107219,
      "grad_norm": 0.48967793583869934,
      "learning_rate": 0.000203491987562784,
      "loss": 0.0746,
      "step": 7230
    },
    {
      "epoch": 1.0134548002803083,
      "grad_norm": 0.22333772480487823,
      "learning_rate": 0.00020347763692896433,
      "loss": 0.0359,
      "step": 7231
    },
    {
      "epoch": 1.0135949544498948,
      "grad_norm": 0.35928311944007874,
      "learning_rate": 0.0002034632862951447,
      "loss": 0.045,
      "step": 7232
    },
    {
      "epoch": 1.0137351086194815,
      "grad_norm": 1.6389645338058472,
      "learning_rate": 0.00020344893566132504,
      "loss": 0.3991,
      "step": 7233
    },
    {
      "epoch": 1.013875262789068,
      "grad_norm": 1.3229844570159912,
      "learning_rate": 0.00020343458502750537,
      "loss": 0.0578,
      "step": 7234
    },
    {
      "epoch": 1.0140154169586546,
      "grad_norm": 0.017571115866303444,
      "learning_rate": 0.0002034202343936857,
      "loss": 0.002,
      "step": 7235
    },
    {
      "epoch": 1.014155571128241,
      "grad_norm": 0.42118310928344727,
      "learning_rate": 0.00020340588375986605,
      "loss": 0.0795,
      "step": 7236
    },
    {
      "epoch": 1.0142957252978275,
      "grad_norm": 0.326107919216156,
      "learning_rate": 0.00020339153312604638,
      "loss": 0.1243,
      "step": 7237
    },
    {
      "epoch": 1.0144358794674142,
      "grad_norm": 0.3536466956138611,
      "learning_rate": 0.0002033771824922267,
      "loss": 0.0651,
      "step": 7238
    },
    {
      "epoch": 1.0145760336370007,
      "grad_norm": 0.5534151792526245,
      "learning_rate": 0.00020336283185840706,
      "loss": 0.0845,
      "step": 7239
    },
    {
      "epoch": 1.0147161878065873,
      "grad_norm": 0.2486460953950882,
      "learning_rate": 0.0002033484812245874,
      "loss": 0.0369,
      "step": 7240
    },
    {
      "epoch": 1.0148563419761738,
      "grad_norm": 0.44978901743888855,
      "learning_rate": 0.00020333413059076772,
      "loss": 0.0351,
      "step": 7241
    },
    {
      "epoch": 1.0149964961457603,
      "grad_norm": 0.29931503534317017,
      "learning_rate": 0.0002033197799569481,
      "loss": 0.0464,
      "step": 7242
    },
    {
      "epoch": 1.015136650315347,
      "grad_norm": 0.36510804295539856,
      "learning_rate": 0.00020330542932312843,
      "loss": 0.0792,
      "step": 7243
    },
    {
      "epoch": 1.0152768044849334,
      "grad_norm": 0.4880961775779724,
      "learning_rate": 0.00020329107868930876,
      "loss": 0.0824,
      "step": 7244
    },
    {
      "epoch": 1.0154169586545199,
      "grad_norm": 0.16730909049510956,
      "learning_rate": 0.00020327672805548912,
      "loss": 0.031,
      "step": 7245
    },
    {
      "epoch": 1.0155571128241065,
      "grad_norm": 0.2527766227722168,
      "learning_rate": 0.00020326237742166944,
      "loss": 0.0623,
      "step": 7246
    },
    {
      "epoch": 1.015697266993693,
      "grad_norm": 0.4247931241989136,
      "learning_rate": 0.00020324802678784977,
      "loss": 0.0325,
      "step": 7247
    },
    {
      "epoch": 1.0158374211632797,
      "grad_norm": 0.17874020338058472,
      "learning_rate": 0.00020323367615403013,
      "loss": 0.0741,
      "step": 7248
    },
    {
      "epoch": 1.0159775753328661,
      "grad_norm": 0.22037763893604279,
      "learning_rate": 0.00020321932552021046,
      "loss": 0.0345,
      "step": 7249
    },
    {
      "epoch": 1.0161177295024526,
      "grad_norm": 0.18538211286067963,
      "learning_rate": 0.00020320497488639079,
      "loss": 0.0505,
      "step": 7250
    },
    {
      "epoch": 1.0162578836720393,
      "grad_norm": 0.32006552815437317,
      "learning_rate": 0.00020319062425257111,
      "loss": 0.0355,
      "step": 7251
    },
    {
      "epoch": 1.0163980378416257,
      "grad_norm": 0.20736652612686157,
      "learning_rate": 0.0002031762736187515,
      "loss": 0.0389,
      "step": 7252
    },
    {
      "epoch": 1.0165381920112124,
      "grad_norm": 0.23939959704875946,
      "learning_rate": 0.00020316192298493183,
      "loss": 0.0317,
      "step": 7253
    },
    {
      "epoch": 1.0166783461807989,
      "grad_norm": 0.4261023998260498,
      "learning_rate": 0.00020314757235111215,
      "loss": 0.0491,
      "step": 7254
    },
    {
      "epoch": 1.0168185003503853,
      "grad_norm": 0.320877343416214,
      "learning_rate": 0.0002031332217172925,
      "loss": 0.0254,
      "step": 7255
    },
    {
      "epoch": 1.016958654519972,
      "grad_norm": 0.4134514033794403,
      "learning_rate": 0.00020311887108347284,
      "loss": 0.0541,
      "step": 7256
    },
    {
      "epoch": 1.0170988086895585,
      "grad_norm": 0.1331988424062729,
      "learning_rate": 0.00020310452044965317,
      "loss": 0.0114,
      "step": 7257
    },
    {
      "epoch": 1.0172389628591452,
      "grad_norm": 0.31431692838668823,
      "learning_rate": 0.00020309016981583352,
      "loss": 0.0419,
      "step": 7258
    },
    {
      "epoch": 1.0173791170287316,
      "grad_norm": 0.45364242792129517,
      "learning_rate": 0.00020307581918201385,
      "loss": 0.0526,
      "step": 7259
    },
    {
      "epoch": 1.017519271198318,
      "grad_norm": 0.32927805185317993,
      "learning_rate": 0.00020306146854819418,
      "loss": 0.0374,
      "step": 7260
    },
    {
      "epoch": 1.0176594253679048,
      "grad_norm": 0.23745539784431458,
      "learning_rate": 0.00020304711791437454,
      "loss": 0.0373,
      "step": 7261
    },
    {
      "epoch": 1.0177995795374912,
      "grad_norm": 0.23946061730384827,
      "learning_rate": 0.00020303276728055486,
      "loss": 0.0396,
      "step": 7262
    },
    {
      "epoch": 1.0179397337070777,
      "grad_norm": 0.2508072555065155,
      "learning_rate": 0.0002030184166467352,
      "loss": 0.023,
      "step": 7263
    },
    {
      "epoch": 1.0180798878766644,
      "grad_norm": 0.25330471992492676,
      "learning_rate": 0.00020300406601291558,
      "loss": 0.0226,
      "step": 7264
    },
    {
      "epoch": 1.0182200420462508,
      "grad_norm": 0.6070547103881836,
      "learning_rate": 0.0002029897153790959,
      "loss": 0.053,
      "step": 7265
    },
    {
      "epoch": 1.0183601962158375,
      "grad_norm": 0.4620213508605957,
      "learning_rate": 0.00020297536474527623,
      "loss": 0.1009,
      "step": 7266
    },
    {
      "epoch": 1.018500350385424,
      "grad_norm": 1.4101574420928955,
      "learning_rate": 0.0002029610141114566,
      "loss": 0.0963,
      "step": 7267
    },
    {
      "epoch": 1.0186405045550104,
      "grad_norm": 0.28356462717056274,
      "learning_rate": 0.00020294666347763692,
      "loss": 0.033,
      "step": 7268
    },
    {
      "epoch": 1.018780658724597,
      "grad_norm": 0.09166912734508514,
      "learning_rate": 0.00020293231284381725,
      "loss": 0.0133,
      "step": 7269
    },
    {
      "epoch": 1.0189208128941836,
      "grad_norm": 0.44415149092674255,
      "learning_rate": 0.00020291796220999757,
      "loss": 0.074,
      "step": 7270
    },
    {
      "epoch": 1.0190609670637703,
      "grad_norm": 0.6255483031272888,
      "learning_rate": 0.00020290361157617793,
      "loss": 0.1408,
      "step": 7271
    },
    {
      "epoch": 1.0192011212333567,
      "grad_norm": 0.6920514702796936,
      "learning_rate": 0.00020288926094235826,
      "loss": 0.0303,
      "step": 7272
    },
    {
      "epoch": 1.0193412754029432,
      "grad_norm": 0.35557299852371216,
      "learning_rate": 0.0002028749103085386,
      "loss": 0.0479,
      "step": 7273
    },
    {
      "epoch": 1.0194814295725299,
      "grad_norm": 0.3282272219657898,
      "learning_rate": 0.00020286055967471897,
      "loss": 0.0594,
      "step": 7274
    },
    {
      "epoch": 1.0196215837421163,
      "grad_norm": 0.33518266677856445,
      "learning_rate": 0.0002028462090408993,
      "loss": 0.0269,
      "step": 7275
    },
    {
      "epoch": 1.0197617379117028,
      "grad_norm": 0.24703074991703033,
      "learning_rate": 0.00020283185840707963,
      "loss": 0.039,
      "step": 7276
    },
    {
      "epoch": 1.0199018920812895,
      "grad_norm": 0.48286372423171997,
      "learning_rate": 0.00020281750777325998,
      "loss": 0.1049,
      "step": 7277
    },
    {
      "epoch": 1.020042046250876,
      "grad_norm": 0.6484137773513794,
      "learning_rate": 0.0002028031571394403,
      "loss": 0.05,
      "step": 7278
    },
    {
      "epoch": 1.0201822004204626,
      "grad_norm": 0.8242290616035461,
      "learning_rate": 0.00020278880650562064,
      "loss": 0.0527,
      "step": 7279
    },
    {
      "epoch": 1.020322354590049,
      "grad_norm": 0.30259793996810913,
      "learning_rate": 0.000202774455871801,
      "loss": 0.007,
      "step": 7280
    },
    {
      "epoch": 1.0204625087596355,
      "grad_norm": 0.6221671104431152,
      "learning_rate": 0.00020276010523798132,
      "loss": 0.0438,
      "step": 7281
    },
    {
      "epoch": 1.0206026629292222,
      "grad_norm": 0.5225541591644287,
      "learning_rate": 0.00020274575460416165,
      "loss": 0.0313,
      "step": 7282
    },
    {
      "epoch": 1.0207428170988087,
      "grad_norm": 1.2069557905197144,
      "learning_rate": 0.000202731403970342,
      "loss": 0.3316,
      "step": 7283
    },
    {
      "epoch": 1.0208829712683953,
      "grad_norm": 1.5490013360977173,
      "learning_rate": 0.00020271705333652236,
      "loss": 0.0932,
      "step": 7284
    },
    {
      "epoch": 1.0210231254379818,
      "grad_norm": 2.697406053543091,
      "learning_rate": 0.0002027027027027027,
      "loss": 0.3199,
      "step": 7285
    },
    {
      "epoch": 1.0211632796075683,
      "grad_norm": 0.6773514151573181,
      "learning_rate": 0.00020268835206888302,
      "loss": 0.098,
      "step": 7286
    },
    {
      "epoch": 1.021303433777155,
      "grad_norm": 0.2825588881969452,
      "learning_rate": 0.00020267400143506338,
      "loss": 0.055,
      "step": 7287
    },
    {
      "epoch": 1.0214435879467414,
      "grad_norm": 0.2414739429950714,
      "learning_rate": 0.0002026596508012437,
      "loss": 0.0772,
      "step": 7288
    },
    {
      "epoch": 1.0215837421163279,
      "grad_norm": 0.30025947093963623,
      "learning_rate": 0.00020264530016742403,
      "loss": 0.1154,
      "step": 7289
    },
    {
      "epoch": 1.0217238962859145,
      "grad_norm": 0.2881932854652405,
      "learning_rate": 0.0002026309495336044,
      "loss": 0.0444,
      "step": 7290
    },
    {
      "epoch": 1.021864050455501,
      "grad_norm": 0.2947738766670227,
      "learning_rate": 0.00020261659889978472,
      "loss": 0.0352,
      "step": 7291
    },
    {
      "epoch": 1.0220042046250877,
      "grad_norm": 0.30118581652641296,
      "learning_rate": 0.00020260224826596505,
      "loss": 0.0903,
      "step": 7292
    },
    {
      "epoch": 1.0221443587946741,
      "grad_norm": 0.13735143840312958,
      "learning_rate": 0.0002025878976321454,
      "loss": 0.0329,
      "step": 7293
    },
    {
      "epoch": 1.0222845129642606,
      "grad_norm": 0.31322962045669556,
      "learning_rate": 0.00020257354699832573,
      "loss": 0.0957,
      "step": 7294
    },
    {
      "epoch": 1.0224246671338473,
      "grad_norm": 0.4684290885925293,
      "learning_rate": 0.00020255919636450606,
      "loss": 0.1097,
      "step": 7295
    },
    {
      "epoch": 1.0225648213034337,
      "grad_norm": 0.37178564071655273,
      "learning_rate": 0.00020254484573068644,
      "loss": 0.0806,
      "step": 7296
    },
    {
      "epoch": 1.0227049754730204,
      "grad_norm": 0.3892574906349182,
      "learning_rate": 0.00020253049509686677,
      "loss": 0.066,
      "step": 7297
    },
    {
      "epoch": 1.0228451296426069,
      "grad_norm": 0.3241652846336365,
      "learning_rate": 0.0002025161444630471,
      "loss": 0.0801,
      "step": 7298
    },
    {
      "epoch": 1.0229852838121933,
      "grad_norm": 0.17515529692173004,
      "learning_rate": 0.00020250179382922745,
      "loss": 0.0252,
      "step": 7299
    },
    {
      "epoch": 1.02312543798178,
      "grad_norm": 0.32863670587539673,
      "learning_rate": 0.00020248744319540778,
      "loss": 0.0683,
      "step": 7300
    },
    {
      "epoch": 1.0232655921513665,
      "grad_norm": 0.3024722635746002,
      "learning_rate": 0.0002024730925615881,
      "loss": 0.0206,
      "step": 7301
    },
    {
      "epoch": 1.023405746320953,
      "grad_norm": 0.1940188854932785,
      "learning_rate": 0.00020245874192776847,
      "loss": 0.052,
      "step": 7302
    },
    {
      "epoch": 1.0235459004905396,
      "grad_norm": 0.2718174159526825,
      "learning_rate": 0.0002024443912939488,
      "loss": 0.0584,
      "step": 7303
    },
    {
      "epoch": 1.023686054660126,
      "grad_norm": 0.6966903209686279,
      "learning_rate": 0.00020243004066012912,
      "loss": 0.0565,
      "step": 7304
    },
    {
      "epoch": 1.0238262088297128,
      "grad_norm": 0.45669588446617126,
      "learning_rate": 0.00020241569002630945,
      "loss": 0.0405,
      "step": 7305
    },
    {
      "epoch": 1.0239663629992992,
      "grad_norm": 0.4384971857070923,
      "learning_rate": 0.00020240133939248984,
      "loss": 0.1242,
      "step": 7306
    },
    {
      "epoch": 1.0241065171688857,
      "grad_norm": 0.2861323654651642,
      "learning_rate": 0.00020238698875867016,
      "loss": 0.0825,
      "step": 7307
    },
    {
      "epoch": 1.0242466713384724,
      "grad_norm": 0.29005348682403564,
      "learning_rate": 0.0002023726381248505,
      "loss": 0.0257,
      "step": 7308
    },
    {
      "epoch": 1.0243868255080588,
      "grad_norm": 0.5171800851821899,
      "learning_rate": 0.00020235828749103085,
      "loss": 0.0315,
      "step": 7309
    },
    {
      "epoch": 1.0245269796776455,
      "grad_norm": 0.29901498556137085,
      "learning_rate": 0.00020234393685721118,
      "loss": 0.0795,
      "step": 7310
    },
    {
      "epoch": 1.024667133847232,
      "grad_norm": 0.3842817544937134,
      "learning_rate": 0.0002023295862233915,
      "loss": 0.071,
      "step": 7311
    },
    {
      "epoch": 1.0248072880168184,
      "grad_norm": 0.3723081648349762,
      "learning_rate": 0.00020231523558957186,
      "loss": 0.0676,
      "step": 7312
    },
    {
      "epoch": 1.024947442186405,
      "grad_norm": 0.4442616105079651,
      "learning_rate": 0.0002023008849557522,
      "loss": 0.0299,
      "step": 7313
    },
    {
      "epoch": 1.0250875963559916,
      "grad_norm": 0.22426271438598633,
      "learning_rate": 0.00020228653432193252,
      "loss": 0.0996,
      "step": 7314
    },
    {
      "epoch": 1.0252277505255782,
      "grad_norm": 0.5538390874862671,
      "learning_rate": 0.00020227218368811287,
      "loss": 0.073,
      "step": 7315
    },
    {
      "epoch": 1.0253679046951647,
      "grad_norm": 0.29517579078674316,
      "learning_rate": 0.00020225783305429323,
      "loss": 0.0324,
      "step": 7316
    },
    {
      "epoch": 1.0255080588647512,
      "grad_norm": 0.09712222218513489,
      "learning_rate": 0.00020224348242047356,
      "loss": 0.0083,
      "step": 7317
    },
    {
      "epoch": 1.0256482130343378,
      "grad_norm": 0.25485730171203613,
      "learning_rate": 0.00020222913178665391,
      "loss": 0.0437,
      "step": 7318
    },
    {
      "epoch": 1.0257883672039243,
      "grad_norm": 0.34867939352989197,
      "learning_rate": 0.00020221478115283424,
      "loss": 0.0956,
      "step": 7319
    },
    {
      "epoch": 1.0259285213735108,
      "grad_norm": 0.1582653522491455,
      "learning_rate": 0.00020220043051901457,
      "loss": 0.028,
      "step": 7320
    },
    {
      "epoch": 1.0260686755430974,
      "grad_norm": 0.6382333040237427,
      "learning_rate": 0.0002021860798851949,
      "loss": 0.1231,
      "step": 7321
    },
    {
      "epoch": 1.026208829712684,
      "grad_norm": 0.21647104620933533,
      "learning_rate": 0.00020217172925137526,
      "loss": 0.0448,
      "step": 7322
    },
    {
      "epoch": 1.0263489838822706,
      "grad_norm": 0.2099693864583969,
      "learning_rate": 0.00020215737861755558,
      "loss": 0.0378,
      "step": 7323
    },
    {
      "epoch": 1.026489138051857,
      "grad_norm": 0.09777022153139114,
      "learning_rate": 0.0002021430279837359,
      "loss": 0.01,
      "step": 7324
    },
    {
      "epoch": 1.0266292922214435,
      "grad_norm": 0.40246543288230896,
      "learning_rate": 0.00020212867734991627,
      "loss": 0.0796,
      "step": 7325
    },
    {
      "epoch": 1.0267694463910302,
      "grad_norm": 0.6252508163452148,
      "learning_rate": 0.0002021143267160966,
      "loss": 0.0522,
      "step": 7326
    },
    {
      "epoch": 1.0269096005606166,
      "grad_norm": 0.4237343370914459,
      "learning_rate": 0.00020209997608227693,
      "loss": 0.0805,
      "step": 7327
    },
    {
      "epoch": 1.0270497547302033,
      "grad_norm": 0.3577063977718353,
      "learning_rate": 0.0002020856254484573,
      "loss": 0.0744,
      "step": 7328
    },
    {
      "epoch": 1.0271899088997898,
      "grad_norm": 0.33011043071746826,
      "learning_rate": 0.00020207127481463764,
      "loss": 0.0332,
      "step": 7329
    },
    {
      "epoch": 1.0273300630693762,
      "grad_norm": 0.24332870543003082,
      "learning_rate": 0.00020205692418081797,
      "loss": 0.1144,
      "step": 7330
    },
    {
      "epoch": 1.027470217238963,
      "grad_norm": 0.35399290919303894,
      "learning_rate": 0.00020204257354699832,
      "loss": 0.0396,
      "step": 7331
    },
    {
      "epoch": 1.0276103714085494,
      "grad_norm": 0.8833919763565063,
      "learning_rate": 0.00020202822291317865,
      "loss": 0.2596,
      "step": 7332
    },
    {
      "epoch": 1.0277505255781358,
      "grad_norm": 0.24139781296253204,
      "learning_rate": 0.00020201387227935898,
      "loss": 0.0489,
      "step": 7333
    },
    {
      "epoch": 1.0278906797477225,
      "grad_norm": 0.7945934534072876,
      "learning_rate": 0.00020199952164553933,
      "loss": 0.1718,
      "step": 7334
    },
    {
      "epoch": 1.028030833917309,
      "grad_norm": 0.1805168241262436,
      "learning_rate": 0.00020198517101171966,
      "loss": 0.0098,
      "step": 7335
    },
    {
      "epoch": 1.0281709880868957,
      "grad_norm": 0.36715954542160034,
      "learning_rate": 0.0002019708203779,
      "loss": 0.0773,
      "step": 7336
    },
    {
      "epoch": 1.0283111422564821,
      "grad_norm": 0.3491787016391754,
      "learning_rate": 0.00020195646974408032,
      "loss": 0.0584,
      "step": 7337
    },
    {
      "epoch": 1.0284512964260686,
      "grad_norm": 0.31847110390663147,
      "learning_rate": 0.0002019421191102607,
      "loss": 0.0277,
      "step": 7338
    },
    {
      "epoch": 1.0285914505956553,
      "grad_norm": 0.28226909041404724,
      "learning_rate": 0.00020192776847644103,
      "loss": 0.0372,
      "step": 7339
    },
    {
      "epoch": 1.0287316047652417,
      "grad_norm": 0.16133855283260345,
      "learning_rate": 0.00020191341784262136,
      "loss": 0.0221,
      "step": 7340
    },
    {
      "epoch": 1.0288717589348284,
      "grad_norm": 0.5553686022758484,
      "learning_rate": 0.00020189906720880172,
      "loss": 0.0458,
      "step": 7341
    },
    {
      "epoch": 1.0290119131044149,
      "grad_norm": 0.1868080198764801,
      "learning_rate": 0.00020188471657498204,
      "loss": 0.042,
      "step": 7342
    },
    {
      "epoch": 1.0291520672740013,
      "grad_norm": 0.21590116620063782,
      "learning_rate": 0.00020187036594116237,
      "loss": 0.0342,
      "step": 7343
    },
    {
      "epoch": 1.029292221443588,
      "grad_norm": 0.24588674306869507,
      "learning_rate": 0.00020185601530734273,
      "loss": 0.0461,
      "step": 7344
    },
    {
      "epoch": 1.0294323756131745,
      "grad_norm": 0.16774991154670715,
      "learning_rate": 0.00020184166467352306,
      "loss": 0.0422,
      "step": 7345
    },
    {
      "epoch": 1.0295725297827611,
      "grad_norm": 0.7636731863021851,
      "learning_rate": 0.00020182731403970338,
      "loss": 0.0929,
      "step": 7346
    },
    {
      "epoch": 1.0297126839523476,
      "grad_norm": 0.2616930305957794,
      "learning_rate": 0.00020181296340588374,
      "loss": 0.0664,
      "step": 7347
    },
    {
      "epoch": 1.029852838121934,
      "grad_norm": 0.2908239960670471,
      "learning_rate": 0.0002017986127720641,
      "loss": 0.0414,
      "step": 7348
    },
    {
      "epoch": 1.0299929922915207,
      "grad_norm": 0.33808261156082153,
      "learning_rate": 0.00020178426213824442,
      "loss": 0.0454,
      "step": 7349
    },
    {
      "epoch": 1.0301331464611072,
      "grad_norm": 0.18225722014904022,
      "learning_rate": 0.00020176991150442478,
      "loss": 0.0377,
      "step": 7350
    },
    {
      "epoch": 1.0302733006306937,
      "grad_norm": 0.48944491147994995,
      "learning_rate": 0.0002017555608706051,
      "loss": 0.0623,
      "step": 7351
    },
    {
      "epoch": 1.0304134548002803,
      "grad_norm": 0.5488277673721313,
      "learning_rate": 0.00020174121023678544,
      "loss": 0.0939,
      "step": 7352
    },
    {
      "epoch": 1.0305536089698668,
      "grad_norm": 0.24379444122314453,
      "learning_rate": 0.0002017268596029658,
      "loss": 0.0596,
      "step": 7353
    },
    {
      "epoch": 1.0306937631394535,
      "grad_norm": 0.1473238617181778,
      "learning_rate": 0.00020171250896914612,
      "loss": 0.0227,
      "step": 7354
    },
    {
      "epoch": 1.03083391730904,
      "grad_norm": 0.5108892321586609,
      "learning_rate": 0.00020169815833532645,
      "loss": 0.0812,
      "step": 7355
    },
    {
      "epoch": 1.0309740714786264,
      "grad_norm": 0.7706103324890137,
      "learning_rate": 0.00020168380770150678,
      "loss": 0.0395,
      "step": 7356
    },
    {
      "epoch": 1.031114225648213,
      "grad_norm": 0.1537996381521225,
      "learning_rate": 0.00020166945706768713,
      "loss": 0.0344,
      "step": 7357
    },
    {
      "epoch": 1.0312543798177995,
      "grad_norm": 0.4848909378051758,
      "learning_rate": 0.00020165510643386746,
      "loss": 0.0572,
      "step": 7358
    },
    {
      "epoch": 1.0313945339873862,
      "grad_norm": 0.14417120814323425,
      "learning_rate": 0.0002016407558000478,
      "loss": 0.014,
      "step": 7359
    },
    {
      "epoch": 1.0315346881569727,
      "grad_norm": 0.16109015047550201,
      "learning_rate": 0.00020162640516622817,
      "loss": 0.0184,
      "step": 7360
    },
    {
      "epoch": 1.0316748423265591,
      "grad_norm": 0.39149004220962524,
      "learning_rate": 0.0002016120545324085,
      "loss": 0.0537,
      "step": 7361
    },
    {
      "epoch": 1.0318149964961458,
      "grad_norm": 0.47483861446380615,
      "learning_rate": 0.00020159770389858883,
      "loss": 0.0803,
      "step": 7362
    },
    {
      "epoch": 1.0319551506657323,
      "grad_norm": 1.0141165256500244,
      "learning_rate": 0.0002015833532647692,
      "loss": 0.0359,
      "step": 7363
    },
    {
      "epoch": 1.0320953048353187,
      "grad_norm": 0.19929462671279907,
      "learning_rate": 0.00020156900263094952,
      "loss": 0.019,
      "step": 7364
    },
    {
      "epoch": 1.0322354590049054,
      "grad_norm": 0.24317246675491333,
      "learning_rate": 0.00020155465199712984,
      "loss": 0.0098,
      "step": 7365
    },
    {
      "epoch": 1.032375613174492,
      "grad_norm": 0.222087100148201,
      "learning_rate": 0.0002015403013633102,
      "loss": 0.0527,
      "step": 7366
    },
    {
      "epoch": 1.0325157673440786,
      "grad_norm": 0.2788246273994446,
      "learning_rate": 0.00020152595072949053,
      "loss": 0.0519,
      "step": 7367
    },
    {
      "epoch": 1.032655921513665,
      "grad_norm": 0.585625410079956,
      "learning_rate": 0.00020151160009567086,
      "loss": 0.0424,
      "step": 7368
    },
    {
      "epoch": 1.0327960756832515,
      "grad_norm": 0.24283930659294128,
      "learning_rate": 0.00020149724946185124,
      "loss": 0.0479,
      "step": 7369
    },
    {
      "epoch": 1.0329362298528382,
      "grad_norm": 0.10255005955696106,
      "learning_rate": 0.00020148289882803157,
      "loss": 0.0116,
      "step": 7370
    },
    {
      "epoch": 1.0330763840224246,
      "grad_norm": 0.30152273178100586,
      "learning_rate": 0.0002014685481942119,
      "loss": 0.0322,
      "step": 7371
    },
    {
      "epoch": 1.0332165381920113,
      "grad_norm": 0.20395119488239288,
      "learning_rate": 0.00020145419756039223,
      "loss": 0.0074,
      "step": 7372
    },
    {
      "epoch": 1.0333566923615978,
      "grad_norm": 0.15507175028324127,
      "learning_rate": 0.00020143984692657258,
      "loss": 0.052,
      "step": 7373
    },
    {
      "epoch": 1.0334968465311842,
      "grad_norm": 0.280568927526474,
      "learning_rate": 0.0002014254962927529,
      "loss": 0.0194,
      "step": 7374
    },
    {
      "epoch": 1.033637000700771,
      "grad_norm": 0.3317860960960388,
      "learning_rate": 0.00020141114565893324,
      "loss": 0.0712,
      "step": 7375
    },
    {
      "epoch": 1.0337771548703574,
      "grad_norm": 0.8176124691963196,
      "learning_rate": 0.0002013967950251136,
      "loss": 0.0963,
      "step": 7376
    },
    {
      "epoch": 1.0339173090399438,
      "grad_norm": 0.567148745059967,
      "learning_rate": 0.00020138244439129392,
      "loss": 0.0518,
      "step": 7377
    },
    {
      "epoch": 1.0340574632095305,
      "grad_norm": 1.0496882200241089,
      "learning_rate": 0.00020136809375747425,
      "loss": 0.1916,
      "step": 7378
    },
    {
      "epoch": 1.034197617379117,
      "grad_norm": 0.4502559304237366,
      "learning_rate": 0.00020135374312365463,
      "loss": 0.0674,
      "step": 7379
    },
    {
      "epoch": 1.0343377715487037,
      "grad_norm": 0.3167615234851837,
      "learning_rate": 0.00020133939248983496,
      "loss": 0.0424,
      "step": 7380
    },
    {
      "epoch": 1.0344779257182901,
      "grad_norm": 0.6294723749160767,
      "learning_rate": 0.0002013250418560153,
      "loss": 0.1329,
      "step": 7381
    },
    {
      "epoch": 1.0346180798878766,
      "grad_norm": 0.5803903937339783,
      "learning_rate": 0.00020131069122219565,
      "loss": 0.1641,
      "step": 7382
    },
    {
      "epoch": 1.0347582340574633,
      "grad_norm": 0.2547217905521393,
      "learning_rate": 0.00020129634058837598,
      "loss": 0.0563,
      "step": 7383
    },
    {
      "epoch": 1.0348983882270497,
      "grad_norm": 1.5837944746017456,
      "learning_rate": 0.0002012819899545563,
      "loss": 0.1091,
      "step": 7384
    },
    {
      "epoch": 1.0350385423966364,
      "grad_norm": 0.710404098033905,
      "learning_rate": 0.00020126763932073666,
      "loss": 0.0234,
      "step": 7385
    },
    {
      "epoch": 1.0351786965662229,
      "grad_norm": 0.17221421003341675,
      "learning_rate": 0.000201253288686917,
      "loss": 0.036,
      "step": 7386
    },
    {
      "epoch": 1.0353188507358093,
      "grad_norm": 0.25372564792633057,
      "learning_rate": 0.00020123893805309732,
      "loss": 0.0753,
      "step": 7387
    },
    {
      "epoch": 1.035459004905396,
      "grad_norm": 0.23701223731040955,
      "learning_rate": 0.00020122458741927767,
      "loss": 0.0665,
      "step": 7388
    },
    {
      "epoch": 1.0355991590749825,
      "grad_norm": 0.6436792016029358,
      "learning_rate": 0.000201210236785458,
      "loss": 0.0431,
      "step": 7389
    },
    {
      "epoch": 1.035739313244569,
      "grad_norm": 0.5409477949142456,
      "learning_rate": 0.00020119588615163833,
      "loss": 0.0562,
      "step": 7390
    },
    {
      "epoch": 1.0358794674141556,
      "grad_norm": 0.30708518624305725,
      "learning_rate": 0.00020118153551781866,
      "loss": 0.0699,
      "step": 7391
    },
    {
      "epoch": 1.036019621583742,
      "grad_norm": 0.6640605926513672,
      "learning_rate": 0.00020116718488399904,
      "loss": 0.0452,
      "step": 7392
    },
    {
      "epoch": 1.0361597757533287,
      "grad_norm": 0.2592013478279114,
      "learning_rate": 0.00020115283425017937,
      "loss": 0.0622,
      "step": 7393
    },
    {
      "epoch": 1.0362999299229152,
      "grad_norm": 0.24011889100074768,
      "learning_rate": 0.0002011384836163597,
      "loss": 0.0615,
      "step": 7394
    },
    {
      "epoch": 1.0364400840925017,
      "grad_norm": 0.5006332993507385,
      "learning_rate": 0.00020112413298254005,
      "loss": 0.0984,
      "step": 7395
    },
    {
      "epoch": 1.0365802382620883,
      "grad_norm": 0.3818579614162445,
      "learning_rate": 0.00020110978234872038,
      "loss": 0.035,
      "step": 7396
    },
    {
      "epoch": 1.0367203924316748,
      "grad_norm": 0.1948472559452057,
      "learning_rate": 0.0002010954317149007,
      "loss": 0.0164,
      "step": 7397
    },
    {
      "epoch": 1.0368605466012615,
      "grad_norm": 0.13681766390800476,
      "learning_rate": 0.00020108108108108107,
      "loss": 0.0277,
      "step": 7398
    },
    {
      "epoch": 1.037000700770848,
      "grad_norm": 0.8447391986846924,
      "learning_rate": 0.0002010667304472614,
      "loss": 0.1062,
      "step": 7399
    },
    {
      "epoch": 1.0371408549404344,
      "grad_norm": 0.27277272939682007,
      "learning_rate": 0.00020105237981344172,
      "loss": 0.0447,
      "step": 7400
    },
    {
      "epoch": 1.037281009110021,
      "grad_norm": 0.3555588126182556,
      "learning_rate": 0.0002010380291796221,
      "loss": 0.0409,
      "step": 7401
    },
    {
      "epoch": 1.0374211632796075,
      "grad_norm": 0.3866293430328369,
      "learning_rate": 0.00020102367854580243,
      "loss": 0.0483,
      "step": 7402
    },
    {
      "epoch": 1.0375613174491942,
      "grad_norm": 0.5190638899803162,
      "learning_rate": 0.00020100932791198276,
      "loss": 0.1141,
      "step": 7403
    },
    {
      "epoch": 1.0377014716187807,
      "grad_norm": 0.22311319410800934,
      "learning_rate": 0.00020099497727816312,
      "loss": 0.0418,
      "step": 7404
    },
    {
      "epoch": 1.0378416257883671,
      "grad_norm": 0.27752485871315,
      "learning_rate": 0.00020098062664434345,
      "loss": 0.0393,
      "step": 7405
    },
    {
      "epoch": 1.0379817799579538,
      "grad_norm": 0.4159800112247467,
      "learning_rate": 0.00020096627601052378,
      "loss": 0.0438,
      "step": 7406
    },
    {
      "epoch": 1.0381219341275403,
      "grad_norm": 0.4560358226299286,
      "learning_rate": 0.0002009519253767041,
      "loss": 0.0693,
      "step": 7407
    },
    {
      "epoch": 1.0382620882971267,
      "grad_norm": 0.287677139043808,
      "learning_rate": 0.00020093757474288446,
      "loss": 0.0357,
      "step": 7408
    },
    {
      "epoch": 1.0384022424667134,
      "grad_norm": 0.4302176833152771,
      "learning_rate": 0.0002009232241090648,
      "loss": 0.0584,
      "step": 7409
    },
    {
      "epoch": 1.0385423966362999,
      "grad_norm": 0.2120569497346878,
      "learning_rate": 0.00020090887347524512,
      "loss": 0.0721,
      "step": 7410
    },
    {
      "epoch": 1.0386825508058866,
      "grad_norm": 0.4541984498500824,
      "learning_rate": 0.0002008945228414255,
      "loss": 0.0481,
      "step": 7411
    },
    {
      "epoch": 1.038822704975473,
      "grad_norm": 0.5624955296516418,
      "learning_rate": 0.00020088017220760583,
      "loss": 0.0586,
      "step": 7412
    },
    {
      "epoch": 1.0389628591450595,
      "grad_norm": 0.3735843896865845,
      "learning_rate": 0.00020086582157378616,
      "loss": 0.0749,
      "step": 7413
    },
    {
      "epoch": 1.0391030133146462,
      "grad_norm": 0.2377900779247284,
      "learning_rate": 0.0002008514709399665,
      "loss": 0.0463,
      "step": 7414
    },
    {
      "epoch": 1.0392431674842326,
      "grad_norm": 0.3517459034919739,
      "learning_rate": 0.00020083712030614684,
      "loss": 0.031,
      "step": 7415
    },
    {
      "epoch": 1.0393833216538193,
      "grad_norm": 0.3057696521282196,
      "learning_rate": 0.00020082276967232717,
      "loss": 0.0302,
      "step": 7416
    },
    {
      "epoch": 1.0395234758234058,
      "grad_norm": 0.2454061508178711,
      "learning_rate": 0.00020080841903850753,
      "loss": 0.0635,
      "step": 7417
    },
    {
      "epoch": 1.0396636299929922,
      "grad_norm": 0.15972313284873962,
      "learning_rate": 0.00020079406840468785,
      "loss": 0.041,
      "step": 7418
    },
    {
      "epoch": 1.039803784162579,
      "grad_norm": 0.6719050407409668,
      "learning_rate": 0.00020077971777086818,
      "loss": 0.0479,
      "step": 7419
    },
    {
      "epoch": 1.0399439383321654,
      "grad_norm": 0.14499905705451965,
      "learning_rate": 0.00020076536713704854,
      "loss": 0.0142,
      "step": 7420
    },
    {
      "epoch": 1.0400840925017518,
      "grad_norm": 0.50604248046875,
      "learning_rate": 0.00020075101650322887,
      "loss": 0.1007,
      "step": 7421
    },
    {
      "epoch": 1.0402242466713385,
      "grad_norm": 0.11758874356746674,
      "learning_rate": 0.0002007366658694092,
      "loss": 0.0109,
      "step": 7422
    },
    {
      "epoch": 1.040364400840925,
      "grad_norm": 0.40569648146629333,
      "learning_rate": 0.00020072231523558958,
      "loss": 0.0868,
      "step": 7423
    },
    {
      "epoch": 1.0405045550105116,
      "grad_norm": 0.6281216144561768,
      "learning_rate": 0.0002007079646017699,
      "loss": 0.0727,
      "step": 7424
    },
    {
      "epoch": 1.040644709180098,
      "grad_norm": 0.7099785804748535,
      "learning_rate": 0.00020069361396795024,
      "loss": 0.0389,
      "step": 7425
    },
    {
      "epoch": 1.0407848633496846,
      "grad_norm": 0.23996074497699738,
      "learning_rate": 0.00020067926333413056,
      "loss": 0.0507,
      "step": 7426
    },
    {
      "epoch": 1.0409250175192712,
      "grad_norm": 0.8768213987350464,
      "learning_rate": 0.00020066491270031092,
      "loss": 0.0368,
      "step": 7427
    },
    {
      "epoch": 1.0410651716888577,
      "grad_norm": 0.3884669244289398,
      "learning_rate": 0.00020065056206649125,
      "loss": 0.0371,
      "step": 7428
    },
    {
      "epoch": 1.0412053258584444,
      "grad_norm": 0.20927593111991882,
      "learning_rate": 0.00020063621143267158,
      "loss": 0.0215,
      "step": 7429
    },
    {
      "epoch": 1.0413454800280308,
      "grad_norm": 1.5984781980514526,
      "learning_rate": 0.00020062186079885193,
      "loss": 0.1222,
      "step": 7430
    },
    {
      "epoch": 1.0414856341976173,
      "grad_norm": 0.46438372135162354,
      "learning_rate": 0.00020060751016503226,
      "loss": 0.0608,
      "step": 7431
    },
    {
      "epoch": 1.041625788367204,
      "grad_norm": 1.4514050483703613,
      "learning_rate": 0.0002005931595312126,
      "loss": 0.3426,
      "step": 7432
    },
    {
      "epoch": 1.0417659425367904,
      "grad_norm": 0.8924177885055542,
      "learning_rate": 0.00020057880889739297,
      "loss": 0.1756,
      "step": 7433
    },
    {
      "epoch": 1.0419060967063771,
      "grad_norm": 0.2706003189086914,
      "learning_rate": 0.0002005644582635733,
      "loss": 0.0963,
      "step": 7434
    },
    {
      "epoch": 1.0420462508759636,
      "grad_norm": 0.4359434247016907,
      "learning_rate": 0.00020055010762975363,
      "loss": 0.0721,
      "step": 7435
    },
    {
      "epoch": 1.04218640504555,
      "grad_norm": 0.3326599895954132,
      "learning_rate": 0.00020053575699593399,
      "loss": 0.062,
      "step": 7436
    },
    {
      "epoch": 1.0423265592151367,
      "grad_norm": 0.3369787931442261,
      "learning_rate": 0.00020052140636211431,
      "loss": 0.057,
      "step": 7437
    },
    {
      "epoch": 1.0424667133847232,
      "grad_norm": 0.3704361021518707,
      "learning_rate": 0.00020050705572829464,
      "loss": 0.0996,
      "step": 7438
    },
    {
      "epoch": 1.0426068675543096,
      "grad_norm": 0.41580623388290405,
      "learning_rate": 0.000200492705094475,
      "loss": 0.0671,
      "step": 7439
    },
    {
      "epoch": 1.0427470217238963,
      "grad_norm": 0.1372150480747223,
      "learning_rate": 0.00020047835446065533,
      "loss": 0.0289,
      "step": 7440
    },
    {
      "epoch": 1.0428871758934828,
      "grad_norm": 0.26816439628601074,
      "learning_rate": 0.00020046400382683566,
      "loss": 0.0315,
      "step": 7441
    },
    {
      "epoch": 1.0430273300630695,
      "grad_norm": 1.0764726400375366,
      "learning_rate": 0.00020044965319301598,
      "loss": 0.0418,
      "step": 7442
    },
    {
      "epoch": 1.043167484232656,
      "grad_norm": 0.1662338525056839,
      "learning_rate": 0.00020043530255919637,
      "loss": 0.0161,
      "step": 7443
    },
    {
      "epoch": 1.0433076384022424,
      "grad_norm": 0.4620538651943207,
      "learning_rate": 0.0002004209519253767,
      "loss": 0.0826,
      "step": 7444
    },
    {
      "epoch": 1.043447792571829,
      "grad_norm": 0.3773845136165619,
      "learning_rate": 0.00020040660129155702,
      "loss": 0.0507,
      "step": 7445
    },
    {
      "epoch": 1.0435879467414155,
      "grad_norm": 0.20116060972213745,
      "learning_rate": 0.00020039225065773738,
      "loss": 0.0558,
      "step": 7446
    },
    {
      "epoch": 1.043728100911002,
      "grad_norm": 0.5692968964576721,
      "learning_rate": 0.0002003779000239177,
      "loss": 0.0814,
      "step": 7447
    },
    {
      "epoch": 1.0438682550805887,
      "grad_norm": 0.44019076228141785,
      "learning_rate": 0.00020036354939009804,
      "loss": 0.0509,
      "step": 7448
    },
    {
      "epoch": 1.0440084092501751,
      "grad_norm": 0.9188428521156311,
      "learning_rate": 0.0002003491987562784,
      "loss": 0.1231,
      "step": 7449
    },
    {
      "epoch": 1.0441485634197618,
      "grad_norm": 0.4651030898094177,
      "learning_rate": 0.00020033484812245872,
      "loss": 0.1281,
      "step": 7450
    },
    {
      "epoch": 1.0442887175893483,
      "grad_norm": 0.3135788142681122,
      "learning_rate": 0.00020032049748863905,
      "loss": 0.0541,
      "step": 7451
    },
    {
      "epoch": 1.0444288717589347,
      "grad_norm": 0.7074897289276123,
      "learning_rate": 0.0002003061468548194,
      "loss": 0.0506,
      "step": 7452
    },
    {
      "epoch": 1.0445690259285214,
      "grad_norm": 0.625097930431366,
      "learning_rate": 0.00020029179622099973,
      "loss": 0.044,
      "step": 7453
    },
    {
      "epoch": 1.0447091800981079,
      "grad_norm": 0.18113072216510773,
      "learning_rate": 0.00020027744558718006,
      "loss": 0.0126,
      "step": 7454
    },
    {
      "epoch": 1.0448493342676946,
      "grad_norm": 0.20055146515369415,
      "learning_rate": 0.00020026309495336044,
      "loss": 0.0405,
      "step": 7455
    },
    {
      "epoch": 1.044989488437281,
      "grad_norm": 0.3123377859592438,
      "learning_rate": 0.00020024874431954077,
      "loss": 0.0427,
      "step": 7456
    },
    {
      "epoch": 1.0451296426068675,
      "grad_norm": 0.16642262041568756,
      "learning_rate": 0.0002002343936857211,
      "loss": 0.0177,
      "step": 7457
    },
    {
      "epoch": 1.0452697967764542,
      "grad_norm": 0.46752092242240906,
      "learning_rate": 0.00020022004305190146,
      "loss": 0.0661,
      "step": 7458
    },
    {
      "epoch": 1.0454099509460406,
      "grad_norm": 0.1605590581893921,
      "learning_rate": 0.00020020569241808179,
      "loss": 0.0173,
      "step": 7459
    },
    {
      "epoch": 1.0455501051156273,
      "grad_norm": 0.3142889738082886,
      "learning_rate": 0.00020019134178426211,
      "loss": 0.0781,
      "step": 7460
    },
    {
      "epoch": 1.0456902592852138,
      "grad_norm": 0.33982208371162415,
      "learning_rate": 0.00020017699115044244,
      "loss": 0.0686,
      "step": 7461
    },
    {
      "epoch": 1.0458304134548002,
      "grad_norm": 0.3422391414642334,
      "learning_rate": 0.0002001626405166228,
      "loss": 0.0231,
      "step": 7462
    },
    {
      "epoch": 1.045970567624387,
      "grad_norm": 0.3262426257133484,
      "learning_rate": 0.00020014828988280313,
      "loss": 0.0301,
      "step": 7463
    },
    {
      "epoch": 1.0461107217939734,
      "grad_norm": 0.3322065770626068,
      "learning_rate": 0.00020013393924898346,
      "loss": 0.0195,
      "step": 7464
    },
    {
      "epoch": 1.0462508759635598,
      "grad_norm": 0.2396252453327179,
      "learning_rate": 0.00020011958861516384,
      "loss": 0.037,
      "step": 7465
    },
    {
      "epoch": 1.0463910301331465,
      "grad_norm": 0.43130582571029663,
      "learning_rate": 0.00020010523798134417,
      "loss": 0.1024,
      "step": 7466
    },
    {
      "epoch": 1.046531184302733,
      "grad_norm": 0.25824931263923645,
      "learning_rate": 0.0002000908873475245,
      "loss": 0.038,
      "step": 7467
    },
    {
      "epoch": 1.0466713384723196,
      "grad_norm": 0.3817158639431,
      "learning_rate": 0.00020007653671370485,
      "loss": 0.0726,
      "step": 7468
    },
    {
      "epoch": 1.046811492641906,
      "grad_norm": 0.343127578496933,
      "learning_rate": 0.00020006218607988518,
      "loss": 0.0389,
      "step": 7469
    },
    {
      "epoch": 1.0469516468114926,
      "grad_norm": 0.2909266948699951,
      "learning_rate": 0.0002000478354460655,
      "loss": 0.0344,
      "step": 7470
    },
    {
      "epoch": 1.0470918009810792,
      "grad_norm": 0.5257437825202942,
      "learning_rate": 0.00020003348481224586,
      "loss": 0.0377,
      "step": 7471
    },
    {
      "epoch": 1.0472319551506657,
      "grad_norm": 0.2117527425289154,
      "learning_rate": 0.0002000191341784262,
      "loss": 0.0222,
      "step": 7472
    },
    {
      "epoch": 1.0473721093202524,
      "grad_norm": 0.5073584318161011,
      "learning_rate": 0.00020000478354460652,
      "loss": 0.0679,
      "step": 7473
    },
    {
      "epoch": 1.0475122634898388,
      "grad_norm": 0.38086339831352234,
      "learning_rate": 0.0001999904329107869,
      "loss": 0.0707,
      "step": 7474
    },
    {
      "epoch": 1.0476524176594253,
      "grad_norm": 0.49791890382766724,
      "learning_rate": 0.00019997608227696723,
      "loss": 0.0547,
      "step": 7475
    },
    {
      "epoch": 1.047792571829012,
      "grad_norm": 0.6414896845817566,
      "learning_rate": 0.00019996173164314756,
      "loss": 0.0596,
      "step": 7476
    },
    {
      "epoch": 1.0479327259985984,
      "grad_norm": 0.4338603913784027,
      "learning_rate": 0.0001999473810093279,
      "loss": 0.0865,
      "step": 7477
    },
    {
      "epoch": 1.048072880168185,
      "grad_norm": 0.27509358525276184,
      "learning_rate": 0.00019993303037550825,
      "loss": 0.0128,
      "step": 7478
    },
    {
      "epoch": 1.0482130343377716,
      "grad_norm": 0.3782193660736084,
      "learning_rate": 0.00019991867974168857,
      "loss": 0.0641,
      "step": 7479
    },
    {
      "epoch": 1.048353188507358,
      "grad_norm": 0.2564517557621002,
      "learning_rate": 0.0001999043291078689,
      "loss": 0.0288,
      "step": 7480
    },
    {
      "epoch": 1.0484933426769447,
      "grad_norm": 1.3186359405517578,
      "learning_rate": 0.00019988997847404926,
      "loss": 0.2088,
      "step": 7481
    },
    {
      "epoch": 1.0486334968465312,
      "grad_norm": 0.845447301864624,
      "learning_rate": 0.0001998756278402296,
      "loss": 0.2101,
      "step": 7482
    },
    {
      "epoch": 1.0487736510161176,
      "grad_norm": 0.6468475461006165,
      "learning_rate": 0.00019986127720640992,
      "loss": 0.1327,
      "step": 7483
    },
    {
      "epoch": 1.0489138051857043,
      "grad_norm": 0.4572700262069702,
      "learning_rate": 0.00019984692657259027,
      "loss": 0.1116,
      "step": 7484
    },
    {
      "epoch": 1.0490539593552908,
      "grad_norm": 1.019961953163147,
      "learning_rate": 0.0001998325759387706,
      "loss": 0.1179,
      "step": 7485
    },
    {
      "epoch": 1.0491941135248775,
      "grad_norm": 0.49574682116508484,
      "learning_rate": 0.00019981822530495093,
      "loss": 0.0721,
      "step": 7486
    },
    {
      "epoch": 1.049334267694464,
      "grad_norm": 0.23519475758075714,
      "learning_rate": 0.0001998038746711313,
      "loss": 0.0593,
      "step": 7487
    },
    {
      "epoch": 1.0494744218640504,
      "grad_norm": 0.3164590299129486,
      "learning_rate": 0.00019978952403731164,
      "loss": 0.0543,
      "step": 7488
    },
    {
      "epoch": 1.049614576033637,
      "grad_norm": 0.46372711658477783,
      "learning_rate": 0.00019977517340349197,
      "loss": 0.0745,
      "step": 7489
    },
    {
      "epoch": 1.0497547302032235,
      "grad_norm": 0.26876315474510193,
      "learning_rate": 0.00019976082276967232,
      "loss": 0.0449,
      "step": 7490
    },
    {
      "epoch": 1.0498948843728102,
      "grad_norm": 0.2763366997241974,
      "learning_rate": 0.00019974647213585265,
      "loss": 0.0854,
      "step": 7491
    },
    {
      "epoch": 1.0500350385423967,
      "grad_norm": 0.4557197093963623,
      "learning_rate": 0.00019973212150203298,
      "loss": 0.0333,
      "step": 7492
    },
    {
      "epoch": 1.0501751927119831,
      "grad_norm": 0.41645553708076477,
      "learning_rate": 0.00019971777086821334,
      "loss": 0.0405,
      "step": 7493
    },
    {
      "epoch": 1.0503153468815698,
      "grad_norm": 0.43728384375572205,
      "learning_rate": 0.00019970342023439367,
      "loss": 0.1349,
      "step": 7494
    },
    {
      "epoch": 1.0504555010511563,
      "grad_norm": 0.48241326212882996,
      "learning_rate": 0.000199689069600574,
      "loss": 0.0746,
      "step": 7495
    },
    {
      "epoch": 1.0505956552207427,
      "grad_norm": 0.5028722286224365,
      "learning_rate": 0.00019967471896675432,
      "loss": 0.1016,
      "step": 7496
    },
    {
      "epoch": 1.0507358093903294,
      "grad_norm": 0.22248029708862305,
      "learning_rate": 0.0001996603683329347,
      "loss": 0.0455,
      "step": 7497
    },
    {
      "epoch": 1.0508759635599159,
      "grad_norm": 0.8588648438453674,
      "learning_rate": 0.00019964601769911503,
      "loss": 0.0895,
      "step": 7498
    },
    {
      "epoch": 1.0510161177295025,
      "grad_norm": 0.23754124343395233,
      "learning_rate": 0.00019963166706529536,
      "loss": 0.0526,
      "step": 7499
    },
    {
      "epoch": 1.051156271899089,
      "grad_norm": 0.7451086640357971,
      "learning_rate": 0.00019961731643147572,
      "loss": 0.1106,
      "step": 7500
    },
    {
      "epoch": 1.0512964260686755,
      "grad_norm": 0.2748192548751831,
      "learning_rate": 0.00019960296579765605,
      "loss": 0.0362,
      "step": 7501
    },
    {
      "epoch": 1.0514365802382621,
      "grad_norm": 0.21557137370109558,
      "learning_rate": 0.00019958861516383638,
      "loss": 0.0248,
      "step": 7502
    },
    {
      "epoch": 1.0515767344078486,
      "grad_norm": 0.4813844859600067,
      "learning_rate": 0.00019957426453001673,
      "loss": 0.0718,
      "step": 7503
    },
    {
      "epoch": 1.0517168885774353,
      "grad_norm": 0.32053476572036743,
      "learning_rate": 0.00019955991389619706,
      "loss": 0.0413,
      "step": 7504
    },
    {
      "epoch": 1.0518570427470217,
      "grad_norm": 0.40857240557670593,
      "learning_rate": 0.0001995455632623774,
      "loss": 0.0341,
      "step": 7505
    },
    {
      "epoch": 1.0519971969166082,
      "grad_norm": 0.3459523916244507,
      "learning_rate": 0.00019953121262855777,
      "loss": 0.0734,
      "step": 7506
    },
    {
      "epoch": 1.0521373510861949,
      "grad_norm": 0.4750889241695404,
      "learning_rate": 0.0001995168619947381,
      "loss": 0.1062,
      "step": 7507
    },
    {
      "epoch": 1.0522775052557813,
      "grad_norm": 0.6610357165336609,
      "learning_rate": 0.00019950251136091843,
      "loss": 0.0614,
      "step": 7508
    },
    {
      "epoch": 1.0524176594253678,
      "grad_norm": 0.27166077494621277,
      "learning_rate": 0.00019948816072709878,
      "loss": 0.0519,
      "step": 7509
    },
    {
      "epoch": 1.0525578135949545,
      "grad_norm": 0.4206123948097229,
      "learning_rate": 0.0001994738100932791,
      "loss": 0.0409,
      "step": 7510
    },
    {
      "epoch": 1.052697967764541,
      "grad_norm": 0.6885419487953186,
      "learning_rate": 0.00019945945945945944,
      "loss": 0.0312,
      "step": 7511
    },
    {
      "epoch": 1.0528381219341276,
      "grad_norm": 0.6111764907836914,
      "learning_rate": 0.00019944510882563977,
      "loss": 0.1358,
      "step": 7512
    },
    {
      "epoch": 1.052978276103714,
      "grad_norm": 0.6657649874687195,
      "learning_rate": 0.00019943075819182012,
      "loss": 0.0707,
      "step": 7513
    },
    {
      "epoch": 1.0531184302733005,
      "grad_norm": 0.2693290710449219,
      "learning_rate": 0.00019941640755800045,
      "loss": 0.0492,
      "step": 7514
    },
    {
      "epoch": 1.0532585844428872,
      "grad_norm": 0.7509901523590088,
      "learning_rate": 0.00019940205692418078,
      "loss": 0.0916,
      "step": 7515
    },
    {
      "epoch": 1.0533987386124737,
      "grad_norm": 0.30057698488235474,
      "learning_rate": 0.00019938770629036114,
      "loss": 0.0445,
      "step": 7516
    },
    {
      "epoch": 1.0535388927820604,
      "grad_norm": 0.22511519491672516,
      "learning_rate": 0.00019937335565654147,
      "loss": 0.0421,
      "step": 7517
    },
    {
      "epoch": 1.0536790469516468,
      "grad_norm": 0.6416477560997009,
      "learning_rate": 0.0001993590050227218,
      "loss": 0.136,
      "step": 7518
    },
    {
      "epoch": 1.0538192011212333,
      "grad_norm": 0.5774193406105042,
      "learning_rate": 0.00019934465438890218,
      "loss": 0.0618,
      "step": 7519
    },
    {
      "epoch": 1.05395935529082,
      "grad_norm": 0.7175807356834412,
      "learning_rate": 0.0001993303037550825,
      "loss": 0.0705,
      "step": 7520
    },
    {
      "epoch": 1.0540995094604064,
      "grad_norm": 0.3576360046863556,
      "learning_rate": 0.00019931595312126283,
      "loss": 0.047,
      "step": 7521
    },
    {
      "epoch": 1.054239663629993,
      "grad_norm": 0.08990313112735748,
      "learning_rate": 0.0001993016024874432,
      "loss": 0.0137,
      "step": 7522
    },
    {
      "epoch": 1.0543798177995796,
      "grad_norm": 0.27882933616638184,
      "learning_rate": 0.00019928725185362352,
      "loss": 0.0481,
      "step": 7523
    },
    {
      "epoch": 1.054519971969166,
      "grad_norm": 0.39496538043022156,
      "learning_rate": 0.00019927290121980385,
      "loss": 0.0291,
      "step": 7524
    },
    {
      "epoch": 1.0546601261387527,
      "grad_norm": 0.5221301913261414,
      "learning_rate": 0.0001992585505859842,
      "loss": 0.0382,
      "step": 7525
    },
    {
      "epoch": 1.0548002803083392,
      "grad_norm": 0.1998799443244934,
      "learning_rate": 0.00019924419995216453,
      "loss": 0.0313,
      "step": 7526
    },
    {
      "epoch": 1.0549404344779256,
      "grad_norm": 0.4970436692237854,
      "learning_rate": 0.00019922984931834486,
      "loss": 0.0562,
      "step": 7527
    },
    {
      "epoch": 1.0550805886475123,
      "grad_norm": 0.8129107356071472,
      "learning_rate": 0.00019921549868452524,
      "loss": 0.0532,
      "step": 7528
    },
    {
      "epoch": 1.0552207428170988,
      "grad_norm": 0.2570417523384094,
      "learning_rate": 0.00019920114805070557,
      "loss": 0.0115,
      "step": 7529
    },
    {
      "epoch": 1.0553608969866854,
      "grad_norm": 0.9175505638122559,
      "learning_rate": 0.0001991867974168859,
      "loss": 0.0298,
      "step": 7530
    },
    {
      "epoch": 1.055501051156272,
      "grad_norm": 0.515227198600769,
      "learning_rate": 0.00019917244678306623,
      "loss": 0.0461,
      "step": 7531
    },
    {
      "epoch": 1.0556412053258584,
      "grad_norm": 0.5605230331420898,
      "learning_rate": 0.00019915809614924658,
      "loss": 0.0309,
      "step": 7532
    },
    {
      "epoch": 1.055781359495445,
      "grad_norm": 0.16247159242630005,
      "learning_rate": 0.0001991437455154269,
      "loss": 0.0262,
      "step": 7533
    },
    {
      "epoch": 1.0559215136650315,
      "grad_norm": 0.09054696559906006,
      "learning_rate": 0.00019912939488160724,
      "loss": 0.0047,
      "step": 7534
    },
    {
      "epoch": 1.056061667834618,
      "grad_norm": 2.6213529109954834,
      "learning_rate": 0.0001991150442477876,
      "loss": 0.1639,
      "step": 7535
    },
    {
      "epoch": 1.0562018220042046,
      "grad_norm": 0.3176807463169098,
      "learning_rate": 0.00019910069361396793,
      "loss": 0.0372,
      "step": 7536
    },
    {
      "epoch": 1.056341976173791,
      "grad_norm": 0.5382075905799866,
      "learning_rate": 0.00019908634298014825,
      "loss": 0.0535,
      "step": 7537
    },
    {
      "epoch": 1.0564821303433778,
      "grad_norm": 0.3053722381591797,
      "learning_rate": 0.00019907199234632864,
      "loss": 0.0845,
      "step": 7538
    },
    {
      "epoch": 1.0566222845129642,
      "grad_norm": 0.29730188846588135,
      "learning_rate": 0.00019905764171250897,
      "loss": 0.038,
      "step": 7539
    },
    {
      "epoch": 1.0567624386825507,
      "grad_norm": 0.526772141456604,
      "learning_rate": 0.0001990432910786893,
      "loss": 0.1425,
      "step": 7540
    },
    {
      "epoch": 1.0569025928521374,
      "grad_norm": 0.32177266478538513,
      "learning_rate": 0.00019902894044486965,
      "loss": 0.0505,
      "step": 7541
    },
    {
      "epoch": 1.0570427470217238,
      "grad_norm": 0.47809478640556335,
      "learning_rate": 0.00019901458981104998,
      "loss": 0.0524,
      "step": 7542
    },
    {
      "epoch": 1.0571829011913105,
      "grad_norm": 0.4227425754070282,
      "learning_rate": 0.0001990002391772303,
      "loss": 0.1108,
      "step": 7543
    },
    {
      "epoch": 1.057323055360897,
      "grad_norm": 0.4794295132160187,
      "learning_rate": 0.00019898588854341066,
      "loss": 0.028,
      "step": 7544
    },
    {
      "epoch": 1.0574632095304835,
      "grad_norm": 0.820785403251648,
      "learning_rate": 0.000198971537909591,
      "loss": 0.0414,
      "step": 7545
    },
    {
      "epoch": 1.0576033637000701,
      "grad_norm": 0.5381726026535034,
      "learning_rate": 0.00019895718727577132,
      "loss": 0.0589,
      "step": 7546
    },
    {
      "epoch": 1.0577435178696566,
      "grad_norm": 0.27617841958999634,
      "learning_rate": 0.00019894283664195165,
      "loss": 0.0799,
      "step": 7547
    },
    {
      "epoch": 1.0578836720392433,
      "grad_norm": 0.5296957492828369,
      "learning_rate": 0.000198928486008132,
      "loss": 0.0759,
      "step": 7548
    },
    {
      "epoch": 1.0580238262088297,
      "grad_norm": 0.2712271213531494,
      "learning_rate": 0.00019891413537431233,
      "loss": 0.0332,
      "step": 7549
    },
    {
      "epoch": 1.0581639803784162,
      "grad_norm": 0.3844560980796814,
      "learning_rate": 0.00019889978474049266,
      "loss": 0.0684,
      "step": 7550
    },
    {
      "epoch": 1.0583041345480029,
      "grad_norm": 0.3186555802822113,
      "learning_rate": 0.00019888543410667304,
      "loss": 0.0272,
      "step": 7551
    },
    {
      "epoch": 1.0584442887175893,
      "grad_norm": 0.31765928864479065,
      "learning_rate": 0.00019887108347285337,
      "loss": 0.0446,
      "step": 7552
    },
    {
      "epoch": 1.0585844428871758,
      "grad_norm": 0.32476410269737244,
      "learning_rate": 0.0001988567328390337,
      "loss": 0.1209,
      "step": 7553
    },
    {
      "epoch": 1.0587245970567625,
      "grad_norm": 0.32868948578834534,
      "learning_rate": 0.00019884238220521406,
      "loss": 0.1,
      "step": 7554
    },
    {
      "epoch": 1.058864751226349,
      "grad_norm": 0.4197101593017578,
      "learning_rate": 0.00019882803157139439,
      "loss": 0.0482,
      "step": 7555
    },
    {
      "epoch": 1.0590049053959356,
      "grad_norm": 0.1931259036064148,
      "learning_rate": 0.00019881368093757471,
      "loss": 0.0225,
      "step": 7556
    },
    {
      "epoch": 1.059145059565522,
      "grad_norm": 0.34077540040016174,
      "learning_rate": 0.00019879933030375507,
      "loss": 0.0584,
      "step": 7557
    },
    {
      "epoch": 1.0592852137351085,
      "grad_norm": 0.43230900168418884,
      "learning_rate": 0.0001987849796699354,
      "loss": 0.0663,
      "step": 7558
    },
    {
      "epoch": 1.0594253679046952,
      "grad_norm": 0.22732526063919067,
      "learning_rate": 0.00019877062903611573,
      "loss": 0.0348,
      "step": 7559
    },
    {
      "epoch": 1.0595655220742817,
      "grad_norm": 0.1841212511062622,
      "learning_rate": 0.0001987562784022961,
      "loss": 0.0544,
      "step": 7560
    },
    {
      "epoch": 1.0597056762438684,
      "grad_norm": 0.37000972032546997,
      "learning_rate": 0.00019874192776847644,
      "loss": 0.0283,
      "step": 7561
    },
    {
      "epoch": 1.0598458304134548,
      "grad_norm": 0.48312994837760925,
      "learning_rate": 0.00019872757713465677,
      "loss": 0.0841,
      "step": 7562
    },
    {
      "epoch": 1.0599859845830413,
      "grad_norm": 0.34391459822654724,
      "learning_rate": 0.00019871322650083712,
      "loss": 0.0462,
      "step": 7563
    },
    {
      "epoch": 1.060126138752628,
      "grad_norm": 0.15284235775470734,
      "learning_rate": 0.00019869887586701745,
      "loss": 0.0268,
      "step": 7564
    },
    {
      "epoch": 1.0602662929222144,
      "grad_norm": 0.19502505660057068,
      "learning_rate": 0.00019868452523319778,
      "loss": 0.0452,
      "step": 7565
    },
    {
      "epoch": 1.0604064470918009,
      "grad_norm": 0.2463318258523941,
      "learning_rate": 0.0001986701745993781,
      "loss": 0.0377,
      "step": 7566
    },
    {
      "epoch": 1.0605466012613876,
      "grad_norm": 0.21984611451625824,
      "learning_rate": 0.00019865582396555846,
      "loss": 0.0315,
      "step": 7567
    },
    {
      "epoch": 1.060686755430974,
      "grad_norm": 0.8172303438186646,
      "learning_rate": 0.0001986414733317388,
      "loss": 0.0665,
      "step": 7568
    },
    {
      "epoch": 1.0608269096005607,
      "grad_norm": 0.37418031692504883,
      "learning_rate": 0.00019862712269791912,
      "loss": 0.0839,
      "step": 7569
    },
    {
      "epoch": 1.0609670637701472,
      "grad_norm": 0.4457027018070221,
      "learning_rate": 0.0001986127720640995,
      "loss": 0.0736,
      "step": 7570
    },
    {
      "epoch": 1.0611072179397336,
      "grad_norm": 0.13124918937683105,
      "learning_rate": 0.00019859842143027983,
      "loss": 0.023,
      "step": 7571
    },
    {
      "epoch": 1.0612473721093203,
      "grad_norm": 0.33823636174201965,
      "learning_rate": 0.00019858407079646016,
      "loss": 0.0567,
      "step": 7572
    },
    {
      "epoch": 1.0613875262789068,
      "grad_norm": 0.1908726990222931,
      "learning_rate": 0.00019856972016264052,
      "loss": 0.0155,
      "step": 7573
    },
    {
      "epoch": 1.0615276804484934,
      "grad_norm": 0.3268665671348572,
      "learning_rate": 0.00019855536952882084,
      "loss": 0.0683,
      "step": 7574
    },
    {
      "epoch": 1.06166783461808,
      "grad_norm": 0.20980527997016907,
      "learning_rate": 0.00019854101889500117,
      "loss": 0.0563,
      "step": 7575
    },
    {
      "epoch": 1.0618079887876664,
      "grad_norm": 0.30090105533599854,
      "learning_rate": 0.00019852666826118153,
      "loss": 0.0627,
      "step": 7576
    },
    {
      "epoch": 1.061948142957253,
      "grad_norm": 0.09535747021436691,
      "learning_rate": 0.00019851231762736186,
      "loss": 0.0103,
      "step": 7577
    },
    {
      "epoch": 1.0620882971268395,
      "grad_norm": 0.40273454785346985,
      "learning_rate": 0.00019849796699354219,
      "loss": 0.0397,
      "step": 7578
    },
    {
      "epoch": 1.0622284512964262,
      "grad_norm": 0.7150191068649292,
      "learning_rate": 0.00019848361635972254,
      "loss": 0.0584,
      "step": 7579
    },
    {
      "epoch": 1.0623686054660126,
      "grad_norm": 0.4053855538368225,
      "learning_rate": 0.00019846926572590287,
      "loss": 0.0281,
      "step": 7580
    },
    {
      "epoch": 1.062508759635599,
      "grad_norm": 2.127323865890503,
      "learning_rate": 0.0001984549150920832,
      "loss": 0.396,
      "step": 7581
    },
    {
      "epoch": 1.0626489138051858,
      "grad_norm": 0.8696365356445312,
      "learning_rate": 0.00019844056445826353,
      "loss": 0.0997,
      "step": 7582
    },
    {
      "epoch": 1.0627890679747722,
      "grad_norm": 3.740506410598755,
      "learning_rate": 0.0001984262138244439,
      "loss": 0.0472,
      "step": 7583
    },
    {
      "epoch": 1.0629292221443587,
      "grad_norm": 0.8195126056671143,
      "learning_rate": 0.00019841186319062424,
      "loss": 0.0616,
      "step": 7584
    },
    {
      "epoch": 1.0630693763139454,
      "grad_norm": 1.1001120805740356,
      "learning_rate": 0.00019839751255680457,
      "loss": 0.0576,
      "step": 7585
    },
    {
      "epoch": 1.0632095304835318,
      "grad_norm": 0.44824814796447754,
      "learning_rate": 0.00019838316192298492,
      "loss": 0.0456,
      "step": 7586
    },
    {
      "epoch": 1.0633496846531185,
      "grad_norm": 0.3747692108154297,
      "learning_rate": 0.00019836881128916525,
      "loss": 0.0553,
      "step": 7587
    },
    {
      "epoch": 1.063489838822705,
      "grad_norm": 0.26847440004348755,
      "learning_rate": 0.00019835446065534558,
      "loss": 0.0653,
      "step": 7588
    },
    {
      "epoch": 1.0636299929922914,
      "grad_norm": 0.42787301540374756,
      "learning_rate": 0.00019834011002152594,
      "loss": 0.032,
      "step": 7589
    },
    {
      "epoch": 1.0637701471618781,
      "grad_norm": 0.374841570854187,
      "learning_rate": 0.00019832575938770626,
      "loss": 0.1007,
      "step": 7590
    },
    {
      "epoch": 1.0639103013314646,
      "grad_norm": 0.1404145509004593,
      "learning_rate": 0.0001983114087538866,
      "loss": 0.0213,
      "step": 7591
    },
    {
      "epoch": 1.064050455501051,
      "grad_norm": 0.2259984165430069,
      "learning_rate": 0.00019829705812006698,
      "loss": 0.0411,
      "step": 7592
    },
    {
      "epoch": 1.0641906096706377,
      "grad_norm": 0.2627427279949188,
      "learning_rate": 0.0001982827074862473,
      "loss": 0.1163,
      "step": 7593
    },
    {
      "epoch": 1.0643307638402242,
      "grad_norm": 0.3677605092525482,
      "learning_rate": 0.00019826835685242763,
      "loss": 0.0646,
      "step": 7594
    },
    {
      "epoch": 1.0644709180098109,
      "grad_norm": 0.17421786487102509,
      "learning_rate": 0.000198254006218608,
      "loss": 0.0329,
      "step": 7595
    },
    {
      "epoch": 1.0646110721793973,
      "grad_norm": 0.6553385257720947,
      "learning_rate": 0.00019823965558478832,
      "loss": 0.0739,
      "step": 7596
    },
    {
      "epoch": 1.0647512263489838,
      "grad_norm": 0.3015502989292145,
      "learning_rate": 0.00019822530495096865,
      "loss": 0.0246,
      "step": 7597
    },
    {
      "epoch": 1.0648913805185705,
      "grad_norm": 0.3092505931854248,
      "learning_rate": 0.000198210954317149,
      "loss": 0.0493,
      "step": 7598
    },
    {
      "epoch": 1.065031534688157,
      "grad_norm": 0.5699924230575562,
      "learning_rate": 0.00019819660368332933,
      "loss": 0.0277,
      "step": 7599
    },
    {
      "epoch": 1.0651716888577436,
      "grad_norm": 0.15097761154174805,
      "learning_rate": 0.00019818225304950966,
      "loss": 0.0257,
      "step": 7600
    },
    {
      "epoch": 1.06531184302733,
      "grad_norm": 0.19576691091060638,
      "learning_rate": 0.00019816790241569,
      "loss": 0.0409,
      "step": 7601
    },
    {
      "epoch": 1.0654519971969165,
      "grad_norm": 0.7171152830123901,
      "learning_rate": 0.00019815355178187037,
      "loss": 0.1016,
      "step": 7602
    },
    {
      "epoch": 1.0655921513665032,
      "grad_norm": 0.2911851108074188,
      "learning_rate": 0.0001981392011480507,
      "loss": 0.0496,
      "step": 7603
    },
    {
      "epoch": 1.0657323055360897,
      "grad_norm": 0.358695924282074,
      "learning_rate": 0.00019812485051423103,
      "loss": 0.0968,
      "step": 7604
    },
    {
      "epoch": 1.0658724597056763,
      "grad_norm": 0.29031676054000854,
      "learning_rate": 0.00019811049988041138,
      "loss": 0.1149,
      "step": 7605
    },
    {
      "epoch": 1.0660126138752628,
      "grad_norm": 0.12392277270555496,
      "learning_rate": 0.0001980961492465917,
      "loss": 0.0092,
      "step": 7606
    },
    {
      "epoch": 1.0661527680448493,
      "grad_norm": 0.25931262969970703,
      "learning_rate": 0.00019808179861277204,
      "loss": 0.0324,
      "step": 7607
    },
    {
      "epoch": 1.066292922214436,
      "grad_norm": 0.8556021451950073,
      "learning_rate": 0.0001980674479789524,
      "loss": 0.1004,
      "step": 7608
    },
    {
      "epoch": 1.0664330763840224,
      "grad_norm": 1.0676041841506958,
      "learning_rate": 0.00019805309734513272,
      "loss": 0.0768,
      "step": 7609
    },
    {
      "epoch": 1.066573230553609,
      "grad_norm": 0.23332707583904266,
      "learning_rate": 0.00019803874671131305,
      "loss": 0.0287,
      "step": 7610
    },
    {
      "epoch": 1.0667133847231955,
      "grad_norm": 0.39812928438186646,
      "learning_rate": 0.0001980243960774934,
      "loss": 0.0494,
      "step": 7611
    },
    {
      "epoch": 1.066853538892782,
      "grad_norm": 0.3797602653503418,
      "learning_rate": 0.00019801004544367374,
      "loss": 0.0758,
      "step": 7612
    },
    {
      "epoch": 1.0669936930623687,
      "grad_norm": 0.4300213158130646,
      "learning_rate": 0.00019799569480985406,
      "loss": 0.0419,
      "step": 7613
    },
    {
      "epoch": 1.0671338472319551,
      "grad_norm": 0.21468813717365265,
      "learning_rate": 0.00019798134417603445,
      "loss": 0.0265,
      "step": 7614
    },
    {
      "epoch": 1.0672740014015416,
      "grad_norm": 0.20800012350082397,
      "learning_rate": 0.00019796699354221478,
      "loss": 0.0183,
      "step": 7615
    },
    {
      "epoch": 1.0674141555711283,
      "grad_norm": 0.6823247671127319,
      "learning_rate": 0.0001979526429083951,
      "loss": 0.0838,
      "step": 7616
    },
    {
      "epoch": 1.0675543097407147,
      "grad_norm": 0.19142048060894012,
      "learning_rate": 0.00019793829227457543,
      "loss": 0.0333,
      "step": 7617
    },
    {
      "epoch": 1.0676944639103014,
      "grad_norm": 0.1455041915178299,
      "learning_rate": 0.0001979239416407558,
      "loss": 0.0817,
      "step": 7618
    },
    {
      "epoch": 1.0678346180798879,
      "grad_norm": 0.3570077419281006,
      "learning_rate": 0.00019790959100693612,
      "loss": 0.1126,
      "step": 7619
    },
    {
      "epoch": 1.0679747722494743,
      "grad_norm": 0.42099517583847046,
      "learning_rate": 0.00019789524037311645,
      "loss": 0.1019,
      "step": 7620
    },
    {
      "epoch": 1.068114926419061,
      "grad_norm": 0.5669971704483032,
      "learning_rate": 0.0001978808897392968,
      "loss": 0.0572,
      "step": 7621
    },
    {
      "epoch": 1.0682550805886475,
      "grad_norm": 1.097241997718811,
      "learning_rate": 0.00019786653910547713,
      "loss": 0.138,
      "step": 7622
    },
    {
      "epoch": 1.068395234758234,
      "grad_norm": 0.4841913878917694,
      "learning_rate": 0.00019785218847165746,
      "loss": 0.0641,
      "step": 7623
    },
    {
      "epoch": 1.0685353889278206,
      "grad_norm": 0.36787909269332886,
      "learning_rate": 0.00019783783783783784,
      "loss": 0.0671,
      "step": 7624
    },
    {
      "epoch": 1.068675543097407,
      "grad_norm": 0.13150766491889954,
      "learning_rate": 0.00019782348720401817,
      "loss": 0.016,
      "step": 7625
    },
    {
      "epoch": 1.0688156972669938,
      "grad_norm": 0.7408350110054016,
      "learning_rate": 0.0001978091365701985,
      "loss": 0.0767,
      "step": 7626
    },
    {
      "epoch": 1.0689558514365802,
      "grad_norm": 0.5474209189414978,
      "learning_rate": 0.00019779478593637885,
      "loss": 0.1179,
      "step": 7627
    },
    {
      "epoch": 1.0690960056061667,
      "grad_norm": 0.7485910654067993,
      "learning_rate": 0.00019778043530255918,
      "loss": 0.0969,
      "step": 7628
    },
    {
      "epoch": 1.0692361597757534,
      "grad_norm": 0.3601694107055664,
      "learning_rate": 0.0001977660846687395,
      "loss": 0.0306,
      "step": 7629
    },
    {
      "epoch": 1.0693763139453398,
      "grad_norm": 0.25154393911361694,
      "learning_rate": 0.00019775173403491987,
      "loss": 0.0272,
      "step": 7630
    },
    {
      "epoch": 1.0695164681149265,
      "grad_norm": 1.0999586582183838,
      "learning_rate": 0.0001977373834011002,
      "loss": 0.0459,
      "step": 7631
    },
    {
      "epoch": 1.069656622284513,
      "grad_norm": 0.3171786367893219,
      "learning_rate": 0.00019772303276728052,
      "loss": 0.0382,
      "step": 7632
    },
    {
      "epoch": 1.0697967764540994,
      "grad_norm": 0.4888435900211334,
      "learning_rate": 0.00019770868213346085,
      "loss": 0.0423,
      "step": 7633
    },
    {
      "epoch": 1.069936930623686,
      "grad_norm": 0.5235338807106018,
      "learning_rate": 0.00019769433149964124,
      "loss": 0.1306,
      "step": 7634
    },
    {
      "epoch": 1.0700770847932726,
      "grad_norm": 2.3171980381011963,
      "learning_rate": 0.00019767998086582156,
      "loss": 0.5611,
      "step": 7635
    },
    {
      "epoch": 1.0702172389628593,
      "grad_norm": 0.20193396508693695,
      "learning_rate": 0.0001976656302320019,
      "loss": 0.0258,
      "step": 7636
    },
    {
      "epoch": 1.0703573931324457,
      "grad_norm": 0.2289632111787796,
      "learning_rate": 0.00019765127959818225,
      "loss": 0.0456,
      "step": 7637
    },
    {
      "epoch": 1.0704975473020322,
      "grad_norm": 0.22437684237957,
      "learning_rate": 0.00019763692896436258,
      "loss": 0.0272,
      "step": 7638
    },
    {
      "epoch": 1.0706377014716189,
      "grad_norm": 0.21627455949783325,
      "learning_rate": 0.0001976225783305429,
      "loss": 0.0282,
      "step": 7639
    },
    {
      "epoch": 1.0707778556412053,
      "grad_norm": 0.08288053423166275,
      "learning_rate": 0.00019760822769672326,
      "loss": 0.013,
      "step": 7640
    },
    {
      "epoch": 1.0709180098107918,
      "grad_norm": 0.3561341464519501,
      "learning_rate": 0.0001975938770629036,
      "loss": 0.0621,
      "step": 7641
    },
    {
      "epoch": 1.0710581639803785,
      "grad_norm": 0.31880494952201843,
      "learning_rate": 0.00019757952642908392,
      "loss": 0.0349,
      "step": 7642
    },
    {
      "epoch": 1.071198318149965,
      "grad_norm": 0.3106725513935089,
      "learning_rate": 0.00019756517579526427,
      "loss": 0.0388,
      "step": 7643
    },
    {
      "epoch": 1.0713384723195516,
      "grad_norm": 0.4200275242328644,
      "learning_rate": 0.0001975508251614446,
      "loss": 0.0316,
      "step": 7644
    },
    {
      "epoch": 1.071478626489138,
      "grad_norm": 0.29874706268310547,
      "learning_rate": 0.00019753647452762493,
      "loss": 0.0868,
      "step": 7645
    },
    {
      "epoch": 1.0716187806587245,
      "grad_norm": 0.3218427896499634,
      "learning_rate": 0.00019752212389380531,
      "loss": 0.0502,
      "step": 7646
    },
    {
      "epoch": 1.0717589348283112,
      "grad_norm": 0.16986088454723358,
      "learning_rate": 0.00019750777325998564,
      "loss": 0.0656,
      "step": 7647
    },
    {
      "epoch": 1.0718990889978977,
      "grad_norm": 0.26179036498069763,
      "learning_rate": 0.00019749342262616597,
      "loss": 0.0381,
      "step": 7648
    },
    {
      "epoch": 1.0720392431674843,
      "grad_norm": 0.2574778199195862,
      "learning_rate": 0.00019747907199234633,
      "loss": 0.0806,
      "step": 7649
    },
    {
      "epoch": 1.0721793973370708,
      "grad_norm": 0.43426260352134705,
      "learning_rate": 0.00019746472135852666,
      "loss": 0.0273,
      "step": 7650
    },
    {
      "epoch": 1.0723195515066573,
      "grad_norm": 0.3414384424686432,
      "learning_rate": 0.00019745037072470698,
      "loss": 0.0833,
      "step": 7651
    },
    {
      "epoch": 1.072459705676244,
      "grad_norm": 0.6683964133262634,
      "learning_rate": 0.0001974360200908873,
      "loss": 0.0194,
      "step": 7652
    },
    {
      "epoch": 1.0725998598458304,
      "grad_norm": 0.083673395216465,
      "learning_rate": 0.00019742166945706767,
      "loss": 0.0047,
      "step": 7653
    },
    {
      "epoch": 1.0727400140154169,
      "grad_norm": 0.33309340476989746,
      "learning_rate": 0.000197407318823248,
      "loss": 0.0496,
      "step": 7654
    },
    {
      "epoch": 1.0728801681850035,
      "grad_norm": 0.20718157291412354,
      "learning_rate": 0.00019739296818942833,
      "loss": 0.056,
      "step": 7655
    },
    {
      "epoch": 1.07302032235459,
      "grad_norm": 0.4584257900714874,
      "learning_rate": 0.0001973786175556087,
      "loss": 0.052,
      "step": 7656
    },
    {
      "epoch": 1.0731604765241767,
      "grad_norm": 0.4084530472755432,
      "learning_rate": 0.00019736426692178904,
      "loss": 0.0861,
      "step": 7657
    },
    {
      "epoch": 1.0733006306937631,
      "grad_norm": 0.3616637885570526,
      "learning_rate": 0.00019734991628796937,
      "loss": 0.0548,
      "step": 7658
    },
    {
      "epoch": 1.0734407848633496,
      "grad_norm": 0.4114520847797394,
      "learning_rate": 0.00019733556565414972,
      "loss": 0.1039,
      "step": 7659
    },
    {
      "epoch": 1.0735809390329363,
      "grad_norm": 0.1908227503299713,
      "learning_rate": 0.00019732121502033005,
      "loss": 0.0641,
      "step": 7660
    },
    {
      "epoch": 1.0737210932025227,
      "grad_norm": 0.4970937967300415,
      "learning_rate": 0.00019730686438651038,
      "loss": 0.0416,
      "step": 7661
    },
    {
      "epoch": 1.0738612473721094,
      "grad_norm": 0.5051129460334778,
      "learning_rate": 0.00019729251375269073,
      "loss": 0.1104,
      "step": 7662
    },
    {
      "epoch": 1.0740014015416959,
      "grad_norm": 0.6620528101921082,
      "learning_rate": 0.00019727816311887106,
      "loss": 0.1118,
      "step": 7663
    },
    {
      "epoch": 1.0741415557112823,
      "grad_norm": 0.34434640407562256,
      "learning_rate": 0.0001972638124850514,
      "loss": 0.0755,
      "step": 7664
    },
    {
      "epoch": 1.074281709880869,
      "grad_norm": 0.22941954433918,
      "learning_rate": 0.00019724946185123177,
      "loss": 0.0314,
      "step": 7665
    },
    {
      "epoch": 1.0744218640504555,
      "grad_norm": 0.15672756731510162,
      "learning_rate": 0.0001972351112174121,
      "loss": 0.0237,
      "step": 7666
    },
    {
      "epoch": 1.0745620182200422,
      "grad_norm": 0.377066433429718,
      "learning_rate": 0.00019722076058359243,
      "loss": 0.0608,
      "step": 7667
    },
    {
      "epoch": 1.0747021723896286,
      "grad_norm": 0.4135933816432953,
      "learning_rate": 0.00019720640994977276,
      "loss": 0.0582,
      "step": 7668
    },
    {
      "epoch": 1.074842326559215,
      "grad_norm": 0.4028638005256653,
      "learning_rate": 0.00019719205931595311,
      "loss": 0.1164,
      "step": 7669
    },
    {
      "epoch": 1.0749824807288018,
      "grad_norm": 0.512250542640686,
      "learning_rate": 0.00019717770868213344,
      "loss": 0.0757,
      "step": 7670
    },
    {
      "epoch": 1.0751226348983882,
      "grad_norm": 0.2980792224407196,
      "learning_rate": 0.00019716335804831377,
      "loss": 0.0346,
      "step": 7671
    },
    {
      "epoch": 1.0752627890679747,
      "grad_norm": 0.17340050637722015,
      "learning_rate": 0.00019714900741449413,
      "loss": 0.0503,
      "step": 7672
    },
    {
      "epoch": 1.0754029432375614,
      "grad_norm": 0.20955890417099,
      "learning_rate": 0.00019713465678067446,
      "loss": 0.0172,
      "step": 7673
    },
    {
      "epoch": 1.0755430974071478,
      "grad_norm": 0.19779478013515472,
      "learning_rate": 0.00019712030614685478,
      "loss": 0.0434,
      "step": 7674
    },
    {
      "epoch": 1.0756832515767345,
      "grad_norm": 0.6693371534347534,
      "learning_rate": 0.00019710595551303514,
      "loss": 0.0626,
      "step": 7675
    },
    {
      "epoch": 1.075823405746321,
      "grad_norm": 0.18301236629486084,
      "learning_rate": 0.00019709160487921547,
      "loss": 0.0267,
      "step": 7676
    },
    {
      "epoch": 1.0759635599159074,
      "grad_norm": 0.42358770966529846,
      "learning_rate": 0.0001970772542453958,
      "loss": 0.0521,
      "step": 7677
    },
    {
      "epoch": 1.076103714085494,
      "grad_norm": 0.6924291253089905,
      "learning_rate": 0.00019706290361157618,
      "loss": 0.1468,
      "step": 7678
    },
    {
      "epoch": 1.0762438682550806,
      "grad_norm": 0.45006266236305237,
      "learning_rate": 0.0001970485529777565,
      "loss": 0.0456,
      "step": 7679
    },
    {
      "epoch": 1.076384022424667,
      "grad_norm": 0.25100284814834595,
      "learning_rate": 0.00019703420234393684,
      "loss": 0.0472,
      "step": 7680
    },
    {
      "epoch": 1.0765241765942537,
      "grad_norm": 0.32928958535194397,
      "learning_rate": 0.0001970198517101172,
      "loss": 0.0582,
      "step": 7681
    },
    {
      "epoch": 1.0766643307638402,
      "grad_norm": 0.5857794880867004,
      "learning_rate": 0.00019700550107629752,
      "loss": 0.0667,
      "step": 7682
    },
    {
      "epoch": 1.0768044849334268,
      "grad_norm": 0.5003965497016907,
      "learning_rate": 0.00019699115044247785,
      "loss": 0.0284,
      "step": 7683
    },
    {
      "epoch": 1.0769446391030133,
      "grad_norm": 1.3723543882369995,
      "learning_rate": 0.0001969767998086582,
      "loss": 0.2033,
      "step": 7684
    },
    {
      "epoch": 1.0770847932725998,
      "grad_norm": 1.1034142971038818,
      "learning_rate": 0.00019696244917483853,
      "loss": 0.2817,
      "step": 7685
    },
    {
      "epoch": 1.0772249474421864,
      "grad_norm": 0.365693062543869,
      "learning_rate": 0.00019694809854101886,
      "loss": 0.0403,
      "step": 7686
    },
    {
      "epoch": 1.077365101611773,
      "grad_norm": 0.3229908049106598,
      "learning_rate": 0.0001969337479071992,
      "loss": 0.0405,
      "step": 7687
    },
    {
      "epoch": 1.0775052557813596,
      "grad_norm": 0.18721123039722443,
      "learning_rate": 0.00019691939727337957,
      "loss": 0.0457,
      "step": 7688
    },
    {
      "epoch": 1.077645409950946,
      "grad_norm": 0.32023507356643677,
      "learning_rate": 0.0001969050466395599,
      "loss": 0.0677,
      "step": 7689
    },
    {
      "epoch": 1.0777855641205325,
      "grad_norm": 0.41388630867004395,
      "learning_rate": 0.00019689069600574023,
      "loss": 0.0711,
      "step": 7690
    },
    {
      "epoch": 1.0779257182901192,
      "grad_norm": 0.3030456006526947,
      "learning_rate": 0.0001968763453719206,
      "loss": 0.0592,
      "step": 7691
    },
    {
      "epoch": 1.0780658724597056,
      "grad_norm": 0.21968378126621246,
      "learning_rate": 0.00019686199473810092,
      "loss": 0.0574,
      "step": 7692
    },
    {
      "epoch": 1.0782060266292923,
      "grad_norm": 0.22468826174736023,
      "learning_rate": 0.00019684764410428124,
      "loss": 0.0322,
      "step": 7693
    },
    {
      "epoch": 1.0783461807988788,
      "grad_norm": 0.10458522289991379,
      "learning_rate": 0.0001968332934704616,
      "loss": 0.026,
      "step": 7694
    },
    {
      "epoch": 1.0784863349684652,
      "grad_norm": 0.3012867867946625,
      "learning_rate": 0.00019681894283664193,
      "loss": 0.0701,
      "step": 7695
    },
    {
      "epoch": 1.078626489138052,
      "grad_norm": 0.2631513178348541,
      "learning_rate": 0.00019680459220282226,
      "loss": 0.0667,
      "step": 7696
    },
    {
      "epoch": 1.0787666433076384,
      "grad_norm": 0.10674197971820831,
      "learning_rate": 0.00019679024156900264,
      "loss": 0.0202,
      "step": 7697
    },
    {
      "epoch": 1.0789067974772248,
      "grad_norm": 0.4610738158226013,
      "learning_rate": 0.00019677589093518297,
      "loss": 0.1531,
      "step": 7698
    },
    {
      "epoch": 1.0790469516468115,
      "grad_norm": 0.3037176728248596,
      "learning_rate": 0.0001967615403013633,
      "loss": 0.071,
      "step": 7699
    },
    {
      "epoch": 1.079187105816398,
      "grad_norm": 0.5380765199661255,
      "learning_rate": 0.00019674718966754365,
      "loss": 0.1136,
      "step": 7700
    },
    {
      "epoch": 1.0793272599859847,
      "grad_norm": 0.22383515536785126,
      "learning_rate": 0.00019673283903372398,
      "loss": 0.042,
      "step": 7701
    },
    {
      "epoch": 1.0794674141555711,
      "grad_norm": 0.2810443937778473,
      "learning_rate": 0.0001967184883999043,
      "loss": 0.0217,
      "step": 7702
    },
    {
      "epoch": 1.0796075683251576,
      "grad_norm": 0.25943776965141296,
      "learning_rate": 0.00019670413776608464,
      "loss": 0.0357,
      "step": 7703
    },
    {
      "epoch": 1.0797477224947443,
      "grad_norm": 0.22084593772888184,
      "learning_rate": 0.000196689787132265,
      "loss": 0.0358,
      "step": 7704
    },
    {
      "epoch": 1.0798878766643307,
      "grad_norm": 0.16578315198421478,
      "learning_rate": 0.00019667543649844532,
      "loss": 0.0247,
      "step": 7705
    },
    {
      "epoch": 1.0800280308339174,
      "grad_norm": 0.3578493595123291,
      "learning_rate": 0.00019666108586462565,
      "loss": 0.0651,
      "step": 7706
    },
    {
      "epoch": 1.0801681850035039,
      "grad_norm": 0.20319566130638123,
      "learning_rate": 0.000196646735230806,
      "loss": 0.0118,
      "step": 7707
    },
    {
      "epoch": 1.0803083391730903,
      "grad_norm": 0.2713966369628906,
      "learning_rate": 0.00019663238459698634,
      "loss": 0.0499,
      "step": 7708
    },
    {
      "epoch": 1.080448493342677,
      "grad_norm": 0.25646093487739563,
      "learning_rate": 0.00019661803396316666,
      "loss": 0.032,
      "step": 7709
    },
    {
      "epoch": 1.0805886475122635,
      "grad_norm": 0.2926417887210846,
      "learning_rate": 0.00019660368332934705,
      "loss": 0.1006,
      "step": 7710
    },
    {
      "epoch": 1.08072880168185,
      "grad_norm": 0.2752074897289276,
      "learning_rate": 0.00019658933269552738,
      "loss": 0.016,
      "step": 7711
    },
    {
      "epoch": 1.0808689558514366,
      "grad_norm": 0.5022668838500977,
      "learning_rate": 0.0001965749820617077,
      "loss": 0.0525,
      "step": 7712
    },
    {
      "epoch": 1.081009110021023,
      "grad_norm": 0.19097664952278137,
      "learning_rate": 0.00019656063142788806,
      "loss": 0.0542,
      "step": 7713
    },
    {
      "epoch": 1.0811492641906097,
      "grad_norm": 0.2498018890619278,
      "learning_rate": 0.0001965462807940684,
      "loss": 0.0966,
      "step": 7714
    },
    {
      "epoch": 1.0812894183601962,
      "grad_norm": 0.13516223430633545,
      "learning_rate": 0.00019653193016024872,
      "loss": 0.0223,
      "step": 7715
    },
    {
      "epoch": 1.0814295725297827,
      "grad_norm": 0.15037253499031067,
      "learning_rate": 0.00019651757952642907,
      "loss": 0.0085,
      "step": 7716
    },
    {
      "epoch": 1.0815697266993693,
      "grad_norm": 0.9176134467124939,
      "learning_rate": 0.0001965032288926094,
      "loss": 0.0726,
      "step": 7717
    },
    {
      "epoch": 1.0817098808689558,
      "grad_norm": 0.12279194593429565,
      "learning_rate": 0.00019648887825878973,
      "loss": 0.0105,
      "step": 7718
    },
    {
      "epoch": 1.0818500350385425,
      "grad_norm": 0.38189974427223206,
      "learning_rate": 0.0001964745276249701,
      "loss": 0.0754,
      "step": 7719
    },
    {
      "epoch": 1.081990189208129,
      "grad_norm": 0.4557897448539734,
      "learning_rate": 0.00019646017699115044,
      "loss": 0.0729,
      "step": 7720
    },
    {
      "epoch": 1.0821303433777154,
      "grad_norm": 0.3637048304080963,
      "learning_rate": 0.00019644582635733077,
      "loss": 0.0796,
      "step": 7721
    },
    {
      "epoch": 1.082270497547302,
      "grad_norm": 0.7959785461425781,
      "learning_rate": 0.0001964314757235111,
      "loss": 0.0657,
      "step": 7722
    },
    {
      "epoch": 1.0824106517168885,
      "grad_norm": 0.7466673254966736,
      "learning_rate": 0.00019641712508969145,
      "loss": 0.184,
      "step": 7723
    },
    {
      "epoch": 1.0825508058864752,
      "grad_norm": 0.2976071834564209,
      "learning_rate": 0.00019640277445587178,
      "loss": 0.0381,
      "step": 7724
    },
    {
      "epoch": 1.0826909600560617,
      "grad_norm": 0.9213812947273254,
      "learning_rate": 0.0001963884238220521,
      "loss": 0.0828,
      "step": 7725
    },
    {
      "epoch": 1.0828311142256482,
      "grad_norm": 0.6381317973136902,
      "learning_rate": 0.00019637407318823247,
      "loss": 0.0963,
      "step": 7726
    },
    {
      "epoch": 1.0829712683952348,
      "grad_norm": 0.7232535481452942,
      "learning_rate": 0.0001963597225544128,
      "loss": 0.0343,
      "step": 7727
    },
    {
      "epoch": 1.0831114225648213,
      "grad_norm": 0.4582957625389099,
      "learning_rate": 0.00019634537192059312,
      "loss": 0.0791,
      "step": 7728
    },
    {
      "epoch": 1.0832515767344078,
      "grad_norm": 0.3746282458305359,
      "learning_rate": 0.0001963310212867735,
      "loss": 0.0533,
      "step": 7729
    },
    {
      "epoch": 1.0833917309039944,
      "grad_norm": 0.4384494125843048,
      "learning_rate": 0.00019631667065295383,
      "loss": 0.0516,
      "step": 7730
    },
    {
      "epoch": 1.083531885073581,
      "grad_norm": 0.3319662809371948,
      "learning_rate": 0.00019630232001913416,
      "loss": 0.029,
      "step": 7731
    },
    {
      "epoch": 1.0836720392431676,
      "grad_norm": 0.7180753350257874,
      "learning_rate": 0.00019628796938531452,
      "loss": 0.1695,
      "step": 7732
    },
    {
      "epoch": 1.083812193412754,
      "grad_norm": 1.7276380062103271,
      "learning_rate": 0.00019627361875149485,
      "loss": 0.1131,
      "step": 7733
    },
    {
      "epoch": 1.0839523475823405,
      "grad_norm": 0.6530478000640869,
      "learning_rate": 0.00019625926811767518,
      "loss": 0.1136,
      "step": 7734
    },
    {
      "epoch": 1.0840925017519272,
      "grad_norm": 0.9716345071792603,
      "learning_rate": 0.00019624491748385553,
      "loss": 0.1691,
      "step": 7735
    },
    {
      "epoch": 1.0842326559215136,
      "grad_norm": 0.5784167051315308,
      "learning_rate": 0.00019623056685003586,
      "loss": 0.0718,
      "step": 7736
    },
    {
      "epoch": 1.0843728100911003,
      "grad_norm": 0.21612343192100525,
      "learning_rate": 0.0001962162162162162,
      "loss": 0.0363,
      "step": 7737
    },
    {
      "epoch": 1.0845129642606868,
      "grad_norm": 0.1700899749994278,
      "learning_rate": 0.00019620186558239652,
      "loss": 0.0255,
      "step": 7738
    },
    {
      "epoch": 1.0846531184302732,
      "grad_norm": 0.17835497856140137,
      "learning_rate": 0.00019618751494857687,
      "loss": 0.0176,
      "step": 7739
    },
    {
      "epoch": 1.08479327259986,
      "grad_norm": 0.3192072808742523,
      "learning_rate": 0.0001961731643147572,
      "loss": 0.0844,
      "step": 7740
    },
    {
      "epoch": 1.0849334267694464,
      "grad_norm": 0.20011286437511444,
      "learning_rate": 0.00019615881368093753,
      "loss": 0.035,
      "step": 7741
    },
    {
      "epoch": 1.0850735809390328,
      "grad_norm": 0.257863312959671,
      "learning_rate": 0.0001961444630471179,
      "loss": 0.0545,
      "step": 7742
    },
    {
      "epoch": 1.0852137351086195,
      "grad_norm": 0.2425508350133896,
      "learning_rate": 0.00019613011241329824,
      "loss": 0.0618,
      "step": 7743
    },
    {
      "epoch": 1.085353889278206,
      "grad_norm": 0.0386434942483902,
      "learning_rate": 0.00019611576177947857,
      "loss": 0.0057,
      "step": 7744
    },
    {
      "epoch": 1.0854940434477927,
      "grad_norm": 0.37644556164741516,
      "learning_rate": 0.00019610141114565893,
      "loss": 0.0628,
      "step": 7745
    },
    {
      "epoch": 1.0856341976173791,
      "grad_norm": 0.2679113745689392,
      "learning_rate": 0.00019608706051183925,
      "loss": 0.0394,
      "step": 7746
    },
    {
      "epoch": 1.0857743517869656,
      "grad_norm": 0.3189074397087097,
      "learning_rate": 0.00019607270987801958,
      "loss": 0.0614,
      "step": 7747
    },
    {
      "epoch": 1.0859145059565523,
      "grad_norm": 0.3783037066459656,
      "learning_rate": 0.00019605835924419994,
      "loss": 0.0644,
      "step": 7748
    },
    {
      "epoch": 1.0860546601261387,
      "grad_norm": 0.29174134135246277,
      "learning_rate": 0.00019604400861038027,
      "loss": 0.0538,
      "step": 7749
    },
    {
      "epoch": 1.0861948142957254,
      "grad_norm": 0.4524042010307312,
      "learning_rate": 0.0001960296579765606,
      "loss": 0.0855,
      "step": 7750
    },
    {
      "epoch": 1.0863349684653119,
      "grad_norm": 0.18859432637691498,
      "learning_rate": 0.00019601530734274098,
      "loss": 0.0343,
      "step": 7751
    },
    {
      "epoch": 1.0864751226348983,
      "grad_norm": 0.42731592059135437,
      "learning_rate": 0.0001960009567089213,
      "loss": 0.0333,
      "step": 7752
    },
    {
      "epoch": 1.086615276804485,
      "grad_norm": 0.292182058095932,
      "learning_rate": 0.00019598660607510164,
      "loss": 0.0481,
      "step": 7753
    },
    {
      "epoch": 1.0867554309740715,
      "grad_norm": 0.267025351524353,
      "learning_rate": 0.000195972255441282,
      "loss": 0.0493,
      "step": 7754
    },
    {
      "epoch": 1.0868955851436581,
      "grad_norm": 0.23354393243789673,
      "learning_rate": 0.00019595790480746232,
      "loss": 0.0209,
      "step": 7755
    },
    {
      "epoch": 1.0870357393132446,
      "grad_norm": 0.46044641733169556,
      "learning_rate": 0.00019594355417364265,
      "loss": 0.0509,
      "step": 7756
    },
    {
      "epoch": 1.087175893482831,
      "grad_norm": 0.29496869444847107,
      "learning_rate": 0.00019592920353982298,
      "loss": 0.0704,
      "step": 7757
    },
    {
      "epoch": 1.0873160476524177,
      "grad_norm": 0.16183353960514069,
      "learning_rate": 0.00019591485290600333,
      "loss": 0.0588,
      "step": 7758
    },
    {
      "epoch": 1.0874562018220042,
      "grad_norm": 0.17188377678394318,
      "learning_rate": 0.00019590050227218366,
      "loss": 0.0265,
      "step": 7759
    },
    {
      "epoch": 1.0875963559915907,
      "grad_norm": 0.5238999128341675,
      "learning_rate": 0.000195886151638364,
      "loss": 0.0577,
      "step": 7760
    },
    {
      "epoch": 1.0877365101611773,
      "grad_norm": 0.35187384486198425,
      "learning_rate": 0.00019587180100454437,
      "loss": 0.077,
      "step": 7761
    },
    {
      "epoch": 1.0878766643307638,
      "grad_norm": 0.5250095725059509,
      "learning_rate": 0.0001958574503707247,
      "loss": 0.0401,
      "step": 7762
    },
    {
      "epoch": 1.0880168185003505,
      "grad_norm": 0.28186026215553284,
      "learning_rate": 0.00019584309973690503,
      "loss": 0.0677,
      "step": 7763
    },
    {
      "epoch": 1.088156972669937,
      "grad_norm": 0.4049895405769348,
      "learning_rate": 0.00019582874910308539,
      "loss": 0.0458,
      "step": 7764
    },
    {
      "epoch": 1.0882971268395234,
      "grad_norm": 0.15566539764404297,
      "learning_rate": 0.00019581439846926571,
      "loss": 0.0241,
      "step": 7765
    },
    {
      "epoch": 1.08843728100911,
      "grad_norm": 0.4208690822124481,
      "learning_rate": 0.00019580004783544604,
      "loss": 0.0654,
      "step": 7766
    },
    {
      "epoch": 1.0885774351786965,
      "grad_norm": 0.34259408712387085,
      "learning_rate": 0.0001957856972016264,
      "loss": 0.0704,
      "step": 7767
    },
    {
      "epoch": 1.088717589348283,
      "grad_norm": 0.21498975157737732,
      "learning_rate": 0.00019577134656780673,
      "loss": 0.0431,
      "step": 7768
    },
    {
      "epoch": 1.0888577435178697,
      "grad_norm": 0.1400747001171112,
      "learning_rate": 0.00019575699593398706,
      "loss": 0.0317,
      "step": 7769
    },
    {
      "epoch": 1.0889978976874561,
      "grad_norm": 0.24307838082313538,
      "learning_rate": 0.0001957426453001674,
      "loss": 0.0196,
      "step": 7770
    },
    {
      "epoch": 1.0891380518570428,
      "grad_norm": 0.2923641800880432,
      "learning_rate": 0.00019572829466634774,
      "loss": 0.0406,
      "step": 7771
    },
    {
      "epoch": 1.0892782060266293,
      "grad_norm": 0.1953319013118744,
      "learning_rate": 0.00019571394403252807,
      "loss": 0.036,
      "step": 7772
    },
    {
      "epoch": 1.0894183601962157,
      "grad_norm": 0.8618971705436707,
      "learning_rate": 0.0001956995933987084,
      "loss": 0.0673,
      "step": 7773
    },
    {
      "epoch": 1.0895585143658024,
      "grad_norm": 0.4419419467449188,
      "learning_rate": 0.00019568524276488878,
      "loss": 0.0836,
      "step": 7774
    },
    {
      "epoch": 1.0896986685353889,
      "grad_norm": 0.08663631975650787,
      "learning_rate": 0.0001956708921310691,
      "loss": 0.0101,
      "step": 7775
    },
    {
      "epoch": 1.0898388227049756,
      "grad_norm": 0.4783826768398285,
      "learning_rate": 0.00019565654149724944,
      "loss": 0.1525,
      "step": 7776
    },
    {
      "epoch": 1.089978976874562,
      "grad_norm": 0.4597350060939789,
      "learning_rate": 0.0001956421908634298,
      "loss": 0.0287,
      "step": 7777
    },
    {
      "epoch": 1.0901191310441485,
      "grad_norm": 0.6316526532173157,
      "learning_rate": 0.00019562784022961012,
      "loss": 0.0976,
      "step": 7778
    },
    {
      "epoch": 1.0902592852137352,
      "grad_norm": 0.042646557092666626,
      "learning_rate": 0.00019561348959579045,
      "loss": 0.0051,
      "step": 7779
    },
    {
      "epoch": 1.0903994393833216,
      "grad_norm": 0.428884893655777,
      "learning_rate": 0.0001955991389619708,
      "loss": 0.1054,
      "step": 7780
    },
    {
      "epoch": 1.0905395935529083,
      "grad_norm": 0.7622066140174866,
      "learning_rate": 0.00019558478832815113,
      "loss": 0.0609,
      "step": 7781
    },
    {
      "epoch": 1.0906797477224948,
      "grad_norm": 2.064103364944458,
      "learning_rate": 0.00019557043769433146,
      "loss": 0.1038,
      "step": 7782
    },
    {
      "epoch": 1.0908199018920812,
      "grad_norm": 0.23380877077579498,
      "learning_rate": 0.00019555608706051184,
      "loss": 0.0207,
      "step": 7783
    },
    {
      "epoch": 1.090960056061668,
      "grad_norm": 1.3618347644805908,
      "learning_rate": 0.00019554173642669217,
      "loss": 0.0982,
      "step": 7784
    },
    {
      "epoch": 1.0911002102312544,
      "grad_norm": 1.470130443572998,
      "learning_rate": 0.0001955273857928725,
      "loss": 0.0433,
      "step": 7785
    },
    {
      "epoch": 1.0912403644008408,
      "grad_norm": 0.39666977524757385,
      "learning_rate": 0.00019551303515905286,
      "loss": 0.075,
      "step": 7786
    },
    {
      "epoch": 1.0913805185704275,
      "grad_norm": 0.361255407333374,
      "learning_rate": 0.00019549868452523319,
      "loss": 0.0794,
      "step": 7787
    },
    {
      "epoch": 1.091520672740014,
      "grad_norm": 0.19289886951446533,
      "learning_rate": 0.00019548433389141351,
      "loss": 0.0383,
      "step": 7788
    },
    {
      "epoch": 1.0916608269096006,
      "grad_norm": 0.2669353187084198,
      "learning_rate": 0.00019546998325759387,
      "loss": 0.0654,
      "step": 7789
    },
    {
      "epoch": 1.091800981079187,
      "grad_norm": 0.41004490852355957,
      "learning_rate": 0.0001954556326237742,
      "loss": 0.0621,
      "step": 7790
    },
    {
      "epoch": 1.0919411352487736,
      "grad_norm": 0.3903586268424988,
      "learning_rate": 0.00019544128198995453,
      "loss": 0.0591,
      "step": 7791
    },
    {
      "epoch": 1.0920812894183602,
      "grad_norm": 1.1857318878173828,
      "learning_rate": 0.00019542693135613486,
      "loss": 0.0433,
      "step": 7792
    },
    {
      "epoch": 1.0922214435879467,
      "grad_norm": 0.34951847791671753,
      "learning_rate": 0.00019541258072231524,
      "loss": 0.0382,
      "step": 7793
    },
    {
      "epoch": 1.0923615977575334,
      "grad_norm": 0.4225844740867615,
      "learning_rate": 0.00019539823008849557,
      "loss": 0.0547,
      "step": 7794
    },
    {
      "epoch": 1.0925017519271198,
      "grad_norm": 0.4104465842247009,
      "learning_rate": 0.0001953838794546759,
      "loss": 0.0685,
      "step": 7795
    },
    {
      "epoch": 1.0926419060967063,
      "grad_norm": 0.26032739877700806,
      "learning_rate": 0.00019536952882085625,
      "loss": 0.0495,
      "step": 7796
    },
    {
      "epoch": 1.092782060266293,
      "grad_norm": 0.3200533092021942,
      "learning_rate": 0.00019535517818703658,
      "loss": 0.0552,
      "step": 7797
    },
    {
      "epoch": 1.0929222144358794,
      "grad_norm": 0.28809860348701477,
      "learning_rate": 0.0001953408275532169,
      "loss": 0.0291,
      "step": 7798
    },
    {
      "epoch": 1.093062368605466,
      "grad_norm": 0.6311838030815125,
      "learning_rate": 0.00019532647691939726,
      "loss": 0.0824,
      "step": 7799
    },
    {
      "epoch": 1.0932025227750526,
      "grad_norm": 0.13497351109981537,
      "learning_rate": 0.0001953121262855776,
      "loss": 0.0185,
      "step": 7800
    },
    {
      "epoch": 1.093342676944639,
      "grad_norm": 0.538755476474762,
      "learning_rate": 0.00019529777565175792,
      "loss": 0.1242,
      "step": 7801
    },
    {
      "epoch": 1.0934828311142257,
      "grad_norm": 0.12650761008262634,
      "learning_rate": 0.00019528342501793828,
      "loss": 0.0085,
      "step": 7802
    },
    {
      "epoch": 1.0936229852838122,
      "grad_norm": 0.3782743215560913,
      "learning_rate": 0.0001952690743841186,
      "loss": 0.0395,
      "step": 7803
    },
    {
      "epoch": 1.0937631394533986,
      "grad_norm": 0.3198573887348175,
      "learning_rate": 0.00019525472375029893,
      "loss": 0.0527,
      "step": 7804
    },
    {
      "epoch": 1.0939032936229853,
      "grad_norm": 0.35259202122688293,
      "learning_rate": 0.00019524037311647932,
      "loss": 0.0415,
      "step": 7805
    },
    {
      "epoch": 1.0940434477925718,
      "grad_norm": 0.3345377743244171,
      "learning_rate": 0.00019522602248265965,
      "loss": 0.0512,
      "step": 7806
    },
    {
      "epoch": 1.0941836019621585,
      "grad_norm": 0.11956437677145004,
      "learning_rate": 0.00019521167184883997,
      "loss": 0.0085,
      "step": 7807
    },
    {
      "epoch": 1.094323756131745,
      "grad_norm": 0.2903411388397217,
      "learning_rate": 0.0001951973212150203,
      "loss": 0.0612,
      "step": 7808
    },
    {
      "epoch": 1.0944639103013314,
      "grad_norm": 0.35868003964424133,
      "learning_rate": 0.00019518297058120066,
      "loss": 0.0244,
      "step": 7809
    },
    {
      "epoch": 1.094604064470918,
      "grad_norm": 0.19841048121452332,
      "learning_rate": 0.000195168619947381,
      "loss": 0.0485,
      "step": 7810
    },
    {
      "epoch": 1.0947442186405045,
      "grad_norm": 0.5561944842338562,
      "learning_rate": 0.00019515426931356132,
      "loss": 0.0864,
      "step": 7811
    },
    {
      "epoch": 1.0948843728100912,
      "grad_norm": 0.37782612442970276,
      "learning_rate": 0.00019513991867974167,
      "loss": 0.059,
      "step": 7812
    },
    {
      "epoch": 1.0950245269796777,
      "grad_norm": 0.33996427059173584,
      "learning_rate": 0.000195125568045922,
      "loss": 0.0857,
      "step": 7813
    },
    {
      "epoch": 1.0951646811492641,
      "grad_norm": 0.460025429725647,
      "learning_rate": 0.00019511121741210233,
      "loss": 0.0561,
      "step": 7814
    },
    {
      "epoch": 1.0953048353188508,
      "grad_norm": 0.11655977368354797,
      "learning_rate": 0.0001950968667782827,
      "loss": 0.0262,
      "step": 7815
    },
    {
      "epoch": 1.0954449894884373,
      "grad_norm": 0.2663809359073639,
      "learning_rate": 0.00019508251614446304,
      "loss": 0.0381,
      "step": 7816
    },
    {
      "epoch": 1.0955851436580237,
      "grad_norm": 0.48363324999809265,
      "learning_rate": 0.00019506816551064337,
      "loss": 0.0624,
      "step": 7817
    },
    {
      "epoch": 1.0957252978276104,
      "grad_norm": 1.2686469554901123,
      "learning_rate": 0.00019505381487682372,
      "loss": 0.0867,
      "step": 7818
    },
    {
      "epoch": 1.0958654519971969,
      "grad_norm": 0.3008817136287689,
      "learning_rate": 0.00019503946424300405,
      "loss": 0.0313,
      "step": 7819
    },
    {
      "epoch": 1.0960056061667836,
      "grad_norm": 0.6708664298057556,
      "learning_rate": 0.00019502511360918438,
      "loss": 0.1013,
      "step": 7820
    },
    {
      "epoch": 1.09614576033637,
      "grad_norm": 0.5214825868606567,
      "learning_rate": 0.00019501076297536474,
      "loss": 0.0689,
      "step": 7821
    },
    {
      "epoch": 1.0962859145059565,
      "grad_norm": 0.3507421314716339,
      "learning_rate": 0.00019499641234154507,
      "loss": 0.0655,
      "step": 7822
    },
    {
      "epoch": 1.0964260686755432,
      "grad_norm": 0.2852841317653656,
      "learning_rate": 0.0001949820617077254,
      "loss": 0.0591,
      "step": 7823
    },
    {
      "epoch": 1.0965662228451296,
      "grad_norm": 0.13078312575817108,
      "learning_rate": 0.00019496771107390578,
      "loss": 0.0128,
      "step": 7824
    },
    {
      "epoch": 1.096706377014716,
      "grad_norm": 0.43701568245887756,
      "learning_rate": 0.0001949533604400861,
      "loss": 0.0255,
      "step": 7825
    },
    {
      "epoch": 1.0968465311843028,
      "grad_norm": 0.7900677919387817,
      "learning_rate": 0.00019493900980626643,
      "loss": 0.1769,
      "step": 7826
    },
    {
      "epoch": 1.0969866853538892,
      "grad_norm": 0.0890909880399704,
      "learning_rate": 0.00019492465917244676,
      "loss": 0.0095,
      "step": 7827
    },
    {
      "epoch": 1.097126839523476,
      "grad_norm": 0.9859515428543091,
      "learning_rate": 0.00019491030853862712,
      "loss": 0.1583,
      "step": 7828
    },
    {
      "epoch": 1.0972669936930624,
      "grad_norm": 0.23981811106204987,
      "learning_rate": 0.00019489595790480745,
      "loss": 0.0145,
      "step": 7829
    },
    {
      "epoch": 1.0974071478626488,
      "grad_norm": 1.265547513961792,
      "learning_rate": 0.00019488160727098777,
      "loss": 0.2236,
      "step": 7830
    },
    {
      "epoch": 1.0975473020322355,
      "grad_norm": 0.7176455855369568,
      "learning_rate": 0.00019486725663716813,
      "loss": 0.0666,
      "step": 7831
    },
    {
      "epoch": 1.097687456201822,
      "grad_norm": 0.6914899349212646,
      "learning_rate": 0.00019485290600334846,
      "loss": 0.0834,
      "step": 7832
    },
    {
      "epoch": 1.0978276103714086,
      "grad_norm": 0.7753621935844421,
      "learning_rate": 0.0001948385553695288,
      "loss": 0.1039,
      "step": 7833
    },
    {
      "epoch": 1.097967764540995,
      "grad_norm": 0.7670572996139526,
      "learning_rate": 0.00019482420473570914,
      "loss": 0.0345,
      "step": 7834
    },
    {
      "epoch": 1.0981079187105816,
      "grad_norm": 0.8203732967376709,
      "learning_rate": 0.00019480985410188947,
      "loss": 0.0438,
      "step": 7835
    },
    {
      "epoch": 1.0982480728801682,
      "grad_norm": 0.294342577457428,
      "learning_rate": 0.0001947955034680698,
      "loss": 0.0478,
      "step": 7836
    },
    {
      "epoch": 1.0983882270497547,
      "grad_norm": 0.3223698139190674,
      "learning_rate": 0.00019478115283425018,
      "loss": 0.0473,
      "step": 7837
    },
    {
      "epoch": 1.0985283812193414,
      "grad_norm": 0.4793541729450226,
      "learning_rate": 0.0001947668022004305,
      "loss": 0.0868,
      "step": 7838
    },
    {
      "epoch": 1.0986685353889278,
      "grad_norm": 0.29572466015815735,
      "learning_rate": 0.00019475245156661084,
      "loss": 0.0739,
      "step": 7839
    },
    {
      "epoch": 1.0988086895585143,
      "grad_norm": 0.18741720914840698,
      "learning_rate": 0.0001947381009327912,
      "loss": 0.0351,
      "step": 7840
    },
    {
      "epoch": 1.098948843728101,
      "grad_norm": 0.3839675784111023,
      "learning_rate": 0.00019472375029897152,
      "loss": 0.0686,
      "step": 7841
    },
    {
      "epoch": 1.0990889978976874,
      "grad_norm": 0.29997843503952026,
      "learning_rate": 0.00019470939966515185,
      "loss": 0.0819,
      "step": 7842
    },
    {
      "epoch": 1.0992291520672741,
      "grad_norm": 0.3279254138469696,
      "learning_rate": 0.00019469504903133218,
      "loss": 0.046,
      "step": 7843
    },
    {
      "epoch": 1.0993693062368606,
      "grad_norm": 0.47016963362693787,
      "learning_rate": 0.00019468069839751254,
      "loss": 0.0352,
      "step": 7844
    },
    {
      "epoch": 1.099509460406447,
      "grad_norm": 0.2271425575017929,
      "learning_rate": 0.00019466634776369287,
      "loss": 0.0291,
      "step": 7845
    },
    {
      "epoch": 1.0996496145760337,
      "grad_norm": 0.1966121643781662,
      "learning_rate": 0.0001946519971298732,
      "loss": 0.0528,
      "step": 7846
    },
    {
      "epoch": 1.0997897687456202,
      "grad_norm": 0.168510302901268,
      "learning_rate": 0.00019463764649605358,
      "loss": 0.0165,
      "step": 7847
    },
    {
      "epoch": 1.0999299229152066,
      "grad_norm": 0.4008927047252655,
      "learning_rate": 0.0001946232958622339,
      "loss": 0.0504,
      "step": 7848
    },
    {
      "epoch": 1.1000700770847933,
      "grad_norm": 0.21007902920246124,
      "learning_rate": 0.00019460894522841423,
      "loss": 0.0661,
      "step": 7849
    },
    {
      "epoch": 1.1002102312543798,
      "grad_norm": 0.4335603713989258,
      "learning_rate": 0.0001945945945945946,
      "loss": 0.0645,
      "step": 7850
    },
    {
      "epoch": 1.1003503854239665,
      "grad_norm": 0.11960671842098236,
      "learning_rate": 0.00019458024396077492,
      "loss": 0.0088,
      "step": 7851
    },
    {
      "epoch": 1.100490539593553,
      "grad_norm": 0.2624993324279785,
      "learning_rate": 0.00019456589332695525,
      "loss": 0.0466,
      "step": 7852
    },
    {
      "epoch": 1.1006306937631394,
      "grad_norm": 0.26366522908210754,
      "learning_rate": 0.0001945515426931356,
      "loss": 0.0244,
      "step": 7853
    },
    {
      "epoch": 1.100770847932726,
      "grad_norm": 0.19341231882572174,
      "learning_rate": 0.00019453719205931593,
      "loss": 0.0441,
      "step": 7854
    },
    {
      "epoch": 1.1009110021023125,
      "grad_norm": 0.16881239414215088,
      "learning_rate": 0.00019452284142549626,
      "loss": 0.0164,
      "step": 7855
    },
    {
      "epoch": 1.101051156271899,
      "grad_norm": 0.13859426975250244,
      "learning_rate": 0.00019450849079167664,
      "loss": 0.0238,
      "step": 7856
    },
    {
      "epoch": 1.1011913104414857,
      "grad_norm": 0.4294334053993225,
      "learning_rate": 0.00019449414015785697,
      "loss": 0.1345,
      "step": 7857
    },
    {
      "epoch": 1.1013314646110721,
      "grad_norm": 0.8766242861747742,
      "learning_rate": 0.0001944797895240373,
      "loss": 0.0764,
      "step": 7858
    },
    {
      "epoch": 1.1014716187806588,
      "grad_norm": 0.3806171119213104,
      "learning_rate": 0.00019446543889021766,
      "loss": 0.057,
      "step": 7859
    },
    {
      "epoch": 1.1016117729502453,
      "grad_norm": 0.7184103727340698,
      "learning_rate": 0.00019445108825639798,
      "loss": 0.0152,
      "step": 7860
    },
    {
      "epoch": 1.1017519271198317,
      "grad_norm": 0.41738948225975037,
      "learning_rate": 0.0001944367376225783,
      "loss": 0.0465,
      "step": 7861
    },
    {
      "epoch": 1.1018920812894184,
      "grad_norm": 0.3431372046470642,
      "learning_rate": 0.00019442238698875864,
      "loss": 0.0433,
      "step": 7862
    },
    {
      "epoch": 1.1020322354590049,
      "grad_norm": 0.1914280205965042,
      "learning_rate": 0.000194408036354939,
      "loss": 0.024,
      "step": 7863
    },
    {
      "epoch": 1.1021723896285915,
      "grad_norm": 0.2511115074157715,
      "learning_rate": 0.00019439368572111933,
      "loss": 0.0319,
      "step": 7864
    },
    {
      "epoch": 1.102312543798178,
      "grad_norm": 0.7820152640342712,
      "learning_rate": 0.00019437933508729965,
      "loss": 0.0716,
      "step": 7865
    },
    {
      "epoch": 1.1024526979677645,
      "grad_norm": 0.6725873351097107,
      "learning_rate": 0.00019436498445348,
      "loss": 0.0646,
      "step": 7866
    },
    {
      "epoch": 1.1025928521373511,
      "grad_norm": 0.2968751788139343,
      "learning_rate": 0.00019435063381966034,
      "loss": 0.1052,
      "step": 7867
    },
    {
      "epoch": 1.1027330063069376,
      "grad_norm": 0.5666132569313049,
      "learning_rate": 0.00019433628318584067,
      "loss": 0.0496,
      "step": 7868
    },
    {
      "epoch": 1.1028731604765243,
      "grad_norm": 0.698685348033905,
      "learning_rate": 0.00019432193255202105,
      "loss": 0.0556,
      "step": 7869
    },
    {
      "epoch": 1.1030133146461107,
      "grad_norm": 0.3263324797153473,
      "learning_rate": 0.00019430758191820138,
      "loss": 0.0543,
      "step": 7870
    },
    {
      "epoch": 1.1031534688156972,
      "grad_norm": 0.36369356513023376,
      "learning_rate": 0.0001942932312843817,
      "loss": 0.0692,
      "step": 7871
    },
    {
      "epoch": 1.1032936229852839,
      "grad_norm": 0.2374008446931839,
      "learning_rate": 0.00019427888065056206,
      "loss": 0.022,
      "step": 7872
    },
    {
      "epoch": 1.1034337771548703,
      "grad_norm": 0.4797550439834595,
      "learning_rate": 0.0001942645300167424,
      "loss": 0.0587,
      "step": 7873
    },
    {
      "epoch": 1.1035739313244568,
      "grad_norm": 0.27886685729026794,
      "learning_rate": 0.00019425017938292272,
      "loss": 0.0417,
      "step": 7874
    },
    {
      "epoch": 1.1037140854940435,
      "grad_norm": 0.17206478118896484,
      "learning_rate": 0.00019423582874910308,
      "loss": 0.0143,
      "step": 7875
    },
    {
      "epoch": 1.10385423966363,
      "grad_norm": 0.15134984254837036,
      "learning_rate": 0.0001942214781152834,
      "loss": 0.0127,
      "step": 7876
    },
    {
      "epoch": 1.1039943938332166,
      "grad_norm": 0.6880931854248047,
      "learning_rate": 0.00019420712748146373,
      "loss": 0.0744,
      "step": 7877
    },
    {
      "epoch": 1.104134548002803,
      "grad_norm": 0.12138552963733673,
      "learning_rate": 0.00019419277684764406,
      "loss": 0.0127,
      "step": 7878
    },
    {
      "epoch": 1.1042747021723895,
      "grad_norm": 0.052256833761930466,
      "learning_rate": 0.00019417842621382444,
      "loss": 0.0037,
      "step": 7879
    },
    {
      "epoch": 1.1044148563419762,
      "grad_norm": 0.8468590974807739,
      "learning_rate": 0.00019416407558000477,
      "loss": 0.1146,
      "step": 7880
    },
    {
      "epoch": 1.1045550105115627,
      "grad_norm": 0.3638436496257782,
      "learning_rate": 0.0001941497249461851,
      "loss": 0.0854,
      "step": 7881
    },
    {
      "epoch": 1.1046951646811494,
      "grad_norm": 0.3517143428325653,
      "learning_rate": 0.00019413537431236546,
      "loss": 0.0199,
      "step": 7882
    },
    {
      "epoch": 1.1048353188507358,
      "grad_norm": 1.5876296758651733,
      "learning_rate": 0.00019412102367854578,
      "loss": 0.1463,
      "step": 7883
    },
    {
      "epoch": 1.1049754730203223,
      "grad_norm": 1.5338149070739746,
      "learning_rate": 0.0001941066730447261,
      "loss": 0.0748,
      "step": 7884
    },
    {
      "epoch": 1.105115627189909,
      "grad_norm": 3.5674824714660645,
      "learning_rate": 0.00019409232241090647,
      "loss": 0.3716,
      "step": 7885
    },
    {
      "epoch": 1.1052557813594954,
      "grad_norm": 0.20747163891792297,
      "learning_rate": 0.0001940779717770868,
      "loss": 0.0271,
      "step": 7886
    },
    {
      "epoch": 1.1053959355290819,
      "grad_norm": 0.28685444593429565,
      "learning_rate": 0.00019406362114326713,
      "loss": 0.0294,
      "step": 7887
    },
    {
      "epoch": 1.1055360896986686,
      "grad_norm": 0.2830279469490051,
      "learning_rate": 0.0001940492705094475,
      "loss": 0.0578,
      "step": 7888
    },
    {
      "epoch": 1.105676243868255,
      "grad_norm": 0.5834911465644836,
      "learning_rate": 0.00019403491987562784,
      "loss": 0.0955,
      "step": 7889
    },
    {
      "epoch": 1.1058163980378417,
      "grad_norm": 0.35689404606819153,
      "learning_rate": 0.00019402056924180817,
      "loss": 0.0254,
      "step": 7890
    },
    {
      "epoch": 1.1059565522074282,
      "grad_norm": 0.15401045978069305,
      "learning_rate": 0.00019400621860798852,
      "loss": 0.0338,
      "step": 7891
    },
    {
      "epoch": 1.1060967063770146,
      "grad_norm": 0.3949108123779297,
      "learning_rate": 0.00019399186797416885,
      "loss": 0.066,
      "step": 7892
    },
    {
      "epoch": 1.1062368605466013,
      "grad_norm": 0.5664955973625183,
      "learning_rate": 0.00019397751734034918,
      "loss": 0.0946,
      "step": 7893
    },
    {
      "epoch": 1.1063770147161878,
      "grad_norm": 0.1789037585258484,
      "learning_rate": 0.0001939631667065295,
      "loss": 0.043,
      "step": 7894
    },
    {
      "epoch": 1.1065171688857744,
      "grad_norm": 0.18390746414661407,
      "learning_rate": 0.00019394881607270986,
      "loss": 0.0346,
      "step": 7895
    },
    {
      "epoch": 1.106657323055361,
      "grad_norm": 0.8283275365829468,
      "learning_rate": 0.0001939344654388902,
      "loss": 0.1155,
      "step": 7896
    },
    {
      "epoch": 1.1067974772249474,
      "grad_norm": 0.17174625396728516,
      "learning_rate": 0.00019392011480507052,
      "loss": 0.024,
      "step": 7897
    },
    {
      "epoch": 1.106937631394534,
      "grad_norm": 0.7722519636154175,
      "learning_rate": 0.00019390576417125088,
      "loss": 0.0368,
      "step": 7898
    },
    {
      "epoch": 1.1070777855641205,
      "grad_norm": 0.5638456344604492,
      "learning_rate": 0.0001938914135374312,
      "loss": 0.0785,
      "step": 7899
    },
    {
      "epoch": 1.1072179397337072,
      "grad_norm": 0.2936611473560333,
      "learning_rate": 0.00019387706290361153,
      "loss": 0.0744,
      "step": 7900
    },
    {
      "epoch": 1.1073580939032936,
      "grad_norm": 0.23863860964775085,
      "learning_rate": 0.00019386271226979192,
      "loss": 0.0716,
      "step": 7901
    },
    {
      "epoch": 1.10749824807288,
      "grad_norm": 0.8755089044570923,
      "learning_rate": 0.00019384836163597224,
      "loss": 0.1022,
      "step": 7902
    },
    {
      "epoch": 1.1076384022424668,
      "grad_norm": 0.2535272538661957,
      "learning_rate": 0.00019383401100215257,
      "loss": 0.0487,
      "step": 7903
    },
    {
      "epoch": 1.1077785564120533,
      "grad_norm": 0.23830698430538177,
      "learning_rate": 0.00019381966036833293,
      "loss": 0.023,
      "step": 7904
    },
    {
      "epoch": 1.1079187105816397,
      "grad_norm": 0.37590092420578003,
      "learning_rate": 0.00019380530973451326,
      "loss": 0.0313,
      "step": 7905
    },
    {
      "epoch": 1.1080588647512264,
      "grad_norm": 0.509615957736969,
      "learning_rate": 0.00019379095910069359,
      "loss": 0.0831,
      "step": 7906
    },
    {
      "epoch": 1.1081990189208129,
      "grad_norm": 0.6688640117645264,
      "learning_rate": 0.00019377660846687394,
      "loss": 0.1068,
      "step": 7907
    },
    {
      "epoch": 1.1083391730903995,
      "grad_norm": 0.45249491930007935,
      "learning_rate": 0.00019376225783305427,
      "loss": 0.0703,
      "step": 7908
    },
    {
      "epoch": 1.108479327259986,
      "grad_norm": 0.4076036810874939,
      "learning_rate": 0.0001937479071992346,
      "loss": 0.0682,
      "step": 7909
    },
    {
      "epoch": 1.1086194814295725,
      "grad_norm": 0.30717846751213074,
      "learning_rate": 0.00019373355656541498,
      "loss": 0.0207,
      "step": 7910
    },
    {
      "epoch": 1.1087596355991591,
      "grad_norm": 0.638920783996582,
      "learning_rate": 0.0001937192059315953,
      "loss": 0.0714,
      "step": 7911
    },
    {
      "epoch": 1.1088997897687456,
      "grad_norm": 0.2553742527961731,
      "learning_rate": 0.00019370485529777564,
      "loss": 0.0499,
      "step": 7912
    },
    {
      "epoch": 1.109039943938332,
      "grad_norm": 0.3501196801662445,
      "learning_rate": 0.00019369050466395597,
      "loss": 0.0474,
      "step": 7913
    },
    {
      "epoch": 1.1091800981079187,
      "grad_norm": 0.2916560769081116,
      "learning_rate": 0.00019367615403013632,
      "loss": 0.0264,
      "step": 7914
    },
    {
      "epoch": 1.1093202522775052,
      "grad_norm": 0.5131843686103821,
      "learning_rate": 0.00019366180339631665,
      "loss": 0.0734,
      "step": 7915
    },
    {
      "epoch": 1.1094604064470919,
      "grad_norm": 0.5559123754501343,
      "learning_rate": 0.00019364745276249698,
      "loss": 0.0641,
      "step": 7916
    },
    {
      "epoch": 1.1096005606166783,
      "grad_norm": 0.29364800453186035,
      "learning_rate": 0.00019363310212867734,
      "loss": 0.0599,
      "step": 7917
    },
    {
      "epoch": 1.1097407147862648,
      "grad_norm": 0.3228829801082611,
      "learning_rate": 0.00019361875149485766,
      "loss": 0.0461,
      "step": 7918
    },
    {
      "epoch": 1.1098808689558515,
      "grad_norm": 0.17589092254638672,
      "learning_rate": 0.000193604400861038,
      "loss": 0.0328,
      "step": 7919
    },
    {
      "epoch": 1.110021023125438,
      "grad_norm": 0.500698447227478,
      "learning_rate": 0.00019359005022721838,
      "loss": 0.0575,
      "step": 7920
    },
    {
      "epoch": 1.1101611772950246,
      "grad_norm": 0.17207537591457367,
      "learning_rate": 0.0001935756995933987,
      "loss": 0.0301,
      "step": 7921
    },
    {
      "epoch": 1.110301331464611,
      "grad_norm": 0.3140786588191986,
      "learning_rate": 0.00019356134895957903,
      "loss": 0.035,
      "step": 7922
    },
    {
      "epoch": 1.1104414856341975,
      "grad_norm": 0.12867513298988342,
      "learning_rate": 0.0001935469983257594,
      "loss": 0.0225,
      "step": 7923
    },
    {
      "epoch": 1.1105816398037842,
      "grad_norm": 0.2635933756828308,
      "learning_rate": 0.00019353264769193972,
      "loss": 0.0463,
      "step": 7924
    },
    {
      "epoch": 1.1107217939733707,
      "grad_norm": 0.24610888957977295,
      "learning_rate": 0.00019351829705812005,
      "loss": 0.0354,
      "step": 7925
    },
    {
      "epoch": 1.1108619481429574,
      "grad_norm": 0.3536587357521057,
      "learning_rate": 0.0001935039464243004,
      "loss": 0.0819,
      "step": 7926
    },
    {
      "epoch": 1.1110021023125438,
      "grad_norm": 1.0215599536895752,
      "learning_rate": 0.00019348959579048073,
      "loss": 0.0996,
      "step": 7927
    },
    {
      "epoch": 1.1111422564821303,
      "grad_norm": 0.6937496066093445,
      "learning_rate": 0.00019347524515666106,
      "loss": 0.1081,
      "step": 7928
    },
    {
      "epoch": 1.111282410651717,
      "grad_norm": 0.6525269746780396,
      "learning_rate": 0.0001934608945228414,
      "loss": 0.0953,
      "step": 7929
    },
    {
      "epoch": 1.1114225648213034,
      "grad_norm": 0.6573078632354736,
      "learning_rate": 0.00019344654388902174,
      "loss": 0.0793,
      "step": 7930
    },
    {
      "epoch": 1.1115627189908899,
      "grad_norm": 0.887570858001709,
      "learning_rate": 0.00019343219325520207,
      "loss": 0.158,
      "step": 7931
    },
    {
      "epoch": 1.1117028731604766,
      "grad_norm": 0.6558364033699036,
      "learning_rate": 0.0001934178426213824,
      "loss": 0.0635,
      "step": 7932
    },
    {
      "epoch": 1.111843027330063,
      "grad_norm": 0.5542738437652588,
      "learning_rate": 0.00019340349198756278,
      "loss": 0.0804,
      "step": 7933
    },
    {
      "epoch": 1.1119831814996497,
      "grad_norm": 4.551071643829346,
      "learning_rate": 0.0001933891413537431,
      "loss": 0.1388,
      "step": 7934
    },
    {
      "epoch": 1.1121233356692362,
      "grad_norm": 0.28740426898002625,
      "learning_rate": 0.00019337479071992344,
      "loss": 0.0198,
      "step": 7935
    },
    {
      "epoch": 1.1122634898388226,
      "grad_norm": 0.18654921650886536,
      "learning_rate": 0.0001933604400861038,
      "loss": 0.0505,
      "step": 7936
    },
    {
      "epoch": 1.1124036440084093,
      "grad_norm": 0.22372834384441376,
      "learning_rate": 0.00019334608945228412,
      "loss": 0.0326,
      "step": 7937
    },
    {
      "epoch": 1.1125437981779958,
      "grad_norm": 0.45720720291137695,
      "learning_rate": 0.00019333173881846445,
      "loss": 0.1098,
      "step": 7938
    },
    {
      "epoch": 1.1126839523475824,
      "grad_norm": 0.454596608877182,
      "learning_rate": 0.0001933173881846448,
      "loss": 0.0723,
      "step": 7939
    },
    {
      "epoch": 1.112824106517169,
      "grad_norm": 0.4282210171222687,
      "learning_rate": 0.00019330303755082514,
      "loss": 0.1123,
      "step": 7940
    },
    {
      "epoch": 1.1129642606867554,
      "grad_norm": 0.16677014529705048,
      "learning_rate": 0.00019328868691700546,
      "loss": 0.0358,
      "step": 7941
    },
    {
      "epoch": 1.113104414856342,
      "grad_norm": 0.23977968096733093,
      "learning_rate": 0.00019327433628318585,
      "loss": 0.0483,
      "step": 7942
    },
    {
      "epoch": 1.1132445690259285,
      "grad_norm": 0.13637958467006683,
      "learning_rate": 0.00019325998564936618,
      "loss": 0.0165,
      "step": 7943
    },
    {
      "epoch": 1.113384723195515,
      "grad_norm": 0.29663246870040894,
      "learning_rate": 0.0001932456350155465,
      "loss": 0.0327,
      "step": 7944
    },
    {
      "epoch": 1.1135248773651016,
      "grad_norm": 0.1970030665397644,
      "learning_rate": 0.00019323128438172686,
      "loss": 0.0436,
      "step": 7945
    },
    {
      "epoch": 1.113665031534688,
      "grad_norm": 0.3472856283187866,
      "learning_rate": 0.0001932169337479072,
      "loss": 0.0412,
      "step": 7946
    },
    {
      "epoch": 1.1138051857042748,
      "grad_norm": 0.18037283420562744,
      "learning_rate": 0.00019320258311408752,
      "loss": 0.047,
      "step": 7947
    },
    {
      "epoch": 1.1139453398738612,
      "grad_norm": 0.275040864944458,
      "learning_rate": 0.00019318823248026785,
      "loss": 0.0368,
      "step": 7948
    },
    {
      "epoch": 1.1140854940434477,
      "grad_norm": 0.30739516019821167,
      "learning_rate": 0.0001931738818464482,
      "loss": 0.0384,
      "step": 7949
    },
    {
      "epoch": 1.1142256482130344,
      "grad_norm": 0.5042182803153992,
      "learning_rate": 0.00019315953121262853,
      "loss": 0.1226,
      "step": 7950
    },
    {
      "epoch": 1.1143658023826208,
      "grad_norm": 0.22179976105690002,
      "learning_rate": 0.00019314518057880886,
      "loss": 0.0448,
      "step": 7951
    },
    {
      "epoch": 1.1145059565522075,
      "grad_norm": 0.5325523018836975,
      "learning_rate": 0.00019313082994498924,
      "loss": 0.0851,
      "step": 7952
    },
    {
      "epoch": 1.114646110721794,
      "grad_norm": 0.36900416016578674,
      "learning_rate": 0.00019311647931116957,
      "loss": 0.0827,
      "step": 7953
    },
    {
      "epoch": 1.1147862648913804,
      "grad_norm": 0.406634658575058,
      "learning_rate": 0.0001931021286773499,
      "loss": 0.0681,
      "step": 7954
    },
    {
      "epoch": 1.1149264190609671,
      "grad_norm": 0.4363675117492676,
      "learning_rate": 0.00019308777804353025,
      "loss": 0.0941,
      "step": 7955
    },
    {
      "epoch": 1.1150665732305536,
      "grad_norm": 0.29236674308776855,
      "learning_rate": 0.00019307342740971058,
      "loss": 0.0554,
      "step": 7956
    },
    {
      "epoch": 1.1152067274001403,
      "grad_norm": 0.1862139105796814,
      "learning_rate": 0.0001930590767758909,
      "loss": 0.0119,
      "step": 7957
    },
    {
      "epoch": 1.1153468815697267,
      "grad_norm": 0.31448566913604736,
      "learning_rate": 0.00019304472614207127,
      "loss": 0.0381,
      "step": 7958
    },
    {
      "epoch": 1.1154870357393132,
      "grad_norm": 0.47054511308670044,
      "learning_rate": 0.0001930303755082516,
      "loss": 0.0672,
      "step": 7959
    },
    {
      "epoch": 1.1156271899088999,
      "grad_norm": 0.35249945521354675,
      "learning_rate": 0.00019301602487443192,
      "loss": 0.0507,
      "step": 7960
    },
    {
      "epoch": 1.1157673440784863,
      "grad_norm": 0.44620150327682495,
      "learning_rate": 0.00019300167424061228,
      "loss": 0.1127,
      "step": 7961
    },
    {
      "epoch": 1.1159074982480728,
      "grad_norm": 0.2217017561197281,
      "learning_rate": 0.0001929873236067926,
      "loss": 0.0537,
      "step": 7962
    },
    {
      "epoch": 1.1160476524176595,
      "grad_norm": 0.6119962930679321,
      "learning_rate": 0.00019297297297297294,
      "loss": 0.0887,
      "step": 7963
    },
    {
      "epoch": 1.116187806587246,
      "grad_norm": 0.8834139108657837,
      "learning_rate": 0.0001929586223391533,
      "loss": 0.097,
      "step": 7964
    },
    {
      "epoch": 1.1163279607568326,
      "grad_norm": 0.3024270832538605,
      "learning_rate": 0.00019294427170533365,
      "loss": 0.0527,
      "step": 7965
    },
    {
      "epoch": 1.116468114926419,
      "grad_norm": 0.14427095651626587,
      "learning_rate": 0.00019292992107151398,
      "loss": 0.038,
      "step": 7966
    },
    {
      "epoch": 1.1166082690960055,
      "grad_norm": 0.17836223542690277,
      "learning_rate": 0.0001929155704376943,
      "loss": 0.0412,
      "step": 7967
    },
    {
      "epoch": 1.1167484232655922,
      "grad_norm": 0.19944356381893158,
      "learning_rate": 0.00019290121980387466,
      "loss": 0.0431,
      "step": 7968
    },
    {
      "epoch": 1.1168885774351787,
      "grad_norm": 0.34181687235832214,
      "learning_rate": 0.000192886869170055,
      "loss": 0.0626,
      "step": 7969
    },
    {
      "epoch": 1.1170287316047651,
      "grad_norm": 0.27578863501548767,
      "learning_rate": 0.00019287251853623532,
      "loss": 0.0718,
      "step": 7970
    },
    {
      "epoch": 1.1171688857743518,
      "grad_norm": 0.4260280728340149,
      "learning_rate": 0.00019285816790241567,
      "loss": 0.1472,
      "step": 7971
    },
    {
      "epoch": 1.1173090399439383,
      "grad_norm": 0.9106219410896301,
      "learning_rate": 0.000192843817268596,
      "loss": 0.1447,
      "step": 7972
    },
    {
      "epoch": 1.117449194113525,
      "grad_norm": 0.27554360032081604,
      "learning_rate": 0.00019282946663477633,
      "loss": 0.0625,
      "step": 7973
    },
    {
      "epoch": 1.1175893482831114,
      "grad_norm": 0.1750875562429428,
      "learning_rate": 0.00019281511600095671,
      "loss": 0.0558,
      "step": 7974
    },
    {
      "epoch": 1.1177295024526979,
      "grad_norm": 0.24339143931865692,
      "learning_rate": 0.00019280076536713704,
      "loss": 0.0211,
      "step": 7975
    },
    {
      "epoch": 1.1178696566222845,
      "grad_norm": 0.4519350826740265,
      "learning_rate": 0.00019278641473331737,
      "loss": 0.0541,
      "step": 7976
    },
    {
      "epoch": 1.118009810791871,
      "grad_norm": 0.7721365094184875,
      "learning_rate": 0.00019277206409949773,
      "loss": 0.0833,
      "step": 7977
    },
    {
      "epoch": 1.1181499649614577,
      "grad_norm": 0.7545231580734253,
      "learning_rate": 0.00019275771346567806,
      "loss": 0.1039,
      "step": 7978
    },
    {
      "epoch": 1.1182901191310441,
      "grad_norm": 0.9306256771087646,
      "learning_rate": 0.00019274336283185838,
      "loss": 0.0362,
      "step": 7979
    },
    {
      "epoch": 1.1184302733006306,
      "grad_norm": 1.333509087562561,
      "learning_rate": 0.00019272901219803874,
      "loss": 0.2061,
      "step": 7980
    },
    {
      "epoch": 1.1185704274702173,
      "grad_norm": 0.5061262845993042,
      "learning_rate": 0.00019271466156421907,
      "loss": 0.0393,
      "step": 7981
    },
    {
      "epoch": 1.1187105816398037,
      "grad_norm": 0.589859127998352,
      "learning_rate": 0.0001927003109303994,
      "loss": 0.0785,
      "step": 7982
    },
    {
      "epoch": 1.1188507358093904,
      "grad_norm": 0.5165272355079651,
      "learning_rate": 0.00019268596029657973,
      "loss": 0.0717,
      "step": 7983
    },
    {
      "epoch": 1.1189908899789769,
      "grad_norm": 0.7961622476577759,
      "learning_rate": 0.0001926716096627601,
      "loss": 0.0349,
      "step": 7984
    },
    {
      "epoch": 1.1191310441485633,
      "grad_norm": 3.2725977897644043,
      "learning_rate": 0.00019265725902894044,
      "loss": 0.2943,
      "step": 7985
    },
    {
      "epoch": 1.11927119831815,
      "grad_norm": 0.44366204738616943,
      "learning_rate": 0.00019264290839512076,
      "loss": 0.0958,
      "step": 7986
    },
    {
      "epoch": 1.1194113524877365,
      "grad_norm": 0.5367948412895203,
      "learning_rate": 0.00019262855776130112,
      "loss": 0.0859,
      "step": 7987
    },
    {
      "epoch": 1.1195515066573232,
      "grad_norm": 0.18503437936306,
      "learning_rate": 0.00019261420712748145,
      "loss": 0.0378,
      "step": 7988
    },
    {
      "epoch": 1.1196916608269096,
      "grad_norm": 0.2635434567928314,
      "learning_rate": 0.00019259985649366178,
      "loss": 0.044,
      "step": 7989
    },
    {
      "epoch": 1.119831814996496,
      "grad_norm": 0.5049416422843933,
      "learning_rate": 0.00019258550585984213,
      "loss": 0.071,
      "step": 7990
    },
    {
      "epoch": 1.1199719691660828,
      "grad_norm": 0.5209174156188965,
      "learning_rate": 0.00019257115522602246,
      "loss": 0.0555,
      "step": 7991
    },
    {
      "epoch": 1.1201121233356692,
      "grad_norm": 0.2113497406244278,
      "learning_rate": 0.0001925568045922028,
      "loss": 0.0416,
      "step": 7992
    },
    {
      "epoch": 1.1202522775052557,
      "grad_norm": 0.28367385268211365,
      "learning_rate": 0.00019254245395838315,
      "loss": 0.0612,
      "step": 7993
    },
    {
      "epoch": 1.1203924316748424,
      "grad_norm": 0.40710994601249695,
      "learning_rate": 0.00019252810332456347,
      "loss": 0.0761,
      "step": 7994
    },
    {
      "epoch": 1.1205325858444288,
      "grad_norm": 0.15141141414642334,
      "learning_rate": 0.0001925137526907438,
      "loss": 0.0289,
      "step": 7995
    },
    {
      "epoch": 1.1206727400140155,
      "grad_norm": 0.45504826307296753,
      "learning_rate": 0.00019249940205692419,
      "loss": 0.103,
      "step": 7996
    },
    {
      "epoch": 1.120812894183602,
      "grad_norm": 0.17744044959545135,
      "learning_rate": 0.00019248505142310451,
      "loss": 0.0408,
      "step": 7997
    },
    {
      "epoch": 1.1209530483531884,
      "grad_norm": 0.5326001644134521,
      "learning_rate": 0.00019247070078928484,
      "loss": 0.0352,
      "step": 7998
    },
    {
      "epoch": 1.1210932025227751,
      "grad_norm": 0.3257143795490265,
      "learning_rate": 0.00019245635015546517,
      "loss": 0.0911,
      "step": 7999
    },
    {
      "epoch": 1.1212333566923616,
      "grad_norm": 0.46379753947257996,
      "learning_rate": 0.00019244199952164553,
      "loss": 0.0607,
      "step": 8000
    },
    {
      "epoch": 1.121373510861948,
      "grad_norm": 0.2637377083301544,
      "learning_rate": 0.00019242764888782586,
      "loss": 0.0261,
      "step": 8001
    },
    {
      "epoch": 1.1215136650315347,
      "grad_norm": 0.32528358697891235,
      "learning_rate": 0.00019241329825400618,
      "loss": 0.0957,
      "step": 8002
    },
    {
      "epoch": 1.1216538192011212,
      "grad_norm": 0.44165828824043274,
      "learning_rate": 0.00019239894762018654,
      "loss": 0.051,
      "step": 8003
    },
    {
      "epoch": 1.1217939733707079,
      "grad_norm": 0.1660914570093155,
      "learning_rate": 0.00019238459698636687,
      "loss": 0.043,
      "step": 8004
    },
    {
      "epoch": 1.1219341275402943,
      "grad_norm": 0.48632216453552246,
      "learning_rate": 0.0001923702463525472,
      "loss": 0.113,
      "step": 8005
    },
    {
      "epoch": 1.1220742817098808,
      "grad_norm": 0.2467777580022812,
      "learning_rate": 0.00019235589571872758,
      "loss": 0.0516,
      "step": 8006
    },
    {
      "epoch": 1.1222144358794675,
      "grad_norm": 0.24975550174713135,
      "learning_rate": 0.0001923415450849079,
      "loss": 0.0632,
      "step": 8007
    },
    {
      "epoch": 1.122354590049054,
      "grad_norm": 0.6065748929977417,
      "learning_rate": 0.00019232719445108824,
      "loss": 0.0513,
      "step": 8008
    },
    {
      "epoch": 1.1224947442186406,
      "grad_norm": 0.44722145795822144,
      "learning_rate": 0.0001923128438172686,
      "loss": 0.0666,
      "step": 8009
    },
    {
      "epoch": 1.122634898388227,
      "grad_norm": 0.509182333946228,
      "learning_rate": 0.00019229849318344892,
      "loss": 0.0876,
      "step": 8010
    },
    {
      "epoch": 1.1227750525578135,
      "grad_norm": 0.38631001114845276,
      "learning_rate": 0.00019228414254962925,
      "loss": 0.0518,
      "step": 8011
    },
    {
      "epoch": 1.1229152067274002,
      "grad_norm": 0.20089133083820343,
      "learning_rate": 0.0001922697919158096,
      "loss": 0.0443,
      "step": 8012
    },
    {
      "epoch": 1.1230553608969867,
      "grad_norm": 0.22899366915225983,
      "learning_rate": 0.00019225544128198993,
      "loss": 0.0357,
      "step": 8013
    },
    {
      "epoch": 1.1231955150665733,
      "grad_norm": 0.5111515522003174,
      "learning_rate": 0.00019224109064817026,
      "loss": 0.0426,
      "step": 8014
    },
    {
      "epoch": 1.1233356692361598,
      "grad_norm": 0.42809003591537476,
      "learning_rate": 0.00019222674001435065,
      "loss": 0.0551,
      "step": 8015
    },
    {
      "epoch": 1.1234758234057463,
      "grad_norm": 0.17355439066886902,
      "learning_rate": 0.00019221238938053097,
      "loss": 0.0069,
      "step": 8016
    },
    {
      "epoch": 1.123615977575333,
      "grad_norm": 0.3530888259410858,
      "learning_rate": 0.0001921980387467113,
      "loss": 0.0597,
      "step": 8017
    },
    {
      "epoch": 1.1237561317449194,
      "grad_norm": 0.781139075756073,
      "learning_rate": 0.00019218368811289163,
      "loss": 0.0634,
      "step": 8018
    },
    {
      "epoch": 1.1238962859145059,
      "grad_norm": 0.14461249113082886,
      "learning_rate": 0.000192169337479072,
      "loss": 0.018,
      "step": 8019
    },
    {
      "epoch": 1.1240364400840925,
      "grad_norm": 0.5108649730682373,
      "learning_rate": 0.00019215498684525232,
      "loss": 0.066,
      "step": 8020
    },
    {
      "epoch": 1.124176594253679,
      "grad_norm": 0.2528609037399292,
      "learning_rate": 0.00019214063621143264,
      "loss": 0.0335,
      "step": 8021
    },
    {
      "epoch": 1.1243167484232657,
      "grad_norm": 0.6112862229347229,
      "learning_rate": 0.000192126285577613,
      "loss": 0.1046,
      "step": 8022
    },
    {
      "epoch": 1.1244569025928521,
      "grad_norm": 0.8291991353034973,
      "learning_rate": 0.00019211193494379333,
      "loss": 0.1284,
      "step": 8023
    },
    {
      "epoch": 1.1245970567624386,
      "grad_norm": 0.23615464568138123,
      "learning_rate": 0.00019209758430997366,
      "loss": 0.0133,
      "step": 8024
    },
    {
      "epoch": 1.1247372109320253,
      "grad_norm": 0.3060218393802643,
      "learning_rate": 0.000192083233676154,
      "loss": 0.0492,
      "step": 8025
    },
    {
      "epoch": 1.1248773651016117,
      "grad_norm": 0.657585859298706,
      "learning_rate": 0.00019206888304233434,
      "loss": 0.0412,
      "step": 8026
    },
    {
      "epoch": 1.1250175192711982,
      "grad_norm": 0.2563890218734741,
      "learning_rate": 0.00019205453240851467,
      "loss": 0.1012,
      "step": 8027
    },
    {
      "epoch": 1.1251576734407849,
      "grad_norm": 0.14099502563476562,
      "learning_rate": 0.00019204018177469505,
      "loss": 0.0127,
      "step": 8028
    },
    {
      "epoch": 1.1252978276103713,
      "grad_norm": 0.21189579367637634,
      "learning_rate": 0.00019202583114087538,
      "loss": 0.0688,
      "step": 8029
    },
    {
      "epoch": 1.125437981779958,
      "grad_norm": 0.4861416220664978,
      "learning_rate": 0.0001920114805070557,
      "loss": 0.0504,
      "step": 8030
    },
    {
      "epoch": 1.1255781359495445,
      "grad_norm": 2.024507522583008,
      "learning_rate": 0.00019199712987323607,
      "loss": 0.2907,
      "step": 8031
    },
    {
      "epoch": 1.125718290119131,
      "grad_norm": 0.5642987489700317,
      "learning_rate": 0.0001919827792394164,
      "loss": 0.0506,
      "step": 8032
    },
    {
      "epoch": 1.1258584442887176,
      "grad_norm": 0.8888179063796997,
      "learning_rate": 0.00019196842860559672,
      "loss": 0.0248,
      "step": 8033
    },
    {
      "epoch": 1.125998598458304,
      "grad_norm": 1.257251262664795,
      "learning_rate": 0.00019195407797177705,
      "loss": 0.1059,
      "step": 8034
    },
    {
      "epoch": 1.1261387526278908,
      "grad_norm": 1.0725864171981812,
      "learning_rate": 0.0001919397273379574,
      "loss": 0.043,
      "step": 8035
    },
    {
      "epoch": 1.1262789067974772,
      "grad_norm": 0.442417711019516,
      "learning_rate": 0.00019192537670413774,
      "loss": 0.0393,
      "step": 8036
    },
    {
      "epoch": 1.1264190609670637,
      "grad_norm": 0.3742845058441162,
      "learning_rate": 0.00019191102607031806,
      "loss": 0.0757,
      "step": 8037
    },
    {
      "epoch": 1.1265592151366504,
      "grad_norm": 0.08734431862831116,
      "learning_rate": 0.00019189667543649845,
      "loss": 0.0102,
      "step": 8038
    },
    {
      "epoch": 1.1266993693062368,
      "grad_norm": 0.22829271852970123,
      "learning_rate": 0.00019188232480267877,
      "loss": 0.0508,
      "step": 8039
    },
    {
      "epoch": 1.1268395234758235,
      "grad_norm": 0.19670654833316803,
      "learning_rate": 0.0001918679741688591,
      "loss": 0.0405,
      "step": 8040
    },
    {
      "epoch": 1.12697967764541,
      "grad_norm": 0.5248310565948486,
      "learning_rate": 0.00019185362353503946,
      "loss": 0.0467,
      "step": 8041
    },
    {
      "epoch": 1.1271198318149964,
      "grad_norm": 0.29423487186431885,
      "learning_rate": 0.0001918392729012198,
      "loss": 0.0381,
      "step": 8042
    },
    {
      "epoch": 1.127259985984583,
      "grad_norm": 1.0417219400405884,
      "learning_rate": 0.00019182492226740012,
      "loss": 0.1391,
      "step": 8043
    },
    {
      "epoch": 1.1274001401541696,
      "grad_norm": 0.16028207540512085,
      "learning_rate": 0.00019181057163358047,
      "loss": 0.0337,
      "step": 8044
    },
    {
      "epoch": 1.1275402943237562,
      "grad_norm": 0.5439640879631042,
      "learning_rate": 0.0001917962209997608,
      "loss": 0.0855,
      "step": 8045
    },
    {
      "epoch": 1.1276804484933427,
      "grad_norm": 0.16845403611660004,
      "learning_rate": 0.00019178187036594113,
      "loss": 0.0128,
      "step": 8046
    },
    {
      "epoch": 1.1278206026629292,
      "grad_norm": 0.36725983023643494,
      "learning_rate": 0.0001917675197321215,
      "loss": 0.1029,
      "step": 8047
    },
    {
      "epoch": 1.1279607568325158,
      "grad_norm": 1.0634560585021973,
      "learning_rate": 0.00019175316909830184,
      "loss": 0.0718,
      "step": 8048
    },
    {
      "epoch": 1.1281009110021023,
      "grad_norm": 0.21387803554534912,
      "learning_rate": 0.00019173881846448217,
      "loss": 0.035,
      "step": 8049
    },
    {
      "epoch": 1.1282410651716888,
      "grad_norm": 0.19767367839813232,
      "learning_rate": 0.00019172446783066252,
      "loss": 0.0196,
      "step": 8050
    },
    {
      "epoch": 1.1283812193412754,
      "grad_norm": 1.1467914581298828,
      "learning_rate": 0.00019171011719684285,
      "loss": 0.08,
      "step": 8051
    },
    {
      "epoch": 1.128521373510862,
      "grad_norm": 0.3003193140029907,
      "learning_rate": 0.00019169576656302318,
      "loss": 0.0917,
      "step": 8052
    },
    {
      "epoch": 1.1286615276804486,
      "grad_norm": 0.225721538066864,
      "learning_rate": 0.0001916814159292035,
      "loss": 0.0278,
      "step": 8053
    },
    {
      "epoch": 1.128801681850035,
      "grad_norm": 0.23106655478477478,
      "learning_rate": 0.00019166706529538387,
      "loss": 0.0658,
      "step": 8054
    },
    {
      "epoch": 1.1289418360196215,
      "grad_norm": 0.25449538230895996,
      "learning_rate": 0.0001916527146615642,
      "loss": 0.0753,
      "step": 8055
    },
    {
      "epoch": 1.1290819901892082,
      "grad_norm": 0.15561680495738983,
      "learning_rate": 0.00019163836402774452,
      "loss": 0.0262,
      "step": 8056
    },
    {
      "epoch": 1.1292221443587946,
      "grad_norm": 0.652627170085907,
      "learning_rate": 0.00019162401339392488,
      "loss": 0.0833,
      "step": 8057
    },
    {
      "epoch": 1.129362298528381,
      "grad_norm": 0.3154574930667877,
      "learning_rate": 0.0001916096627601052,
      "loss": 0.0731,
      "step": 8058
    },
    {
      "epoch": 1.1295024526979678,
      "grad_norm": 0.24196955561637878,
      "learning_rate": 0.00019159531212628556,
      "loss": 0.0379,
      "step": 8059
    },
    {
      "epoch": 1.1296426068675542,
      "grad_norm": 0.17953385412693024,
      "learning_rate": 0.00019158096149246592,
      "loss": 0.0229,
      "step": 8060
    },
    {
      "epoch": 1.129782761037141,
      "grad_norm": 0.18981239199638367,
      "learning_rate": 0.00019156661085864625,
      "loss": 0.0461,
      "step": 8061
    },
    {
      "epoch": 1.1299229152067274,
      "grad_norm": 0.2708427608013153,
      "learning_rate": 0.00019155226022482658,
      "loss": 0.0288,
      "step": 8062
    },
    {
      "epoch": 1.1300630693763138,
      "grad_norm": 0.31338393688201904,
      "learning_rate": 0.00019153790959100693,
      "loss": 0.0517,
      "step": 8063
    },
    {
      "epoch": 1.1302032235459005,
      "grad_norm": 0.5928025841712952,
      "learning_rate": 0.00019152355895718726,
      "loss": 0.0772,
      "step": 8064
    },
    {
      "epoch": 1.130343377715487,
      "grad_norm": 0.23111417889595032,
      "learning_rate": 0.0001915092083233676,
      "loss": 0.0423,
      "step": 8065
    },
    {
      "epoch": 1.1304835318850737,
      "grad_norm": 0.2650432586669922,
      "learning_rate": 0.00019149485768954794,
      "loss": 0.0617,
      "step": 8066
    },
    {
      "epoch": 1.1306236860546601,
      "grad_norm": 0.15775763988494873,
      "learning_rate": 0.00019148050705572827,
      "loss": 0.0336,
      "step": 8067
    },
    {
      "epoch": 1.1307638402242466,
      "grad_norm": 0.09262767434120178,
      "learning_rate": 0.0001914661564219086,
      "loss": 0.0148,
      "step": 8068
    },
    {
      "epoch": 1.1309039943938333,
      "grad_norm": 0.2704620957374573,
      "learning_rate": 0.00019145180578808893,
      "loss": 0.0375,
      "step": 8069
    },
    {
      "epoch": 1.1310441485634197,
      "grad_norm": 0.8019313216209412,
      "learning_rate": 0.0001914374551542693,
      "loss": 0.0685,
      "step": 8070
    },
    {
      "epoch": 1.1311843027330064,
      "grad_norm": 0.30213186144828796,
      "learning_rate": 0.00019142310452044964,
      "loss": 0.0569,
      "step": 8071
    },
    {
      "epoch": 1.1313244569025929,
      "grad_norm": 0.226015105843544,
      "learning_rate": 0.00019140875388662997,
      "loss": 0.059,
      "step": 8072
    },
    {
      "epoch": 1.1314646110721793,
      "grad_norm": 0.395021915435791,
      "learning_rate": 0.00019139440325281033,
      "loss": 0.0757,
      "step": 8073
    },
    {
      "epoch": 1.131604765241766,
      "grad_norm": 0.41723376512527466,
      "learning_rate": 0.00019138005261899065,
      "loss": 0.0577,
      "step": 8074
    },
    {
      "epoch": 1.1317449194113525,
      "grad_norm": 0.334320992231369,
      "learning_rate": 0.00019136570198517098,
      "loss": 0.0782,
      "step": 8075
    },
    {
      "epoch": 1.1318850735809391,
      "grad_norm": 2.3480494022369385,
      "learning_rate": 0.00019135135135135134,
      "loss": 0.0793,
      "step": 8076
    },
    {
      "epoch": 1.1320252277505256,
      "grad_norm": 0.09011318534612656,
      "learning_rate": 0.00019133700071753167,
      "loss": 0.0053,
      "step": 8077
    },
    {
      "epoch": 1.132165381920112,
      "grad_norm": 0.6127850413322449,
      "learning_rate": 0.000191322650083712,
      "loss": 0.0646,
      "step": 8078
    },
    {
      "epoch": 1.1323055360896987,
      "grad_norm": 0.39642518758773804,
      "learning_rate": 0.00019130829944989238,
      "loss": 0.0551,
      "step": 8079
    },
    {
      "epoch": 1.1324456902592852,
      "grad_norm": 0.4842112362384796,
      "learning_rate": 0.0001912939488160727,
      "loss": 0.0541,
      "step": 8080
    },
    {
      "epoch": 1.1325858444288717,
      "grad_norm": 0.46259811520576477,
      "learning_rate": 0.00019127959818225304,
      "loss": 0.0708,
      "step": 8081
    },
    {
      "epoch": 1.1327259985984584,
      "grad_norm": 0.3893522024154663,
      "learning_rate": 0.0001912652475484334,
      "loss": 0.048,
      "step": 8082
    },
    {
      "epoch": 1.1328661527680448,
      "grad_norm": 0.1922399252653122,
      "learning_rate": 0.00019125089691461372,
      "loss": 0.0164,
      "step": 8083
    },
    {
      "epoch": 1.1330063069376315,
      "grad_norm": 0.7262095212936401,
      "learning_rate": 0.00019123654628079405,
      "loss": 0.0681,
      "step": 8084
    },
    {
      "epoch": 1.133146461107218,
      "grad_norm": 0.9022505283355713,
      "learning_rate": 0.0001912221956469744,
      "loss": 0.0686,
      "step": 8085
    },
    {
      "epoch": 1.1332866152768044,
      "grad_norm": 0.26339420676231384,
      "learning_rate": 0.00019120784501315473,
      "loss": 0.0581,
      "step": 8086
    },
    {
      "epoch": 1.133426769446391,
      "grad_norm": 0.32881686091423035,
      "learning_rate": 0.00019119349437933506,
      "loss": 0.0758,
      "step": 8087
    },
    {
      "epoch": 1.1335669236159776,
      "grad_norm": 0.2719397246837616,
      "learning_rate": 0.0001911791437455154,
      "loss": 0.0706,
      "step": 8088
    },
    {
      "epoch": 1.133707077785564,
      "grad_norm": 0.5079146027565002,
      "learning_rate": 0.00019116479311169575,
      "loss": 0.0606,
      "step": 8089
    },
    {
      "epoch": 1.1338472319551507,
      "grad_norm": 0.26316285133361816,
      "learning_rate": 0.00019115044247787607,
      "loss": 0.0551,
      "step": 8090
    },
    {
      "epoch": 1.1339873861247372,
      "grad_norm": 0.4014352560043335,
      "learning_rate": 0.00019113609184405643,
      "loss": 0.0522,
      "step": 8091
    },
    {
      "epoch": 1.1341275402943238,
      "grad_norm": 0.23115786910057068,
      "learning_rate": 0.00019112174121023678,
      "loss": 0.0373,
      "step": 8092
    },
    {
      "epoch": 1.1342676944639103,
      "grad_norm": 0.503904402256012,
      "learning_rate": 0.0001911073905764171,
      "loss": 0.0575,
      "step": 8093
    },
    {
      "epoch": 1.1344078486334968,
      "grad_norm": 0.47989892959594727,
      "learning_rate": 0.00019109303994259744,
      "loss": 0.1067,
      "step": 8094
    },
    {
      "epoch": 1.1345480028030834,
      "grad_norm": 0.481784850358963,
      "learning_rate": 0.0001910786893087778,
      "loss": 0.0952,
      "step": 8095
    },
    {
      "epoch": 1.13468815697267,
      "grad_norm": 0.4192765951156616,
      "learning_rate": 0.00019106433867495813,
      "loss": 0.0654,
      "step": 8096
    },
    {
      "epoch": 1.1348283111422566,
      "grad_norm": 0.6675271987915039,
      "learning_rate": 0.00019104998804113845,
      "loss": 0.0704,
      "step": 8097
    },
    {
      "epoch": 1.134968465311843,
      "grad_norm": 0.412428081035614,
      "learning_rate": 0.0001910356374073188,
      "loss": 0.0439,
      "step": 8098
    },
    {
      "epoch": 1.1351086194814295,
      "grad_norm": 0.37966421246528625,
      "learning_rate": 0.00019102128677349914,
      "loss": 0.0687,
      "step": 8099
    },
    {
      "epoch": 1.1352487736510162,
      "grad_norm": 0.1752105951309204,
      "learning_rate": 0.00019100693613967947,
      "loss": 0.0225,
      "step": 8100
    },
    {
      "epoch": 1.1353889278206026,
      "grad_norm": 0.5338507294654846,
      "learning_rate": 0.00019099258550585985,
      "loss": 0.1215,
      "step": 8101
    },
    {
      "epoch": 1.1355290819901893,
      "grad_norm": 0.21000151336193085,
      "learning_rate": 0.00019097823487204018,
      "loss": 0.0234,
      "step": 8102
    },
    {
      "epoch": 1.1356692361597758,
      "grad_norm": 0.27362069487571716,
      "learning_rate": 0.0001909638842382205,
      "loss": 0.0452,
      "step": 8103
    },
    {
      "epoch": 1.1358093903293622,
      "grad_norm": 0.2722977101802826,
      "learning_rate": 0.00019094953360440084,
      "loss": 0.0671,
      "step": 8104
    },
    {
      "epoch": 1.135949544498949,
      "grad_norm": 0.27142515778541565,
      "learning_rate": 0.0001909351829705812,
      "loss": 0.0696,
      "step": 8105
    },
    {
      "epoch": 1.1360896986685354,
      "grad_norm": 0.22372335195541382,
      "learning_rate": 0.00019092083233676152,
      "loss": 0.0917,
      "step": 8106
    },
    {
      "epoch": 1.136229852838122,
      "grad_norm": 0.4673496186733246,
      "learning_rate": 0.00019090648170294185,
      "loss": 0.033,
      "step": 8107
    },
    {
      "epoch": 1.1363700070077085,
      "grad_norm": 0.5011157989501953,
      "learning_rate": 0.0001908921310691222,
      "loss": 0.1164,
      "step": 8108
    },
    {
      "epoch": 1.136510161177295,
      "grad_norm": 0.45621436834335327,
      "learning_rate": 0.00019087778043530253,
      "loss": 0.0963,
      "step": 8109
    },
    {
      "epoch": 1.1366503153468817,
      "grad_norm": 0.5107046961784363,
      "learning_rate": 0.00019086342980148286,
      "loss": 0.0454,
      "step": 8110
    },
    {
      "epoch": 1.1367904695164681,
      "grad_norm": 0.5959237217903137,
      "learning_rate": 0.00019084907916766324,
      "loss": 0.0414,
      "step": 8111
    },
    {
      "epoch": 1.1369306236860546,
      "grad_norm": 0.16816528141498566,
      "learning_rate": 0.00019083472853384357,
      "loss": 0.0741,
      "step": 8112
    },
    {
      "epoch": 1.1370707778556413,
      "grad_norm": 0.8550013899803162,
      "learning_rate": 0.0001908203779000239,
      "loss": 0.0755,
      "step": 8113
    },
    {
      "epoch": 1.1372109320252277,
      "grad_norm": 0.15414147078990936,
      "learning_rate": 0.00019080602726620426,
      "loss": 0.0201,
      "step": 8114
    },
    {
      "epoch": 1.1373510861948142,
      "grad_norm": 0.3081057667732239,
      "learning_rate": 0.00019079167663238459,
      "loss": 0.1005,
      "step": 8115
    },
    {
      "epoch": 1.1374912403644009,
      "grad_norm": 0.4586223065853119,
      "learning_rate": 0.00019077732599856491,
      "loss": 0.0672,
      "step": 8116
    },
    {
      "epoch": 1.1376313945339873,
      "grad_norm": 0.2728267312049866,
      "learning_rate": 0.00019076297536474527,
      "loss": 0.0363,
      "step": 8117
    },
    {
      "epoch": 1.137771548703574,
      "grad_norm": 0.5043738484382629,
      "learning_rate": 0.0001907486247309256,
      "loss": 0.0628,
      "step": 8118
    },
    {
      "epoch": 1.1379117028731605,
      "grad_norm": 0.5502161383628845,
      "learning_rate": 0.00019073427409710593,
      "loss": 0.0788,
      "step": 8119
    },
    {
      "epoch": 1.138051857042747,
      "grad_norm": 0.43950599431991577,
      "learning_rate": 0.00019071992346328628,
      "loss": 0.0771,
      "step": 8120
    },
    {
      "epoch": 1.1381920112123336,
      "grad_norm": 0.6934971809387207,
      "learning_rate": 0.0001907055728294666,
      "loss": 0.0444,
      "step": 8121
    },
    {
      "epoch": 1.13833216538192,
      "grad_norm": 0.5587944388389587,
      "learning_rate": 0.00019069122219564694,
      "loss": 0.0785,
      "step": 8122
    },
    {
      "epoch": 1.1384723195515067,
      "grad_norm": 0.47870492935180664,
      "learning_rate": 0.0001906768715618273,
      "loss": 0.144,
      "step": 8123
    },
    {
      "epoch": 1.1386124737210932,
      "grad_norm": 0.6334477066993713,
      "learning_rate": 0.00019066252092800765,
      "loss": 0.0308,
      "step": 8124
    },
    {
      "epoch": 1.1387526278906797,
      "grad_norm": 0.5581763386726379,
      "learning_rate": 0.00019064817029418798,
      "loss": 0.1099,
      "step": 8125
    },
    {
      "epoch": 1.1388927820602663,
      "grad_norm": 0.23852522671222687,
      "learning_rate": 0.0001906338196603683,
      "loss": 0.0163,
      "step": 8126
    },
    {
      "epoch": 1.1390329362298528,
      "grad_norm": 0.6680789589881897,
      "learning_rate": 0.00019061946902654866,
      "loss": 0.139,
      "step": 8127
    },
    {
      "epoch": 1.1391730903994395,
      "grad_norm": 0.33995699882507324,
      "learning_rate": 0.000190605118392729,
      "loss": 0.0485,
      "step": 8128
    },
    {
      "epoch": 1.139313244569026,
      "grad_norm": 1.069642186164856,
      "learning_rate": 0.00019059076775890932,
      "loss": 0.0834,
      "step": 8129
    },
    {
      "epoch": 1.1394533987386124,
      "grad_norm": 0.2828660309314728,
      "learning_rate": 0.00019057641712508968,
      "loss": 0.0181,
      "step": 8130
    },
    {
      "epoch": 1.139593552908199,
      "grad_norm": 0.4177997410297394,
      "learning_rate": 0.00019056206649127,
      "loss": 0.0594,
      "step": 8131
    },
    {
      "epoch": 1.1397337070777855,
      "grad_norm": 0.9500121474266052,
      "learning_rate": 0.00019054771585745033,
      "loss": 0.1209,
      "step": 8132
    },
    {
      "epoch": 1.1398738612473722,
      "grad_norm": 0.3758149743080139,
      "learning_rate": 0.00019053336522363072,
      "loss": 0.0555,
      "step": 8133
    },
    {
      "epoch": 1.1400140154169587,
      "grad_norm": 1.231550693511963,
      "learning_rate": 0.00019051901458981105,
      "loss": 0.2025,
      "step": 8134
    },
    {
      "epoch": 1.1401541695865451,
      "grad_norm": 2.562213659286499,
      "learning_rate": 0.00019050466395599137,
      "loss": 0.236,
      "step": 8135
    },
    {
      "epoch": 1.1402943237561318,
      "grad_norm": 0.251033753156662,
      "learning_rate": 0.00019049031332217173,
      "loss": 0.0665,
      "step": 8136
    },
    {
      "epoch": 1.1404344779257183,
      "grad_norm": 0.34847134351730347,
      "learning_rate": 0.00019047596268835206,
      "loss": 0.0961,
      "step": 8137
    },
    {
      "epoch": 1.1405746320953047,
      "grad_norm": 0.13728486001491547,
      "learning_rate": 0.0001904616120545324,
      "loss": 0.0156,
      "step": 8138
    },
    {
      "epoch": 1.1407147862648914,
      "grad_norm": 0.30178892612457275,
      "learning_rate": 0.00019044726142071272,
      "loss": 0.0441,
      "step": 8139
    },
    {
      "epoch": 1.1408549404344779,
      "grad_norm": 0.19602464139461517,
      "learning_rate": 0.00019043291078689307,
      "loss": 0.077,
      "step": 8140
    },
    {
      "epoch": 1.1409950946040646,
      "grad_norm": 0.35875406861305237,
      "learning_rate": 0.0001904185601530734,
      "loss": 0.1188,
      "step": 8141
    },
    {
      "epoch": 1.141135248773651,
      "grad_norm": 0.21018241345882416,
      "learning_rate": 0.00019040420951925373,
      "loss": 0.047,
      "step": 8142
    },
    {
      "epoch": 1.1412754029432375,
      "grad_norm": 0.23383651673793793,
      "learning_rate": 0.0001903898588854341,
      "loss": 0.059,
      "step": 8143
    },
    {
      "epoch": 1.1414155571128242,
      "grad_norm": 1.4485735893249512,
      "learning_rate": 0.00019037550825161444,
      "loss": 0.0811,
      "step": 8144
    },
    {
      "epoch": 1.1415557112824106,
      "grad_norm": 0.28622177243232727,
      "learning_rate": 0.00019036115761779477,
      "loss": 0.0326,
      "step": 8145
    },
    {
      "epoch": 1.141695865451997,
      "grad_norm": 0.33646684885025024,
      "learning_rate": 0.00019034680698397512,
      "loss": 0.1203,
      "step": 8146
    },
    {
      "epoch": 1.1418360196215838,
      "grad_norm": 0.3636622428894043,
      "learning_rate": 0.00019033245635015545,
      "loss": 0.0671,
      "step": 8147
    },
    {
      "epoch": 1.1419761737911702,
      "grad_norm": 0.3842368721961975,
      "learning_rate": 0.00019031810571633578,
      "loss": 0.0555,
      "step": 8148
    },
    {
      "epoch": 1.142116327960757,
      "grad_norm": 0.24999134242534637,
      "learning_rate": 0.00019030375508251614,
      "loss": 0.0442,
      "step": 8149
    },
    {
      "epoch": 1.1422564821303434,
      "grad_norm": 0.2663787007331848,
      "learning_rate": 0.00019028940444869646,
      "loss": 0.059,
      "step": 8150
    },
    {
      "epoch": 1.1423966362999298,
      "grad_norm": 0.4734601378440857,
      "learning_rate": 0.0001902750538148768,
      "loss": 0.0942,
      "step": 8151
    },
    {
      "epoch": 1.1425367904695165,
      "grad_norm": 0.22639179229736328,
      "learning_rate": 0.00019026070318105715,
      "loss": 0.0732,
      "step": 8152
    },
    {
      "epoch": 1.142676944639103,
      "grad_norm": 0.24627842009067535,
      "learning_rate": 0.00019024635254723748,
      "loss": 0.0638,
      "step": 8153
    },
    {
      "epoch": 1.1428170988086896,
      "grad_norm": 0.4628268778324127,
      "learning_rate": 0.0001902320019134178,
      "loss": 0.0695,
      "step": 8154
    },
    {
      "epoch": 1.142957252978276,
      "grad_norm": 0.3266357183456421,
      "learning_rate": 0.0001902176512795982,
      "loss": 0.03,
      "step": 8155
    },
    {
      "epoch": 1.1430974071478626,
      "grad_norm": 0.14849551022052765,
      "learning_rate": 0.00019020330064577852,
      "loss": 0.0157,
      "step": 8156
    },
    {
      "epoch": 1.1432375613174492,
      "grad_norm": 0.2778032422065735,
      "learning_rate": 0.00019018895001195885,
      "loss": 0.0326,
      "step": 8157
    },
    {
      "epoch": 1.1433777154870357,
      "grad_norm": 0.306439608335495,
      "learning_rate": 0.00019017459937813917,
      "loss": 0.1054,
      "step": 8158
    },
    {
      "epoch": 1.1435178696566224,
      "grad_norm": 0.5741012096405029,
      "learning_rate": 0.00019016024874431953,
      "loss": 0.03,
      "step": 8159
    },
    {
      "epoch": 1.1436580238262088,
      "grad_norm": 0.15136685967445374,
      "learning_rate": 0.00019014589811049986,
      "loss": 0.0376,
      "step": 8160
    },
    {
      "epoch": 1.1437981779957953,
      "grad_norm": 0.3661526143550873,
      "learning_rate": 0.0001901315474766802,
      "loss": 0.097,
      "step": 8161
    },
    {
      "epoch": 1.143938332165382,
      "grad_norm": 0.18994368612766266,
      "learning_rate": 0.00019011719684286054,
      "loss": 0.0457,
      "step": 8162
    },
    {
      "epoch": 1.1440784863349684,
      "grad_norm": 0.23135820031166077,
      "learning_rate": 0.00019010284620904087,
      "loss": 0.0941,
      "step": 8163
    },
    {
      "epoch": 1.1442186405045551,
      "grad_norm": 0.37540772557258606,
      "learning_rate": 0.0001900884955752212,
      "loss": 0.0744,
      "step": 8164
    },
    {
      "epoch": 1.1443587946741416,
      "grad_norm": 0.3024245500564575,
      "learning_rate": 0.00019007414494140158,
      "loss": 0.0838,
      "step": 8165
    },
    {
      "epoch": 1.144498948843728,
      "grad_norm": 0.41323837637901306,
      "learning_rate": 0.0001900597943075819,
      "loss": 0.0772,
      "step": 8166
    },
    {
      "epoch": 1.1446391030133147,
      "grad_norm": 0.4566798806190491,
      "learning_rate": 0.00019004544367376224,
      "loss": 0.1013,
      "step": 8167
    },
    {
      "epoch": 1.1447792571829012,
      "grad_norm": 0.26838138699531555,
      "learning_rate": 0.0001900310930399426,
      "loss": 0.0465,
      "step": 8168
    },
    {
      "epoch": 1.1449194113524876,
      "grad_norm": 0.2764984965324402,
      "learning_rate": 0.00019001674240612292,
      "loss": 0.0377,
      "step": 8169
    },
    {
      "epoch": 1.1450595655220743,
      "grad_norm": 0.08905649185180664,
      "learning_rate": 0.00019000239177230325,
      "loss": 0.0074,
      "step": 8170
    },
    {
      "epoch": 1.1451997196916608,
      "grad_norm": 0.67715984582901,
      "learning_rate": 0.0001899880411384836,
      "loss": 0.1194,
      "step": 8171
    },
    {
      "epoch": 1.1453398738612472,
      "grad_norm": 0.3427574038505554,
      "learning_rate": 0.00018997369050466394,
      "loss": 0.0222,
      "step": 8172
    },
    {
      "epoch": 1.145480028030834,
      "grad_norm": 0.9477463364601135,
      "learning_rate": 0.00018995933987084427,
      "loss": 0.0771,
      "step": 8173
    },
    {
      "epoch": 1.1456201822004204,
      "grad_norm": 0.26231104135513306,
      "learning_rate": 0.0001899449892370246,
      "loss": 0.0653,
      "step": 8174
    },
    {
      "epoch": 1.145760336370007,
      "grad_norm": 0.23415350914001465,
      "learning_rate": 0.00018993063860320498,
      "loss": 0.0207,
      "step": 8175
    },
    {
      "epoch": 1.1459004905395935,
      "grad_norm": 0.5579625964164734,
      "learning_rate": 0.0001899162879693853,
      "loss": 0.0907,
      "step": 8176
    },
    {
      "epoch": 1.14604064470918,
      "grad_norm": 0.3386024832725525,
      "learning_rate": 0.00018990193733556563,
      "loss": 0.0437,
      "step": 8177
    },
    {
      "epoch": 1.1461807988787667,
      "grad_norm": 0.8833440542221069,
      "learning_rate": 0.000189887586701746,
      "loss": 0.0743,
      "step": 8178
    },
    {
      "epoch": 1.1463209530483531,
      "grad_norm": 0.18453925848007202,
      "learning_rate": 0.00018987323606792632,
      "loss": 0.0448,
      "step": 8179
    },
    {
      "epoch": 1.1464611072179398,
      "grad_norm": 0.8642560243606567,
      "learning_rate": 0.00018985888543410665,
      "loss": 0.0703,
      "step": 8180
    },
    {
      "epoch": 1.1466012613875263,
      "grad_norm": 0.07260432094335556,
      "learning_rate": 0.000189844534800287,
      "loss": 0.005,
      "step": 8181
    },
    {
      "epoch": 1.1467414155571127,
      "grad_norm": 0.3032330572605133,
      "learning_rate": 0.00018983018416646733,
      "loss": 0.0254,
      "step": 8182
    },
    {
      "epoch": 1.1468815697266994,
      "grad_norm": 0.4182407557964325,
      "learning_rate": 0.00018981583353264766,
      "loss": 0.0774,
      "step": 8183
    },
    {
      "epoch": 1.1470217238962859,
      "grad_norm": 1.464682936668396,
      "learning_rate": 0.00018980148289882802,
      "loss": 0.3925,
      "step": 8184
    },
    {
      "epoch": 1.1471618780658726,
      "grad_norm": 0.05969484522938728,
      "learning_rate": 0.00018978713226500834,
      "loss": 0.0052,
      "step": 8185
    },
    {
      "epoch": 1.147302032235459,
      "grad_norm": 0.3523111939430237,
      "learning_rate": 0.0001897727816311887,
      "loss": 0.0266,
      "step": 8186
    },
    {
      "epoch": 1.1474421864050455,
      "grad_norm": 0.15141288936138153,
      "learning_rate": 0.00018975843099736906,
      "loss": 0.046,
      "step": 8187
    },
    {
      "epoch": 1.1475823405746322,
      "grad_norm": 0.5671917200088501,
      "learning_rate": 0.00018974408036354938,
      "loss": 0.0723,
      "step": 8188
    },
    {
      "epoch": 1.1477224947442186,
      "grad_norm": 0.18358641862869263,
      "learning_rate": 0.0001897297297297297,
      "loss": 0.0496,
      "step": 8189
    },
    {
      "epoch": 1.1478626489138053,
      "grad_norm": 0.2935771644115448,
      "learning_rate": 0.00018971537909591004,
      "loss": 0.0712,
      "step": 8190
    },
    {
      "epoch": 1.1480028030833918,
      "grad_norm": 0.31260305643081665,
      "learning_rate": 0.0001897010284620904,
      "loss": 0.0713,
      "step": 8191
    },
    {
      "epoch": 1.1481429572529782,
      "grad_norm": 0.1767255663871765,
      "learning_rate": 0.00018968667782827073,
      "loss": 0.0674,
      "step": 8192
    },
    {
      "epoch": 1.148283111422565,
      "grad_norm": 0.39744818210601807,
      "learning_rate": 0.00018967232719445105,
      "loss": 0.03,
      "step": 8193
    },
    {
      "epoch": 1.1484232655921514,
      "grad_norm": 0.2499329000711441,
      "learning_rate": 0.0001896579765606314,
      "loss": 0.0623,
      "step": 8194
    },
    {
      "epoch": 1.148563419761738,
      "grad_norm": 0.2834676504135132,
      "learning_rate": 0.00018964362592681174,
      "loss": 0.0832,
      "step": 8195
    },
    {
      "epoch": 1.1487035739313245,
      "grad_norm": 0.5464600920677185,
      "learning_rate": 0.00018962927529299207,
      "loss": 0.0887,
      "step": 8196
    },
    {
      "epoch": 1.148843728100911,
      "grad_norm": 0.3822512924671173,
      "learning_rate": 0.00018961492465917245,
      "loss": 0.0716,
      "step": 8197
    },
    {
      "epoch": 1.1489838822704976,
      "grad_norm": 0.5086199641227722,
      "learning_rate": 0.00018960057402535278,
      "loss": 0.0486,
      "step": 8198
    },
    {
      "epoch": 1.149124036440084,
      "grad_norm": 0.4189874529838562,
      "learning_rate": 0.0001895862233915331,
      "loss": 0.1297,
      "step": 8199
    },
    {
      "epoch": 1.1492641906096706,
      "grad_norm": 0.28184911608695984,
      "learning_rate": 0.00018957187275771346,
      "loss": 0.0563,
      "step": 8200
    },
    {
      "epoch": 1.1494043447792572,
      "grad_norm": 0.5149058103561401,
      "learning_rate": 0.0001895575221238938,
      "loss": 0.0231,
      "step": 8201
    },
    {
      "epoch": 1.1495444989488437,
      "grad_norm": 0.08912201225757599,
      "learning_rate": 0.00018954317149007412,
      "loss": 0.0138,
      "step": 8202
    },
    {
      "epoch": 1.1496846531184302,
      "grad_norm": 0.10554853826761246,
      "learning_rate": 0.00018952882085625447,
      "loss": 0.0151,
      "step": 8203
    },
    {
      "epoch": 1.1498248072880168,
      "grad_norm": 0.5522861480712891,
      "learning_rate": 0.0001895144702224348,
      "loss": 0.0921,
      "step": 8204
    },
    {
      "epoch": 1.1499649614576033,
      "grad_norm": 0.2820811867713928,
      "learning_rate": 0.00018950011958861513,
      "loss": 0.0671,
      "step": 8205
    },
    {
      "epoch": 1.15010511562719,
      "grad_norm": 0.5302562713623047,
      "learning_rate": 0.00018948576895479551,
      "loss": 0.077,
      "step": 8206
    },
    {
      "epoch": 1.1502452697967764,
      "grad_norm": 1.1579928398132324,
      "learning_rate": 0.00018947141832097584,
      "loss": 0.0676,
      "step": 8207
    },
    {
      "epoch": 1.150385423966363,
      "grad_norm": 0.27457666397094727,
      "learning_rate": 0.00018945706768715617,
      "loss": 0.0567,
      "step": 8208
    },
    {
      "epoch": 1.1505255781359496,
      "grad_norm": 0.23893170058727264,
      "learning_rate": 0.0001894427170533365,
      "loss": 0.0504,
      "step": 8209
    },
    {
      "epoch": 1.150665732305536,
      "grad_norm": 0.33306390047073364,
      "learning_rate": 0.00018942836641951686,
      "loss": 0.017,
      "step": 8210
    },
    {
      "epoch": 1.1508058864751227,
      "grad_norm": 0.23821240663528442,
      "learning_rate": 0.00018941401578569718,
      "loss": 0.0285,
      "step": 8211
    },
    {
      "epoch": 1.1509460406447092,
      "grad_norm": 0.3634144067764282,
      "learning_rate": 0.0001893996651518775,
      "loss": 0.1039,
      "step": 8212
    },
    {
      "epoch": 1.1510861948142956,
      "grad_norm": 0.1651085466146469,
      "learning_rate": 0.00018938531451805787,
      "loss": 0.0247,
      "step": 8213
    },
    {
      "epoch": 1.1512263489838823,
      "grad_norm": 1.0651239156723022,
      "learning_rate": 0.0001893709638842382,
      "loss": 0.139,
      "step": 8214
    },
    {
      "epoch": 1.1513665031534688,
      "grad_norm": 0.4140488803386688,
      "learning_rate": 0.00018935661325041853,
      "loss": 0.1118,
      "step": 8215
    },
    {
      "epoch": 1.1515066573230555,
      "grad_norm": 0.18126502633094788,
      "learning_rate": 0.00018934226261659888,
      "loss": 0.0246,
      "step": 8216
    },
    {
      "epoch": 1.151646811492642,
      "grad_norm": 0.13416564464569092,
      "learning_rate": 0.0001893279119827792,
      "loss": 0.0128,
      "step": 8217
    },
    {
      "epoch": 1.1517869656622284,
      "grad_norm": 0.3202526569366455,
      "learning_rate": 0.00018931356134895957,
      "loss": 0.0589,
      "step": 8218
    },
    {
      "epoch": 1.151927119831815,
      "grad_norm": 0.6767866015434265,
      "learning_rate": 0.00018929921071513992,
      "loss": 0.0451,
      "step": 8219
    },
    {
      "epoch": 1.1520672740014015,
      "grad_norm": 0.2485506534576416,
      "learning_rate": 0.00018928486008132025,
      "loss": 0.0775,
      "step": 8220
    },
    {
      "epoch": 1.1522074281709882,
      "grad_norm": 0.4857812821865082,
      "learning_rate": 0.00018927050944750058,
      "loss": 0.0518,
      "step": 8221
    },
    {
      "epoch": 1.1523475823405747,
      "grad_norm": 0.19365179538726807,
      "learning_rate": 0.00018925615881368093,
      "loss": 0.0312,
      "step": 8222
    },
    {
      "epoch": 1.1524877365101611,
      "grad_norm": 0.3440621495246887,
      "learning_rate": 0.00018924180817986126,
      "loss": 0.081,
      "step": 8223
    },
    {
      "epoch": 1.1526278906797478,
      "grad_norm": 0.12351280450820923,
      "learning_rate": 0.0001892274575460416,
      "loss": 0.0102,
      "step": 8224
    },
    {
      "epoch": 1.1527680448493343,
      "grad_norm": 0.4448149800300598,
      "learning_rate": 0.00018921310691222192,
      "loss": 0.0713,
      "step": 8225
    },
    {
      "epoch": 1.1529081990189207,
      "grad_norm": 2.7275681495666504,
      "learning_rate": 0.00018919875627840228,
      "loss": 0.0143,
      "step": 8226
    },
    {
      "epoch": 1.1530483531885074,
      "grad_norm": 0.2122349590063095,
      "learning_rate": 0.0001891844056445826,
      "loss": 0.044,
      "step": 8227
    },
    {
      "epoch": 1.1531885073580939,
      "grad_norm": 0.30623915791511536,
      "learning_rate": 0.00018917005501076293,
      "loss": 0.0596,
      "step": 8228
    },
    {
      "epoch": 1.1533286615276805,
      "grad_norm": 0.6875832676887512,
      "learning_rate": 0.00018915570437694332,
      "loss": 0.0323,
      "step": 8229
    },
    {
      "epoch": 1.153468815697267,
      "grad_norm": 1.5201548337936401,
      "learning_rate": 0.00018914135374312364,
      "loss": 0.0834,
      "step": 8230
    },
    {
      "epoch": 1.1536089698668535,
      "grad_norm": 0.23452061414718628,
      "learning_rate": 0.00018912700310930397,
      "loss": 0.011,
      "step": 8231
    },
    {
      "epoch": 1.1537491240364401,
      "grad_norm": 0.5297860503196716,
      "learning_rate": 0.00018911265247548433,
      "loss": 0.1174,
      "step": 8232
    },
    {
      "epoch": 1.1538892782060266,
      "grad_norm": 1.8183493614196777,
      "learning_rate": 0.00018909830184166466,
      "loss": 0.1645,
      "step": 8233
    },
    {
      "epoch": 1.154029432375613,
      "grad_norm": 1.5141992568969727,
      "learning_rate": 0.00018908395120784499,
      "loss": 0.1216,
      "step": 8234
    },
    {
      "epoch": 1.1541695865451997,
      "grad_norm": 4.588138103485107,
      "learning_rate": 0.00018906960057402534,
      "loss": 0.1987,
      "step": 8235
    },
    {
      "epoch": 1.1543097407147862,
      "grad_norm": 0.18887551128864288,
      "learning_rate": 0.00018905524994020567,
      "loss": 0.032,
      "step": 8236
    },
    {
      "epoch": 1.1544498948843729,
      "grad_norm": 0.3860364556312561,
      "learning_rate": 0.000189040899306386,
      "loss": 0.0982,
      "step": 8237
    },
    {
      "epoch": 1.1545900490539593,
      "grad_norm": 0.25927311182022095,
      "learning_rate": 0.00018902654867256638,
      "loss": 0.0407,
      "step": 8238
    },
    {
      "epoch": 1.1547302032235458,
      "grad_norm": 0.283090740442276,
      "learning_rate": 0.0001890121980387467,
      "loss": 0.0431,
      "step": 8239
    },
    {
      "epoch": 1.1548703573931325,
      "grad_norm": 0.14458709955215454,
      "learning_rate": 0.00018899784740492704,
      "loss": 0.029,
      "step": 8240
    },
    {
      "epoch": 1.155010511562719,
      "grad_norm": 0.2459026426076889,
      "learning_rate": 0.0001889834967711074,
      "loss": 0.0367,
      "step": 8241
    },
    {
      "epoch": 1.1551506657323056,
      "grad_norm": 0.4028928875923157,
      "learning_rate": 0.00018896914613728772,
      "loss": 0.083,
      "step": 8242
    },
    {
      "epoch": 1.155290819901892,
      "grad_norm": 0.2877585291862488,
      "learning_rate": 0.00018895479550346805,
      "loss": 0.0807,
      "step": 8243
    },
    {
      "epoch": 1.1554309740714785,
      "grad_norm": 0.31825652718544006,
      "learning_rate": 0.00018894044486964838,
      "loss": 0.0599,
      "step": 8244
    },
    {
      "epoch": 1.1555711282410652,
      "grad_norm": 0.456478089094162,
      "learning_rate": 0.00018892609423582874,
      "loss": 0.1627,
      "step": 8245
    },
    {
      "epoch": 1.1557112824106517,
      "grad_norm": 0.41934704780578613,
      "learning_rate": 0.00018891174360200906,
      "loss": 0.1298,
      "step": 8246
    },
    {
      "epoch": 1.1558514365802384,
      "grad_norm": 0.38486745953559875,
      "learning_rate": 0.0001888973929681894,
      "loss": 0.0391,
      "step": 8247
    },
    {
      "epoch": 1.1559915907498248,
      "grad_norm": 0.3698129653930664,
      "learning_rate": 0.00018888304233436975,
      "loss": 0.1231,
      "step": 8248
    },
    {
      "epoch": 1.1561317449194113,
      "grad_norm": 0.08605755865573883,
      "learning_rate": 0.00018886869170055008,
      "loss": 0.0105,
      "step": 8249
    },
    {
      "epoch": 1.156271899088998,
      "grad_norm": 0.2706579864025116,
      "learning_rate": 0.00018885434106673043,
      "loss": 0.051,
      "step": 8250
    },
    {
      "epoch": 1.1564120532585844,
      "grad_norm": 0.5422663688659668,
      "learning_rate": 0.0001888399904329108,
      "loss": 0.0802,
      "step": 8251
    },
    {
      "epoch": 1.156552207428171,
      "grad_norm": 0.29781830310821533,
      "learning_rate": 0.00018882563979909112,
      "loss": 0.0758,
      "step": 8252
    },
    {
      "epoch": 1.1566923615977576,
      "grad_norm": 0.15586911141872406,
      "learning_rate": 0.00018881128916527144,
      "loss": 0.0263,
      "step": 8253
    },
    {
      "epoch": 1.156832515767344,
      "grad_norm": 0.2306089550256729,
      "learning_rate": 0.0001887969385314518,
      "loss": 0.0436,
      "step": 8254
    },
    {
      "epoch": 1.1569726699369307,
      "grad_norm": 0.44029954075813293,
      "learning_rate": 0.00018878258789763213,
      "loss": 0.0375,
      "step": 8255
    },
    {
      "epoch": 1.1571128241065172,
      "grad_norm": 0.2494993805885315,
      "learning_rate": 0.00018876823726381246,
      "loss": 0.0483,
      "step": 8256
    },
    {
      "epoch": 1.1572529782761036,
      "grad_norm": 0.3793202042579651,
      "learning_rate": 0.0001887538866299928,
      "loss": 0.0335,
      "step": 8257
    },
    {
      "epoch": 1.1573931324456903,
      "grad_norm": 0.3121330440044403,
      "learning_rate": 0.00018873953599617314,
      "loss": 0.0555,
      "step": 8258
    },
    {
      "epoch": 1.1575332866152768,
      "grad_norm": 0.45819664001464844,
      "learning_rate": 0.00018872518536235347,
      "loss": 0.0836,
      "step": 8259
    },
    {
      "epoch": 1.1576734407848632,
      "grad_norm": 0.1707604080438614,
      "learning_rate": 0.0001887108347285338,
      "loss": 0.0171,
      "step": 8260
    },
    {
      "epoch": 1.15781359495445,
      "grad_norm": 0.3753352463245392,
      "learning_rate": 0.00018869648409471418,
      "loss": 0.0768,
      "step": 8261
    },
    {
      "epoch": 1.1579537491240364,
      "grad_norm": 0.31892135739326477,
      "learning_rate": 0.0001886821334608945,
      "loss": 0.0795,
      "step": 8262
    },
    {
      "epoch": 1.158093903293623,
      "grad_norm": 0.1381869912147522,
      "learning_rate": 0.00018866778282707484,
      "loss": 0.0153,
      "step": 8263
    },
    {
      "epoch": 1.1582340574632095,
      "grad_norm": 0.32289040088653564,
      "learning_rate": 0.0001886534321932552,
      "loss": 0.0499,
      "step": 8264
    },
    {
      "epoch": 1.158374211632796,
      "grad_norm": 0.15772201120853424,
      "learning_rate": 0.00018863908155943552,
      "loss": 0.0117,
      "step": 8265
    },
    {
      "epoch": 1.1585143658023827,
      "grad_norm": 0.4177266061306,
      "learning_rate": 0.00018862473092561585,
      "loss": 0.0557,
      "step": 8266
    },
    {
      "epoch": 1.158654519971969,
      "grad_norm": 0.3156067430973053,
      "learning_rate": 0.0001886103802917962,
      "loss": 0.0222,
      "step": 8267
    },
    {
      "epoch": 1.1587946741415558,
      "grad_norm": 0.2939659059047699,
      "learning_rate": 0.00018859602965797654,
      "loss": 0.065,
      "step": 8268
    },
    {
      "epoch": 1.1589348283111423,
      "grad_norm": 0.6546409726142883,
      "learning_rate": 0.00018858167902415686,
      "loss": 0.0761,
      "step": 8269
    },
    {
      "epoch": 1.1590749824807287,
      "grad_norm": 0.29267361760139465,
      "learning_rate": 0.00018856732839033725,
      "loss": 0.0435,
      "step": 8270
    },
    {
      "epoch": 1.1592151366503154,
      "grad_norm": 0.250063955783844,
      "learning_rate": 0.00018855297775651758,
      "loss": 0.0232,
      "step": 8271
    },
    {
      "epoch": 1.1593552908199019,
      "grad_norm": 0.37106165289878845,
      "learning_rate": 0.0001885386271226979,
      "loss": 0.1344,
      "step": 8272
    },
    {
      "epoch": 1.1594954449894885,
      "grad_norm": 0.7054430842399597,
      "learning_rate": 0.00018852427648887826,
      "loss": 0.241,
      "step": 8273
    },
    {
      "epoch": 1.159635599159075,
      "grad_norm": 0.653161346912384,
      "learning_rate": 0.0001885099258550586,
      "loss": 0.0986,
      "step": 8274
    },
    {
      "epoch": 1.1597757533286615,
      "grad_norm": 0.5201935768127441,
      "learning_rate": 0.00018849557522123892,
      "loss": 0.0774,
      "step": 8275
    },
    {
      "epoch": 1.1599159074982481,
      "grad_norm": 0.33321377635002136,
      "learning_rate": 0.00018848122458741927,
      "loss": 0.0748,
      "step": 8276
    },
    {
      "epoch": 1.1600560616678346,
      "grad_norm": 0.8500821590423584,
      "learning_rate": 0.0001884668739535996,
      "loss": 0.0301,
      "step": 8277
    },
    {
      "epoch": 1.1601962158374213,
      "grad_norm": 0.8910528421401978,
      "learning_rate": 0.00018845252331977993,
      "loss": 0.3395,
      "step": 8278
    },
    {
      "epoch": 1.1603363700070077,
      "grad_norm": 0.37445971369743347,
      "learning_rate": 0.00018843817268596026,
      "loss": 0.0631,
      "step": 8279
    },
    {
      "epoch": 1.1604765241765942,
      "grad_norm": 0.6641671657562256,
      "learning_rate": 0.00018842382205214061,
      "loss": 0.081,
      "step": 8280
    },
    {
      "epoch": 1.1606166783461809,
      "grad_norm": 0.955694317817688,
      "learning_rate": 0.00018840947141832097,
      "loss": 0.0942,
      "step": 8281
    },
    {
      "epoch": 1.1607568325157673,
      "grad_norm": 1.1868804693222046,
      "learning_rate": 0.0001883951207845013,
      "loss": 0.1858,
      "step": 8282
    },
    {
      "epoch": 1.1608969866853538,
      "grad_norm": 1.3917365074157715,
      "learning_rate": 0.00018838077015068165,
      "loss": 0.0717,
      "step": 8283
    },
    {
      "epoch": 1.1610371408549405,
      "grad_norm": 1.6709376573562622,
      "learning_rate": 0.00018836641951686198,
      "loss": 0.0724,
      "step": 8284
    },
    {
      "epoch": 1.161177295024527,
      "grad_norm": 8.957598686218262,
      "learning_rate": 0.0001883520688830423,
      "loss": 0.202,
      "step": 8285
    },
    {
      "epoch": 1.1613174491941136,
      "grad_norm": 0.15242385864257812,
      "learning_rate": 0.00018833771824922267,
      "loss": 0.0228,
      "step": 8286
    },
    {
      "epoch": 1.1614576033637,
      "grad_norm": 0.34917619824409485,
      "learning_rate": 0.000188323367615403,
      "loss": 0.0438,
      "step": 8287
    },
    {
      "epoch": 1.1615977575332865,
      "grad_norm": 0.578653872013092,
      "learning_rate": 0.00018830901698158332,
      "loss": 0.0637,
      "step": 8288
    },
    {
      "epoch": 1.1617379117028732,
      "grad_norm": 0.7125800848007202,
      "learning_rate": 0.00018829466634776368,
      "loss": 0.1002,
      "step": 8289
    },
    {
      "epoch": 1.1618780658724597,
      "grad_norm": 0.3307386636734009,
      "learning_rate": 0.000188280315713944,
      "loss": 0.0591,
      "step": 8290
    },
    {
      "epoch": 1.1620182200420461,
      "grad_norm": 0.24913068115711212,
      "learning_rate": 0.00018826596508012434,
      "loss": 0.057,
      "step": 8291
    },
    {
      "epoch": 1.1621583742116328,
      "grad_norm": 0.4080732464790344,
      "learning_rate": 0.00018825161444630472,
      "loss": 0.0486,
      "step": 8292
    },
    {
      "epoch": 1.1622985283812193,
      "grad_norm": 0.31905221939086914,
      "learning_rate": 0.00018823726381248505,
      "loss": 0.0509,
      "step": 8293
    },
    {
      "epoch": 1.162438682550806,
      "grad_norm": 0.2527974843978882,
      "learning_rate": 0.00018822291317866538,
      "loss": 0.0619,
      "step": 8294
    },
    {
      "epoch": 1.1625788367203924,
      "grad_norm": 0.24787454307079315,
      "learning_rate": 0.0001882085625448457,
      "loss": 0.0331,
      "step": 8295
    },
    {
      "epoch": 1.1627189908899789,
      "grad_norm": 0.27092069387435913,
      "learning_rate": 0.00018819421191102606,
      "loss": 0.0689,
      "step": 8296
    },
    {
      "epoch": 1.1628591450595656,
      "grad_norm": 0.18213745951652527,
      "learning_rate": 0.0001881798612772064,
      "loss": 0.0494,
      "step": 8297
    },
    {
      "epoch": 1.162999299229152,
      "grad_norm": 0.43573054671287537,
      "learning_rate": 0.00018816551064338672,
      "loss": 0.0777,
      "step": 8298
    },
    {
      "epoch": 1.1631394533987387,
      "grad_norm": 0.42315080761909485,
      "learning_rate": 0.00018815116000956707,
      "loss": 0.057,
      "step": 8299
    },
    {
      "epoch": 1.1632796075683252,
      "grad_norm": 1.2821844816207886,
      "learning_rate": 0.0001881368093757474,
      "loss": 0.0505,
      "step": 8300
    },
    {
      "epoch": 1.1634197617379116,
      "grad_norm": 0.3277112543582916,
      "learning_rate": 0.00018812245874192773,
      "loss": 0.0574,
      "step": 8301
    },
    {
      "epoch": 1.1635599159074983,
      "grad_norm": 0.41321098804473877,
      "learning_rate": 0.00018810810810810811,
      "loss": 0.113,
      "step": 8302
    },
    {
      "epoch": 1.1637000700770848,
      "grad_norm": 0.2844173014163971,
      "learning_rate": 0.00018809375747428844,
      "loss": 0.038,
      "step": 8303
    },
    {
      "epoch": 1.1638402242466714,
      "grad_norm": 0.31971997022628784,
      "learning_rate": 0.00018807940684046877,
      "loss": 0.0555,
      "step": 8304
    },
    {
      "epoch": 1.163980378416258,
      "grad_norm": 0.3507886826992035,
      "learning_rate": 0.00018806505620664913,
      "loss": 0.0465,
      "step": 8305
    },
    {
      "epoch": 1.1641205325858444,
      "grad_norm": 0.20255932211875916,
      "learning_rate": 0.00018805070557282945,
      "loss": 0.0162,
      "step": 8306
    },
    {
      "epoch": 1.164260686755431,
      "grad_norm": 0.33217859268188477,
      "learning_rate": 0.00018803635493900978,
      "loss": 0.0386,
      "step": 8307
    },
    {
      "epoch": 1.1644008409250175,
      "grad_norm": 0.19839653372764587,
      "learning_rate": 0.00018802200430519014,
      "loss": 0.0474,
      "step": 8308
    },
    {
      "epoch": 1.1645409950946042,
      "grad_norm": 0.19316379725933075,
      "learning_rate": 0.00018800765367137047,
      "loss": 0.0393,
      "step": 8309
    },
    {
      "epoch": 1.1646811492641906,
      "grad_norm": 0.19684062898159027,
      "learning_rate": 0.0001879933030375508,
      "loss": 0.0252,
      "step": 8310
    },
    {
      "epoch": 1.164821303433777,
      "grad_norm": 0.2286255657672882,
      "learning_rate": 0.00018797895240373115,
      "loss": 0.0473,
      "step": 8311
    },
    {
      "epoch": 1.1649614576033638,
      "grad_norm": 0.18859437108039856,
      "learning_rate": 0.00018796460176991148,
      "loss": 0.0309,
      "step": 8312
    },
    {
      "epoch": 1.1651016117729502,
      "grad_norm": 0.1314937174320221,
      "learning_rate": 0.00018795025113609184,
      "loss": 0.021,
      "step": 8313
    },
    {
      "epoch": 1.1652417659425367,
      "grad_norm": 0.369870662689209,
      "learning_rate": 0.00018793590050227216,
      "loss": 0.0622,
      "step": 8314
    },
    {
      "epoch": 1.1653819201121234,
      "grad_norm": 0.7222849726676941,
      "learning_rate": 0.00018792154986845252,
      "loss": 0.0494,
      "step": 8315
    },
    {
      "epoch": 1.1655220742817098,
      "grad_norm": 0.3878611624240875,
      "learning_rate": 0.00018790719923463285,
      "loss": 0.0789,
      "step": 8316
    },
    {
      "epoch": 1.1656622284512965,
      "grad_norm": 0.22645603120326996,
      "learning_rate": 0.00018789284860081318,
      "loss": 0.0144,
      "step": 8317
    },
    {
      "epoch": 1.165802382620883,
      "grad_norm": 0.21751196682453156,
      "learning_rate": 0.00018787849796699353,
      "loss": 0.0375,
      "step": 8318
    },
    {
      "epoch": 1.1659425367904694,
      "grad_norm": 0.2437049299478531,
      "learning_rate": 0.00018786414733317386,
      "loss": 0.0217,
      "step": 8319
    },
    {
      "epoch": 1.1660826909600561,
      "grad_norm": 0.14584419131278992,
      "learning_rate": 0.0001878497966993542,
      "loss": 0.0241,
      "step": 8320
    },
    {
      "epoch": 1.1662228451296426,
      "grad_norm": 0.15523898601531982,
      "learning_rate": 0.00018783544606553455,
      "loss": 0.0232,
      "step": 8321
    },
    {
      "epoch": 1.166362999299229,
      "grad_norm": 0.2595839202404022,
      "learning_rate": 0.00018782109543171487,
      "loss": 0.047,
      "step": 8322
    },
    {
      "epoch": 1.1665031534688157,
      "grad_norm": 0.5731723308563232,
      "learning_rate": 0.0001878067447978952,
      "loss": 0.0375,
      "step": 8323
    },
    {
      "epoch": 1.1666433076384022,
      "grad_norm": 0.7627415657043457,
      "learning_rate": 0.00018779239416407559,
      "loss": 0.0688,
      "step": 8324
    },
    {
      "epoch": 1.1667834618079889,
      "grad_norm": 1.139765977859497,
      "learning_rate": 0.00018777804353025591,
      "loss": 0.0877,
      "step": 8325
    },
    {
      "epoch": 1.1669236159775753,
      "grad_norm": 0.40040647983551025,
      "learning_rate": 0.00018776369289643624,
      "loss": 0.0584,
      "step": 8326
    },
    {
      "epoch": 1.1670637701471618,
      "grad_norm": 0.5922824740409851,
      "learning_rate": 0.0001877493422626166,
      "loss": 0.0681,
      "step": 8327
    },
    {
      "epoch": 1.1672039243167485,
      "grad_norm": 0.2685219347476959,
      "learning_rate": 0.00018773499162879693,
      "loss": 0.0293,
      "step": 8328
    },
    {
      "epoch": 1.167344078486335,
      "grad_norm": 1.1396485567092896,
      "learning_rate": 0.00018772064099497726,
      "loss": 0.0492,
      "step": 8329
    },
    {
      "epoch": 1.1674842326559216,
      "grad_norm": 0.43605372309684753,
      "learning_rate": 0.00018770629036115758,
      "loss": 0.0354,
      "step": 8330
    },
    {
      "epoch": 1.167624386825508,
      "grad_norm": 0.26172560453414917,
      "learning_rate": 0.00018769193972733794,
      "loss": 0.0257,
      "step": 8331
    },
    {
      "epoch": 1.1677645409950945,
      "grad_norm": 0.2245074361562729,
      "learning_rate": 0.00018767758909351827,
      "loss": 0.0195,
      "step": 8332
    },
    {
      "epoch": 1.1679046951646812,
      "grad_norm": 1.2434290647506714,
      "learning_rate": 0.0001876632384596986,
      "loss": 0.0706,
      "step": 8333
    },
    {
      "epoch": 1.1680448493342677,
      "grad_norm": 1.3934446573257446,
      "learning_rate": 0.00018764888782587898,
      "loss": 0.2143,
      "step": 8334
    },
    {
      "epoch": 1.1681850035038543,
      "grad_norm": 1.9912471771240234,
      "learning_rate": 0.0001876345371920593,
      "loss": 0.1416,
      "step": 8335
    },
    {
      "epoch": 1.1683251576734408,
      "grad_norm": 0.16852135956287384,
      "learning_rate": 0.00018762018655823964,
      "loss": 0.0163,
      "step": 8336
    },
    {
      "epoch": 1.1684653118430273,
      "grad_norm": 0.4382365047931671,
      "learning_rate": 0.00018760583592442,
      "loss": 0.0735,
      "step": 8337
    },
    {
      "epoch": 1.168605466012614,
      "grad_norm": 0.34233397245407104,
      "learning_rate": 0.00018759148529060032,
      "loss": 0.0371,
      "step": 8338
    },
    {
      "epoch": 1.1687456201822004,
      "grad_norm": 0.36186596751213074,
      "learning_rate": 0.00018757713465678065,
      "loss": 0.0356,
      "step": 8339
    },
    {
      "epoch": 1.168885774351787,
      "grad_norm": 0.27241790294647217,
      "learning_rate": 0.000187562784022961,
      "loss": 0.0855,
      "step": 8340
    },
    {
      "epoch": 1.1690259285213735,
      "grad_norm": 0.24945935606956482,
      "learning_rate": 0.00018754843338914133,
      "loss": 0.0484,
      "step": 8341
    },
    {
      "epoch": 1.16916608269096,
      "grad_norm": 0.2609570324420929,
      "learning_rate": 0.00018753408275532166,
      "loss": 0.044,
      "step": 8342
    },
    {
      "epoch": 1.1693062368605467,
      "grad_norm": 0.548702597618103,
      "learning_rate": 0.00018751973212150202,
      "loss": 0.0445,
      "step": 8343
    },
    {
      "epoch": 1.1694463910301331,
      "grad_norm": 0.3048005700111389,
      "learning_rate": 0.00018750538148768235,
      "loss": 0.0204,
      "step": 8344
    },
    {
      "epoch": 1.1695865451997196,
      "grad_norm": 0.7119125127792358,
      "learning_rate": 0.0001874910308538627,
      "loss": 0.0389,
      "step": 8345
    },
    {
      "epoch": 1.1697266993693063,
      "grad_norm": 0.31920552253723145,
      "learning_rate": 0.00018747668022004306,
      "loss": 0.073,
      "step": 8346
    },
    {
      "epoch": 1.1698668535388927,
      "grad_norm": 0.8130448460578918,
      "learning_rate": 0.0001874623295862234,
      "loss": 0.08,
      "step": 8347
    },
    {
      "epoch": 1.1700070077084792,
      "grad_norm": 0.2658419609069824,
      "learning_rate": 0.00018744797895240372,
      "loss": 0.0444,
      "step": 8348
    },
    {
      "epoch": 1.170147161878066,
      "grad_norm": 0.11541197448968887,
      "learning_rate": 0.00018743362831858404,
      "loss": 0.0073,
      "step": 8349
    },
    {
      "epoch": 1.1702873160476523,
      "grad_norm": 0.41286951303482056,
      "learning_rate": 0.0001874192776847644,
      "loss": 0.0716,
      "step": 8350
    },
    {
      "epoch": 1.170427470217239,
      "grad_norm": 0.38991743326187134,
      "learning_rate": 0.00018740492705094473,
      "loss": 0.0308,
      "step": 8351
    },
    {
      "epoch": 1.1705676243868255,
      "grad_norm": 0.13846030831336975,
      "learning_rate": 0.00018739057641712506,
      "loss": 0.0212,
      "step": 8352
    },
    {
      "epoch": 1.170707778556412,
      "grad_norm": 0.6411053538322449,
      "learning_rate": 0.0001873762257833054,
      "loss": 0.1269,
      "step": 8353
    },
    {
      "epoch": 1.1708479327259986,
      "grad_norm": 0.25690263509750366,
      "learning_rate": 0.00018736187514948574,
      "loss": 0.0428,
      "step": 8354
    },
    {
      "epoch": 1.170988086895585,
      "grad_norm": 0.18725794553756714,
      "learning_rate": 0.00018734752451566607,
      "loss": 0.0313,
      "step": 8355
    },
    {
      "epoch": 1.1711282410651718,
      "grad_norm": 0.21428105235099792,
      "learning_rate": 0.00018733317388184645,
      "loss": 0.0185,
      "step": 8356
    },
    {
      "epoch": 1.1712683952347582,
      "grad_norm": 0.25514960289001465,
      "learning_rate": 0.00018731882324802678,
      "loss": 0.0334,
      "step": 8357
    },
    {
      "epoch": 1.1714085494043447,
      "grad_norm": 0.43981075286865234,
      "learning_rate": 0.0001873044726142071,
      "loss": 0.0703,
      "step": 8358
    },
    {
      "epoch": 1.1715487035739314,
      "grad_norm": 0.4248395562171936,
      "learning_rate": 0.00018729012198038746,
      "loss": 0.1012,
      "step": 8359
    },
    {
      "epoch": 1.1716888577435178,
      "grad_norm": 0.604216456413269,
      "learning_rate": 0.0001872757713465678,
      "loss": 0.0676,
      "step": 8360
    },
    {
      "epoch": 1.1718290119131045,
      "grad_norm": 0.08519989252090454,
      "learning_rate": 0.00018726142071274812,
      "loss": 0.0073,
      "step": 8361
    },
    {
      "epoch": 1.171969166082691,
      "grad_norm": 0.4951140880584717,
      "learning_rate": 0.00018724707007892848,
      "loss": 0.1452,
      "step": 8362
    },
    {
      "epoch": 1.1721093202522774,
      "grad_norm": 0.3014000654220581,
      "learning_rate": 0.0001872327194451088,
      "loss": 0.0519,
      "step": 8363
    },
    {
      "epoch": 1.1722494744218641,
      "grad_norm": 0.291948527097702,
      "learning_rate": 0.00018721836881128913,
      "loss": 0.0306,
      "step": 8364
    },
    {
      "epoch": 1.1723896285914506,
      "grad_norm": 0.21021240949630737,
      "learning_rate": 0.00018720401817746946,
      "loss": 0.0423,
      "step": 8365
    },
    {
      "epoch": 1.1725297827610373,
      "grad_norm": 1.003395438194275,
      "learning_rate": 0.00018718966754364985,
      "loss": 0.1596,
      "step": 8366
    },
    {
      "epoch": 1.1726699369306237,
      "grad_norm": 1.0240802764892578,
      "learning_rate": 0.00018717531690983017,
      "loss": 0.0725,
      "step": 8367
    },
    {
      "epoch": 1.1728100911002102,
      "grad_norm": 0.6086975932121277,
      "learning_rate": 0.0001871609662760105,
      "loss": 0.0853,
      "step": 8368
    },
    {
      "epoch": 1.1729502452697969,
      "grad_norm": 0.4363154172897339,
      "learning_rate": 0.00018714661564219086,
      "loss": 0.0451,
      "step": 8369
    },
    {
      "epoch": 1.1730903994393833,
      "grad_norm": 0.28637152910232544,
      "learning_rate": 0.0001871322650083712,
      "loss": 0.0507,
      "step": 8370
    },
    {
      "epoch": 1.1732305536089698,
      "grad_norm": 0.33646801114082336,
      "learning_rate": 0.00018711791437455152,
      "loss": 0.0356,
      "step": 8371
    },
    {
      "epoch": 1.1733707077785565,
      "grad_norm": 0.1964593231678009,
      "learning_rate": 0.00018710356374073187,
      "loss": 0.0188,
      "step": 8372
    },
    {
      "epoch": 1.173510861948143,
      "grad_norm": 0.6283341646194458,
      "learning_rate": 0.0001870892131069122,
      "loss": 0.059,
      "step": 8373
    },
    {
      "epoch": 1.1736510161177296,
      "grad_norm": 0.6729354858398438,
      "learning_rate": 0.00018707486247309253,
      "loss": 0.0466,
      "step": 8374
    },
    {
      "epoch": 1.173791170287316,
      "grad_norm": 0.3486192226409912,
      "learning_rate": 0.00018706051183927288,
      "loss": 0.0163,
      "step": 8375
    },
    {
      "epoch": 1.1739313244569025,
      "grad_norm": 0.14129824936389923,
      "learning_rate": 0.0001870461612054532,
      "loss": 0.0325,
      "step": 8376
    },
    {
      "epoch": 1.1740714786264892,
      "grad_norm": 0.34836235642433167,
      "learning_rate": 0.00018703181057163357,
      "loss": 0.0172,
      "step": 8377
    },
    {
      "epoch": 1.1742116327960757,
      "grad_norm": 0.6925960183143616,
      "learning_rate": 0.00018701745993781392,
      "loss": 0.1112,
      "step": 8378
    },
    {
      "epoch": 1.1743517869656621,
      "grad_norm": 0.3958178162574768,
      "learning_rate": 0.00018700310930399425,
      "loss": 0.1184,
      "step": 8379
    },
    {
      "epoch": 1.1744919411352488,
      "grad_norm": 0.13694526255130768,
      "learning_rate": 0.00018698875867017458,
      "loss": 0.0094,
      "step": 8380
    },
    {
      "epoch": 1.1746320953048353,
      "grad_norm": 0.7940552830696106,
      "learning_rate": 0.00018697440803635494,
      "loss": 0.0294,
      "step": 8381
    },
    {
      "epoch": 1.174772249474422,
      "grad_norm": 0.8132178783416748,
      "learning_rate": 0.00018696005740253527,
      "loss": 0.1169,
      "step": 8382
    },
    {
      "epoch": 1.1749124036440084,
      "grad_norm": 0.23389962315559387,
      "learning_rate": 0.0001869457067687156,
      "loss": 0.055,
      "step": 8383
    },
    {
      "epoch": 1.1750525578135949,
      "grad_norm": 0.90741366147995,
      "learning_rate": 0.00018693135613489592,
      "loss": 0.2185,
      "step": 8384
    },
    {
      "epoch": 1.1751927119831815,
      "grad_norm": 3.7466557025909424,
      "learning_rate": 0.00018691700550107628,
      "loss": 0.2281,
      "step": 8385
    },
    {
      "epoch": 1.175332866152768,
      "grad_norm": 0.19903984665870667,
      "learning_rate": 0.0001869026548672566,
      "loss": 0.0249,
      "step": 8386
    },
    {
      "epoch": 1.1754730203223547,
      "grad_norm": 0.26661404967308044,
      "learning_rate": 0.00018688830423343694,
      "loss": 0.1125,
      "step": 8387
    },
    {
      "epoch": 1.1756131744919411,
      "grad_norm": 0.4420032799243927,
      "learning_rate": 0.00018687395359961732,
      "loss": 0.0712,
      "step": 8388
    },
    {
      "epoch": 1.1757533286615276,
      "grad_norm": 0.19613388180732727,
      "learning_rate": 0.00018685960296579765,
      "loss": 0.02,
      "step": 8389
    },
    {
      "epoch": 1.1758934828311143,
      "grad_norm": 0.1935284435749054,
      "learning_rate": 0.00018684525233197798,
      "loss": 0.0478,
      "step": 8390
    },
    {
      "epoch": 1.1760336370007007,
      "grad_norm": 0.3692995309829712,
      "learning_rate": 0.00018683090169815833,
      "loss": 0.0551,
      "step": 8391
    },
    {
      "epoch": 1.1761737911702874,
      "grad_norm": 0.20738908648490906,
      "learning_rate": 0.00018681655106433866,
      "loss": 0.0446,
      "step": 8392
    },
    {
      "epoch": 1.1763139453398739,
      "grad_norm": 1.241309404373169,
      "learning_rate": 0.000186802200430519,
      "loss": 0.1116,
      "step": 8393
    },
    {
      "epoch": 1.1764540995094603,
      "grad_norm": 0.5874578952789307,
      "learning_rate": 0.00018678784979669934,
      "loss": 0.079,
      "step": 8394
    },
    {
      "epoch": 1.176594253679047,
      "grad_norm": 0.19788005948066711,
      "learning_rate": 0.00018677349916287967,
      "loss": 0.0494,
      "step": 8395
    },
    {
      "epoch": 1.1767344078486335,
      "grad_norm": 0.31705984473228455,
      "learning_rate": 0.00018675914852906,
      "loss": 0.0703,
      "step": 8396
    },
    {
      "epoch": 1.1768745620182202,
      "grad_norm": 0.2618476152420044,
      "learning_rate": 0.00018674479789524038,
      "loss": 0.0282,
      "step": 8397
    },
    {
      "epoch": 1.1770147161878066,
      "grad_norm": 0.3796525001525879,
      "learning_rate": 0.0001867304472614207,
      "loss": 0.0952,
      "step": 8398
    },
    {
      "epoch": 1.177154870357393,
      "grad_norm": 0.2857600450515747,
      "learning_rate": 0.00018671609662760104,
      "loss": 0.0259,
      "step": 8399
    },
    {
      "epoch": 1.1772950245269798,
      "grad_norm": 0.4316628575325012,
      "learning_rate": 0.00018670174599378137,
      "loss": 0.0495,
      "step": 8400
    },
    {
      "epoch": 1.1774351786965662,
      "grad_norm": 0.3237719237804413,
      "learning_rate": 0.00018668739535996173,
      "loss": 0.0424,
      "step": 8401
    },
    {
      "epoch": 1.1775753328661527,
      "grad_norm": 0.5039601922035217,
      "learning_rate": 0.00018667304472614205,
      "loss": 0.1105,
      "step": 8402
    },
    {
      "epoch": 1.1777154870357394,
      "grad_norm": 0.44629162549972534,
      "learning_rate": 0.00018665869409232238,
      "loss": 0.0719,
      "step": 8403
    },
    {
      "epoch": 1.1778556412053258,
      "grad_norm": 0.878272533416748,
      "learning_rate": 0.00018664434345850274,
      "loss": 0.1124,
      "step": 8404
    },
    {
      "epoch": 1.1779957953749123,
      "grad_norm": 0.35363179445266724,
      "learning_rate": 0.00018662999282468307,
      "loss": 0.0515,
      "step": 8405
    },
    {
      "epoch": 1.178135949544499,
      "grad_norm": 0.205809086561203,
      "learning_rate": 0.0001866156421908634,
      "loss": 0.0342,
      "step": 8406
    },
    {
      "epoch": 1.1782761037140854,
      "grad_norm": 0.2189025729894638,
      "learning_rate": 0.00018660129155704375,
      "loss": 0.0411,
      "step": 8407
    },
    {
      "epoch": 1.178416257883672,
      "grad_norm": 0.4029066264629364,
      "learning_rate": 0.0001865869409232241,
      "loss": 0.0777,
      "step": 8408
    },
    {
      "epoch": 1.1785564120532586,
      "grad_norm": 0.3204864263534546,
      "learning_rate": 0.00018657259028940444,
      "loss": 0.0622,
      "step": 8409
    },
    {
      "epoch": 1.178696566222845,
      "grad_norm": 0.1740313470363617,
      "learning_rate": 0.0001865582396555848,
      "loss": 0.0365,
      "step": 8410
    },
    {
      "epoch": 1.1788367203924317,
      "grad_norm": 0.2346617728471756,
      "learning_rate": 0.00018654388902176512,
      "loss": 0.0189,
      "step": 8411
    },
    {
      "epoch": 1.1789768745620182,
      "grad_norm": 0.37026694416999817,
      "learning_rate": 0.00018652953838794545,
      "loss": 0.0437,
      "step": 8412
    },
    {
      "epoch": 1.1791170287316048,
      "grad_norm": 0.25204741954803467,
      "learning_rate": 0.0001865151877541258,
      "loss": 0.0261,
      "step": 8413
    },
    {
      "epoch": 1.1792571829011913,
      "grad_norm": 0.26381635665893555,
      "learning_rate": 0.00018650083712030613,
      "loss": 0.0279,
      "step": 8414
    },
    {
      "epoch": 1.1793973370707778,
      "grad_norm": 0.34110718965530396,
      "learning_rate": 0.00018648648648648646,
      "loss": 0.0529,
      "step": 8415
    },
    {
      "epoch": 1.1795374912403644,
      "grad_norm": 0.38720956444740295,
      "learning_rate": 0.00018647213585266682,
      "loss": 0.08,
      "step": 8416
    },
    {
      "epoch": 1.179677645409951,
      "grad_norm": 0.1541094034910202,
      "learning_rate": 0.00018645778521884714,
      "loss": 0.0334,
      "step": 8417
    },
    {
      "epoch": 1.1798177995795376,
      "grad_norm": 0.3178797960281372,
      "learning_rate": 0.00018644343458502747,
      "loss": 0.0846,
      "step": 8418
    },
    {
      "epoch": 1.179957953749124,
      "grad_norm": 0.33582767844200134,
      "learning_rate": 0.0001864290839512078,
      "loss": 0.0617,
      "step": 8419
    },
    {
      "epoch": 1.1800981079187105,
      "grad_norm": 1.152430534362793,
      "learning_rate": 0.00018641473331738818,
      "loss": 0.0881,
      "step": 8420
    },
    {
      "epoch": 1.1802382620882972,
      "grad_norm": 0.2558451294898987,
      "learning_rate": 0.0001864003826835685,
      "loss": 0.0279,
      "step": 8421
    },
    {
      "epoch": 1.1803784162578836,
      "grad_norm": 0.6255281567573547,
      "learning_rate": 0.00018638603204974884,
      "loss": 0.0413,
      "step": 8422
    },
    {
      "epoch": 1.1805185704274703,
      "grad_norm": 0.2184334248304367,
      "learning_rate": 0.0001863716814159292,
      "loss": 0.0407,
      "step": 8423
    },
    {
      "epoch": 1.1806587245970568,
      "grad_norm": 0.30584102869033813,
      "learning_rate": 0.00018635733078210953,
      "loss": 0.0305,
      "step": 8424
    },
    {
      "epoch": 1.1807988787666432,
      "grad_norm": 1.3026423454284668,
      "learning_rate": 0.00018634298014828985,
      "loss": 0.0955,
      "step": 8425
    },
    {
      "epoch": 1.18093903293623,
      "grad_norm": 0.41045674681663513,
      "learning_rate": 0.0001863286295144702,
      "loss": 0.0263,
      "step": 8426
    },
    {
      "epoch": 1.1810791871058164,
      "grad_norm": 0.592070460319519,
      "learning_rate": 0.00018631427888065054,
      "loss": 0.0917,
      "step": 8427
    },
    {
      "epoch": 1.181219341275403,
      "grad_norm": 0.34400054812431335,
      "learning_rate": 0.00018629992824683087,
      "loss": 0.0533,
      "step": 8428
    },
    {
      "epoch": 1.1813594954449895,
      "grad_norm": 0.4891303777694702,
      "learning_rate": 0.00018628557761301125,
      "loss": 0.0814,
      "step": 8429
    },
    {
      "epoch": 1.181499649614576,
      "grad_norm": 0.49518412351608276,
      "learning_rate": 0.00018627122697919158,
      "loss": 0.1097,
      "step": 8430
    },
    {
      "epoch": 1.1816398037841627,
      "grad_norm": 1.4780369997024536,
      "learning_rate": 0.0001862568763453719,
      "loss": 0.2873,
      "step": 8431
    },
    {
      "epoch": 1.1817799579537491,
      "grad_norm": 1.0688928365707397,
      "learning_rate": 0.00018624252571155226,
      "loss": 0.0758,
      "step": 8432
    },
    {
      "epoch": 1.1819201121233356,
      "grad_norm": 2.4572336673736572,
      "learning_rate": 0.0001862281750777326,
      "loss": 0.0754,
      "step": 8433
    },
    {
      "epoch": 1.1820602662929223,
      "grad_norm": 0.5639296770095825,
      "learning_rate": 0.00018621382444391292,
      "loss": 0.1319,
      "step": 8434
    },
    {
      "epoch": 1.1822004204625087,
      "grad_norm": 1.2623234987258911,
      "learning_rate": 0.00018619947381009325,
      "loss": 0.3551,
      "step": 8435
    },
    {
      "epoch": 1.1823405746320952,
      "grad_norm": 0.4160192310810089,
      "learning_rate": 0.0001861851231762736,
      "loss": 0.0809,
      "step": 8436
    },
    {
      "epoch": 1.1824807288016819,
      "grad_norm": 0.4047568142414093,
      "learning_rate": 0.00018617077254245393,
      "loss": 0.0623,
      "step": 8437
    },
    {
      "epoch": 1.1826208829712683,
      "grad_norm": 0.5796579122543335,
      "learning_rate": 0.00018615642190863426,
      "loss": 0.0833,
      "step": 8438
    },
    {
      "epoch": 1.182761037140855,
      "grad_norm": 0.2815111577510834,
      "learning_rate": 0.00018614207127481462,
      "loss": 0.0456,
      "step": 8439
    },
    {
      "epoch": 1.1829011913104415,
      "grad_norm": 0.4378122091293335,
      "learning_rate": 0.00018612772064099497,
      "loss": 0.1344,
      "step": 8440
    },
    {
      "epoch": 1.183041345480028,
      "grad_norm": 0.22896511852741241,
      "learning_rate": 0.0001861133700071753,
      "loss": 0.0526,
      "step": 8441
    },
    {
      "epoch": 1.1831814996496146,
      "grad_norm": 0.1659672111272812,
      "learning_rate": 0.00018609901937335566,
      "loss": 0.0403,
      "step": 8442
    },
    {
      "epoch": 1.183321653819201,
      "grad_norm": 0.42920631170272827,
      "learning_rate": 0.00018608466873953599,
      "loss": 0.1015,
      "step": 8443
    },
    {
      "epoch": 1.1834618079887878,
      "grad_norm": 1.0234919786453247,
      "learning_rate": 0.00018607031810571631,
      "loss": 0.1248,
      "step": 8444
    },
    {
      "epoch": 1.1836019621583742,
      "grad_norm": 0.18251116573810577,
      "learning_rate": 0.00018605596747189667,
      "loss": 0.0376,
      "step": 8445
    },
    {
      "epoch": 1.1837421163279607,
      "grad_norm": 0.1405799686908722,
      "learning_rate": 0.000186041616838077,
      "loss": 0.0383,
      "step": 8446
    },
    {
      "epoch": 1.1838822704975474,
      "grad_norm": 0.5530314445495605,
      "learning_rate": 0.00018602726620425733,
      "loss": 0.1039,
      "step": 8447
    },
    {
      "epoch": 1.1840224246671338,
      "grad_norm": 0.6715810894966125,
      "learning_rate": 0.00018601291557043768,
      "loss": 0.0951,
      "step": 8448
    },
    {
      "epoch": 1.1841625788367205,
      "grad_norm": 0.2891225516796112,
      "learning_rate": 0.000185998564936618,
      "loss": 0.0608,
      "step": 8449
    },
    {
      "epoch": 1.184302733006307,
      "grad_norm": 0.4518788456916809,
      "learning_rate": 0.00018598421430279834,
      "loss": 0.0631,
      "step": 8450
    },
    {
      "epoch": 1.1844428871758934,
      "grad_norm": 0.1502719223499298,
      "learning_rate": 0.00018596986366897867,
      "loss": 0.0279,
      "step": 8451
    },
    {
      "epoch": 1.18458304134548,
      "grad_norm": 0.44686177372932434,
      "learning_rate": 0.00018595551303515905,
      "loss": 0.0337,
      "step": 8452
    },
    {
      "epoch": 1.1847231955150666,
      "grad_norm": 0.33555567264556885,
      "learning_rate": 0.00018594116240133938,
      "loss": 0.0704,
      "step": 8453
    },
    {
      "epoch": 1.1848633496846532,
      "grad_norm": 0.34301823377609253,
      "learning_rate": 0.0001859268117675197,
      "loss": 0.0785,
      "step": 8454
    },
    {
      "epoch": 1.1850035038542397,
      "grad_norm": 0.3517075181007385,
      "learning_rate": 0.00018591246113370006,
      "loss": 0.0901,
      "step": 8455
    },
    {
      "epoch": 1.1851436580238262,
      "grad_norm": 0.2612588703632355,
      "learning_rate": 0.0001858981104998804,
      "loss": 0.0465,
      "step": 8456
    },
    {
      "epoch": 1.1852838121934128,
      "grad_norm": 0.18803571164608002,
      "learning_rate": 0.00018588375986606072,
      "loss": 0.0575,
      "step": 8457
    },
    {
      "epoch": 1.1854239663629993,
      "grad_norm": 0.29370224475860596,
      "learning_rate": 0.00018586940923224108,
      "loss": 0.076,
      "step": 8458
    },
    {
      "epoch": 1.1855641205325858,
      "grad_norm": 0.6146436929702759,
      "learning_rate": 0.0001858550585984214,
      "loss": 0.092,
      "step": 8459
    },
    {
      "epoch": 1.1857042747021724,
      "grad_norm": 0.43006423115730286,
      "learning_rate": 0.00018584070796460173,
      "loss": 0.1221,
      "step": 8460
    },
    {
      "epoch": 1.185844428871759,
      "grad_norm": 0.3662368953227997,
      "learning_rate": 0.00018582635733078212,
      "loss": 0.0991,
      "step": 8461
    },
    {
      "epoch": 1.1859845830413456,
      "grad_norm": 0.3738935589790344,
      "learning_rate": 0.00018581200669696245,
      "loss": 0.0412,
      "step": 8462
    },
    {
      "epoch": 1.186124737210932,
      "grad_norm": 0.4793298542499542,
      "learning_rate": 0.00018579765606314277,
      "loss": 0.0792,
      "step": 8463
    },
    {
      "epoch": 1.1862648913805185,
      "grad_norm": 1.356055736541748,
      "learning_rate": 0.00018578330542932313,
      "loss": 0.0612,
      "step": 8464
    },
    {
      "epoch": 1.1864050455501052,
      "grad_norm": 0.4037184417247772,
      "learning_rate": 0.00018576895479550346,
      "loss": 0.0798,
      "step": 8465
    },
    {
      "epoch": 1.1865451997196916,
      "grad_norm": 0.17769403755664825,
      "learning_rate": 0.00018575460416168379,
      "loss": 0.0253,
      "step": 8466
    },
    {
      "epoch": 1.186685353889278,
      "grad_norm": 0.39587733149528503,
      "learning_rate": 0.00018574025352786414,
      "loss": 0.0369,
      "step": 8467
    },
    {
      "epoch": 1.1868255080588648,
      "grad_norm": 1.1012765169143677,
      "learning_rate": 0.00018572590289404447,
      "loss": 0.0902,
      "step": 8468
    },
    {
      "epoch": 1.1869656622284512,
      "grad_norm": 0.3209918141365051,
      "learning_rate": 0.0001857115522602248,
      "loss": 0.0731,
      "step": 8469
    },
    {
      "epoch": 1.187105816398038,
      "grad_norm": 0.3216061294078827,
      "learning_rate": 0.00018569720162640513,
      "loss": 0.0393,
      "step": 8470
    },
    {
      "epoch": 1.1872459705676244,
      "grad_norm": 0.22348041832447052,
      "learning_rate": 0.00018568285099258548,
      "loss": 0.019,
      "step": 8471
    },
    {
      "epoch": 1.1873861247372108,
      "grad_norm": 0.7250317931175232,
      "learning_rate": 0.00018566850035876584,
      "loss": 0.0225,
      "step": 8472
    },
    {
      "epoch": 1.1875262789067975,
      "grad_norm": 0.2691590487957001,
      "learning_rate": 0.00018565414972494617,
      "loss": 0.026,
      "step": 8473
    },
    {
      "epoch": 1.187666433076384,
      "grad_norm": 0.521014928817749,
      "learning_rate": 0.00018563979909112652,
      "loss": 0.0303,
      "step": 8474
    },
    {
      "epoch": 1.1878065872459707,
      "grad_norm": 0.789585292339325,
      "learning_rate": 0.00018562544845730685,
      "loss": 0.132,
      "step": 8475
    },
    {
      "epoch": 1.1879467414155571,
      "grad_norm": 0.7490088939666748,
      "learning_rate": 0.00018561109782348718,
      "loss": 0.203,
      "step": 8476
    },
    {
      "epoch": 1.1880868955851436,
      "grad_norm": 0.23827721178531647,
      "learning_rate": 0.00018559674718966754,
      "loss": 0.0175,
      "step": 8477
    },
    {
      "epoch": 1.1882270497547303,
      "grad_norm": 0.3921199440956116,
      "learning_rate": 0.00018558239655584786,
      "loss": 0.0483,
      "step": 8478
    },
    {
      "epoch": 1.1883672039243167,
      "grad_norm": 0.28024962544441223,
      "learning_rate": 0.0001855680459220282,
      "loss": 0.0404,
      "step": 8479
    },
    {
      "epoch": 1.1885073580939034,
      "grad_norm": 0.9474804997444153,
      "learning_rate": 0.00018555369528820855,
      "loss": 0.0999,
      "step": 8480
    },
    {
      "epoch": 1.1886475122634899,
      "grad_norm": 0.2688314914703369,
      "learning_rate": 0.00018553934465438888,
      "loss": 0.0143,
      "step": 8481
    },
    {
      "epoch": 1.1887876664330763,
      "grad_norm": 0.7894964218139648,
      "learning_rate": 0.0001855249940205692,
      "loss": 0.1318,
      "step": 8482
    },
    {
      "epoch": 1.188927820602663,
      "grad_norm": 0.8433345556259155,
      "learning_rate": 0.0001855106433867496,
      "loss": 0.1358,
      "step": 8483
    },
    {
      "epoch": 1.1890679747722495,
      "grad_norm": 2.712667465209961,
      "learning_rate": 0.00018549629275292992,
      "loss": 0.0983,
      "step": 8484
    },
    {
      "epoch": 1.1892081289418361,
      "grad_norm": 4.782211780548096,
      "learning_rate": 0.00018548194211911025,
      "loss": 0.1831,
      "step": 8485
    },
    {
      "epoch": 1.1893482831114226,
      "grad_norm": 0.3351871967315674,
      "learning_rate": 0.00018546759148529057,
      "loss": 0.1222,
      "step": 8486
    },
    {
      "epoch": 1.189488437281009,
      "grad_norm": 0.45225581526756287,
      "learning_rate": 0.00018545324085147093,
      "loss": 0.1117,
      "step": 8487
    },
    {
      "epoch": 1.1896285914505957,
      "grad_norm": 0.3638037443161011,
      "learning_rate": 0.00018543889021765126,
      "loss": 0.0836,
      "step": 8488
    },
    {
      "epoch": 1.1897687456201822,
      "grad_norm": 0.2246861606836319,
      "learning_rate": 0.0001854245395838316,
      "loss": 0.0247,
      "step": 8489
    },
    {
      "epoch": 1.1899088997897687,
      "grad_norm": 0.48111027479171753,
      "learning_rate": 0.00018541018895001194,
      "loss": 0.0955,
      "step": 8490
    },
    {
      "epoch": 1.1900490539593553,
      "grad_norm": 0.24523086845874786,
      "learning_rate": 0.00018539583831619227,
      "loss": 0.0215,
      "step": 8491
    },
    {
      "epoch": 1.1901892081289418,
      "grad_norm": 0.22513946890830994,
      "learning_rate": 0.0001853814876823726,
      "loss": 0.0175,
      "step": 8492
    },
    {
      "epoch": 1.1903293622985283,
      "grad_norm": 0.1555069535970688,
      "learning_rate": 0.00018536713704855298,
      "loss": 0.0578,
      "step": 8493
    },
    {
      "epoch": 1.190469516468115,
      "grad_norm": 0.44876307249069214,
      "learning_rate": 0.0001853527864147333,
      "loss": 0.0658,
      "step": 8494
    },
    {
      "epoch": 1.1906096706377014,
      "grad_norm": 0.4169052839279175,
      "learning_rate": 0.00018533843578091364,
      "loss": 0.0635,
      "step": 8495
    },
    {
      "epoch": 1.190749824807288,
      "grad_norm": 0.38028332591056824,
      "learning_rate": 0.000185324085147094,
      "loss": 0.0667,
      "step": 8496
    },
    {
      "epoch": 1.1908899789768745,
      "grad_norm": 0.21552136540412903,
      "learning_rate": 0.00018530973451327432,
      "loss": 0.0427,
      "step": 8497
    },
    {
      "epoch": 1.191030133146461,
      "grad_norm": 0.2731516659259796,
      "learning_rate": 0.00018529538387945465,
      "loss": 0.0326,
      "step": 8498
    },
    {
      "epoch": 1.1911702873160477,
      "grad_norm": 0.4314786493778229,
      "learning_rate": 0.000185281033245635,
      "loss": 0.0403,
      "step": 8499
    },
    {
      "epoch": 1.1913104414856341,
      "grad_norm": 0.32413384318351746,
      "learning_rate": 0.00018526668261181534,
      "loss": 0.0761,
      "step": 8500
    },
    {
      "epoch": 1.1914505956552208,
      "grad_norm": 0.48024535179138184,
      "learning_rate": 0.00018525233197799567,
      "loss": 0.0367,
      "step": 8501
    },
    {
      "epoch": 1.1915907498248073,
      "grad_norm": 0.2946139872074127,
      "learning_rate": 0.00018523798134417602,
      "loss": 0.0575,
      "step": 8502
    },
    {
      "epoch": 1.1917309039943937,
      "grad_norm": 0.5404660105705261,
      "learning_rate": 0.00018522363071035638,
      "loss": 0.1124,
      "step": 8503
    },
    {
      "epoch": 1.1918710581639804,
      "grad_norm": 0.0834084302186966,
      "learning_rate": 0.0001852092800765367,
      "loss": 0.0112,
      "step": 8504
    },
    {
      "epoch": 1.1920112123335669,
      "grad_norm": 0.44430938363075256,
      "learning_rate": 0.00018519492944271703,
      "loss": 0.1216,
      "step": 8505
    },
    {
      "epoch": 1.1921513665031536,
      "grad_norm": 0.4010743498802185,
      "learning_rate": 0.0001851805788088974,
      "loss": 0.0565,
      "step": 8506
    },
    {
      "epoch": 1.19229152067274,
      "grad_norm": 0.7161933183670044,
      "learning_rate": 0.00018516622817507772,
      "loss": 0.1479,
      "step": 8507
    },
    {
      "epoch": 1.1924316748423265,
      "grad_norm": 0.38962656259536743,
      "learning_rate": 0.00018515187754125805,
      "loss": 0.1049,
      "step": 8508
    },
    {
      "epoch": 1.1925718290119132,
      "grad_norm": 0.7673728466033936,
      "learning_rate": 0.0001851375269074384,
      "loss": 0.1296,
      "step": 8509
    },
    {
      "epoch": 1.1927119831814996,
      "grad_norm": 0.2714522182941437,
      "learning_rate": 0.00018512317627361873,
      "loss": 0.0388,
      "step": 8510
    },
    {
      "epoch": 1.1928521373510863,
      "grad_norm": 0.34488505125045776,
      "learning_rate": 0.00018510882563979906,
      "loss": 0.0458,
      "step": 8511
    },
    {
      "epoch": 1.1929922915206728,
      "grad_norm": 0.19427450001239777,
      "learning_rate": 0.00018509447500597942,
      "loss": 0.0691,
      "step": 8512
    },
    {
      "epoch": 1.1931324456902592,
      "grad_norm": 0.1503702700138092,
      "learning_rate": 0.00018508012437215974,
      "loss": 0.0245,
      "step": 8513
    },
    {
      "epoch": 1.193272599859846,
      "grad_norm": 0.39883527159690857,
      "learning_rate": 0.00018506577373834007,
      "loss": 0.0724,
      "step": 8514
    },
    {
      "epoch": 1.1934127540294324,
      "grad_norm": 0.5355033278465271,
      "learning_rate": 0.00018505142310452046,
      "loss": 0.0616,
      "step": 8515
    },
    {
      "epoch": 1.1935529081990188,
      "grad_norm": 0.28657329082489014,
      "learning_rate": 0.00018503707247070078,
      "loss": 0.0659,
      "step": 8516
    },
    {
      "epoch": 1.1936930623686055,
      "grad_norm": 1.1676677465438843,
      "learning_rate": 0.0001850227218368811,
      "loss": 0.1472,
      "step": 8517
    },
    {
      "epoch": 1.193833216538192,
      "grad_norm": 0.2774539291858673,
      "learning_rate": 0.00018500837120306147,
      "loss": 0.0504,
      "step": 8518
    },
    {
      "epoch": 1.1939733707077786,
      "grad_norm": 0.39250561594963074,
      "learning_rate": 0.0001849940205692418,
      "loss": 0.0723,
      "step": 8519
    },
    {
      "epoch": 1.194113524877365,
      "grad_norm": 0.4060557782649994,
      "learning_rate": 0.00018497966993542212,
      "loss": 0.0913,
      "step": 8520
    },
    {
      "epoch": 1.1942536790469516,
      "grad_norm": 0.32384955883026123,
      "learning_rate": 0.00018496531930160245,
      "loss": 0.0929,
      "step": 8521
    },
    {
      "epoch": 1.1943938332165382,
      "grad_norm": 0.24844740331172943,
      "learning_rate": 0.0001849509686677828,
      "loss": 0.0717,
      "step": 8522
    },
    {
      "epoch": 1.1945339873861247,
      "grad_norm": 0.3311687409877777,
      "learning_rate": 0.00018493661803396314,
      "loss": 0.0908,
      "step": 8523
    },
    {
      "epoch": 1.1946741415557112,
      "grad_norm": 0.09977242350578308,
      "learning_rate": 0.00018492226740014347,
      "loss": 0.0139,
      "step": 8524
    },
    {
      "epoch": 1.1948142957252978,
      "grad_norm": 0.28865495324134827,
      "learning_rate": 0.00018490791676632385,
      "loss": 0.0388,
      "step": 8525
    },
    {
      "epoch": 1.1949544498948843,
      "grad_norm": 0.3077395558357239,
      "learning_rate": 0.00018489356613250418,
      "loss": 0.0196,
      "step": 8526
    },
    {
      "epoch": 1.195094604064471,
      "grad_norm": 0.3865760564804077,
      "learning_rate": 0.0001848792154986845,
      "loss": 0.066,
      "step": 8527
    },
    {
      "epoch": 1.1952347582340574,
      "grad_norm": 0.5432672500610352,
      "learning_rate": 0.00018486486486486486,
      "loss": 0.1328,
      "step": 8528
    },
    {
      "epoch": 1.195374912403644,
      "grad_norm": 0.35212069749832153,
      "learning_rate": 0.0001848505142310452,
      "loss": 0.1249,
      "step": 8529
    },
    {
      "epoch": 1.1955150665732306,
      "grad_norm": 0.19249378144741058,
      "learning_rate": 0.00018483616359722552,
      "loss": 0.0123,
      "step": 8530
    },
    {
      "epoch": 1.195655220742817,
      "grad_norm": 0.5189899802207947,
      "learning_rate": 0.00018482181296340587,
      "loss": 0.1196,
      "step": 8531
    },
    {
      "epoch": 1.1957953749124037,
      "grad_norm": 0.4813257157802582,
      "learning_rate": 0.0001848074623295862,
      "loss": 0.1401,
      "step": 8532
    },
    {
      "epoch": 1.1959355290819902,
      "grad_norm": 1.005050539970398,
      "learning_rate": 0.00018479311169576653,
      "loss": 0.1695,
      "step": 8533
    },
    {
      "epoch": 1.1960756832515766,
      "grad_norm": 0.3431726098060608,
      "learning_rate": 0.0001847787610619469,
      "loss": 0.0902,
      "step": 8534
    },
    {
      "epoch": 1.1962158374211633,
      "grad_norm": 1.6111829280853271,
      "learning_rate": 0.00018476441042812724,
      "loss": 0.1166,
      "step": 8535
    },
    {
      "epoch": 1.1963559915907498,
      "grad_norm": 0.4205247461795807,
      "learning_rate": 0.00018475005979430757,
      "loss": 0.0779,
      "step": 8536
    },
    {
      "epoch": 1.1964961457603365,
      "grad_norm": 0.2543555498123169,
      "learning_rate": 0.00018473570916048793,
      "loss": 0.0734,
      "step": 8537
    },
    {
      "epoch": 1.196636299929923,
      "grad_norm": 0.3554800748825073,
      "learning_rate": 0.00018472135852666826,
      "loss": 0.0883,
      "step": 8538
    },
    {
      "epoch": 1.1967764540995094,
      "grad_norm": 0.18363718688488007,
      "learning_rate": 0.00018470700789284858,
      "loss": 0.0463,
      "step": 8539
    },
    {
      "epoch": 1.196916608269096,
      "grad_norm": 0.15836459398269653,
      "learning_rate": 0.0001846926572590289,
      "loss": 0.0536,
      "step": 8540
    },
    {
      "epoch": 1.1970567624386825,
      "grad_norm": 0.31539079546928406,
      "learning_rate": 0.00018467830662520927,
      "loss": 0.0786,
      "step": 8541
    },
    {
      "epoch": 1.1971969166082692,
      "grad_norm": 0.23454636335372925,
      "learning_rate": 0.0001846639559913896,
      "loss": 0.0512,
      "step": 8542
    },
    {
      "epoch": 1.1973370707778557,
      "grad_norm": 0.3851999342441559,
      "learning_rate": 0.00018464960535756993,
      "loss": 0.1011,
      "step": 8543
    },
    {
      "epoch": 1.1974772249474421,
      "grad_norm": 0.3170848786830902,
      "learning_rate": 0.00018463525472375028,
      "loss": 0.038,
      "step": 8544
    },
    {
      "epoch": 1.1976173791170288,
      "grad_norm": 0.34854915738105774,
      "learning_rate": 0.0001846209040899306,
      "loss": 0.0681,
      "step": 8545
    },
    {
      "epoch": 1.1977575332866153,
      "grad_norm": 0.21731990575790405,
      "learning_rate": 0.00018460655345611094,
      "loss": 0.0468,
      "step": 8546
    },
    {
      "epoch": 1.1978976874562017,
      "grad_norm": 0.3786073625087738,
      "learning_rate": 0.00018459220282229132,
      "loss": 0.0663,
      "step": 8547
    },
    {
      "epoch": 1.1980378416257884,
      "grad_norm": 0.1978093832731247,
      "learning_rate": 0.00018457785218847165,
      "loss": 0.0482,
      "step": 8548
    },
    {
      "epoch": 1.1981779957953749,
      "grad_norm": 0.23888733983039856,
      "learning_rate": 0.00018456350155465198,
      "loss": 0.0472,
      "step": 8549
    },
    {
      "epoch": 1.1983181499649616,
      "grad_norm": 0.45601579546928406,
      "learning_rate": 0.00018454915092083233,
      "loss": 0.0751,
      "step": 8550
    },
    {
      "epoch": 1.198458304134548,
      "grad_norm": 0.49066832661628723,
      "learning_rate": 0.00018453480028701266,
      "loss": 0.0865,
      "step": 8551
    },
    {
      "epoch": 1.1985984583041345,
      "grad_norm": 0.1663818657398224,
      "learning_rate": 0.000184520449653193,
      "loss": 0.0206,
      "step": 8552
    },
    {
      "epoch": 1.1987386124737212,
      "grad_norm": 0.7273907661437988,
      "learning_rate": 0.00018450609901937335,
      "loss": 0.0451,
      "step": 8553
    },
    {
      "epoch": 1.1988787666433076,
      "grad_norm": 0.16610193252563477,
      "learning_rate": 0.00018449174838555368,
      "loss": 0.0394,
      "step": 8554
    },
    {
      "epoch": 1.199018920812894,
      "grad_norm": 0.3918784260749817,
      "learning_rate": 0.000184477397751734,
      "loss": 0.0634,
      "step": 8555
    },
    {
      "epoch": 1.1991590749824808,
      "grad_norm": 0.09749762713909149,
      "learning_rate": 0.00018446304711791433,
      "loss": 0.0283,
      "step": 8556
    },
    {
      "epoch": 1.1992992291520672,
      "grad_norm": 0.19365651905536652,
      "learning_rate": 0.00018444869648409472,
      "loss": 0.0109,
      "step": 8557
    },
    {
      "epoch": 1.199439383321654,
      "grad_norm": 0.6888707280158997,
      "learning_rate": 0.00018443434585027504,
      "loss": 0.0822,
      "step": 8558
    },
    {
      "epoch": 1.1995795374912404,
      "grad_norm": 0.3675529658794403,
      "learning_rate": 0.00018441999521645537,
      "loss": 0.0381,
      "step": 8559
    },
    {
      "epoch": 1.1997196916608268,
      "grad_norm": 0.285459965467453,
      "learning_rate": 0.00018440564458263573,
      "loss": 0.0694,
      "step": 8560
    },
    {
      "epoch": 1.1998598458304135,
      "grad_norm": 0.3074505925178528,
      "learning_rate": 0.00018439129394881606,
      "loss": 0.0825,
      "step": 8561
    },
    {
      "epoch": 1.2,
      "grad_norm": 0.4485792815685272,
      "learning_rate": 0.00018437694331499639,
      "loss": 0.077,
      "step": 8562
    },
    {
      "epoch": 1.2001401541695866,
      "grad_norm": 0.6382713913917542,
      "learning_rate": 0.00018436259268117674,
      "loss": 0.1037,
      "step": 8563
    },
    {
      "epoch": 1.200280308339173,
      "grad_norm": 0.7484086751937866,
      "learning_rate": 0.00018434824204735707,
      "loss": 0.1019,
      "step": 8564
    },
    {
      "epoch": 1.2004204625087596,
      "grad_norm": 0.6526056528091431,
      "learning_rate": 0.0001843338914135374,
      "loss": 0.0725,
      "step": 8565
    },
    {
      "epoch": 1.2005606166783462,
      "grad_norm": 0.4769851565361023,
      "learning_rate": 0.00018431954077971775,
      "loss": 0.1073,
      "step": 8566
    },
    {
      "epoch": 1.2007007708479327,
      "grad_norm": 0.3522621691226959,
      "learning_rate": 0.0001843051901458981,
      "loss": 0.0949,
      "step": 8567
    },
    {
      "epoch": 1.2008409250175194,
      "grad_norm": 0.11805888265371323,
      "learning_rate": 0.00018429083951207844,
      "loss": 0.0192,
      "step": 8568
    },
    {
      "epoch": 1.2009810791871058,
      "grad_norm": 0.32674145698547363,
      "learning_rate": 0.0001842764888782588,
      "loss": 0.0906,
      "step": 8569
    },
    {
      "epoch": 1.2011212333566923,
      "grad_norm": 0.39941170811653137,
      "learning_rate": 0.00018426213824443912,
      "loss": 0.0395,
      "step": 8570
    },
    {
      "epoch": 1.201261387526279,
      "grad_norm": 0.3969572186470032,
      "learning_rate": 0.00018424778761061945,
      "loss": 0.0927,
      "step": 8571
    },
    {
      "epoch": 1.2014015416958654,
      "grad_norm": 0.5477501749992371,
      "learning_rate": 0.0001842334369767998,
      "loss": 0.0626,
      "step": 8572
    },
    {
      "epoch": 1.2015416958654521,
      "grad_norm": 0.43994542956352234,
      "learning_rate": 0.00018421908634298013,
      "loss": 0.1254,
      "step": 8573
    },
    {
      "epoch": 1.2016818500350386,
      "grad_norm": 0.3910580277442932,
      "learning_rate": 0.00018420473570916046,
      "loss": 0.032,
      "step": 8574
    },
    {
      "epoch": 1.201822004204625,
      "grad_norm": 0.5842429995536804,
      "learning_rate": 0.0001841903850753408,
      "loss": 0.0464,
      "step": 8575
    },
    {
      "epoch": 1.2019621583742117,
      "grad_norm": 0.1773810237646103,
      "learning_rate": 0.00018417603444152115,
      "loss": 0.0329,
      "step": 8576
    },
    {
      "epoch": 1.2021023125437982,
      "grad_norm": 0.4404746890068054,
      "learning_rate": 0.00018416168380770148,
      "loss": 0.058,
      "step": 8577
    },
    {
      "epoch": 1.2022424667133846,
      "grad_norm": 0.7999029755592346,
      "learning_rate": 0.0001841473331738818,
      "loss": 0.2951,
      "step": 8578
    },
    {
      "epoch": 1.2023826208829713,
      "grad_norm": 0.30653145909309387,
      "learning_rate": 0.0001841329825400622,
      "loss": 0.0962,
      "step": 8579
    },
    {
      "epoch": 1.2025227750525578,
      "grad_norm": 0.2000439316034317,
      "learning_rate": 0.00018411863190624252,
      "loss": 0.0672,
      "step": 8580
    },
    {
      "epoch": 1.2026629292221442,
      "grad_norm": 0.5974790453910828,
      "learning_rate": 0.00018410428127242284,
      "loss": 0.1004,
      "step": 8581
    },
    {
      "epoch": 1.202803083391731,
      "grad_norm": 0.6584644913673401,
      "learning_rate": 0.0001840899306386032,
      "loss": 0.0443,
      "step": 8582
    },
    {
      "epoch": 1.2029432375613174,
      "grad_norm": 0.09968796372413635,
      "learning_rate": 0.00018407558000478353,
      "loss": 0.019,
      "step": 8583
    },
    {
      "epoch": 1.203083391730904,
      "grad_norm": 0.8579974174499512,
      "learning_rate": 0.00018406122937096386,
      "loss": 0.1454,
      "step": 8584
    },
    {
      "epoch": 1.2032235459004905,
      "grad_norm": 1.0567073822021484,
      "learning_rate": 0.0001840468787371442,
      "loss": 0.1917,
      "step": 8585
    },
    {
      "epoch": 1.203363700070077,
      "grad_norm": 0.20587460696697235,
      "learning_rate": 0.00018403252810332454,
      "loss": 0.0284,
      "step": 8586
    },
    {
      "epoch": 1.2035038542396637,
      "grad_norm": 0.21338842809200287,
      "learning_rate": 0.00018401817746950487,
      "loss": 0.0574,
      "step": 8587
    },
    {
      "epoch": 1.2036440084092501,
      "grad_norm": 0.2831384241580963,
      "learning_rate": 0.00018400382683568525,
      "loss": 0.0335,
      "step": 8588
    },
    {
      "epoch": 1.2037841625788368,
      "grad_norm": 0.24759230017662048,
      "learning_rate": 0.00018398947620186558,
      "loss": 0.0482,
      "step": 8589
    },
    {
      "epoch": 1.2039243167484233,
      "grad_norm": 0.18353457748889923,
      "learning_rate": 0.0001839751255680459,
      "loss": 0.0466,
      "step": 8590
    },
    {
      "epoch": 1.2040644709180097,
      "grad_norm": 0.2520129084587097,
      "learning_rate": 0.00018396077493422624,
      "loss": 0.0859,
      "step": 8591
    },
    {
      "epoch": 1.2042046250875964,
      "grad_norm": 0.4092220664024353,
      "learning_rate": 0.0001839464243004066,
      "loss": 0.0803,
      "step": 8592
    },
    {
      "epoch": 1.2043447792571829,
      "grad_norm": 0.20996958017349243,
      "learning_rate": 0.00018393207366658692,
      "loss": 0.0617,
      "step": 8593
    },
    {
      "epoch": 1.2044849334267695,
      "grad_norm": 0.25399351119995117,
      "learning_rate": 0.00018391772303276725,
      "loss": 0.0463,
      "step": 8594
    },
    {
      "epoch": 1.204625087596356,
      "grad_norm": 0.2289029359817505,
      "learning_rate": 0.0001839033723989476,
      "loss": 0.0401,
      "step": 8595
    },
    {
      "epoch": 1.2047652417659425,
      "grad_norm": 0.18452684581279755,
      "learning_rate": 0.00018388902176512794,
      "loss": 0.029,
      "step": 8596
    },
    {
      "epoch": 1.2049053959355291,
      "grad_norm": 0.24377408623695374,
      "learning_rate": 0.00018387467113130826,
      "loss": 0.0401,
      "step": 8597
    },
    {
      "epoch": 1.2050455501051156,
      "grad_norm": 0.3590956926345825,
      "learning_rate": 0.00018386032049748862,
      "loss": 0.0608,
      "step": 8598
    },
    {
      "epoch": 1.2051857042747023,
      "grad_norm": 0.44609805941581726,
      "learning_rate": 0.00018384596986366898,
      "loss": 0.0478,
      "step": 8599
    },
    {
      "epoch": 1.2053258584442887,
      "grad_norm": 0.4489005506038666,
      "learning_rate": 0.0001838316192298493,
      "loss": 0.0228,
      "step": 8600
    },
    {
      "epoch": 1.2054660126138752,
      "grad_norm": 0.275067538022995,
      "learning_rate": 0.00018381726859602966,
      "loss": 0.047,
      "step": 8601
    },
    {
      "epoch": 1.2056061667834619,
      "grad_norm": 0.4546056091785431,
      "learning_rate": 0.00018380291796221,
      "loss": 0.0335,
      "step": 8602
    },
    {
      "epoch": 1.2057463209530483,
      "grad_norm": 0.6824021339416504,
      "learning_rate": 0.00018378856732839032,
      "loss": 0.074,
      "step": 8603
    },
    {
      "epoch": 1.2058864751226348,
      "grad_norm": 0.38738110661506653,
      "learning_rate": 0.00018377421669457067,
      "loss": 0.0656,
      "step": 8604
    },
    {
      "epoch": 1.2060266292922215,
      "grad_norm": 0.6033779978752136,
      "learning_rate": 0.000183759866060751,
      "loss": 0.0683,
      "step": 8605
    },
    {
      "epoch": 1.206166783461808,
      "grad_norm": 0.18526332080364227,
      "learning_rate": 0.00018374551542693133,
      "loss": 0.0207,
      "step": 8606
    },
    {
      "epoch": 1.2063069376313946,
      "grad_norm": 0.4811161756515503,
      "learning_rate": 0.00018373116479311169,
      "loss": 0.0477,
      "step": 8607
    },
    {
      "epoch": 1.206447091800981,
      "grad_norm": 0.2796858251094818,
      "learning_rate": 0.00018371681415929201,
      "loss": 0.0683,
      "step": 8608
    },
    {
      "epoch": 1.2065872459705675,
      "grad_norm": 0.5424320101737976,
      "learning_rate": 0.00018370246352547234,
      "loss": 0.0892,
      "step": 8609
    },
    {
      "epoch": 1.2067274001401542,
      "grad_norm": 0.712868332862854,
      "learning_rate": 0.00018368811289165267,
      "loss": 0.1111,
      "step": 8610
    },
    {
      "epoch": 1.2068675543097407,
      "grad_norm": 0.2624228596687317,
      "learning_rate": 0.00018367376225783305,
      "loss": 0.05,
      "step": 8611
    },
    {
      "epoch": 1.2070077084793271,
      "grad_norm": 0.3732443153858185,
      "learning_rate": 0.00018365941162401338,
      "loss": 0.0601,
      "step": 8612
    },
    {
      "epoch": 1.2071478626489138,
      "grad_norm": 0.47797560691833496,
      "learning_rate": 0.0001836450609901937,
      "loss": 0.0772,
      "step": 8613
    },
    {
      "epoch": 1.2072880168185003,
      "grad_norm": 1.25688636302948,
      "learning_rate": 0.00018363071035637407,
      "loss": 0.0382,
      "step": 8614
    },
    {
      "epoch": 1.207428170988087,
      "grad_norm": 0.3314021825790405,
      "learning_rate": 0.0001836163597225544,
      "loss": 0.0912,
      "step": 8615
    },
    {
      "epoch": 1.2075683251576734,
      "grad_norm": 0.6147961616516113,
      "learning_rate": 0.00018360200908873472,
      "loss": 0.2393,
      "step": 8616
    },
    {
      "epoch": 1.2077084793272599,
      "grad_norm": 0.2781028151512146,
      "learning_rate": 0.00018358765845491508,
      "loss": 0.0307,
      "step": 8617
    },
    {
      "epoch": 1.2078486334968466,
      "grad_norm": 0.285336971282959,
      "learning_rate": 0.0001835733078210954,
      "loss": 0.035,
      "step": 8618
    },
    {
      "epoch": 1.207988787666433,
      "grad_norm": 0.33116811513900757,
      "learning_rate": 0.00018355895718727574,
      "loss": 0.0207,
      "step": 8619
    },
    {
      "epoch": 1.2081289418360197,
      "grad_norm": 0.3539915680885315,
      "learning_rate": 0.00018354460655345612,
      "loss": 0.0645,
      "step": 8620
    },
    {
      "epoch": 1.2082690960056062,
      "grad_norm": 0.233585923910141,
      "learning_rate": 0.00018353025591963645,
      "loss": 0.0165,
      "step": 8621
    },
    {
      "epoch": 1.2084092501751926,
      "grad_norm": 0.28798067569732666,
      "learning_rate": 0.00018351590528581678,
      "loss": 0.1027,
      "step": 8622
    },
    {
      "epoch": 1.2085494043447793,
      "grad_norm": 0.41744330525398254,
      "learning_rate": 0.00018350155465199713,
      "loss": 0.0484,
      "step": 8623
    },
    {
      "epoch": 1.2086895585143658,
      "grad_norm": 0.21721336245536804,
      "learning_rate": 0.00018348720401817746,
      "loss": 0.0511,
      "step": 8624
    },
    {
      "epoch": 1.2088297126839525,
      "grad_norm": 0.156241774559021,
      "learning_rate": 0.0001834728533843578,
      "loss": 0.0305,
      "step": 8625
    },
    {
      "epoch": 1.208969866853539,
      "grad_norm": 0.34080907702445984,
      "learning_rate": 0.00018345850275053812,
      "loss": 0.0402,
      "step": 8626
    },
    {
      "epoch": 1.2091100210231254,
      "grad_norm": 0.7670702338218689,
      "learning_rate": 0.00018344415211671847,
      "loss": 0.1638,
      "step": 8627
    },
    {
      "epoch": 1.209250175192712,
      "grad_norm": 1.6789828538894653,
      "learning_rate": 0.0001834298014828988,
      "loss": 0.1111,
      "step": 8628
    },
    {
      "epoch": 1.2093903293622985,
      "grad_norm": 0.5524570941925049,
      "learning_rate": 0.00018341545084907913,
      "loss": 0.0938,
      "step": 8629
    },
    {
      "epoch": 1.2095304835318852,
      "grad_norm": 0.8256645202636719,
      "learning_rate": 0.0001834011002152595,
      "loss": 0.0885,
      "step": 8630
    },
    {
      "epoch": 1.2096706377014717,
      "grad_norm": 1.3773860931396484,
      "learning_rate": 0.00018338674958143984,
      "loss": 0.1457,
      "step": 8631
    },
    {
      "epoch": 1.209810791871058,
      "grad_norm": 1.313921570777893,
      "learning_rate": 0.00018337239894762017,
      "loss": 0.2281,
      "step": 8632
    },
    {
      "epoch": 1.2099509460406448,
      "grad_norm": 1.0833861827850342,
      "learning_rate": 0.00018335804831380053,
      "loss": 0.0567,
      "step": 8633
    },
    {
      "epoch": 1.2100911002102313,
      "grad_norm": 3.6338839530944824,
      "learning_rate": 0.00018334369767998085,
      "loss": 0.12,
      "step": 8634
    },
    {
      "epoch": 1.2102312543798177,
      "grad_norm": 0.7777293920516968,
      "learning_rate": 0.00018332934704616118,
      "loss": 0.1219,
      "step": 8635
    },
    {
      "epoch": 1.2103714085494044,
      "grad_norm": 0.2840065062046051,
      "learning_rate": 0.00018331499641234154,
      "loss": 0.0825,
      "step": 8636
    },
    {
      "epoch": 1.2105115627189909,
      "grad_norm": 0.3750748038291931,
      "learning_rate": 0.00018330064577852187,
      "loss": 0.061,
      "step": 8637
    },
    {
      "epoch": 1.2106517168885773,
      "grad_norm": 0.45787206292152405,
      "learning_rate": 0.0001832862951447022,
      "loss": 0.1067,
      "step": 8638
    },
    {
      "epoch": 1.210791871058164,
      "grad_norm": 0.18101774156093597,
      "learning_rate": 0.00018327194451088255,
      "loss": 0.0281,
      "step": 8639
    },
    {
      "epoch": 1.2109320252277505,
      "grad_norm": 0.45391911268234253,
      "learning_rate": 0.00018325759387706288,
      "loss": 0.0682,
      "step": 8640
    },
    {
      "epoch": 1.2110721793973371,
      "grad_norm": 0.2658536732196808,
      "learning_rate": 0.0001832432432432432,
      "loss": 0.0229,
      "step": 8641
    },
    {
      "epoch": 1.2112123335669236,
      "grad_norm": 0.4262392818927765,
      "learning_rate": 0.0001832288926094236,
      "loss": 0.073,
      "step": 8642
    },
    {
      "epoch": 1.21135248773651,
      "grad_norm": 0.4265669882297516,
      "learning_rate": 0.00018321454197560392,
      "loss": 0.0621,
      "step": 8643
    },
    {
      "epoch": 1.2114926419060967,
      "grad_norm": 0.3030928075313568,
      "learning_rate": 0.00018320019134178425,
      "loss": 0.049,
      "step": 8644
    },
    {
      "epoch": 1.2116327960756832,
      "grad_norm": 0.23425814509391785,
      "learning_rate": 0.00018318584070796458,
      "loss": 0.0282,
      "step": 8645
    },
    {
      "epoch": 1.2117729502452699,
      "grad_norm": 0.42376530170440674,
      "learning_rate": 0.00018317149007414493,
      "loss": 0.0784,
      "step": 8646
    },
    {
      "epoch": 1.2119131044148563,
      "grad_norm": 0.11200292408466339,
      "learning_rate": 0.00018315713944032526,
      "loss": 0.0149,
      "step": 8647
    },
    {
      "epoch": 1.2120532585844428,
      "grad_norm": 0.29357296228408813,
      "learning_rate": 0.0001831427888065056,
      "loss": 0.0387,
      "step": 8648
    },
    {
      "epoch": 1.2121934127540295,
      "grad_norm": 0.3336953818798065,
      "learning_rate": 0.00018312843817268595,
      "loss": 0.1397,
      "step": 8649
    },
    {
      "epoch": 1.212333566923616,
      "grad_norm": 0.3192681670188904,
      "learning_rate": 0.00018311408753886627,
      "loss": 0.0746,
      "step": 8650
    },
    {
      "epoch": 1.2124737210932026,
      "grad_norm": 0.20528243482112885,
      "learning_rate": 0.0001830997369050466,
      "loss": 0.0257,
      "step": 8651
    },
    {
      "epoch": 1.212613875262789,
      "grad_norm": 0.16346094012260437,
      "learning_rate": 0.00018308538627122699,
      "loss": 0.0247,
      "step": 8652
    },
    {
      "epoch": 1.2127540294323755,
      "grad_norm": 0.21894079446792603,
      "learning_rate": 0.00018307103563740731,
      "loss": 0.0609,
      "step": 8653
    },
    {
      "epoch": 1.2128941836019622,
      "grad_norm": 0.26416367292404175,
      "learning_rate": 0.00018305668500358764,
      "loss": 0.0317,
      "step": 8654
    },
    {
      "epoch": 1.2130343377715487,
      "grad_norm": 0.4925996959209442,
      "learning_rate": 0.000183042334369768,
      "loss": 0.071,
      "step": 8655
    },
    {
      "epoch": 1.2131744919411354,
      "grad_norm": 0.20673345029354095,
      "learning_rate": 0.00018302798373594833,
      "loss": 0.0171,
      "step": 8656
    },
    {
      "epoch": 1.2133146461107218,
      "grad_norm": 0.43811532855033875,
      "learning_rate": 0.00018301363310212866,
      "loss": 0.0478,
      "step": 8657
    },
    {
      "epoch": 1.2134548002803083,
      "grad_norm": 0.2935405671596527,
      "learning_rate": 0.000182999282468309,
      "loss": 0.0571,
      "step": 8658
    },
    {
      "epoch": 1.213594954449895,
      "grad_norm": 0.2003026008605957,
      "learning_rate": 0.00018298493183448934,
      "loss": 0.028,
      "step": 8659
    },
    {
      "epoch": 1.2137351086194814,
      "grad_norm": 0.20540444552898407,
      "learning_rate": 0.00018297058120066967,
      "loss": 0.0426,
      "step": 8660
    },
    {
      "epoch": 1.213875262789068,
      "grad_norm": 0.34198060631752014,
      "learning_rate": 0.00018295623056685,
      "loss": 0.0591,
      "step": 8661
    },
    {
      "epoch": 1.2140154169586546,
      "grad_norm": 0.5858361124992371,
      "learning_rate": 0.00018294187993303038,
      "loss": 0.0414,
      "step": 8662
    },
    {
      "epoch": 1.214155571128241,
      "grad_norm": 0.33844229578971863,
      "learning_rate": 0.0001829275292992107,
      "loss": 0.0306,
      "step": 8663
    },
    {
      "epoch": 1.2142957252978277,
      "grad_norm": 0.43569090962409973,
      "learning_rate": 0.00018291317866539104,
      "loss": 0.0271,
      "step": 8664
    },
    {
      "epoch": 1.2144358794674142,
      "grad_norm": 0.36916837096214294,
      "learning_rate": 0.0001828988280315714,
      "loss": 0.0828,
      "step": 8665
    },
    {
      "epoch": 1.2145760336370006,
      "grad_norm": 0.33154740929603577,
      "learning_rate": 0.00018288447739775172,
      "loss": 0.039,
      "step": 8666
    },
    {
      "epoch": 1.2147161878065873,
      "grad_norm": 0.24683552980422974,
      "learning_rate": 0.00018287012676393205,
      "loss": 0.0844,
      "step": 8667
    },
    {
      "epoch": 1.2148563419761738,
      "grad_norm": 0.09857519716024399,
      "learning_rate": 0.0001828557761301124,
      "loss": 0.0132,
      "step": 8668
    },
    {
      "epoch": 1.2149964961457602,
      "grad_norm": 0.6895578503608704,
      "learning_rate": 0.00018284142549629273,
      "loss": 0.0485,
      "step": 8669
    },
    {
      "epoch": 1.215136650315347,
      "grad_norm": 0.36764654517173767,
      "learning_rate": 0.00018282707486247306,
      "loss": 0.0796,
      "step": 8670
    },
    {
      "epoch": 1.2152768044849334,
      "grad_norm": 0.16565701365470886,
      "learning_rate": 0.00018281272422865342,
      "loss": 0.0278,
      "step": 8671
    },
    {
      "epoch": 1.21541695865452,
      "grad_norm": 0.5505766868591309,
      "learning_rate": 0.00018279837359483375,
      "loss": 0.0809,
      "step": 8672
    },
    {
      "epoch": 1.2155571128241065,
      "grad_norm": 0.4263438582420349,
      "learning_rate": 0.00018278402296101408,
      "loss": 0.0585,
      "step": 8673
    },
    {
      "epoch": 1.215697266993693,
      "grad_norm": 0.5488643050193787,
      "learning_rate": 0.00018276967232719446,
      "loss": 0.0342,
      "step": 8674
    },
    {
      "epoch": 1.2158374211632796,
      "grad_norm": 0.38896697759628296,
      "learning_rate": 0.0001827553216933748,
      "loss": 0.0321,
      "step": 8675
    },
    {
      "epoch": 1.215977575332866,
      "grad_norm": 0.17850738763809204,
      "learning_rate": 0.00018274097105955512,
      "loss": 0.016,
      "step": 8676
    },
    {
      "epoch": 1.2161177295024528,
      "grad_norm": 0.3042354881763458,
      "learning_rate": 0.00018272662042573547,
      "loss": 0.0294,
      "step": 8677
    },
    {
      "epoch": 1.2162578836720392,
      "grad_norm": 0.5600208044052124,
      "learning_rate": 0.0001827122697919158,
      "loss": 0.1301,
      "step": 8678
    },
    {
      "epoch": 1.2163980378416257,
      "grad_norm": 0.7561913132667542,
      "learning_rate": 0.00018269791915809613,
      "loss": 0.1005,
      "step": 8679
    },
    {
      "epoch": 1.2165381920112124,
      "grad_norm": 0.34700363874435425,
      "learning_rate": 0.00018268356852427646,
      "loss": 0.0491,
      "step": 8680
    },
    {
      "epoch": 1.2166783461807988,
      "grad_norm": 0.8128876686096191,
      "learning_rate": 0.0001826692178904568,
      "loss": 0.1313,
      "step": 8681
    },
    {
      "epoch": 1.2168185003503855,
      "grad_norm": 0.40929701924324036,
      "learning_rate": 0.00018265486725663714,
      "loss": 0.0911,
      "step": 8682
    },
    {
      "epoch": 1.216958654519972,
      "grad_norm": 1.4489890336990356,
      "learning_rate": 0.00018264051662281747,
      "loss": 0.2669,
      "step": 8683
    },
    {
      "epoch": 1.2170988086895584,
      "grad_norm": 0.5744010210037231,
      "learning_rate": 0.00018262616598899785,
      "loss": 0.0576,
      "step": 8684
    },
    {
      "epoch": 1.2172389628591451,
      "grad_norm": 1.720779538154602,
      "learning_rate": 0.00018261181535517818,
      "loss": 0.2815,
      "step": 8685
    },
    {
      "epoch": 1.2173791170287316,
      "grad_norm": 0.3878129720687866,
      "learning_rate": 0.0001825974647213585,
      "loss": 0.0638,
      "step": 8686
    },
    {
      "epoch": 1.2175192711983183,
      "grad_norm": 0.344102144241333,
      "learning_rate": 0.00018258311408753886,
      "loss": 0.065,
      "step": 8687
    },
    {
      "epoch": 1.2176594253679047,
      "grad_norm": 0.2713230550289154,
      "learning_rate": 0.0001825687634537192,
      "loss": 0.0451,
      "step": 8688
    },
    {
      "epoch": 1.2177995795374912,
      "grad_norm": 0.4024549424648285,
      "learning_rate": 0.00018255441281989952,
      "loss": 0.0866,
      "step": 8689
    },
    {
      "epoch": 1.2179397337070779,
      "grad_norm": 0.32044848799705505,
      "learning_rate": 0.00018254006218607988,
      "loss": 0.0768,
      "step": 8690
    },
    {
      "epoch": 1.2180798878766643,
      "grad_norm": 0.2760136127471924,
      "learning_rate": 0.0001825257115522602,
      "loss": 0.0335,
      "step": 8691
    },
    {
      "epoch": 1.2182200420462508,
      "grad_norm": 0.20203445851802826,
      "learning_rate": 0.00018251136091844053,
      "loss": 0.0445,
      "step": 8692
    },
    {
      "epoch": 1.2183601962158375,
      "grad_norm": 0.2641170024871826,
      "learning_rate": 0.0001824970102846209,
      "loss": 0.0391,
      "step": 8693
    },
    {
      "epoch": 1.218500350385424,
      "grad_norm": 0.40196365118026733,
      "learning_rate": 0.00018248265965080125,
      "loss": 0.0575,
      "step": 8694
    },
    {
      "epoch": 1.2186405045550106,
      "grad_norm": 0.29076626896858215,
      "learning_rate": 0.00018246830901698157,
      "loss": 0.0579,
      "step": 8695
    },
    {
      "epoch": 1.218780658724597,
      "grad_norm": 0.10989604145288467,
      "learning_rate": 0.0001824539583831619,
      "loss": 0.012,
      "step": 8696
    },
    {
      "epoch": 1.2189208128941835,
      "grad_norm": 0.2224172055721283,
      "learning_rate": 0.00018243960774934226,
      "loss": 0.0659,
      "step": 8697
    },
    {
      "epoch": 1.2190609670637702,
      "grad_norm": 0.2968049943447113,
      "learning_rate": 0.0001824252571155226,
      "loss": 0.0881,
      "step": 8698
    },
    {
      "epoch": 1.2192011212333567,
      "grad_norm": 0.27378183603286743,
      "learning_rate": 0.00018241090648170292,
      "loss": 0.0732,
      "step": 8699
    },
    {
      "epoch": 1.2193412754029431,
      "grad_norm": 0.2966006398200989,
      "learning_rate": 0.00018239655584788327,
      "loss": 0.0758,
      "step": 8700
    },
    {
      "epoch": 1.2194814295725298,
      "grad_norm": 0.27214285731315613,
      "learning_rate": 0.0001823822052140636,
      "loss": 0.0476,
      "step": 8701
    },
    {
      "epoch": 1.2196215837421163,
      "grad_norm": 0.4161134362220764,
      "learning_rate": 0.00018236785458024393,
      "loss": 0.0863,
      "step": 8702
    },
    {
      "epoch": 1.219761737911703,
      "grad_norm": 0.2782464921474457,
      "learning_rate": 0.00018235350394642428,
      "loss": 0.0292,
      "step": 8703
    },
    {
      "epoch": 1.2199018920812894,
      "grad_norm": 0.2341078370809555,
      "learning_rate": 0.0001823391533126046,
      "loss": 0.0389,
      "step": 8704
    },
    {
      "epoch": 1.2200420462508759,
      "grad_norm": 0.2888844311237335,
      "learning_rate": 0.00018232480267878494,
      "loss": 0.092,
      "step": 8705
    },
    {
      "epoch": 1.2201822004204625,
      "grad_norm": 0.3119038939476013,
      "learning_rate": 0.00018231045204496532,
      "loss": 0.0402,
      "step": 8706
    },
    {
      "epoch": 1.220322354590049,
      "grad_norm": 0.2826389670372009,
      "learning_rate": 0.00018229610141114565,
      "loss": 0.0375,
      "step": 8707
    },
    {
      "epoch": 1.2204625087596357,
      "grad_norm": 0.5185393691062927,
      "learning_rate": 0.00018228175077732598,
      "loss": 0.0844,
      "step": 8708
    },
    {
      "epoch": 1.2206026629292221,
      "grad_norm": 0.3574616014957428,
      "learning_rate": 0.00018226740014350634,
      "loss": 0.0656,
      "step": 8709
    },
    {
      "epoch": 1.2207428170988086,
      "grad_norm": 0.6013672947883606,
      "learning_rate": 0.00018225304950968667,
      "loss": 0.0359,
      "step": 8710
    },
    {
      "epoch": 1.2208829712683953,
      "grad_norm": 0.4368615746498108,
      "learning_rate": 0.000182238698875867,
      "loss": 0.0323,
      "step": 8711
    },
    {
      "epoch": 1.2210231254379817,
      "grad_norm": 1.1246472597122192,
      "learning_rate": 0.00018222434824204735,
      "loss": 0.0858,
      "step": 8712
    },
    {
      "epoch": 1.2211632796075684,
      "grad_norm": 0.1949842870235443,
      "learning_rate": 0.00018220999760822768,
      "loss": 0.0154,
      "step": 8713
    },
    {
      "epoch": 1.221303433777155,
      "grad_norm": 0.14524754881858826,
      "learning_rate": 0.000182195646974408,
      "loss": 0.0326,
      "step": 8714
    },
    {
      "epoch": 1.2214435879467413,
      "grad_norm": 0.22695517539978027,
      "learning_rate": 0.00018218129634058834,
      "loss": 0.0218,
      "step": 8715
    },
    {
      "epoch": 1.221583742116328,
      "grad_norm": 0.09179423749446869,
      "learning_rate": 0.00018216694570676872,
      "loss": 0.0189,
      "step": 8716
    },
    {
      "epoch": 1.2217238962859145,
      "grad_norm": 0.2989479601383209,
      "learning_rate": 0.00018215259507294905,
      "loss": 0.057,
      "step": 8717
    },
    {
      "epoch": 1.2218640504555012,
      "grad_norm": 0.3001638650894165,
      "learning_rate": 0.00018213824443912938,
      "loss": 0.0588,
      "step": 8718
    },
    {
      "epoch": 1.2220042046250876,
      "grad_norm": 0.3052922487258911,
      "learning_rate": 0.00018212389380530973,
      "loss": 0.0769,
      "step": 8719
    },
    {
      "epoch": 1.222144358794674,
      "grad_norm": 0.30948060750961304,
      "learning_rate": 0.00018210954317149006,
      "loss": 0.0604,
      "step": 8720
    },
    {
      "epoch": 1.2222845129642608,
      "grad_norm": 0.3422660827636719,
      "learning_rate": 0.0001820951925376704,
      "loss": 0.0605,
      "step": 8721
    },
    {
      "epoch": 1.2224246671338472,
      "grad_norm": 0.43985825777053833,
      "learning_rate": 0.00018208084190385074,
      "loss": 0.0545,
      "step": 8722
    },
    {
      "epoch": 1.2225648213034337,
      "grad_norm": 0.1907966583967209,
      "learning_rate": 0.00018206649127003107,
      "loss": 0.045,
      "step": 8723
    },
    {
      "epoch": 1.2227049754730204,
      "grad_norm": 0.27309638261795044,
      "learning_rate": 0.0001820521406362114,
      "loss": 0.0443,
      "step": 8724
    },
    {
      "epoch": 1.2228451296426068,
      "grad_norm": 0.3246774673461914,
      "learning_rate": 0.00018203779000239176,
      "loss": 0.0429,
      "step": 8725
    },
    {
      "epoch": 1.2229852838121933,
      "grad_norm": 0.0974561870098114,
      "learning_rate": 0.0001820234393685721,
      "loss": 0.015,
      "step": 8726
    },
    {
      "epoch": 1.22312543798178,
      "grad_norm": 0.40235310792922974,
      "learning_rate": 0.00018200908873475244,
      "loss": 0.0406,
      "step": 8727
    },
    {
      "epoch": 1.2232655921513664,
      "grad_norm": 0.6393806338310242,
      "learning_rate": 0.0001819947381009328,
      "loss": 0.2022,
      "step": 8728
    },
    {
      "epoch": 1.2234057463209531,
      "grad_norm": 1.2388378381729126,
      "learning_rate": 0.00018198038746711313,
      "loss": 0.1056,
      "step": 8729
    },
    {
      "epoch": 1.2235459004905396,
      "grad_norm": 0.5445995330810547,
      "learning_rate": 0.00018196603683329345,
      "loss": 0.0872,
      "step": 8730
    },
    {
      "epoch": 1.223686054660126,
      "grad_norm": 0.8528308272361755,
      "learning_rate": 0.00018195168619947378,
      "loss": 0.0245,
      "step": 8731
    },
    {
      "epoch": 1.2238262088297127,
      "grad_norm": 1.624738335609436,
      "learning_rate": 0.00018193733556565414,
      "loss": 0.1435,
      "step": 8732
    },
    {
      "epoch": 1.2239663629992992,
      "grad_norm": 1.2054100036621094,
      "learning_rate": 0.00018192298493183447,
      "loss": 0.0566,
      "step": 8733
    },
    {
      "epoch": 1.2241065171688859,
      "grad_norm": 0.6819858551025391,
      "learning_rate": 0.0001819086342980148,
      "loss": 0.0381,
      "step": 8734
    },
    {
      "epoch": 1.2242466713384723,
      "grad_norm": 0.5789105296134949,
      "learning_rate": 0.00018189428366419515,
      "loss": 0.0779,
      "step": 8735
    },
    {
      "epoch": 1.2243868255080588,
      "grad_norm": 0.24732036888599396,
      "learning_rate": 0.00018187993303037548,
      "loss": 0.0607,
      "step": 8736
    },
    {
      "epoch": 1.2245269796776455,
      "grad_norm": 0.421773761510849,
      "learning_rate": 0.0001818655823965558,
      "loss": 0.0439,
      "step": 8737
    },
    {
      "epoch": 1.224667133847232,
      "grad_norm": 0.24831874668598175,
      "learning_rate": 0.0001818512317627362,
      "loss": 0.0493,
      "step": 8738
    },
    {
      "epoch": 1.2248072880168186,
      "grad_norm": 0.42275428771972656,
      "learning_rate": 0.00018183688112891652,
      "loss": 0.0665,
      "step": 8739
    },
    {
      "epoch": 1.224947442186405,
      "grad_norm": 0.2681684195995331,
      "learning_rate": 0.00018182253049509685,
      "loss": 0.0267,
      "step": 8740
    },
    {
      "epoch": 1.2250875963559915,
      "grad_norm": 0.30987536907196045,
      "learning_rate": 0.0001818081798612772,
      "loss": 0.0793,
      "step": 8741
    },
    {
      "epoch": 1.2252277505255782,
      "grad_norm": 0.319633811712265,
      "learning_rate": 0.00018179382922745753,
      "loss": 0.076,
      "step": 8742
    },
    {
      "epoch": 1.2253679046951647,
      "grad_norm": 0.8478633165359497,
      "learning_rate": 0.00018177947859363786,
      "loss": 0.1082,
      "step": 8743
    },
    {
      "epoch": 1.2255080588647513,
      "grad_norm": 0.3469730019569397,
      "learning_rate": 0.00018176512795981822,
      "loss": 0.0437,
      "step": 8744
    },
    {
      "epoch": 1.2256482130343378,
      "grad_norm": 0.3907775282859802,
      "learning_rate": 0.00018175077732599854,
      "loss": 0.0584,
      "step": 8745
    },
    {
      "epoch": 1.2257883672039243,
      "grad_norm": 0.47700566053390503,
      "learning_rate": 0.00018173642669217887,
      "loss": 0.1019,
      "step": 8746
    },
    {
      "epoch": 1.225928521373511,
      "grad_norm": 0.7770333886146545,
      "learning_rate": 0.0001817220760583592,
      "loss": 0.0189,
      "step": 8747
    },
    {
      "epoch": 1.2260686755430974,
      "grad_norm": 0.42544013261795044,
      "learning_rate": 0.00018170772542453958,
      "loss": 0.0513,
      "step": 8748
    },
    {
      "epoch": 1.2262088297126839,
      "grad_norm": 0.11113908886909485,
      "learning_rate": 0.0001816933747907199,
      "loss": 0.0144,
      "step": 8749
    },
    {
      "epoch": 1.2263489838822705,
      "grad_norm": 0.2252221405506134,
      "learning_rate": 0.00018167902415690024,
      "loss": 0.0115,
      "step": 8750
    },
    {
      "epoch": 1.226489138051857,
      "grad_norm": 0.27279195189476013,
      "learning_rate": 0.0001816646735230806,
      "loss": 0.0405,
      "step": 8751
    },
    {
      "epoch": 1.2266292922214437,
      "grad_norm": 0.16972167789936066,
      "learning_rate": 0.00018165032288926093,
      "loss": 0.0201,
      "step": 8752
    },
    {
      "epoch": 1.2267694463910301,
      "grad_norm": 0.15438127517700195,
      "learning_rate": 0.00018163597225544125,
      "loss": 0.0072,
      "step": 8753
    },
    {
      "epoch": 1.2269096005606166,
      "grad_norm": 0.26638391613960266,
      "learning_rate": 0.0001816216216216216,
      "loss": 0.0326,
      "step": 8754
    },
    {
      "epoch": 1.2270497547302033,
      "grad_norm": 0.47062039375305176,
      "learning_rate": 0.00018160727098780194,
      "loss": 0.1566,
      "step": 8755
    },
    {
      "epoch": 1.2271899088997897,
      "grad_norm": 0.25272372364997864,
      "learning_rate": 0.00018159292035398227,
      "loss": 0.0971,
      "step": 8756
    },
    {
      "epoch": 1.2273300630693762,
      "grad_norm": 0.1324554979801178,
      "learning_rate": 0.00018157856972016265,
      "loss": 0.0201,
      "step": 8757
    },
    {
      "epoch": 1.2274702172389629,
      "grad_norm": 0.27917635440826416,
      "learning_rate": 0.00018156421908634298,
      "loss": 0.0989,
      "step": 8758
    },
    {
      "epoch": 1.2276103714085493,
      "grad_norm": 0.17215456068515778,
      "learning_rate": 0.0001815498684525233,
      "loss": 0.0357,
      "step": 8759
    },
    {
      "epoch": 1.227750525578136,
      "grad_norm": 0.350581556558609,
      "learning_rate": 0.00018153551781870366,
      "loss": 0.063,
      "step": 8760
    },
    {
      "epoch": 1.2278906797477225,
      "grad_norm": 0.293480783700943,
      "learning_rate": 0.000181521167184884,
      "loss": 0.0486,
      "step": 8761
    },
    {
      "epoch": 1.228030833917309,
      "grad_norm": 0.4810590147972107,
      "learning_rate": 0.00018150681655106432,
      "loss": 0.0756,
      "step": 8762
    },
    {
      "epoch": 1.2281709880868956,
      "grad_norm": 0.8719367384910583,
      "learning_rate": 0.00018149246591724468,
      "loss": 0.0653,
      "step": 8763
    },
    {
      "epoch": 1.228311142256482,
      "grad_norm": 0.15654395520687103,
      "learning_rate": 0.000181478115283425,
      "loss": 0.0231,
      "step": 8764
    },
    {
      "epoch": 1.2284512964260688,
      "grad_norm": 0.20144544541835785,
      "learning_rate": 0.00018146376464960533,
      "loss": 0.0468,
      "step": 8765
    },
    {
      "epoch": 1.2285914505956552,
      "grad_norm": 0.08369973301887512,
      "learning_rate": 0.00018144941401578566,
      "loss": 0.0218,
      "step": 8766
    },
    {
      "epoch": 1.2287316047652417,
      "grad_norm": 0.21827556192874908,
      "learning_rate": 0.00018143506338196602,
      "loss": 0.0369,
      "step": 8767
    },
    {
      "epoch": 1.2288717589348284,
      "grad_norm": 0.1105455607175827,
      "learning_rate": 0.00018142071274814635,
      "loss": 0.0202,
      "step": 8768
    },
    {
      "epoch": 1.2290119131044148,
      "grad_norm": 0.20448343455791473,
      "learning_rate": 0.00018140636211432667,
      "loss": 0.0279,
      "step": 8769
    },
    {
      "epoch": 1.2291520672740015,
      "grad_norm": 0.41001424193382263,
      "learning_rate": 0.00018139201148050706,
      "loss": 0.0954,
      "step": 8770
    },
    {
      "epoch": 1.229292221443588,
      "grad_norm": 0.4907464385032654,
      "learning_rate": 0.00018137766084668739,
      "loss": 0.0768,
      "step": 8771
    },
    {
      "epoch": 1.2294323756131744,
      "grad_norm": 0.5235586166381836,
      "learning_rate": 0.00018136331021286771,
      "loss": 0.0687,
      "step": 8772
    },
    {
      "epoch": 1.229572529782761,
      "grad_norm": 0.6149177551269531,
      "learning_rate": 0.00018134895957904807,
      "loss": 0.0506,
      "step": 8773
    },
    {
      "epoch": 1.2297126839523476,
      "grad_norm": 0.944739043712616,
      "learning_rate": 0.0001813346089452284,
      "loss": 0.0479,
      "step": 8774
    },
    {
      "epoch": 1.2298528381219342,
      "grad_norm": 0.18646426498889923,
      "learning_rate": 0.00018132025831140873,
      "loss": 0.0195,
      "step": 8775
    },
    {
      "epoch": 1.2299929922915207,
      "grad_norm": 0.4853493571281433,
      "learning_rate": 0.00018130590767758908,
      "loss": 0.0403,
      "step": 8776
    },
    {
      "epoch": 1.2301331464611072,
      "grad_norm": 0.43826955556869507,
      "learning_rate": 0.0001812915570437694,
      "loss": 0.0586,
      "step": 8777
    },
    {
      "epoch": 1.2302733006306938,
      "grad_norm": 0.39421191811561584,
      "learning_rate": 0.00018127720640994974,
      "loss": 0.0364,
      "step": 8778
    },
    {
      "epoch": 1.2304134548002803,
      "grad_norm": 0.3942345678806305,
      "learning_rate": 0.00018126285577613012,
      "loss": 0.0823,
      "step": 8779
    },
    {
      "epoch": 1.2305536089698668,
      "grad_norm": 0.463264524936676,
      "learning_rate": 0.00018124850514231045,
      "loss": 0.0513,
      "step": 8780
    },
    {
      "epoch": 1.2306937631394534,
      "grad_norm": 0.5555284023284912,
      "learning_rate": 0.00018123415450849078,
      "loss": 0.0493,
      "step": 8781
    },
    {
      "epoch": 1.23083391730904,
      "grad_norm": 0.30923137068748474,
      "learning_rate": 0.0001812198038746711,
      "loss": 0.0875,
      "step": 8782
    },
    {
      "epoch": 1.2309740714786266,
      "grad_norm": 0.3791423439979553,
      "learning_rate": 0.00018120545324085146,
      "loss": 0.0399,
      "step": 8783
    },
    {
      "epoch": 1.231114225648213,
      "grad_norm": 0.5037143230438232,
      "learning_rate": 0.0001811911026070318,
      "loss": 0.0781,
      "step": 8784
    },
    {
      "epoch": 1.2312543798177995,
      "grad_norm": 0.27099910378456116,
      "learning_rate": 0.00018117675197321212,
      "loss": 0.0147,
      "step": 8785
    },
    {
      "epoch": 1.2313945339873862,
      "grad_norm": 0.3733114004135132,
      "learning_rate": 0.00018116240133939248,
      "loss": 0.0733,
      "step": 8786
    },
    {
      "epoch": 1.2315346881569726,
      "grad_norm": 0.43500620126724243,
      "learning_rate": 0.0001811480507055728,
      "loss": 0.0609,
      "step": 8787
    },
    {
      "epoch": 1.231674842326559,
      "grad_norm": 0.4153698980808258,
      "learning_rate": 0.00018113370007175313,
      "loss": 0.0469,
      "step": 8788
    },
    {
      "epoch": 1.2318149964961458,
      "grad_norm": 0.14749528467655182,
      "learning_rate": 0.00018111934943793352,
      "loss": 0.0226,
      "step": 8789
    },
    {
      "epoch": 1.2319551506657322,
      "grad_norm": 0.18205280601978302,
      "learning_rate": 0.00018110499880411384,
      "loss": 0.0201,
      "step": 8790
    },
    {
      "epoch": 1.232095304835319,
      "grad_norm": 0.21400584280490875,
      "learning_rate": 0.00018109064817029417,
      "loss": 0.0303,
      "step": 8791
    },
    {
      "epoch": 1.2322354590049054,
      "grad_norm": 0.31804898381233215,
      "learning_rate": 0.00018107629753647453,
      "loss": 0.042,
      "step": 8792
    },
    {
      "epoch": 1.2323756131744918,
      "grad_norm": 0.2616739273071289,
      "learning_rate": 0.00018106194690265486,
      "loss": 0.0702,
      "step": 8793
    },
    {
      "epoch": 1.2325157673440785,
      "grad_norm": 0.07070483267307281,
      "learning_rate": 0.00018104759626883519,
      "loss": 0.0205,
      "step": 8794
    },
    {
      "epoch": 1.232655921513665,
      "grad_norm": 0.3609790503978729,
      "learning_rate": 0.00018103324563501554,
      "loss": 0.0565,
      "step": 8795
    },
    {
      "epoch": 1.2327960756832517,
      "grad_norm": 0.3016926646232605,
      "learning_rate": 0.00018101889500119587,
      "loss": 0.0397,
      "step": 8796
    },
    {
      "epoch": 1.2329362298528381,
      "grad_norm": 0.5391256213188171,
      "learning_rate": 0.0001810045443673762,
      "loss": 0.1211,
      "step": 8797
    },
    {
      "epoch": 1.2330763840224246,
      "grad_norm": 0.13770680129528046,
      "learning_rate": 0.00018099019373355655,
      "loss": 0.0268,
      "step": 8798
    },
    {
      "epoch": 1.2332165381920113,
      "grad_norm": 0.8050645589828491,
      "learning_rate": 0.00018097584309973688,
      "loss": 0.0787,
      "step": 8799
    },
    {
      "epoch": 1.2333566923615977,
      "grad_norm": 0.5566849112510681,
      "learning_rate": 0.0001809614924659172,
      "loss": 0.0691,
      "step": 8800
    },
    {
      "epoch": 1.2334968465311844,
      "grad_norm": 0.22305220365524292,
      "learning_rate": 0.00018094714183209754,
      "loss": 0.0175,
      "step": 8801
    },
    {
      "epoch": 1.2336370007007709,
      "grad_norm": 0.5556771755218506,
      "learning_rate": 0.00018093279119827792,
      "loss": 0.0589,
      "step": 8802
    },
    {
      "epoch": 1.2337771548703573,
      "grad_norm": 0.24585343897342682,
      "learning_rate": 0.00018091844056445825,
      "loss": 0.0519,
      "step": 8803
    },
    {
      "epoch": 1.233917309039944,
      "grad_norm": 0.1927025318145752,
      "learning_rate": 0.00018090408993063858,
      "loss": 0.018,
      "step": 8804
    },
    {
      "epoch": 1.2340574632095305,
      "grad_norm": 0.18265752494335175,
      "learning_rate": 0.00018088973929681894,
      "loss": 0.0278,
      "step": 8805
    },
    {
      "epoch": 1.2341976173791172,
      "grad_norm": 0.2701992094516754,
      "learning_rate": 0.00018087538866299926,
      "loss": 0.0372,
      "step": 8806
    },
    {
      "epoch": 1.2343377715487036,
      "grad_norm": 0.2678571343421936,
      "learning_rate": 0.0001808610380291796,
      "loss": 0.0504,
      "step": 8807
    },
    {
      "epoch": 1.23447792571829,
      "grad_norm": 0.4660065472126007,
      "learning_rate": 0.00018084668739535995,
      "loss": 0.0378,
      "step": 8808
    },
    {
      "epoch": 1.2346180798878768,
      "grad_norm": 0.39514169096946716,
      "learning_rate": 0.00018083233676154028,
      "loss": 0.106,
      "step": 8809
    },
    {
      "epoch": 1.2347582340574632,
      "grad_norm": 0.8927792310714722,
      "learning_rate": 0.0001808179861277206,
      "loss": 0.0298,
      "step": 8810
    },
    {
      "epoch": 1.2348983882270497,
      "grad_norm": 0.27887874841690063,
      "learning_rate": 0.000180803635493901,
      "loss": 0.0624,
      "step": 8811
    },
    {
      "epoch": 1.2350385423966364,
      "grad_norm": 0.1413101702928543,
      "learning_rate": 0.00018078928486008132,
      "loss": 0.021,
      "step": 8812
    },
    {
      "epoch": 1.2351786965662228,
      "grad_norm": 0.2123580425977707,
      "learning_rate": 0.00018077493422626165,
      "loss": 0.0188,
      "step": 8813
    },
    {
      "epoch": 1.2353188507358093,
      "grad_norm": 1.3969950675964355,
      "learning_rate": 0.000180760583592442,
      "loss": 0.0644,
      "step": 8814
    },
    {
      "epoch": 1.235459004905396,
      "grad_norm": 0.17703315615653992,
      "learning_rate": 0.00018074623295862233,
      "loss": 0.0335,
      "step": 8815
    },
    {
      "epoch": 1.2355991590749824,
      "grad_norm": 0.3511470556259155,
      "learning_rate": 0.00018073188232480266,
      "loss": 0.082,
      "step": 8816
    },
    {
      "epoch": 1.235739313244569,
      "grad_norm": 0.22825947403907776,
      "learning_rate": 0.000180717531690983,
      "loss": 0.0512,
      "step": 8817
    },
    {
      "epoch": 1.2358794674141556,
      "grad_norm": 0.324421763420105,
      "learning_rate": 0.00018070318105716334,
      "loss": 0.061,
      "step": 8818
    },
    {
      "epoch": 1.236019621583742,
      "grad_norm": 0.17748211324214935,
      "learning_rate": 0.00018068883042334367,
      "loss": 0.0192,
      "step": 8819
    },
    {
      "epoch": 1.2361597757533287,
      "grad_norm": 0.25499194860458374,
      "learning_rate": 0.000180674479789524,
      "loss": 0.0622,
      "step": 8820
    },
    {
      "epoch": 1.2362999299229152,
      "grad_norm": 0.48264193534851074,
      "learning_rate": 0.00018066012915570438,
      "loss": 0.098,
      "step": 8821
    },
    {
      "epoch": 1.2364400840925018,
      "grad_norm": 0.7179067134857178,
      "learning_rate": 0.0001806457785218847,
      "loss": 0.1388,
      "step": 8822
    },
    {
      "epoch": 1.2365802382620883,
      "grad_norm": 0.3876977264881134,
      "learning_rate": 0.00018063142788806504,
      "loss": 0.1089,
      "step": 8823
    },
    {
      "epoch": 1.2367203924316748,
      "grad_norm": 0.07642313838005066,
      "learning_rate": 0.0001806170772542454,
      "loss": 0.0135,
      "step": 8824
    },
    {
      "epoch": 1.2368605466012614,
      "grad_norm": 0.18005669116973877,
      "learning_rate": 0.00018060272662042572,
      "loss": 0.0513,
      "step": 8825
    },
    {
      "epoch": 1.237000700770848,
      "grad_norm": 0.1359214037656784,
      "learning_rate": 0.00018058837598660605,
      "loss": 0.0118,
      "step": 8826
    },
    {
      "epoch": 1.2371408549404346,
      "grad_norm": 0.2316037267446518,
      "learning_rate": 0.0001805740253527864,
      "loss": 0.034,
      "step": 8827
    },
    {
      "epoch": 1.237281009110021,
      "grad_norm": 0.27992984652519226,
      "learning_rate": 0.00018055967471896674,
      "loss": 0.0773,
      "step": 8828
    },
    {
      "epoch": 1.2374211632796075,
      "grad_norm": 0.6376937031745911,
      "learning_rate": 0.00018054532408514707,
      "loss": 0.0573,
      "step": 8829
    },
    {
      "epoch": 1.2375613174491942,
      "grad_norm": 0.2907851040363312,
      "learning_rate": 0.00018053097345132742,
      "loss": 0.028,
      "step": 8830
    },
    {
      "epoch": 1.2377014716187806,
      "grad_norm": 1.39239501953125,
      "learning_rate": 0.00018051662281750775,
      "loss": 0.1777,
      "step": 8831
    },
    {
      "epoch": 1.2378416257883673,
      "grad_norm": 0.2886229455471039,
      "learning_rate": 0.00018050227218368808,
      "loss": 0.0526,
      "step": 8832
    },
    {
      "epoch": 1.2379817799579538,
      "grad_norm": 0.3110648989677429,
      "learning_rate": 0.00018048792154986846,
      "loss": 0.0639,
      "step": 8833
    },
    {
      "epoch": 1.2381219341275402,
      "grad_norm": 0.8068387508392334,
      "learning_rate": 0.0001804735709160488,
      "loss": 0.0846,
      "step": 8834
    },
    {
      "epoch": 1.238262088297127,
      "grad_norm": 0.8859265446662903,
      "learning_rate": 0.00018045922028222912,
      "loss": 0.0605,
      "step": 8835
    },
    {
      "epoch": 1.2384022424667134,
      "grad_norm": 0.15297624468803406,
      "learning_rate": 0.00018044486964840945,
      "loss": 0.0231,
      "step": 8836
    },
    {
      "epoch": 1.2385423966362998,
      "grad_norm": 1.0882353782653809,
      "learning_rate": 0.0001804305190145898,
      "loss": 0.0669,
      "step": 8837
    },
    {
      "epoch": 1.2386825508058865,
      "grad_norm": 0.2530595660209656,
      "learning_rate": 0.00018041616838077013,
      "loss": 0.0414,
      "step": 8838
    },
    {
      "epoch": 1.238822704975473,
      "grad_norm": 0.267008513212204,
      "learning_rate": 0.00018040181774695046,
      "loss": 0.0713,
      "step": 8839
    },
    {
      "epoch": 1.2389628591450597,
      "grad_norm": 0.2525866627693176,
      "learning_rate": 0.00018038746711313081,
      "loss": 0.0406,
      "step": 8840
    },
    {
      "epoch": 1.2391030133146461,
      "grad_norm": 0.20825164020061493,
      "learning_rate": 0.00018037311647931114,
      "loss": 0.0176,
      "step": 8841
    },
    {
      "epoch": 1.2392431674842326,
      "grad_norm": 1.2770270109176636,
      "learning_rate": 0.00018035876584549147,
      "loss": 0.147,
      "step": 8842
    },
    {
      "epoch": 1.2393833216538193,
      "grad_norm": 0.5578776597976685,
      "learning_rate": 0.00018034441521167185,
      "loss": 0.063,
      "step": 8843
    },
    {
      "epoch": 1.2395234758234057,
      "grad_norm": 0.382220596075058,
      "learning_rate": 0.00018033006457785218,
      "loss": 0.0606,
      "step": 8844
    },
    {
      "epoch": 1.2396636299929922,
      "grad_norm": 0.39881905913352966,
      "learning_rate": 0.0001803157139440325,
      "loss": 0.0883,
      "step": 8845
    },
    {
      "epoch": 1.2398037841625789,
      "grad_norm": 0.484540194272995,
      "learning_rate": 0.00018030136331021287,
      "loss": 0.0612,
      "step": 8846
    },
    {
      "epoch": 1.2399439383321653,
      "grad_norm": 0.20494407415390015,
      "learning_rate": 0.0001802870126763932,
      "loss": 0.05,
      "step": 8847
    },
    {
      "epoch": 1.240084092501752,
      "grad_norm": 0.36396971344947815,
      "learning_rate": 0.00018027266204257352,
      "loss": 0.0245,
      "step": 8848
    },
    {
      "epoch": 1.2402242466713385,
      "grad_norm": 0.2862990200519562,
      "learning_rate": 0.00018025831140875388,
      "loss": 0.0446,
      "step": 8849
    },
    {
      "epoch": 1.240364400840925,
      "grad_norm": 0.25687506794929504,
      "learning_rate": 0.0001802439607749342,
      "loss": 0.0289,
      "step": 8850
    },
    {
      "epoch": 1.2405045550105116,
      "grad_norm": 1.0851212739944458,
      "learning_rate": 0.00018022961014111454,
      "loss": 0.0603,
      "step": 8851
    },
    {
      "epoch": 1.240644709180098,
      "grad_norm": 0.35455313324928284,
      "learning_rate": 0.00018021525950729487,
      "loss": 0.0599,
      "step": 8852
    },
    {
      "epoch": 1.2407848633496847,
      "grad_norm": 0.3976340889930725,
      "learning_rate": 0.00018020090887347525,
      "loss": 0.0301,
      "step": 8853
    },
    {
      "epoch": 1.2409250175192712,
      "grad_norm": 0.2038656622171402,
      "learning_rate": 0.00018018655823965558,
      "loss": 0.0315,
      "step": 8854
    },
    {
      "epoch": 1.2410651716888577,
      "grad_norm": 0.4963422417640686,
      "learning_rate": 0.0001801722076058359,
      "loss": 0.0247,
      "step": 8855
    },
    {
      "epoch": 1.2412053258584443,
      "grad_norm": 0.5993987321853638,
      "learning_rate": 0.00018015785697201626,
      "loss": 0.1168,
      "step": 8856
    },
    {
      "epoch": 1.2413454800280308,
      "grad_norm": 0.5678585171699524,
      "learning_rate": 0.0001801435063381966,
      "loss": 0.0911,
      "step": 8857
    },
    {
      "epoch": 1.2414856341976175,
      "grad_norm": 0.526900053024292,
      "learning_rate": 0.00018012915570437692,
      "loss": 0.0335,
      "step": 8858
    },
    {
      "epoch": 1.241625788367204,
      "grad_norm": 0.11840403079986572,
      "learning_rate": 0.00018011480507055727,
      "loss": 0.0114,
      "step": 8859
    },
    {
      "epoch": 1.2417659425367904,
      "grad_norm": 0.8445833921432495,
      "learning_rate": 0.0001801004544367376,
      "loss": 0.0814,
      "step": 8860
    },
    {
      "epoch": 1.241906096706377,
      "grad_norm": 0.42567649483680725,
      "learning_rate": 0.00018008610380291793,
      "loss": 0.0342,
      "step": 8861
    },
    {
      "epoch": 1.2420462508759635,
      "grad_norm": 0.26783329248428345,
      "learning_rate": 0.0001800717531690983,
      "loss": 0.0235,
      "step": 8862
    },
    {
      "epoch": 1.2421864050455502,
      "grad_norm": 0.21952690184116364,
      "learning_rate": 0.00018005740253527862,
      "loss": 0.0294,
      "step": 8863
    },
    {
      "epoch": 1.2423265592151367,
      "grad_norm": 0.4130716919898987,
      "learning_rate": 0.00018004305190145894,
      "loss": 0.0625,
      "step": 8864
    },
    {
      "epoch": 1.2424667133847231,
      "grad_norm": 0.6658200025558472,
      "learning_rate": 0.00018002870126763933,
      "loss": 0.0734,
      "step": 8865
    },
    {
      "epoch": 1.2426068675543098,
      "grad_norm": 0.45264697074890137,
      "learning_rate": 0.00018001435063381966,
      "loss": 0.074,
      "step": 8866
    },
    {
      "epoch": 1.2427470217238963,
      "grad_norm": 0.39087289571762085,
      "learning_rate": 0.00017999999999999998,
      "loss": 0.0637,
      "step": 8867
    },
    {
      "epoch": 1.2428871758934827,
      "grad_norm": 0.06972356140613556,
      "learning_rate": 0.00017998564936618034,
      "loss": 0.008,
      "step": 8868
    },
    {
      "epoch": 1.2430273300630694,
      "grad_norm": 0.226225346326828,
      "learning_rate": 0.00017997129873236067,
      "loss": 0.0189,
      "step": 8869
    },
    {
      "epoch": 1.2431674842326559,
      "grad_norm": 0.4632932245731354,
      "learning_rate": 0.000179956948098541,
      "loss": 0.1249,
      "step": 8870
    },
    {
      "epoch": 1.2433076384022423,
      "grad_norm": 0.22286300361156464,
      "learning_rate": 0.00017994259746472133,
      "loss": 0.0312,
      "step": 8871
    },
    {
      "epoch": 1.243447792571829,
      "grad_norm": 0.3477002680301666,
      "learning_rate": 0.00017992824683090168,
      "loss": 0.0499,
      "step": 8872
    },
    {
      "epoch": 1.2435879467414155,
      "grad_norm": 0.07345622032880783,
      "learning_rate": 0.000179913896197082,
      "loss": 0.0045,
      "step": 8873
    },
    {
      "epoch": 1.2437281009110022,
      "grad_norm": 0.15686769783496857,
      "learning_rate": 0.00017989954556326234,
      "loss": 0.0168,
      "step": 8874
    },
    {
      "epoch": 1.2438682550805886,
      "grad_norm": 0.23002737760543823,
      "learning_rate": 0.00017988519492944272,
      "loss": 0.0397,
      "step": 8875
    },
    {
      "epoch": 1.244008409250175,
      "grad_norm": 0.45163851976394653,
      "learning_rate": 0.00017987084429562305,
      "loss": 0.0591,
      "step": 8876
    },
    {
      "epoch": 1.2441485634197618,
      "grad_norm": 0.42541858553886414,
      "learning_rate": 0.00017985649366180338,
      "loss": 0.0628,
      "step": 8877
    },
    {
      "epoch": 1.2442887175893482,
      "grad_norm": 0.24296577274799347,
      "learning_rate": 0.00017984214302798373,
      "loss": 0.0329,
      "step": 8878
    },
    {
      "epoch": 1.244428871758935,
      "grad_norm": 0.39703062176704407,
      "learning_rate": 0.00017982779239416406,
      "loss": 0.1151,
      "step": 8879
    },
    {
      "epoch": 1.2445690259285214,
      "grad_norm": 0.5747311115264893,
      "learning_rate": 0.0001798134417603444,
      "loss": 0.1525,
      "step": 8880
    },
    {
      "epoch": 1.2447091800981078,
      "grad_norm": 0.883503258228302,
      "learning_rate": 0.00017979909112652475,
      "loss": 0.0701,
      "step": 8881
    },
    {
      "epoch": 1.2448493342676945,
      "grad_norm": 1.8566811084747314,
      "learning_rate": 0.00017978474049270508,
      "loss": 0.3588,
      "step": 8882
    },
    {
      "epoch": 1.244989488437281,
      "grad_norm": 1.2332398891448975,
      "learning_rate": 0.0001797703898588854,
      "loss": 0.1362,
      "step": 8883
    },
    {
      "epoch": 1.2451296426068676,
      "grad_norm": 1.036044955253601,
      "learning_rate": 0.0001797560392250658,
      "loss": 0.0943,
      "step": 8884
    },
    {
      "epoch": 1.245269796776454,
      "grad_norm": 1.9353384971618652,
      "learning_rate": 0.00017974168859124612,
      "loss": 0.2039,
      "step": 8885
    },
    {
      "epoch": 1.2454099509460406,
      "grad_norm": 0.17843778431415558,
      "learning_rate": 0.00017972733795742644,
      "loss": 0.0542,
      "step": 8886
    },
    {
      "epoch": 1.2455501051156272,
      "grad_norm": 0.26558083295822144,
      "learning_rate": 0.00017971298732360677,
      "loss": 0.0735,
      "step": 8887
    },
    {
      "epoch": 1.2456902592852137,
      "grad_norm": 0.6807878017425537,
      "learning_rate": 0.00017969863668978713,
      "loss": 0.0464,
      "step": 8888
    },
    {
      "epoch": 1.2458304134548004,
      "grad_norm": 0.08774295449256897,
      "learning_rate": 0.00017968428605596746,
      "loss": 0.0097,
      "step": 8889
    },
    {
      "epoch": 1.2459705676243868,
      "grad_norm": 0.2471190243959427,
      "learning_rate": 0.00017966993542214779,
      "loss": 0.0883,
      "step": 8890
    },
    {
      "epoch": 1.2461107217939733,
      "grad_norm": 0.3901362419128418,
      "learning_rate": 0.00017965558478832814,
      "loss": 0.0822,
      "step": 8891
    },
    {
      "epoch": 1.24625087596356,
      "grad_norm": 0.40733176469802856,
      "learning_rate": 0.00017964123415450847,
      "loss": 0.0521,
      "step": 8892
    },
    {
      "epoch": 1.2463910301331464,
      "grad_norm": 0.27577829360961914,
      "learning_rate": 0.0001796268835206888,
      "loss": 0.1004,
      "step": 8893
    },
    {
      "epoch": 1.2465311843027331,
      "grad_norm": 0.28781938552856445,
      "learning_rate": 0.00017961253288686915,
      "loss": 0.0328,
      "step": 8894
    },
    {
      "epoch": 1.2466713384723196,
      "grad_norm": 0.40174397826194763,
      "learning_rate": 0.00017959818225304948,
      "loss": 0.096,
      "step": 8895
    },
    {
      "epoch": 1.246811492641906,
      "grad_norm": 0.3050895035266876,
      "learning_rate": 0.0001795838316192298,
      "loss": 0.0639,
      "step": 8896
    },
    {
      "epoch": 1.2469516468114927,
      "grad_norm": 0.47422388195991516,
      "learning_rate": 0.0001795694809854102,
      "loss": 0.0352,
      "step": 8897
    },
    {
      "epoch": 1.2470918009810792,
      "grad_norm": 0.30920857191085815,
      "learning_rate": 0.00017955513035159052,
      "loss": 0.0835,
      "step": 8898
    },
    {
      "epoch": 1.2472319551506656,
      "grad_norm": 0.158649280667305,
      "learning_rate": 0.00017954077971777085,
      "loss": 0.0686,
      "step": 8899
    },
    {
      "epoch": 1.2473721093202523,
      "grad_norm": 0.4309064447879791,
      "learning_rate": 0.0001795264290839512,
      "loss": 0.0483,
      "step": 8900
    },
    {
      "epoch": 1.2475122634898388,
      "grad_norm": 0.17623929679393768,
      "learning_rate": 0.00017951207845013153,
      "loss": 0.0278,
      "step": 8901
    },
    {
      "epoch": 1.2476524176594253,
      "grad_norm": 0.35155707597732544,
      "learning_rate": 0.00017949772781631186,
      "loss": 0.06,
      "step": 8902
    },
    {
      "epoch": 1.247792571829012,
      "grad_norm": 0.3256426453590393,
      "learning_rate": 0.00017948337718249222,
      "loss": 0.0959,
      "step": 8903
    },
    {
      "epoch": 1.2479327259985984,
      "grad_norm": 0.23376812040805817,
      "learning_rate": 0.00017946902654867255,
      "loss": 0.049,
      "step": 8904
    },
    {
      "epoch": 1.248072880168185,
      "grad_norm": 0.27109384536743164,
      "learning_rate": 0.00017945467591485288,
      "loss": 0.029,
      "step": 8905
    },
    {
      "epoch": 1.2482130343377715,
      "grad_norm": 0.4933944642543793,
      "learning_rate": 0.0001794403252810332,
      "loss": 0.025,
      "step": 8906
    },
    {
      "epoch": 1.248353188507358,
      "grad_norm": 0.23296017944812775,
      "learning_rate": 0.0001794259746472136,
      "loss": 0.0127,
      "step": 8907
    },
    {
      "epoch": 1.2484933426769447,
      "grad_norm": 0.19686123728752136,
      "learning_rate": 0.00017941162401339392,
      "loss": 0.0456,
      "step": 8908
    },
    {
      "epoch": 1.2486334968465311,
      "grad_norm": 0.38507455587387085,
      "learning_rate": 0.00017939727337957424,
      "loss": 0.094,
      "step": 8909
    },
    {
      "epoch": 1.2487736510161178,
      "grad_norm": 0.6581763029098511,
      "learning_rate": 0.0001793829227457546,
      "loss": 0.1131,
      "step": 8910
    },
    {
      "epoch": 1.2489138051857043,
      "grad_norm": 0.88583904504776,
      "learning_rate": 0.00017936857211193493,
      "loss": 0.1287,
      "step": 8911
    },
    {
      "epoch": 1.2490539593552907,
      "grad_norm": 0.47335192561149597,
      "learning_rate": 0.00017935422147811526,
      "loss": 0.0606,
      "step": 8912
    },
    {
      "epoch": 1.2491941135248774,
      "grad_norm": 0.2666143774986267,
      "learning_rate": 0.0001793398708442956,
      "loss": 0.0301,
      "step": 8913
    },
    {
      "epoch": 1.2493342676944639,
      "grad_norm": 1.089593529701233,
      "learning_rate": 0.00017932552021047594,
      "loss": 0.1099,
      "step": 8914
    },
    {
      "epoch": 1.2494744218640506,
      "grad_norm": 0.5098910927772522,
      "learning_rate": 0.00017931116957665627,
      "loss": 0.0365,
      "step": 8915
    },
    {
      "epoch": 1.249614576033637,
      "grad_norm": 0.4775100648403168,
      "learning_rate": 0.00017929681894283665,
      "loss": 0.089,
      "step": 8916
    },
    {
      "epoch": 1.2497547302032235,
      "grad_norm": 0.2448941022157669,
      "learning_rate": 0.00017928246830901698,
      "loss": 0.0157,
      "step": 8917
    },
    {
      "epoch": 1.2498948843728102,
      "grad_norm": 0.20178437232971191,
      "learning_rate": 0.0001792681176751973,
      "loss": 0.0281,
      "step": 8918
    },
    {
      "epoch": 1.2500350385423966,
      "grad_norm": 0.34960028529167175,
      "learning_rate": 0.00017925376704137767,
      "loss": 0.0958,
      "step": 8919
    },
    {
      "epoch": 1.2501751927119833,
      "grad_norm": 0.40998774766921997,
      "learning_rate": 0.000179239416407558,
      "loss": 0.0343,
      "step": 8920
    },
    {
      "epoch": 1.2503153468815698,
      "grad_norm": 0.19427341222763062,
      "learning_rate": 0.00017922506577373832,
      "loss": 0.0281,
      "step": 8921
    },
    {
      "epoch": 1.2504555010511562,
      "grad_norm": 0.34853291511535645,
      "learning_rate": 0.00017921071513991865,
      "loss": 0.0499,
      "step": 8922
    },
    {
      "epoch": 1.250595655220743,
      "grad_norm": 0.4753531217575073,
      "learning_rate": 0.000179196364506099,
      "loss": 0.0781,
      "step": 8923
    },
    {
      "epoch": 1.2507358093903294,
      "grad_norm": 0.1523970067501068,
      "learning_rate": 0.00017918201387227934,
      "loss": 0.0227,
      "step": 8924
    },
    {
      "epoch": 1.250875963559916,
      "grad_norm": 0.9778902530670166,
      "learning_rate": 0.00017916766323845966,
      "loss": 0.091,
      "step": 8925
    },
    {
      "epoch": 1.2510161177295025,
      "grad_norm": 0.3031432628631592,
      "learning_rate": 0.00017915331260464002,
      "loss": 0.0337,
      "step": 8926
    },
    {
      "epoch": 1.251156271899089,
      "grad_norm": 0.3461599051952362,
      "learning_rate": 0.00017913896197082035,
      "loss": 0.0534,
      "step": 8927
    },
    {
      "epoch": 1.2512964260686754,
      "grad_norm": 0.31594088673591614,
      "learning_rate": 0.00017912461133700068,
      "loss": 0.1118,
      "step": 8928
    },
    {
      "epoch": 1.251436580238262,
      "grad_norm": 0.6019078493118286,
      "learning_rate": 0.00017911026070318106,
      "loss": 0.0644,
      "step": 8929
    },
    {
      "epoch": 1.2515767344078486,
      "grad_norm": 0.4162902534008026,
      "learning_rate": 0.0001790959100693614,
      "loss": 0.0463,
      "step": 8930
    },
    {
      "epoch": 1.2517168885774352,
      "grad_norm": 0.5430227518081665,
      "learning_rate": 0.00017908155943554172,
      "loss": 0.0466,
      "step": 8931
    },
    {
      "epoch": 1.2518570427470217,
      "grad_norm": 0.23210391402244568,
      "learning_rate": 0.00017906720880172207,
      "loss": 0.0348,
      "step": 8932
    },
    {
      "epoch": 1.2519971969166082,
      "grad_norm": 0.2866744101047516,
      "learning_rate": 0.0001790528581679024,
      "loss": 0.0144,
      "step": 8933
    },
    {
      "epoch": 1.2521373510861948,
      "grad_norm": 0.8020691871643066,
      "learning_rate": 0.00017903850753408273,
      "loss": 0.0412,
      "step": 8934
    },
    {
      "epoch": 1.2522775052557813,
      "grad_norm": 0.8091253638267517,
      "learning_rate": 0.00017902415690026309,
      "loss": 0.0223,
      "step": 8935
    },
    {
      "epoch": 1.252417659425368,
      "grad_norm": 0.4563453793525696,
      "learning_rate": 0.00017900980626644341,
      "loss": 0.1206,
      "step": 8936
    },
    {
      "epoch": 1.2525578135949544,
      "grad_norm": 0.2362118512392044,
      "learning_rate": 0.00017899545563262374,
      "loss": 0.0311,
      "step": 8937
    },
    {
      "epoch": 1.252697967764541,
      "grad_norm": 0.6482368111610413,
      "learning_rate": 0.00017898110499880413,
      "loss": 0.081,
      "step": 8938
    },
    {
      "epoch": 1.2528381219341276,
      "grad_norm": 0.6267337799072266,
      "learning_rate": 0.00017896675436498445,
      "loss": 0.0566,
      "step": 8939
    },
    {
      "epoch": 1.252978276103714,
      "grad_norm": 0.24068769812583923,
      "learning_rate": 0.00017895240373116478,
      "loss": 0.0264,
      "step": 8940
    },
    {
      "epoch": 1.2531184302733007,
      "grad_norm": 0.24023745954036713,
      "learning_rate": 0.0001789380530973451,
      "loss": 0.0366,
      "step": 8941
    },
    {
      "epoch": 1.2532585844428872,
      "grad_norm": 0.21197065711021423,
      "learning_rate": 0.00017892370246352547,
      "loss": 0.03,
      "step": 8942
    },
    {
      "epoch": 1.2533987386124736,
      "grad_norm": 0.6333866715431213,
      "learning_rate": 0.0001789093518297058,
      "loss": 0.0926,
      "step": 8943
    },
    {
      "epoch": 1.2535388927820603,
      "grad_norm": 0.48493653535842896,
      "learning_rate": 0.00017889500119588612,
      "loss": 0.0632,
      "step": 8944
    },
    {
      "epoch": 1.2536790469516468,
      "grad_norm": 0.5835952162742615,
      "learning_rate": 0.00017888065056206648,
      "loss": 0.0293,
      "step": 8945
    },
    {
      "epoch": 1.2538192011212335,
      "grad_norm": 0.4014591872692108,
      "learning_rate": 0.0001788662999282468,
      "loss": 0.0583,
      "step": 8946
    },
    {
      "epoch": 1.25395935529082,
      "grad_norm": 0.6542590856552124,
      "learning_rate": 0.00017885194929442714,
      "loss": 0.0743,
      "step": 8947
    },
    {
      "epoch": 1.2540995094604064,
      "grad_norm": 0.3809979259967804,
      "learning_rate": 0.00017883759866060752,
      "loss": 0.0439,
      "step": 8948
    },
    {
      "epoch": 1.254239663629993,
      "grad_norm": 0.24661223590373993,
      "learning_rate": 0.00017882324802678785,
      "loss": 0.0381,
      "step": 8949
    },
    {
      "epoch": 1.2543798177995795,
      "grad_norm": 0.3975275456905365,
      "learning_rate": 0.00017880889739296818,
      "loss": 0.0631,
      "step": 8950
    },
    {
      "epoch": 1.2545199719691662,
      "grad_norm": 0.5094953775405884,
      "learning_rate": 0.00017879454675914853,
      "loss": 0.0352,
      "step": 8951
    },
    {
      "epoch": 1.2546601261387527,
      "grad_norm": 0.26044341921806335,
      "learning_rate": 0.00017878019612532886,
      "loss": 0.1066,
      "step": 8952
    },
    {
      "epoch": 1.2548002803083391,
      "grad_norm": 0.4691629707813263,
      "learning_rate": 0.0001787658454915092,
      "loss": 0.0721,
      "step": 8953
    },
    {
      "epoch": 1.2549404344779256,
      "grad_norm": 0.520023763179779,
      "learning_rate": 0.00017875149485768954,
      "loss": 0.079,
      "step": 8954
    },
    {
      "epoch": 1.2550805886475123,
      "grad_norm": 0.3155807852745056,
      "learning_rate": 0.00017873714422386987,
      "loss": 0.0304,
      "step": 8955
    },
    {
      "epoch": 1.255220742817099,
      "grad_norm": 0.3651512563228607,
      "learning_rate": 0.0001787227935900502,
      "loss": 0.0376,
      "step": 8956
    },
    {
      "epoch": 1.2553608969866854,
      "grad_norm": 0.44446155428886414,
      "learning_rate": 0.00017870844295623053,
      "loss": 0.0644,
      "step": 8957
    },
    {
      "epoch": 1.2555010511562719,
      "grad_norm": 0.4956394135951996,
      "learning_rate": 0.00017869409232241089,
      "loss": 0.0369,
      "step": 8958
    },
    {
      "epoch": 1.2556412053258583,
      "grad_norm": 0.13711056113243103,
      "learning_rate": 0.00017867974168859121,
      "loss": 0.0345,
      "step": 8959
    },
    {
      "epoch": 1.255781359495445,
      "grad_norm": 0.47741255164146423,
      "learning_rate": 0.00017866539105477154,
      "loss": 0.0522,
      "step": 8960
    },
    {
      "epoch": 1.2559215136650315,
      "grad_norm": 0.10002173483371735,
      "learning_rate": 0.00017865104042095193,
      "loss": 0.019,
      "step": 8961
    },
    {
      "epoch": 1.2560616678346181,
      "grad_norm": 1.0469988584518433,
      "learning_rate": 0.00017863668978713225,
      "loss": 0.126,
      "step": 8962
    },
    {
      "epoch": 1.2562018220042046,
      "grad_norm": 0.2660384178161621,
      "learning_rate": 0.00017862233915331258,
      "loss": 0.0703,
      "step": 8963
    },
    {
      "epoch": 1.256341976173791,
      "grad_norm": 0.40686583518981934,
      "learning_rate": 0.00017860798851949294,
      "loss": 0.035,
      "step": 8964
    },
    {
      "epoch": 1.2564821303433777,
      "grad_norm": 0.439657062292099,
      "learning_rate": 0.00017859363788567327,
      "loss": 0.0597,
      "step": 8965
    },
    {
      "epoch": 1.2566222845129642,
      "grad_norm": 0.5489710569381714,
      "learning_rate": 0.0001785792872518536,
      "loss": 0.0567,
      "step": 8966
    },
    {
      "epoch": 1.2567624386825509,
      "grad_norm": 0.5363398194313049,
      "learning_rate": 0.00017856493661803395,
      "loss": 0.1139,
      "step": 8967
    },
    {
      "epoch": 1.2569025928521373,
      "grad_norm": 0.44276735186576843,
      "learning_rate": 0.00017855058598421428,
      "loss": 0.0638,
      "step": 8968
    },
    {
      "epoch": 1.2570427470217238,
      "grad_norm": 0.3831653296947479,
      "learning_rate": 0.0001785362353503946,
      "loss": 0.035,
      "step": 8969
    },
    {
      "epoch": 1.2571829011913105,
      "grad_norm": 0.5723667740821838,
      "learning_rate": 0.000178521884716575,
      "loss": 0.091,
      "step": 8970
    },
    {
      "epoch": 1.257323055360897,
      "grad_norm": 0.47155535221099854,
      "learning_rate": 0.00017850753408275532,
      "loss": 0.0504,
      "step": 8971
    },
    {
      "epoch": 1.2574632095304836,
      "grad_norm": 0.43353211879730225,
      "learning_rate": 0.00017849318344893565,
      "loss": 0.0905,
      "step": 8972
    },
    {
      "epoch": 1.25760336370007,
      "grad_norm": 0.20206907391548157,
      "learning_rate": 0.000178478832815116,
      "loss": 0.0852,
      "step": 8973
    },
    {
      "epoch": 1.2577435178696565,
      "grad_norm": 0.27788031101226807,
      "learning_rate": 0.00017846448218129633,
      "loss": 0.0211,
      "step": 8974
    },
    {
      "epoch": 1.2578836720392432,
      "grad_norm": 0.7037582993507385,
      "learning_rate": 0.00017845013154747666,
      "loss": 0.091,
      "step": 8975
    },
    {
      "epoch": 1.2580238262088297,
      "grad_norm": 0.30100011825561523,
      "learning_rate": 0.000178435780913657,
      "loss": 0.067,
      "step": 8976
    },
    {
      "epoch": 1.2581639803784164,
      "grad_norm": 0.5694476366043091,
      "learning_rate": 0.00017842143027983735,
      "loss": 0.0646,
      "step": 8977
    },
    {
      "epoch": 1.2583041345480028,
      "grad_norm": 0.2185674011707306,
      "learning_rate": 0.00017840707964601767,
      "loss": 0.0534,
      "step": 8978
    },
    {
      "epoch": 1.2584442887175893,
      "grad_norm": 0.3084188401699066,
      "learning_rate": 0.000178392729012198,
      "loss": 0.0594,
      "step": 8979
    },
    {
      "epoch": 1.258584442887176,
      "grad_norm": 0.708634078502655,
      "learning_rate": 0.00017837837837837839,
      "loss": 0.0991,
      "step": 8980
    },
    {
      "epoch": 1.2587245970567624,
      "grad_norm": 0.6217760443687439,
      "learning_rate": 0.00017836402774455871,
      "loss": 0.0239,
      "step": 8981
    },
    {
      "epoch": 1.258864751226349,
      "grad_norm": 1.0362032651901245,
      "learning_rate": 0.00017834967711073904,
      "loss": 0.2871,
      "step": 8982
    },
    {
      "epoch": 1.2590049053959356,
      "grad_norm": 0.7619487047195435,
      "learning_rate": 0.0001783353264769194,
      "loss": 0.0939,
      "step": 8983
    },
    {
      "epoch": 1.259145059565522,
      "grad_norm": 1.9713295698165894,
      "learning_rate": 0.00017832097584309973,
      "loss": 0.1286,
      "step": 8984
    },
    {
      "epoch": 1.2592852137351085,
      "grad_norm": 0.7887514233589172,
      "learning_rate": 0.00017830662520928006,
      "loss": 0.0936,
      "step": 8985
    },
    {
      "epoch": 1.2594253679046952,
      "grad_norm": 0.27064570784568787,
      "learning_rate": 0.0001782922745754604,
      "loss": 0.0668,
      "step": 8986
    },
    {
      "epoch": 1.2595655220742816,
      "grad_norm": 1.0867239236831665,
      "learning_rate": 0.00017827792394164074,
      "loss": 0.0941,
      "step": 8987
    },
    {
      "epoch": 1.2597056762438683,
      "grad_norm": 0.1598159819841385,
      "learning_rate": 0.00017826357330782107,
      "loss": 0.045,
      "step": 8988
    },
    {
      "epoch": 1.2598458304134548,
      "grad_norm": 0.2231038361787796,
      "learning_rate": 0.00017824922267400142,
      "loss": 0.0298,
      "step": 8989
    },
    {
      "epoch": 1.2599859845830412,
      "grad_norm": 0.38892167806625366,
      "learning_rate": 0.00017823487204018175,
      "loss": 0.0586,
      "step": 8990
    },
    {
      "epoch": 1.260126138752628,
      "grad_norm": 0.2902325391769409,
      "learning_rate": 0.00017822052140636208,
      "loss": 0.0555,
      "step": 8991
    },
    {
      "epoch": 1.2602662929222144,
      "grad_norm": 0.2863599956035614,
      "learning_rate": 0.0001782061707725424,
      "loss": 0.0544,
      "step": 8992
    },
    {
      "epoch": 1.260406447091801,
      "grad_norm": 0.32938963174819946,
      "learning_rate": 0.0001781918201387228,
      "loss": 0.0974,
      "step": 8993
    },
    {
      "epoch": 1.2605466012613875,
      "grad_norm": 0.45764076709747314,
      "learning_rate": 0.00017817746950490312,
      "loss": 0.069,
      "step": 8994
    },
    {
      "epoch": 1.260686755430974,
      "grad_norm": 0.19576597213745117,
      "learning_rate": 0.00017816311887108345,
      "loss": 0.042,
      "step": 8995
    },
    {
      "epoch": 1.2608269096005607,
      "grad_norm": 0.3495793640613556,
      "learning_rate": 0.0001781487682372638,
      "loss": 0.0578,
      "step": 8996
    },
    {
      "epoch": 1.2609670637701471,
      "grad_norm": 0.46037957072257996,
      "learning_rate": 0.00017813441760344413,
      "loss": 0.0744,
      "step": 8997
    },
    {
      "epoch": 1.2611072179397338,
      "grad_norm": 0.19668155908584595,
      "learning_rate": 0.00017812006696962446,
      "loss": 0.0536,
      "step": 8998
    },
    {
      "epoch": 1.2612473721093203,
      "grad_norm": 0.22465582191944122,
      "learning_rate": 0.00017810571633580482,
      "loss": 0.0525,
      "step": 8999
    },
    {
      "epoch": 1.2613875262789067,
      "grad_norm": 0.1850176453590393,
      "learning_rate": 0.00017809136570198515,
      "loss": 0.0195,
      "step": 9000
    },
    {
      "epoch": 1.2615276804484934,
      "grad_norm": 0.2853897511959076,
      "learning_rate": 0.00017807701506816547,
      "loss": 0.0391,
      "step": 9001
    },
    {
      "epoch": 1.2616678346180799,
      "grad_norm": 0.30048874020576477,
      "learning_rate": 0.00017806266443434586,
      "loss": 0.0907,
      "step": 9002
    },
    {
      "epoch": 1.2618079887876665,
      "grad_norm": 0.23137569427490234,
      "learning_rate": 0.00017804831380052619,
      "loss": 0.0342,
      "step": 9003
    },
    {
      "epoch": 1.261948142957253,
      "grad_norm": 0.3478965163230896,
      "learning_rate": 0.00017803396316670651,
      "loss": 0.0954,
      "step": 9004
    },
    {
      "epoch": 1.2620882971268395,
      "grad_norm": 0.3756159543991089,
      "learning_rate": 0.00017801961253288687,
      "loss": 0.0826,
      "step": 9005
    },
    {
      "epoch": 1.2622284512964261,
      "grad_norm": 0.5070664882659912,
      "learning_rate": 0.0001780052618990672,
      "loss": 0.0602,
      "step": 9006
    },
    {
      "epoch": 1.2623686054660126,
      "grad_norm": 0.2849693298339844,
      "learning_rate": 0.00017799091126524753,
      "loss": 0.035,
      "step": 9007
    },
    {
      "epoch": 1.2625087596355993,
      "grad_norm": 0.32464128732681274,
      "learning_rate": 0.00017797656063142786,
      "loss": 0.011,
      "step": 9008
    },
    {
      "epoch": 1.2626489138051857,
      "grad_norm": 0.2038242667913437,
      "learning_rate": 0.0001779622099976082,
      "loss": 0.0303,
      "step": 9009
    },
    {
      "epoch": 1.2627890679747722,
      "grad_norm": 0.3782430589199066,
      "learning_rate": 0.00017794785936378854,
      "loss": 0.0796,
      "step": 9010
    },
    {
      "epoch": 1.2629292221443589,
      "grad_norm": 0.18381468951702118,
      "learning_rate": 0.00017793350872996887,
      "loss": 0.0697,
      "step": 9011
    },
    {
      "epoch": 1.2630693763139453,
      "grad_norm": 0.4218471944332123,
      "learning_rate": 0.00017791915809614925,
      "loss": 0.0878,
      "step": 9012
    },
    {
      "epoch": 1.263209530483532,
      "grad_norm": 0.3217445909976959,
      "learning_rate": 0.00017790480746232958,
      "loss": 0.0379,
      "step": 9013
    },
    {
      "epoch": 1.2633496846531185,
      "grad_norm": 0.3653494417667389,
      "learning_rate": 0.0001778904568285099,
      "loss": 0.0773,
      "step": 9014
    },
    {
      "epoch": 1.263489838822705,
      "grad_norm": 0.42345574498176575,
      "learning_rate": 0.00017787610619469026,
      "loss": 0.0591,
      "step": 9015
    },
    {
      "epoch": 1.2636299929922914,
      "grad_norm": 0.6523240804672241,
      "learning_rate": 0.0001778617555608706,
      "loss": 0.0694,
      "step": 9016
    },
    {
      "epoch": 1.263770147161878,
      "grad_norm": 0.6912359595298767,
      "learning_rate": 0.00017784740492705092,
      "loss": 0.0513,
      "step": 9017
    },
    {
      "epoch": 1.2639103013314645,
      "grad_norm": 0.16744333505630493,
      "learning_rate": 0.00017783305429323128,
      "loss": 0.0167,
      "step": 9018
    },
    {
      "epoch": 1.2640504555010512,
      "grad_norm": 0.7581879496574402,
      "learning_rate": 0.0001778187036594116,
      "loss": 0.0681,
      "step": 9019
    },
    {
      "epoch": 1.2641906096706377,
      "grad_norm": 0.25457116961479187,
      "learning_rate": 0.00017780435302559193,
      "loss": 0.0421,
      "step": 9020
    },
    {
      "epoch": 1.2643307638402241,
      "grad_norm": 0.3509018123149872,
      "learning_rate": 0.0001777900023917723,
      "loss": 0.026,
      "step": 9021
    },
    {
      "epoch": 1.2644709180098108,
      "grad_norm": 0.604300320148468,
      "learning_rate": 0.00017777565175795262,
      "loss": 0.0687,
      "step": 9022
    },
    {
      "epoch": 1.2646110721793973,
      "grad_norm": 0.27976885437965393,
      "learning_rate": 0.00017776130112413295,
      "loss": 0.0617,
      "step": 9023
    },
    {
      "epoch": 1.264751226348984,
      "grad_norm": 0.4046292006969452,
      "learning_rate": 0.00017774695049031333,
      "loss": 0.0935,
      "step": 9024
    },
    {
      "epoch": 1.2648913805185704,
      "grad_norm": 0.5800681114196777,
      "learning_rate": 0.00017773259985649366,
      "loss": 0.147,
      "step": 9025
    },
    {
      "epoch": 1.2650315346881569,
      "grad_norm": 0.4359778165817261,
      "learning_rate": 0.000177718249222674,
      "loss": 0.1021,
      "step": 9026
    },
    {
      "epoch": 1.2651716888577436,
      "grad_norm": 0.13012881577014923,
      "learning_rate": 0.00017770389858885432,
      "loss": 0.0177,
      "step": 9027
    },
    {
      "epoch": 1.26531184302733,
      "grad_norm": 0.8135332465171814,
      "learning_rate": 0.00017768954795503467,
      "loss": 0.1369,
      "step": 9028
    },
    {
      "epoch": 1.2654519971969167,
      "grad_norm": 0.22240281105041504,
      "learning_rate": 0.000177675197321215,
      "loss": 0.0297,
      "step": 9029
    },
    {
      "epoch": 1.2655921513665032,
      "grad_norm": 0.5183936357498169,
      "learning_rate": 0.00017766084668739533,
      "loss": 0.0612,
      "step": 9030
    },
    {
      "epoch": 1.2657323055360896,
      "grad_norm": 0.5014948844909668,
      "learning_rate": 0.00017764649605357568,
      "loss": 0.1044,
      "step": 9031
    },
    {
      "epoch": 1.2658724597056763,
      "grad_norm": 0.7427463531494141,
      "learning_rate": 0.000177632145419756,
      "loss": 0.0269,
      "step": 9032
    },
    {
      "epoch": 1.2660126138752628,
      "grad_norm": 1.4680558443069458,
      "learning_rate": 0.00017761779478593634,
      "loss": 0.227,
      "step": 9033
    },
    {
      "epoch": 1.2661527680448494,
      "grad_norm": 1.5702149868011475,
      "learning_rate": 0.00017760344415211672,
      "loss": 0.1232,
      "step": 9034
    },
    {
      "epoch": 1.266292922214436,
      "grad_norm": 4.549380302429199,
      "learning_rate": 0.00017758909351829705,
      "loss": 0.1444,
      "step": 9035
    },
    {
      "epoch": 1.2664330763840224,
      "grad_norm": 0.14719705283641815,
      "learning_rate": 0.00017757474288447738,
      "loss": 0.0359,
      "step": 9036
    },
    {
      "epoch": 1.266573230553609,
      "grad_norm": 0.27009081840515137,
      "learning_rate": 0.00017756039225065774,
      "loss": 0.0562,
      "step": 9037
    },
    {
      "epoch": 1.2667133847231955,
      "grad_norm": 0.14267082512378693,
      "learning_rate": 0.00017754604161683807,
      "loss": 0.0223,
      "step": 9038
    },
    {
      "epoch": 1.2668535388927822,
      "grad_norm": 0.4080493748188019,
      "learning_rate": 0.0001775316909830184,
      "loss": 0.076,
      "step": 9039
    },
    {
      "epoch": 1.2669936930623686,
      "grad_norm": 0.48317191004753113,
      "learning_rate": 0.00017751734034919875,
      "loss": 0.0924,
      "step": 9040
    },
    {
      "epoch": 1.267133847231955,
      "grad_norm": 0.22761987149715424,
      "learning_rate": 0.00017750298971537908,
      "loss": 0.0284,
      "step": 9041
    },
    {
      "epoch": 1.2672740014015416,
      "grad_norm": 0.41949746012687683,
      "learning_rate": 0.0001774886390815594,
      "loss": 0.0448,
      "step": 9042
    },
    {
      "epoch": 1.2674141555711282,
      "grad_norm": 0.2655579149723053,
      "learning_rate": 0.00017747428844773974,
      "loss": 0.0431,
      "step": 9043
    },
    {
      "epoch": 1.267554309740715,
      "grad_norm": 0.4393776059150696,
      "learning_rate": 0.00017745993781392012,
      "loss": 0.1271,
      "step": 9044
    },
    {
      "epoch": 1.2676944639103014,
      "grad_norm": 0.29652729630470276,
      "learning_rate": 0.00017744558718010045,
      "loss": 0.0562,
      "step": 9045
    },
    {
      "epoch": 1.2678346180798878,
      "grad_norm": 0.36156249046325684,
      "learning_rate": 0.00017743123654628078,
      "loss": 0.0772,
      "step": 9046
    },
    {
      "epoch": 1.2679747722494743,
      "grad_norm": 0.3355903923511505,
      "learning_rate": 0.00017741688591246113,
      "loss": 0.0456,
      "step": 9047
    },
    {
      "epoch": 1.268114926419061,
      "grad_norm": 0.8490182757377625,
      "learning_rate": 0.00017740253527864146,
      "loss": 0.0624,
      "step": 9048
    },
    {
      "epoch": 1.2682550805886474,
      "grad_norm": 0.22248008847236633,
      "learning_rate": 0.0001773881846448218,
      "loss": 0.0286,
      "step": 9049
    },
    {
      "epoch": 1.2683952347582341,
      "grad_norm": 0.27648764848709106,
      "learning_rate": 0.00017737383401100214,
      "loss": 0.0326,
      "step": 9050
    },
    {
      "epoch": 1.2685353889278206,
      "grad_norm": 0.3741937577724457,
      "learning_rate": 0.00017735948337718247,
      "loss": 0.0478,
      "step": 9051
    },
    {
      "epoch": 1.268675543097407,
      "grad_norm": 0.3212883472442627,
      "learning_rate": 0.0001773451327433628,
      "loss": 0.0714,
      "step": 9052
    },
    {
      "epoch": 1.2688156972669937,
      "grad_norm": 0.3490413725376129,
      "learning_rate": 0.00017733078210954316,
      "loss": 0.0613,
      "step": 9053
    },
    {
      "epoch": 1.2689558514365802,
      "grad_norm": 0.21237251162528992,
      "learning_rate": 0.00017731643147572348,
      "loss": 0.0509,
      "step": 9054
    },
    {
      "epoch": 1.2690960056061669,
      "grad_norm": 0.15484821796417236,
      "learning_rate": 0.0001773020808419038,
      "loss": 0.0231,
      "step": 9055
    },
    {
      "epoch": 1.2692361597757533,
      "grad_norm": 0.15340827405452728,
      "learning_rate": 0.0001772877302080842,
      "loss": 0.0151,
      "step": 9056
    },
    {
      "epoch": 1.2693763139453398,
      "grad_norm": 0.40345245599746704,
      "learning_rate": 0.00017727337957426452,
      "loss": 0.0866,
      "step": 9057
    },
    {
      "epoch": 1.2695164681149265,
      "grad_norm": 0.2492057830095291,
      "learning_rate": 0.00017725902894044485,
      "loss": 0.0636,
      "step": 9058
    },
    {
      "epoch": 1.269656622284513,
      "grad_norm": 0.326826274394989,
      "learning_rate": 0.0001772446783066252,
      "loss": 0.0612,
      "step": 9059
    },
    {
      "epoch": 1.2697967764540996,
      "grad_norm": 0.30686521530151367,
      "learning_rate": 0.00017723032767280554,
      "loss": 0.068,
      "step": 9060
    },
    {
      "epoch": 1.269936930623686,
      "grad_norm": 0.38144758343696594,
      "learning_rate": 0.00017721597703898587,
      "loss": 0.086,
      "step": 9061
    },
    {
      "epoch": 1.2700770847932725,
      "grad_norm": 0.18938730657100677,
      "learning_rate": 0.0001772016264051662,
      "loss": 0.0568,
      "step": 9062
    },
    {
      "epoch": 1.2702172389628592,
      "grad_norm": 0.29055100679397583,
      "learning_rate": 0.00017718727577134655,
      "loss": 0.0422,
      "step": 9063
    },
    {
      "epoch": 1.2703573931324457,
      "grad_norm": 0.3560505509376526,
      "learning_rate": 0.00017717292513752688,
      "loss": 0.0544,
      "step": 9064
    },
    {
      "epoch": 1.2704975473020323,
      "grad_norm": 0.302933007478714,
      "learning_rate": 0.0001771585745037072,
      "loss": 0.0456,
      "step": 9065
    },
    {
      "epoch": 1.2706377014716188,
      "grad_norm": 0.34537091851234436,
      "learning_rate": 0.0001771442238698876,
      "loss": 0.0314,
      "step": 9066
    },
    {
      "epoch": 1.2707778556412053,
      "grad_norm": 0.25632351636886597,
      "learning_rate": 0.00017712987323606792,
      "loss": 0.0123,
      "step": 9067
    },
    {
      "epoch": 1.270918009810792,
      "grad_norm": 0.2705901861190796,
      "learning_rate": 0.00017711552260224825,
      "loss": 0.0402,
      "step": 9068
    },
    {
      "epoch": 1.2710581639803784,
      "grad_norm": 1.1916329860687256,
      "learning_rate": 0.0001771011719684286,
      "loss": 0.1206,
      "step": 9069
    },
    {
      "epoch": 1.271198318149965,
      "grad_norm": 0.12040171772241592,
      "learning_rate": 0.00017708682133460893,
      "loss": 0.0273,
      "step": 9070
    },
    {
      "epoch": 1.2713384723195515,
      "grad_norm": 0.24024905264377594,
      "learning_rate": 0.00017707247070078926,
      "loss": 0.0351,
      "step": 9071
    },
    {
      "epoch": 1.271478626489138,
      "grad_norm": 0.3056226074695587,
      "learning_rate": 0.00017705812006696962,
      "loss": 0.0986,
      "step": 9072
    },
    {
      "epoch": 1.2716187806587245,
      "grad_norm": 0.24302147328853607,
      "learning_rate": 0.00017704376943314994,
      "loss": 0.0483,
      "step": 9073
    },
    {
      "epoch": 1.2717589348283111,
      "grad_norm": 0.2792523503303528,
      "learning_rate": 0.00017702941879933027,
      "loss": 0.0354,
      "step": 9074
    },
    {
      "epoch": 1.2718990889978976,
      "grad_norm": 0.2571404278278351,
      "learning_rate": 0.00017701506816551066,
      "loss": 0.0522,
      "step": 9075
    },
    {
      "epoch": 1.2720392431674843,
      "grad_norm": 0.41163089871406555,
      "learning_rate": 0.00017700071753169098,
      "loss": 0.0511,
      "step": 9076
    },
    {
      "epoch": 1.2721793973370707,
      "grad_norm": 0.708112359046936,
      "learning_rate": 0.0001769863668978713,
      "loss": 0.0556,
      "step": 9077
    },
    {
      "epoch": 1.2723195515066572,
      "grad_norm": 0.09369780123233795,
      "learning_rate": 0.00017697201626405164,
      "loss": 0.0086,
      "step": 9078
    },
    {
      "epoch": 1.272459705676244,
      "grad_norm": 0.19607171416282654,
      "learning_rate": 0.000176957665630232,
      "loss": 0.0269,
      "step": 9079
    },
    {
      "epoch": 1.2725998598458304,
      "grad_norm": 0.2334619015455246,
      "learning_rate": 0.00017694331499641233,
      "loss": 0.0163,
      "step": 9080
    },
    {
      "epoch": 1.272740014015417,
      "grad_norm": 0.2039363533258438,
      "learning_rate": 0.00017692896436259265,
      "loss": 0.0094,
      "step": 9081
    },
    {
      "epoch": 1.2728801681850035,
      "grad_norm": 0.16530577838420868,
      "learning_rate": 0.000176914613728773,
      "loss": 0.0213,
      "step": 9082
    },
    {
      "epoch": 1.27302032235459,
      "grad_norm": 2.212944984436035,
      "learning_rate": 0.00017690026309495334,
      "loss": 0.1571,
      "step": 9083
    },
    {
      "epoch": 1.2731604765241766,
      "grad_norm": 0.7792435884475708,
      "learning_rate": 0.00017688591246113367,
      "loss": 0.1337,
      "step": 9084
    },
    {
      "epoch": 1.273300630693763,
      "grad_norm": 1.1687673330307007,
      "learning_rate": 0.00017687156182731402,
      "loss": 0.0895,
      "step": 9085
    },
    {
      "epoch": 1.2734407848633498,
      "grad_norm": 0.17413710057735443,
      "learning_rate": 0.00017685721119349435,
      "loss": 0.0375,
      "step": 9086
    },
    {
      "epoch": 1.2735809390329362,
      "grad_norm": 0.14838114380836487,
      "learning_rate": 0.00017684286055967468,
      "loss": 0.0299,
      "step": 9087
    },
    {
      "epoch": 1.2737210932025227,
      "grad_norm": 0.28675714135169983,
      "learning_rate": 0.00017682850992585506,
      "loss": 0.1071,
      "step": 9088
    },
    {
      "epoch": 1.2738612473721094,
      "grad_norm": 0.5509903430938721,
      "learning_rate": 0.0001768141592920354,
      "loss": 0.1003,
      "step": 9089
    },
    {
      "epoch": 1.2740014015416958,
      "grad_norm": 0.21225732564926147,
      "learning_rate": 0.00017679980865821572,
      "loss": 0.0217,
      "step": 9090
    },
    {
      "epoch": 1.2741415557112825,
      "grad_norm": 0.2884884774684906,
      "learning_rate": 0.00017678545802439608,
      "loss": 0.0405,
      "step": 9091
    },
    {
      "epoch": 1.274281709880869,
      "grad_norm": 0.5166020393371582,
      "learning_rate": 0.0001767711073905764,
      "loss": 0.0833,
      "step": 9092
    },
    {
      "epoch": 1.2744218640504554,
      "grad_norm": 0.4160119891166687,
      "learning_rate": 0.00017675675675675673,
      "loss": 0.0854,
      "step": 9093
    },
    {
      "epoch": 1.2745620182200421,
      "grad_norm": 0.3326933979988098,
      "learning_rate": 0.0001767424061229371,
      "loss": 0.0609,
      "step": 9094
    },
    {
      "epoch": 1.2747021723896286,
      "grad_norm": 0.29613587260246277,
      "learning_rate": 0.00017672805548911742,
      "loss": 0.0371,
      "step": 9095
    },
    {
      "epoch": 1.2748423265592153,
      "grad_norm": 0.3050192892551422,
      "learning_rate": 0.00017671370485529775,
      "loss": 0.0442,
      "step": 9096
    },
    {
      "epoch": 1.2749824807288017,
      "grad_norm": 0.3882840573787689,
      "learning_rate": 0.00017669935422147807,
      "loss": 0.0739,
      "step": 9097
    },
    {
      "epoch": 1.2751226348983882,
      "grad_norm": 0.21806761622428894,
      "learning_rate": 0.00017668500358765846,
      "loss": 0.0278,
      "step": 9098
    },
    {
      "epoch": 1.2752627890679746,
      "grad_norm": 0.34788092970848083,
      "learning_rate": 0.00017667065295383879,
      "loss": 0.0481,
      "step": 9099
    },
    {
      "epoch": 1.2754029432375613,
      "grad_norm": 0.2520679533481598,
      "learning_rate": 0.00017665630232001911,
      "loss": 0.0453,
      "step": 9100
    },
    {
      "epoch": 1.275543097407148,
      "grad_norm": 0.31644484400749207,
      "learning_rate": 0.00017664195168619947,
      "loss": 0.0336,
      "step": 9101
    },
    {
      "epoch": 1.2756832515767345,
      "grad_norm": 0.33437755703926086,
      "learning_rate": 0.0001766276010523798,
      "loss": 0.0343,
      "step": 9102
    },
    {
      "epoch": 1.275823405746321,
      "grad_norm": 0.14972250163555145,
      "learning_rate": 0.00017661325041856013,
      "loss": 0.0179,
      "step": 9103
    },
    {
      "epoch": 1.2759635599159074,
      "grad_norm": 0.3177717924118042,
      "learning_rate": 0.00017659889978474048,
      "loss": 0.0708,
      "step": 9104
    },
    {
      "epoch": 1.276103714085494,
      "grad_norm": 0.7146325707435608,
      "learning_rate": 0.0001765845491509208,
      "loss": 0.0684,
      "step": 9105
    },
    {
      "epoch": 1.2762438682550805,
      "grad_norm": 0.6792933344841003,
      "learning_rate": 0.00017657019851710114,
      "loss": 0.0834,
      "step": 9106
    },
    {
      "epoch": 1.2763840224246672,
      "grad_norm": 0.23644766211509705,
      "learning_rate": 0.00017655584788328152,
      "loss": 0.0456,
      "step": 9107
    },
    {
      "epoch": 1.2765241765942537,
      "grad_norm": 0.4507300853729248,
      "learning_rate": 0.00017654149724946185,
      "loss": 0.0484,
      "step": 9108
    },
    {
      "epoch": 1.2766643307638401,
      "grad_norm": 0.26117345690727234,
      "learning_rate": 0.00017652714661564218,
      "loss": 0.0419,
      "step": 9109
    },
    {
      "epoch": 1.2768044849334268,
      "grad_norm": 0.22464138269424438,
      "learning_rate": 0.00017651279598182253,
      "loss": 0.0266,
      "step": 9110
    },
    {
      "epoch": 1.2769446391030133,
      "grad_norm": 0.4495616853237152,
      "learning_rate": 0.00017649844534800286,
      "loss": 0.0827,
      "step": 9111
    },
    {
      "epoch": 1.2770847932726,
      "grad_norm": 0.7609719634056091,
      "learning_rate": 0.0001764840947141832,
      "loss": 0.0628,
      "step": 9112
    },
    {
      "epoch": 1.2772249474421864,
      "grad_norm": 0.3974210023880005,
      "learning_rate": 0.00017646974408036352,
      "loss": 0.0351,
      "step": 9113
    },
    {
      "epoch": 1.2773651016117729,
      "grad_norm": 0.39843010902404785,
      "learning_rate": 0.00017645539344654388,
      "loss": 0.0332,
      "step": 9114
    },
    {
      "epoch": 1.2775052557813595,
      "grad_norm": 0.4331094026565552,
      "learning_rate": 0.0001764410428127242,
      "loss": 0.0464,
      "step": 9115
    },
    {
      "epoch": 1.277645409950946,
      "grad_norm": 0.9654332399368286,
      "learning_rate": 0.00017642669217890453,
      "loss": 0.0503,
      "step": 9116
    },
    {
      "epoch": 1.2777855641205327,
      "grad_norm": 0.3390435576438904,
      "learning_rate": 0.0001764123415450849,
      "loss": 0.0803,
      "step": 9117
    },
    {
      "epoch": 1.2779257182901191,
      "grad_norm": 0.45222267508506775,
      "learning_rate": 0.00017639799091126522,
      "loss": 0.0597,
      "step": 9118
    },
    {
      "epoch": 1.2780658724597056,
      "grad_norm": 0.32691147923469543,
      "learning_rate": 0.00017638364027744555,
      "loss": 0.0307,
      "step": 9119
    },
    {
      "epoch": 1.2782060266292923,
      "grad_norm": 0.29305174946784973,
      "learning_rate": 0.00017636928964362593,
      "loss": 0.0408,
      "step": 9120
    },
    {
      "epoch": 1.2783461807988787,
      "grad_norm": 0.1483353227376938,
      "learning_rate": 0.00017635493900980626,
      "loss": 0.0304,
      "step": 9121
    },
    {
      "epoch": 1.2784863349684654,
      "grad_norm": 0.15840518474578857,
      "learning_rate": 0.00017634058837598659,
      "loss": 0.0413,
      "step": 9122
    },
    {
      "epoch": 1.2786264891380519,
      "grad_norm": 0.33043789863586426,
      "learning_rate": 0.00017632623774216694,
      "loss": 0.0444,
      "step": 9123
    },
    {
      "epoch": 1.2787666433076383,
      "grad_norm": 0.4495549499988556,
      "learning_rate": 0.00017631188710834727,
      "loss": 0.0679,
      "step": 9124
    },
    {
      "epoch": 1.278906797477225,
      "grad_norm": 0.5658689737319946,
      "learning_rate": 0.0001762975364745276,
      "loss": 0.024,
      "step": 9125
    },
    {
      "epoch": 1.2790469516468115,
      "grad_norm": 0.1985495537519455,
      "learning_rate": 0.00017628318584070795,
      "loss": 0.0353,
      "step": 9126
    },
    {
      "epoch": 1.2791871058163982,
      "grad_norm": 0.06834873557090759,
      "learning_rate": 0.00017626883520688828,
      "loss": 0.0133,
      "step": 9127
    },
    {
      "epoch": 1.2793272599859846,
      "grad_norm": 0.30182531476020813,
      "learning_rate": 0.0001762544845730686,
      "loss": 0.024,
      "step": 9128
    },
    {
      "epoch": 1.279467414155571,
      "grad_norm": 1.2703973054885864,
      "learning_rate": 0.000176240133939249,
      "loss": 0.0592,
      "step": 9129
    },
    {
      "epoch": 1.2796075683251575,
      "grad_norm": 0.5980146527290344,
      "learning_rate": 0.00017622578330542932,
      "loss": 0.0245,
      "step": 9130
    },
    {
      "epoch": 1.2797477224947442,
      "grad_norm": 0.4103550314903259,
      "learning_rate": 0.00017621143267160965,
      "loss": 0.0558,
      "step": 9131
    },
    {
      "epoch": 1.2798878766643307,
      "grad_norm": 0.5899176597595215,
      "learning_rate": 0.00017619708203778998,
      "loss": 0.0378,
      "step": 9132
    },
    {
      "epoch": 1.2800280308339174,
      "grad_norm": 1.5419152975082397,
      "learning_rate": 0.00017618273140397034,
      "loss": 0.1064,
      "step": 9133
    },
    {
      "epoch": 1.2801681850035038,
      "grad_norm": 0.9648589491844177,
      "learning_rate": 0.00017616838077015066,
      "loss": 0.0694,
      "step": 9134
    },
    {
      "epoch": 1.2803083391730903,
      "grad_norm": 1.6360570192337036,
      "learning_rate": 0.000176154030136331,
      "loss": 0.0801,
      "step": 9135
    },
    {
      "epoch": 1.280448493342677,
      "grad_norm": 0.4212769567966461,
      "learning_rate": 0.00017613967950251135,
      "loss": 0.07,
      "step": 9136
    },
    {
      "epoch": 1.2805886475122634,
      "grad_norm": 0.22926774621009827,
      "learning_rate": 0.00017612532886869168,
      "loss": 0.0312,
      "step": 9137
    },
    {
      "epoch": 1.28072880168185,
      "grad_norm": 0.33715131878852844,
      "learning_rate": 0.000176110978234872,
      "loss": 0.0498,
      "step": 9138
    },
    {
      "epoch": 1.2808689558514366,
      "grad_norm": 0.45437347888946533,
      "learning_rate": 0.0001760966276010524,
      "loss": 0.0494,
      "step": 9139
    },
    {
      "epoch": 1.281009110021023,
      "grad_norm": 0.21203671395778656,
      "learning_rate": 0.00017608227696723272,
      "loss": 0.032,
      "step": 9140
    },
    {
      "epoch": 1.2811492641906097,
      "grad_norm": 0.8790839314460754,
      "learning_rate": 0.00017606792633341305,
      "loss": 0.0819,
      "step": 9141
    },
    {
      "epoch": 1.2812894183601962,
      "grad_norm": 0.3380790054798126,
      "learning_rate": 0.0001760535756995934,
      "loss": 0.0404,
      "step": 9142
    },
    {
      "epoch": 1.2814295725297828,
      "grad_norm": 0.585423469543457,
      "learning_rate": 0.00017603922506577373,
      "loss": 0.0592,
      "step": 9143
    },
    {
      "epoch": 1.2815697266993693,
      "grad_norm": 0.7210609912872314,
      "learning_rate": 0.00017602487443195406,
      "loss": 0.1065,
      "step": 9144
    },
    {
      "epoch": 1.2817098808689558,
      "grad_norm": 0.35747218132019043,
      "learning_rate": 0.00017601052379813441,
      "loss": 0.0332,
      "step": 9145
    },
    {
      "epoch": 1.2818500350385424,
      "grad_norm": 0.1638728231191635,
      "learning_rate": 0.00017599617316431474,
      "loss": 0.0191,
      "step": 9146
    },
    {
      "epoch": 1.281990189208129,
      "grad_norm": 0.3132542669773102,
      "learning_rate": 0.00017598182253049507,
      "loss": 0.057,
      "step": 9147
    },
    {
      "epoch": 1.2821303433777156,
      "grad_norm": 0.1341615915298462,
      "learning_rate": 0.0001759674718966754,
      "loss": 0.0125,
      "step": 9148
    },
    {
      "epoch": 1.282270497547302,
      "grad_norm": 0.39787402749061584,
      "learning_rate": 0.00017595312126285576,
      "loss": 0.0773,
      "step": 9149
    },
    {
      "epoch": 1.2824106517168885,
      "grad_norm": 0.5609546303749084,
      "learning_rate": 0.00017593877062903608,
      "loss": 0.1328,
      "step": 9150
    },
    {
      "epoch": 1.2825508058864752,
      "grad_norm": 0.47899529337882996,
      "learning_rate": 0.0001759244199952164,
      "loss": 0.1613,
      "step": 9151
    },
    {
      "epoch": 1.2826909600560616,
      "grad_norm": 0.3914549946784973,
      "learning_rate": 0.0001759100693613968,
      "loss": 0.0673,
      "step": 9152
    },
    {
      "epoch": 1.2828311142256483,
      "grad_norm": 0.20205208659172058,
      "learning_rate": 0.00017589571872757712,
      "loss": 0.0469,
      "step": 9153
    },
    {
      "epoch": 1.2829712683952348,
      "grad_norm": 0.2930958867073059,
      "learning_rate": 0.00017588136809375745,
      "loss": 0.0814,
      "step": 9154
    },
    {
      "epoch": 1.2831114225648212,
      "grad_norm": 0.35807353258132935,
      "learning_rate": 0.0001758670174599378,
      "loss": 0.0589,
      "step": 9155
    },
    {
      "epoch": 1.283251576734408,
      "grad_norm": 0.4420841336250305,
      "learning_rate": 0.00017585266682611814,
      "loss": 0.0514,
      "step": 9156
    },
    {
      "epoch": 1.2833917309039944,
      "grad_norm": 0.3788946568965912,
      "learning_rate": 0.00017583831619229847,
      "loss": 0.0669,
      "step": 9157
    },
    {
      "epoch": 1.283531885073581,
      "grad_norm": 0.21619819104671478,
      "learning_rate": 0.00017582396555847882,
      "loss": 0.0446,
      "step": 9158
    },
    {
      "epoch": 1.2836720392431675,
      "grad_norm": 0.14575551450252533,
      "learning_rate": 0.00017580961492465915,
      "loss": 0.0228,
      "step": 9159
    },
    {
      "epoch": 1.283812193412754,
      "grad_norm": 0.15900026261806488,
      "learning_rate": 0.00017579526429083948,
      "loss": 0.0121,
      "step": 9160
    },
    {
      "epoch": 1.2839523475823404,
      "grad_norm": 0.26260024309158325,
      "learning_rate": 0.00017578091365701986,
      "loss": 0.0175,
      "step": 9161
    },
    {
      "epoch": 1.2840925017519271,
      "grad_norm": 0.3468899726867676,
      "learning_rate": 0.0001757665630232002,
      "loss": 0.107,
      "step": 9162
    },
    {
      "epoch": 1.2842326559215136,
      "grad_norm": 0.317746102809906,
      "learning_rate": 0.00017575221238938052,
      "loss": 0.0493,
      "step": 9163
    },
    {
      "epoch": 1.2843728100911003,
      "grad_norm": 0.21704652905464172,
      "learning_rate": 0.00017573786175556087,
      "loss": 0.0405,
      "step": 9164
    },
    {
      "epoch": 1.2845129642606867,
      "grad_norm": 0.35822945833206177,
      "learning_rate": 0.0001757235111217412,
      "loss": 0.0307,
      "step": 9165
    },
    {
      "epoch": 1.2846531184302732,
      "grad_norm": 0.40714696049690247,
      "learning_rate": 0.00017570916048792153,
      "loss": 0.0754,
      "step": 9166
    },
    {
      "epoch": 1.2847932725998599,
      "grad_norm": 0.2953345775604248,
      "learning_rate": 0.00017569480985410186,
      "loss": 0.0485,
      "step": 9167
    },
    {
      "epoch": 1.2849334267694463,
      "grad_norm": 0.5718219876289368,
      "learning_rate": 0.00017568045922028221,
      "loss": 0.1046,
      "step": 9168
    },
    {
      "epoch": 1.285073580939033,
      "grad_norm": 0.34174638986587524,
      "learning_rate": 0.00017566610858646254,
      "loss": 0.0475,
      "step": 9169
    },
    {
      "epoch": 1.2852137351086195,
      "grad_norm": 0.42920982837677,
      "learning_rate": 0.00017565175795264287,
      "loss": 0.043,
      "step": 9170
    },
    {
      "epoch": 1.285353889278206,
      "grad_norm": 0.8799381256103516,
      "learning_rate": 0.00017563740731882325,
      "loss": 0.1526,
      "step": 9171
    },
    {
      "epoch": 1.2854940434477926,
      "grad_norm": 0.4349725842475891,
      "learning_rate": 0.00017562305668500358,
      "loss": 0.0549,
      "step": 9172
    },
    {
      "epoch": 1.285634197617379,
      "grad_norm": 0.21843469142913818,
      "learning_rate": 0.0001756087060511839,
      "loss": 0.0349,
      "step": 9173
    },
    {
      "epoch": 1.2857743517869658,
      "grad_norm": 0.47788068652153015,
      "learning_rate": 0.00017559435541736427,
      "loss": 0.0503,
      "step": 9174
    },
    {
      "epoch": 1.2859145059565522,
      "grad_norm": 0.48978129029273987,
      "learning_rate": 0.0001755800047835446,
      "loss": 0.1091,
      "step": 9175
    },
    {
      "epoch": 1.2860546601261387,
      "grad_norm": 0.5065164566040039,
      "learning_rate": 0.00017556565414972492,
      "loss": 0.1335,
      "step": 9176
    },
    {
      "epoch": 1.2861948142957254,
      "grad_norm": 0.47257286310195923,
      "learning_rate": 0.00017555130351590528,
      "loss": 0.0516,
      "step": 9177
    },
    {
      "epoch": 1.2863349684653118,
      "grad_norm": 0.3913591504096985,
      "learning_rate": 0.0001755369528820856,
      "loss": 0.0577,
      "step": 9178
    },
    {
      "epoch": 1.2864751226348985,
      "grad_norm": 0.4789082407951355,
      "learning_rate": 0.00017552260224826594,
      "loss": 0.0395,
      "step": 9179
    },
    {
      "epoch": 1.286615276804485,
      "grad_norm": 0.44679173827171326,
      "learning_rate": 0.0001755082516144463,
      "loss": 0.1381,
      "step": 9180
    },
    {
      "epoch": 1.2867554309740714,
      "grad_norm": 0.039233651012182236,
      "learning_rate": 0.00017549390098062662,
      "loss": 0.0037,
      "step": 9181
    },
    {
      "epoch": 1.286895585143658,
      "grad_norm": 1.6388367414474487,
      "learning_rate": 0.00017547955034680695,
      "loss": 0.068,
      "step": 9182
    },
    {
      "epoch": 1.2870357393132446,
      "grad_norm": 1.3405311107635498,
      "learning_rate": 0.00017546519971298728,
      "loss": 0.0605,
      "step": 9183
    },
    {
      "epoch": 1.2871758934828312,
      "grad_norm": 0.23422621190547943,
      "learning_rate": 0.00017545084907916766,
      "loss": 0.0248,
      "step": 9184
    },
    {
      "epoch": 1.2873160476524177,
      "grad_norm": 0.5003237724304199,
      "learning_rate": 0.000175436498445348,
      "loss": 0.0528,
      "step": 9185
    },
    {
      "epoch": 1.2874562018220042,
      "grad_norm": 0.3197651505470276,
      "learning_rate": 0.00017542214781152832,
      "loss": 0.0804,
      "step": 9186
    },
    {
      "epoch": 1.2875963559915906,
      "grad_norm": 0.18552614748477936,
      "learning_rate": 0.00017540779717770867,
      "loss": 0.038,
      "step": 9187
    },
    {
      "epoch": 1.2877365101611773,
      "grad_norm": 0.23271670937538147,
      "learning_rate": 0.000175393446543889,
      "loss": 0.044,
      "step": 9188
    },
    {
      "epoch": 1.287876664330764,
      "grad_norm": 0.11801839619874954,
      "learning_rate": 0.00017537909591006933,
      "loss": 0.0087,
      "step": 9189
    },
    {
      "epoch": 1.2880168185003504,
      "grad_norm": 0.3330506980419159,
      "learning_rate": 0.0001753647452762497,
      "loss": 0.0335,
      "step": 9190
    },
    {
      "epoch": 1.288156972669937,
      "grad_norm": 0.20191267132759094,
      "learning_rate": 0.00017535039464243002,
      "loss": 0.0497,
      "step": 9191
    },
    {
      "epoch": 1.2882971268395234,
      "grad_norm": 1.042723536491394,
      "learning_rate": 0.00017533604400861034,
      "loss": 0.0761,
      "step": 9192
    },
    {
      "epoch": 1.28843728100911,
      "grad_norm": 0.36740317940711975,
      "learning_rate": 0.00017532169337479073,
      "loss": 0.0447,
      "step": 9193
    },
    {
      "epoch": 1.2885774351786965,
      "grad_norm": 0.19223883748054504,
      "learning_rate": 0.00017530734274097106,
      "loss": 0.0348,
      "step": 9194
    },
    {
      "epoch": 1.2887175893482832,
      "grad_norm": 0.24793177843093872,
      "learning_rate": 0.00017529299210715138,
      "loss": 0.0127,
      "step": 9195
    },
    {
      "epoch": 1.2888577435178696,
      "grad_norm": 0.27648964524269104,
      "learning_rate": 0.00017527864147333174,
      "loss": 0.0588,
      "step": 9196
    },
    {
      "epoch": 1.288997897687456,
      "grad_norm": 0.3581071197986603,
      "learning_rate": 0.00017526429083951207,
      "loss": 0.0507,
      "step": 9197
    },
    {
      "epoch": 1.2891380518570428,
      "grad_norm": 0.36401063203811646,
      "learning_rate": 0.0001752499402056924,
      "loss": 0.0628,
      "step": 9198
    },
    {
      "epoch": 1.2892782060266292,
      "grad_norm": 0.27278509736061096,
      "learning_rate": 0.00017523558957187275,
      "loss": 0.0541,
      "step": 9199
    },
    {
      "epoch": 1.289418360196216,
      "grad_norm": 0.6484878063201904,
      "learning_rate": 0.00017522123893805308,
      "loss": 0.0563,
      "step": 9200
    },
    {
      "epoch": 1.2895585143658024,
      "grad_norm": 0.2905210852622986,
      "learning_rate": 0.0001752068883042334,
      "loss": 0.0792,
      "step": 9201
    },
    {
      "epoch": 1.2896986685353888,
      "grad_norm": 0.20685267448425293,
      "learning_rate": 0.00017519253767041374,
      "loss": 0.0491,
      "step": 9202
    },
    {
      "epoch": 1.2898388227049755,
      "grad_norm": 0.6980092525482178,
      "learning_rate": 0.00017517818703659412,
      "loss": 0.0819,
      "step": 9203
    },
    {
      "epoch": 1.289978976874562,
      "grad_norm": 0.358024001121521,
      "learning_rate": 0.00017516383640277445,
      "loss": 0.0815,
      "step": 9204
    },
    {
      "epoch": 1.2901191310441487,
      "grad_norm": 0.2620103359222412,
      "learning_rate": 0.00017514948576895478,
      "loss": 0.0393,
      "step": 9205
    },
    {
      "epoch": 1.2902592852137351,
      "grad_norm": 0.3297736346721649,
      "learning_rate": 0.00017513513513513513,
      "loss": 0.0794,
      "step": 9206
    },
    {
      "epoch": 1.2903994393833216,
      "grad_norm": 0.295920193195343,
      "learning_rate": 0.00017512078450131546,
      "loss": 0.0254,
      "step": 9207
    },
    {
      "epoch": 1.2905395935529083,
      "grad_norm": 0.25152695178985596,
      "learning_rate": 0.0001751064338674958,
      "loss": 0.0389,
      "step": 9208
    },
    {
      "epoch": 1.2906797477224947,
      "grad_norm": 0.35447749495506287,
      "learning_rate": 0.00017509208323367615,
      "loss": 0.0752,
      "step": 9209
    },
    {
      "epoch": 1.2908199018920814,
      "grad_norm": 0.16341543197631836,
      "learning_rate": 0.00017507773259985648,
      "loss": 0.03,
      "step": 9210
    },
    {
      "epoch": 1.2909600560616679,
      "grad_norm": 0.4578576385974884,
      "learning_rate": 0.0001750633819660368,
      "loss": 0.0982,
      "step": 9211
    },
    {
      "epoch": 1.2911002102312543,
      "grad_norm": 0.3087027370929718,
      "learning_rate": 0.00017504903133221716,
      "loss": 0.041,
      "step": 9212
    },
    {
      "epoch": 1.291240364400841,
      "grad_norm": 0.6771711707115173,
      "learning_rate": 0.0001750346806983975,
      "loss": 0.0637,
      "step": 9213
    },
    {
      "epoch": 1.2913805185704275,
      "grad_norm": 0.47814592719078064,
      "learning_rate": 0.00017502033006457782,
      "loss": 0.0763,
      "step": 9214
    },
    {
      "epoch": 1.2915206727400141,
      "grad_norm": 0.17093922197818756,
      "learning_rate": 0.0001750059794307582,
      "loss": 0.0561,
      "step": 9215
    },
    {
      "epoch": 1.2916608269096006,
      "grad_norm": 0.49321088194847107,
      "learning_rate": 0.00017499162879693853,
      "loss": 0.0625,
      "step": 9216
    },
    {
      "epoch": 1.291800981079187,
      "grad_norm": 0.3502731919288635,
      "learning_rate": 0.00017497727816311886,
      "loss": 0.0487,
      "step": 9217
    },
    {
      "epoch": 1.2919411352487735,
      "grad_norm": 0.2070801556110382,
      "learning_rate": 0.00017496292752929918,
      "loss": 0.036,
      "step": 9218
    },
    {
      "epoch": 1.2920812894183602,
      "grad_norm": 0.5243269205093384,
      "learning_rate": 0.00017494857689547954,
      "loss": 0.0737,
      "step": 9219
    },
    {
      "epoch": 1.2922214435879467,
      "grad_norm": 0.2569962739944458,
      "learning_rate": 0.00017493422626165987,
      "loss": 0.0407,
      "step": 9220
    },
    {
      "epoch": 1.2923615977575333,
      "grad_norm": 0.49715694785118103,
      "learning_rate": 0.0001749198756278402,
      "loss": 0.0649,
      "step": 9221
    },
    {
      "epoch": 1.2925017519271198,
      "grad_norm": 0.15350471436977386,
      "learning_rate": 0.00017490552499402055,
      "loss": 0.0219,
      "step": 9222
    },
    {
      "epoch": 1.2926419060967063,
      "grad_norm": 0.5645371675491333,
      "learning_rate": 0.00017489117436020088,
      "loss": 0.0426,
      "step": 9223
    },
    {
      "epoch": 1.292782060266293,
      "grad_norm": 0.4345232844352722,
      "learning_rate": 0.0001748768237263812,
      "loss": 0.0817,
      "step": 9224
    },
    {
      "epoch": 1.2929222144358794,
      "grad_norm": 0.3802756369113922,
      "learning_rate": 0.0001748624730925616,
      "loss": 0.0667,
      "step": 9225
    },
    {
      "epoch": 1.293062368605466,
      "grad_norm": 0.4165038764476776,
      "learning_rate": 0.00017484812245874192,
      "loss": 0.1399,
      "step": 9226
    },
    {
      "epoch": 1.2932025227750525,
      "grad_norm": 0.5550238490104675,
      "learning_rate": 0.00017483377182492225,
      "loss": 0.089,
      "step": 9227
    },
    {
      "epoch": 1.293342676944639,
      "grad_norm": 0.4495529532432556,
      "learning_rate": 0.0001748194211911026,
      "loss": 0.0636,
      "step": 9228
    },
    {
      "epoch": 1.2934828311142257,
      "grad_norm": 0.7648413181304932,
      "learning_rate": 0.00017480507055728293,
      "loss": 0.1098,
      "step": 9229
    },
    {
      "epoch": 1.2936229852838121,
      "grad_norm": 0.2721000015735626,
      "learning_rate": 0.00017479071992346326,
      "loss": 0.0282,
      "step": 9230
    },
    {
      "epoch": 1.2937631394533988,
      "grad_norm": 0.3553127646446228,
      "learning_rate": 0.00017477636928964362,
      "loss": 0.0311,
      "step": 9231
    },
    {
      "epoch": 1.2939032936229853,
      "grad_norm": 0.7984032034873962,
      "learning_rate": 0.00017476201865582395,
      "loss": 0.1414,
      "step": 9232
    },
    {
      "epoch": 1.2940434477925717,
      "grad_norm": 1.4642761945724487,
      "learning_rate": 0.00017474766802200428,
      "loss": 0.2559,
      "step": 9233
    },
    {
      "epoch": 1.2941836019621584,
      "grad_norm": 2.0226428508758545,
      "learning_rate": 0.00017473331738818466,
      "loss": 0.3245,
      "step": 9234
    },
    {
      "epoch": 1.2943237561317449,
      "grad_norm": 1.8149771690368652,
      "learning_rate": 0.000174718966754365,
      "loss": 0.4662,
      "step": 9235
    },
    {
      "epoch": 1.2944639103013316,
      "grad_norm": 0.16745662689208984,
      "learning_rate": 0.00017470461612054532,
      "loss": 0.0176,
      "step": 9236
    },
    {
      "epoch": 1.294604064470918,
      "grad_norm": 0.3042994737625122,
      "learning_rate": 0.00017469026548672564,
      "loss": 0.0775,
      "step": 9237
    },
    {
      "epoch": 1.2947442186405045,
      "grad_norm": 0.3558884263038635,
      "learning_rate": 0.000174675914852906,
      "loss": 0.1116,
      "step": 9238
    },
    {
      "epoch": 1.2948843728100912,
      "grad_norm": 0.2747805118560791,
      "learning_rate": 0.00017466156421908633,
      "loss": 0.0519,
      "step": 9239
    },
    {
      "epoch": 1.2950245269796776,
      "grad_norm": 0.4262479543685913,
      "learning_rate": 0.00017464721358526666,
      "loss": 0.058,
      "step": 9240
    },
    {
      "epoch": 1.2951646811492643,
      "grad_norm": 0.5840110182762146,
      "learning_rate": 0.000174632862951447,
      "loss": 0.0629,
      "step": 9241
    },
    {
      "epoch": 1.2953048353188508,
      "grad_norm": 0.2950681746006012,
      "learning_rate": 0.00017461851231762734,
      "loss": 0.0772,
      "step": 9242
    },
    {
      "epoch": 1.2954449894884372,
      "grad_norm": 0.41101083159446716,
      "learning_rate": 0.00017460416168380767,
      "loss": 0.1177,
      "step": 9243
    },
    {
      "epoch": 1.295585143658024,
      "grad_norm": 0.26323914527893066,
      "learning_rate": 0.00017458981104998803,
      "loss": 0.0484,
      "step": 9244
    },
    {
      "epoch": 1.2957252978276104,
      "grad_norm": 0.2590402066707611,
      "learning_rate": 0.00017457546041616835,
      "loss": 0.0828,
      "step": 9245
    },
    {
      "epoch": 1.295865451997197,
      "grad_norm": 0.5093362927436829,
      "learning_rate": 0.00017456110978234868,
      "loss": 0.0439,
      "step": 9246
    },
    {
      "epoch": 1.2960056061667835,
      "grad_norm": 0.2033235728740692,
      "learning_rate": 0.00017454675914852907,
      "loss": 0.0439,
      "step": 9247
    },
    {
      "epoch": 1.29614576033637,
      "grad_norm": 0.19060920178890228,
      "learning_rate": 0.0001745324085147094,
      "loss": 0.0564,
      "step": 9248
    },
    {
      "epoch": 1.2962859145059564,
      "grad_norm": 0.4060218930244446,
      "learning_rate": 0.00017451805788088972,
      "loss": 0.0381,
      "step": 9249
    },
    {
      "epoch": 1.296426068675543,
      "grad_norm": 0.14771825075149536,
      "learning_rate": 0.00017450370724707008,
      "loss": 0.0491,
      "step": 9250
    },
    {
      "epoch": 1.2965662228451296,
      "grad_norm": 0.19069533050060272,
      "learning_rate": 0.0001744893566132504,
      "loss": 0.0386,
      "step": 9251
    },
    {
      "epoch": 1.2967063770147162,
      "grad_norm": 0.2760177254676819,
      "learning_rate": 0.00017447500597943074,
      "loss": 0.1309,
      "step": 9252
    },
    {
      "epoch": 1.2968465311843027,
      "grad_norm": 0.11279609799385071,
      "learning_rate": 0.00017446065534561106,
      "loss": 0.0091,
      "step": 9253
    },
    {
      "epoch": 1.2969866853538892,
      "grad_norm": 0.7452822923660278,
      "learning_rate": 0.00017444630471179142,
      "loss": 0.1099,
      "step": 9254
    },
    {
      "epoch": 1.2971268395234758,
      "grad_norm": 0.2421618402004242,
      "learning_rate": 0.00017443195407797175,
      "loss": 0.0931,
      "step": 9255
    },
    {
      "epoch": 1.2972669936930623,
      "grad_norm": 0.19737926125526428,
      "learning_rate": 0.00017441760344415208,
      "loss": 0.0291,
      "step": 9256
    },
    {
      "epoch": 1.297407147862649,
      "grad_norm": 0.3796572685241699,
      "learning_rate": 0.00017440325281033246,
      "loss": 0.078,
      "step": 9257
    },
    {
      "epoch": 1.2975473020322355,
      "grad_norm": 0.19226792454719543,
      "learning_rate": 0.0001743889021765128,
      "loss": 0.0564,
      "step": 9258
    },
    {
      "epoch": 1.297687456201822,
      "grad_norm": 0.41613495349884033,
      "learning_rate": 0.00017437455154269312,
      "loss": 0.0576,
      "step": 9259
    },
    {
      "epoch": 1.2978276103714086,
      "grad_norm": 0.21988633275032043,
      "learning_rate": 0.00017436020090887347,
      "loss": 0.0351,
      "step": 9260
    },
    {
      "epoch": 1.297967764540995,
      "grad_norm": 0.26572853326797485,
      "learning_rate": 0.0001743458502750538,
      "loss": 0.0575,
      "step": 9261
    },
    {
      "epoch": 1.2981079187105817,
      "grad_norm": 0.2526540458202362,
      "learning_rate": 0.00017433149964123413,
      "loss": 0.0363,
      "step": 9262
    },
    {
      "epoch": 1.2982480728801682,
      "grad_norm": 0.18137133121490479,
      "learning_rate": 0.00017431714900741449,
      "loss": 0.0285,
      "step": 9263
    },
    {
      "epoch": 1.2983882270497547,
      "grad_norm": 0.7921618819236755,
      "learning_rate": 0.00017430279837359481,
      "loss": 0.0348,
      "step": 9264
    },
    {
      "epoch": 1.2985283812193413,
      "grad_norm": 0.40944787859916687,
      "learning_rate": 0.00017428844773977514,
      "loss": 0.0195,
      "step": 9265
    },
    {
      "epoch": 1.2986685353889278,
      "grad_norm": 0.7854244709014893,
      "learning_rate": 0.00017427409710595552,
      "loss": 0.1102,
      "step": 9266
    },
    {
      "epoch": 1.2988086895585145,
      "grad_norm": 0.23199813067913055,
      "learning_rate": 0.00017425974647213585,
      "loss": 0.0287,
      "step": 9267
    },
    {
      "epoch": 1.298948843728101,
      "grad_norm": 0.34447017312049866,
      "learning_rate": 0.00017424539583831618,
      "loss": 0.0176,
      "step": 9268
    },
    {
      "epoch": 1.2990889978976874,
      "grad_norm": 0.43597227334976196,
      "learning_rate": 0.00017423104520449654,
      "loss": 0.0332,
      "step": 9269
    },
    {
      "epoch": 1.299229152067274,
      "grad_norm": 0.21176619827747345,
      "learning_rate": 0.00017421669457067687,
      "loss": 0.0409,
      "step": 9270
    },
    {
      "epoch": 1.2993693062368605,
      "grad_norm": 0.6289152503013611,
      "learning_rate": 0.0001742023439368572,
      "loss": 0.0835,
      "step": 9271
    },
    {
      "epoch": 1.2995094604064472,
      "grad_norm": 0.48130685091018677,
      "learning_rate": 0.00017418799330303752,
      "loss": 0.0497,
      "step": 9272
    },
    {
      "epoch": 1.2996496145760337,
      "grad_norm": 0.9445815682411194,
      "learning_rate": 0.00017417364266921788,
      "loss": 0.0896,
      "step": 9273
    },
    {
      "epoch": 1.2997897687456201,
      "grad_norm": 0.12780983746051788,
      "learning_rate": 0.0001741592920353982,
      "loss": 0.014,
      "step": 9274
    },
    {
      "epoch": 1.2999299229152066,
      "grad_norm": 0.5552884340286255,
      "learning_rate": 0.00017414494140157854,
      "loss": 0.0891,
      "step": 9275
    },
    {
      "epoch": 1.3000700770847933,
      "grad_norm": 0.45527178049087524,
      "learning_rate": 0.0001741305907677589,
      "loss": 0.0324,
      "step": 9276
    },
    {
      "epoch": 1.30021023125438,
      "grad_norm": 1.0368541479110718,
      "learning_rate": 0.00017411624013393922,
      "loss": 0.0302,
      "step": 9277
    },
    {
      "epoch": 1.3003503854239664,
      "grad_norm": 1.3127162456512451,
      "learning_rate": 0.00017410188950011955,
      "loss": 0.1259,
      "step": 9278
    },
    {
      "epoch": 1.3004905395935529,
      "grad_norm": 0.40234002470970154,
      "learning_rate": 0.00017408753886629993,
      "loss": 0.1897,
      "step": 9279
    },
    {
      "epoch": 1.3006306937631393,
      "grad_norm": 0.031851742416620255,
      "learning_rate": 0.00017407318823248026,
      "loss": 0.0026,
      "step": 9280
    },
    {
      "epoch": 1.300770847932726,
      "grad_norm": 0.3209799528121948,
      "learning_rate": 0.0001740588375986606,
      "loss": 0.0263,
      "step": 9281
    },
    {
      "epoch": 1.3009110021023125,
      "grad_norm": 0.7005457878112793,
      "learning_rate": 0.00017404448696484094,
      "loss": 0.0917,
      "step": 9282
    },
    {
      "epoch": 1.3010511562718992,
      "grad_norm": 1.1309617757797241,
      "learning_rate": 0.00017403013633102127,
      "loss": 0.0651,
      "step": 9283
    },
    {
      "epoch": 1.3011913104414856,
      "grad_norm": 0.5017295479774475,
      "learning_rate": 0.0001740157856972016,
      "loss": 0.0602,
      "step": 9284
    },
    {
      "epoch": 1.301331464611072,
      "grad_norm": 2.6762759685516357,
      "learning_rate": 0.00017400143506338196,
      "loss": 0.0997,
      "step": 9285
    },
    {
      "epoch": 1.3014716187806588,
      "grad_norm": 0.2774607241153717,
      "learning_rate": 0.00017398708442956229,
      "loss": 0.0562,
      "step": 9286
    },
    {
      "epoch": 1.3016117729502452,
      "grad_norm": 0.5338559150695801,
      "learning_rate": 0.00017397273379574261,
      "loss": 0.1086,
      "step": 9287
    },
    {
      "epoch": 1.301751927119832,
      "grad_norm": 0.18976879119873047,
      "learning_rate": 0.00017395838316192294,
      "loss": 0.0413,
      "step": 9288
    },
    {
      "epoch": 1.3018920812894184,
      "grad_norm": 0.4055008292198181,
      "learning_rate": 0.00017394403252810333,
      "loss": 0.0592,
      "step": 9289
    },
    {
      "epoch": 1.3020322354590048,
      "grad_norm": 0.17511360347270966,
      "learning_rate": 0.00017392968189428365,
      "loss": 0.0443,
      "step": 9290
    },
    {
      "epoch": 1.3021723896285915,
      "grad_norm": 0.2881506383419037,
      "learning_rate": 0.00017391533126046398,
      "loss": 0.0295,
      "step": 9291
    },
    {
      "epoch": 1.302312543798178,
      "grad_norm": 0.3340437710285187,
      "learning_rate": 0.00017390098062664434,
      "loss": 0.0528,
      "step": 9292
    },
    {
      "epoch": 1.3024526979677646,
      "grad_norm": 0.27340471744537354,
      "learning_rate": 0.00017388662999282467,
      "loss": 0.0684,
      "step": 9293
    },
    {
      "epoch": 1.302592852137351,
      "grad_norm": 0.20512999594211578,
      "learning_rate": 0.000173872279359005,
      "loss": 0.0392,
      "step": 9294
    },
    {
      "epoch": 1.3027330063069376,
      "grad_norm": 0.4061545729637146,
      "learning_rate": 0.00017385792872518535,
      "loss": 0.0418,
      "step": 9295
    },
    {
      "epoch": 1.3028731604765242,
      "grad_norm": 0.3768601417541504,
      "learning_rate": 0.00017384357809136568,
      "loss": 0.0646,
      "step": 9296
    },
    {
      "epoch": 1.3030133146461107,
      "grad_norm": 0.5274515151977539,
      "learning_rate": 0.000173829227457546,
      "loss": 0.0754,
      "step": 9297
    },
    {
      "epoch": 1.3031534688156974,
      "grad_norm": 0.38769614696502686,
      "learning_rate": 0.0001738148768237264,
      "loss": 0.078,
      "step": 9298
    },
    {
      "epoch": 1.3032936229852838,
      "grad_norm": 0.3846321403980255,
      "learning_rate": 0.00017380052618990672,
      "loss": 0.1279,
      "step": 9299
    },
    {
      "epoch": 1.3034337771548703,
      "grad_norm": 0.19241230189800262,
      "learning_rate": 0.00017378617555608705,
      "loss": 0.0142,
      "step": 9300
    },
    {
      "epoch": 1.303573931324457,
      "grad_norm": 0.4274437427520752,
      "learning_rate": 0.0001737718249222674,
      "loss": 0.0581,
      "step": 9301
    },
    {
      "epoch": 1.3037140854940434,
      "grad_norm": 0.2873820662498474,
      "learning_rate": 0.00017375747428844773,
      "loss": 0.0161,
      "step": 9302
    },
    {
      "epoch": 1.3038542396636301,
      "grad_norm": 0.2634715437889099,
      "learning_rate": 0.00017374312365462806,
      "loss": 0.0654,
      "step": 9303
    },
    {
      "epoch": 1.3039943938332166,
      "grad_norm": 0.3628040850162506,
      "learning_rate": 0.0001737287730208084,
      "loss": 0.0415,
      "step": 9304
    },
    {
      "epoch": 1.304134548002803,
      "grad_norm": 0.3178967237472534,
      "learning_rate": 0.00017371442238698875,
      "loss": 0.0853,
      "step": 9305
    },
    {
      "epoch": 1.3042747021723895,
      "grad_norm": 0.6008825898170471,
      "learning_rate": 0.00017370007175316907,
      "loss": 0.0477,
      "step": 9306
    },
    {
      "epoch": 1.3044148563419762,
      "grad_norm": 0.23644877970218658,
      "learning_rate": 0.0001736857211193494,
      "loss": 0.0708,
      "step": 9307
    },
    {
      "epoch": 1.3045550105115626,
      "grad_norm": 0.26526695489883423,
      "learning_rate": 0.00017367137048552976,
      "loss": 0.0462,
      "step": 9308
    },
    {
      "epoch": 1.3046951646811493,
      "grad_norm": 0.70341956615448,
      "learning_rate": 0.0001736570198517101,
      "loss": 0.0793,
      "step": 9309
    },
    {
      "epoch": 1.3048353188507358,
      "grad_norm": 0.26267218589782715,
      "learning_rate": 0.00017364266921789044,
      "loss": 0.025,
      "step": 9310
    },
    {
      "epoch": 1.3049754730203222,
      "grad_norm": 0.5026963949203491,
      "learning_rate": 0.0001736283185840708,
      "loss": 0.0813,
      "step": 9311
    },
    {
      "epoch": 1.305115627189909,
      "grad_norm": 0.21843931078910828,
      "learning_rate": 0.00017361396795025113,
      "loss": 0.032,
      "step": 9312
    },
    {
      "epoch": 1.3052557813594954,
      "grad_norm": 1.003778338432312,
      "learning_rate": 0.00017359961731643146,
      "loss": 0.0737,
      "step": 9313
    },
    {
      "epoch": 1.305395935529082,
      "grad_norm": 0.2915496230125427,
      "learning_rate": 0.0001735852666826118,
      "loss": 0.0226,
      "step": 9314
    },
    {
      "epoch": 1.3055360896986685,
      "grad_norm": 0.15265758335590363,
      "learning_rate": 0.00017357091604879214,
      "loss": 0.0205,
      "step": 9315
    },
    {
      "epoch": 1.305676243868255,
      "grad_norm": 0.3716699182987213,
      "learning_rate": 0.00017355656541497247,
      "loss": 0.0347,
      "step": 9316
    },
    {
      "epoch": 1.3058163980378417,
      "grad_norm": 0.2023908793926239,
      "learning_rate": 0.00017354221478115282,
      "loss": 0.0311,
      "step": 9317
    },
    {
      "epoch": 1.3059565522074281,
      "grad_norm": 0.21397031843662262,
      "learning_rate": 0.00017352786414733315,
      "loss": 0.0318,
      "step": 9318
    },
    {
      "epoch": 1.3060967063770148,
      "grad_norm": 0.6200441122055054,
      "learning_rate": 0.00017351351351351348,
      "loss": 0.0923,
      "step": 9319
    },
    {
      "epoch": 1.3062368605466013,
      "grad_norm": 1.9982049465179443,
      "learning_rate": 0.00017349916287969386,
      "loss": 0.0432,
      "step": 9320
    },
    {
      "epoch": 1.3063770147161877,
      "grad_norm": 0.3658328056335449,
      "learning_rate": 0.0001734848122458742,
      "loss": 0.0497,
      "step": 9321
    },
    {
      "epoch": 1.3065171688857744,
      "grad_norm": 0.8792640566825867,
      "learning_rate": 0.00017347046161205452,
      "loss": 0.032,
      "step": 9322
    },
    {
      "epoch": 1.3066573230553609,
      "grad_norm": 0.2585294544696808,
      "learning_rate": 0.00017345611097823485,
      "loss": 0.0229,
      "step": 9323
    },
    {
      "epoch": 1.3067974772249475,
      "grad_norm": 0.647038459777832,
      "learning_rate": 0.0001734417603444152,
      "loss": 0.1344,
      "step": 9324
    },
    {
      "epoch": 1.306937631394534,
      "grad_norm": 0.3910023868083954,
      "learning_rate": 0.00017342740971059553,
      "loss": 0.0489,
      "step": 9325
    },
    {
      "epoch": 1.3070777855641205,
      "grad_norm": 0.37376490235328674,
      "learning_rate": 0.00017341305907677586,
      "loss": 0.1398,
      "step": 9326
    },
    {
      "epoch": 1.3072179397337071,
      "grad_norm": 0.2519770562648773,
      "learning_rate": 0.00017339870844295622,
      "loss": 0.052,
      "step": 9327
    },
    {
      "epoch": 1.3073580939032936,
      "grad_norm": 0.5523326992988586,
      "learning_rate": 0.00017338435780913655,
      "loss": 0.0797,
      "step": 9328
    },
    {
      "epoch": 1.3074982480728803,
      "grad_norm": 0.7268182635307312,
      "learning_rate": 0.00017337000717531687,
      "loss": 0.1285,
      "step": 9329
    },
    {
      "epoch": 1.3076384022424667,
      "grad_norm": 0.4166392385959625,
      "learning_rate": 0.00017335565654149726,
      "loss": 0.0693,
      "step": 9330
    },
    {
      "epoch": 1.3077785564120532,
      "grad_norm": 0.22859403491020203,
      "learning_rate": 0.00017334130590767759,
      "loss": 0.0323,
      "step": 9331
    },
    {
      "epoch": 1.3079187105816397,
      "grad_norm": 6.437104225158691,
      "learning_rate": 0.00017332695527385791,
      "loss": 0.061,
      "step": 9332
    },
    {
      "epoch": 1.3080588647512263,
      "grad_norm": 0.2819992005825043,
      "learning_rate": 0.00017331260464003827,
      "loss": 0.018,
      "step": 9333
    },
    {
      "epoch": 1.308199018920813,
      "grad_norm": 0.5006887316703796,
      "learning_rate": 0.0001732982540062186,
      "loss": 0.1007,
      "step": 9334
    },
    {
      "epoch": 1.3083391730903995,
      "grad_norm": 0.27398598194122314,
      "learning_rate": 0.00017328390337239893,
      "loss": 0.0159,
      "step": 9335
    },
    {
      "epoch": 1.308479327259986,
      "grad_norm": 0.5111422538757324,
      "learning_rate": 0.00017326955273857928,
      "loss": 0.0787,
      "step": 9336
    },
    {
      "epoch": 1.3086194814295724,
      "grad_norm": 0.3599962294101715,
      "learning_rate": 0.0001732552021047596,
      "loss": 0.0345,
      "step": 9337
    },
    {
      "epoch": 1.308759635599159,
      "grad_norm": 0.14783678948879242,
      "learning_rate": 0.00017324085147093994,
      "loss": 0.042,
      "step": 9338
    },
    {
      "epoch": 1.3088997897687455,
      "grad_norm": 0.37984761595726013,
      "learning_rate": 0.00017322650083712027,
      "loss": 0.0576,
      "step": 9339
    },
    {
      "epoch": 1.3090399439383322,
      "grad_norm": 0.40532270073890686,
      "learning_rate": 0.00017321215020330062,
      "loss": 0.1169,
      "step": 9340
    },
    {
      "epoch": 1.3091800981079187,
      "grad_norm": 0.3553960621356964,
      "learning_rate": 0.00017319779956948095,
      "loss": 0.0624,
      "step": 9341
    },
    {
      "epoch": 1.3093202522775051,
      "grad_norm": 0.32200583815574646,
      "learning_rate": 0.0001731834489356613,
      "loss": 0.0386,
      "step": 9342
    },
    {
      "epoch": 1.3094604064470918,
      "grad_norm": 0.5956253409385681,
      "learning_rate": 0.00017316909830184166,
      "loss": 0.0726,
      "step": 9343
    },
    {
      "epoch": 1.3096005606166783,
      "grad_norm": 0.4274877607822418,
      "learning_rate": 0.000173154747668022,
      "loss": 0.051,
      "step": 9344
    },
    {
      "epoch": 1.309740714786265,
      "grad_norm": 0.20143482089042664,
      "learning_rate": 0.00017314039703420232,
      "loss": 0.0419,
      "step": 9345
    },
    {
      "epoch": 1.3098808689558514,
      "grad_norm": 0.4330682158470154,
      "learning_rate": 0.00017312604640038268,
      "loss": 0.0895,
      "step": 9346
    },
    {
      "epoch": 1.310021023125438,
      "grad_norm": 0.37343457341194153,
      "learning_rate": 0.000173111695766563,
      "loss": 0.0459,
      "step": 9347
    },
    {
      "epoch": 1.3101611772950246,
      "grad_norm": 0.20155425369739532,
      "learning_rate": 0.00017309734513274333,
      "loss": 0.037,
      "step": 9348
    },
    {
      "epoch": 1.310301331464611,
      "grad_norm": 0.29298028349876404,
      "learning_rate": 0.0001730829944989237,
      "loss": 0.0571,
      "step": 9349
    },
    {
      "epoch": 1.3104414856341977,
      "grad_norm": 0.22942262887954712,
      "learning_rate": 0.00017306864386510402,
      "loss": 0.0369,
      "step": 9350
    },
    {
      "epoch": 1.3105816398037842,
      "grad_norm": 0.2129916250705719,
      "learning_rate": 0.00017305429323128435,
      "loss": 0.0381,
      "step": 9351
    },
    {
      "epoch": 1.3107217939733706,
      "grad_norm": 0.38374507427215576,
      "learning_rate": 0.00017303994259746473,
      "loss": 0.0769,
      "step": 9352
    },
    {
      "epoch": 1.3108619481429573,
      "grad_norm": 0.4300970733165741,
      "learning_rate": 0.00017302559196364506,
      "loss": 0.0656,
      "step": 9353
    },
    {
      "epoch": 1.3110021023125438,
      "grad_norm": 0.12158835679292679,
      "learning_rate": 0.0001730112413298254,
      "loss": 0.0125,
      "step": 9354
    },
    {
      "epoch": 1.3111422564821305,
      "grad_norm": 0.4713352918624878,
      "learning_rate": 0.00017299689069600574,
      "loss": 0.0296,
      "step": 9355
    },
    {
      "epoch": 1.311282410651717,
      "grad_norm": 0.5789030194282532,
      "learning_rate": 0.00017298254006218607,
      "loss": 0.0608,
      "step": 9356
    },
    {
      "epoch": 1.3114225648213034,
      "grad_norm": 0.3154323697090149,
      "learning_rate": 0.0001729681894283664,
      "loss": 0.0494,
      "step": 9357
    },
    {
      "epoch": 1.31156271899089,
      "grad_norm": 0.35313957929611206,
      "learning_rate": 0.00017295383879454673,
      "loss": 0.0409,
      "step": 9358
    },
    {
      "epoch": 1.3117028731604765,
      "grad_norm": 0.4066758453845978,
      "learning_rate": 0.00017293948816072708,
      "loss": 0.0956,
      "step": 9359
    },
    {
      "epoch": 1.3118430273300632,
      "grad_norm": 0.2841693162918091,
      "learning_rate": 0.0001729251375269074,
      "loss": 0.0455,
      "step": 9360
    },
    {
      "epoch": 1.3119831814996497,
      "grad_norm": 0.33462291955947876,
      "learning_rate": 0.00017291078689308774,
      "loss": 0.0682,
      "step": 9361
    },
    {
      "epoch": 1.3121233356692361,
      "grad_norm": 0.6383495330810547,
      "learning_rate": 0.00017289643625926812,
      "loss": 0.0635,
      "step": 9362
    },
    {
      "epoch": 1.3122634898388226,
      "grad_norm": 0.6227530241012573,
      "learning_rate": 0.00017288208562544845,
      "loss": 0.0355,
      "step": 9363
    },
    {
      "epoch": 1.3124036440084093,
      "grad_norm": 0.8222858309745789,
      "learning_rate": 0.00017286773499162878,
      "loss": 0.1387,
      "step": 9364
    },
    {
      "epoch": 1.3125437981779957,
      "grad_norm": 0.43814709782600403,
      "learning_rate": 0.00017285338435780914,
      "loss": 0.0944,
      "step": 9365
    },
    {
      "epoch": 1.3126839523475824,
      "grad_norm": 0.2018711268901825,
      "learning_rate": 0.00017283903372398947,
      "loss": 0.0633,
      "step": 9366
    },
    {
      "epoch": 1.3128241065171689,
      "grad_norm": 0.17528922855854034,
      "learning_rate": 0.0001728246830901698,
      "loss": 0.0257,
      "step": 9367
    },
    {
      "epoch": 1.3129642606867553,
      "grad_norm": 0.47659605741500854,
      "learning_rate": 0.00017281033245635015,
      "loss": 0.0438,
      "step": 9368
    },
    {
      "epoch": 1.313104414856342,
      "grad_norm": 0.3596767783164978,
      "learning_rate": 0.00017279598182253048,
      "loss": 0.0288,
      "step": 9369
    },
    {
      "epoch": 1.3132445690259285,
      "grad_norm": 0.263205885887146,
      "learning_rate": 0.0001727816311887108,
      "loss": 0.0661,
      "step": 9370
    },
    {
      "epoch": 1.3133847231955151,
      "grad_norm": 0.38372817635536194,
      "learning_rate": 0.00017276728055489116,
      "loss": 0.0879,
      "step": 9371
    },
    {
      "epoch": 1.3135248773651016,
      "grad_norm": 0.41303446888923645,
      "learning_rate": 0.0001727529299210715,
      "loss": 0.0586,
      "step": 9372
    },
    {
      "epoch": 1.313665031534688,
      "grad_norm": 0.3223862648010254,
      "learning_rate": 0.00017273857928725182,
      "loss": 0.0484,
      "step": 9373
    },
    {
      "epoch": 1.3138051857042747,
      "grad_norm": 0.15607298910617828,
      "learning_rate": 0.00017272422865343217,
      "loss": 0.0321,
      "step": 9374
    },
    {
      "epoch": 1.3139453398738612,
      "grad_norm": 0.3347414433956146,
      "learning_rate": 0.00017270987801961253,
      "loss": 0.0268,
      "step": 9375
    },
    {
      "epoch": 1.3140854940434479,
      "grad_norm": 0.5157453417778015,
      "learning_rate": 0.00017269552738579286,
      "loss": 0.1011,
      "step": 9376
    },
    {
      "epoch": 1.3142256482130343,
      "grad_norm": 0.4231446087360382,
      "learning_rate": 0.0001726811767519732,
      "loss": 0.0207,
      "step": 9377
    },
    {
      "epoch": 1.3143658023826208,
      "grad_norm": 1.3078078031539917,
      "learning_rate": 0.00017266682611815354,
      "loss": 0.192,
      "step": 9378
    },
    {
      "epoch": 1.3145059565522075,
      "grad_norm": 0.11999952048063278,
      "learning_rate": 0.00017265247548433387,
      "loss": 0.0187,
      "step": 9379
    },
    {
      "epoch": 1.314646110721794,
      "grad_norm": 0.8334781527519226,
      "learning_rate": 0.0001726381248505142,
      "loss": 0.0311,
      "step": 9380
    },
    {
      "epoch": 1.3147862648913806,
      "grad_norm": 0.40437230467796326,
      "learning_rate": 0.00017262377421669456,
      "loss": 0.0528,
      "step": 9381
    },
    {
      "epoch": 1.314926419060967,
      "grad_norm": 0.5065985918045044,
      "learning_rate": 0.00017260942358287488,
      "loss": 0.0933,
      "step": 9382
    },
    {
      "epoch": 1.3150665732305535,
      "grad_norm": 0.3498857915401459,
      "learning_rate": 0.0001725950729490552,
      "loss": 0.015,
      "step": 9383
    },
    {
      "epoch": 1.3152067274001402,
      "grad_norm": 0.6446489095687866,
      "learning_rate": 0.0001725807223152356,
      "loss": 0.0514,
      "step": 9384
    },
    {
      "epoch": 1.3153468815697267,
      "grad_norm": 0.7263616323471069,
      "learning_rate": 0.00017256637168141592,
      "loss": 0.0958,
      "step": 9385
    },
    {
      "epoch": 1.3154870357393134,
      "grad_norm": 0.5223122835159302,
      "learning_rate": 0.00017255202104759625,
      "loss": 0.0604,
      "step": 9386
    },
    {
      "epoch": 1.3156271899088998,
      "grad_norm": 0.19896076619625092,
      "learning_rate": 0.0001725376704137766,
      "loss": 0.022,
      "step": 9387
    },
    {
      "epoch": 1.3157673440784863,
      "grad_norm": 0.5604138970375061,
      "learning_rate": 0.00017252331977995694,
      "loss": 0.0839,
      "step": 9388
    },
    {
      "epoch": 1.315907498248073,
      "grad_norm": 0.26151901483535767,
      "learning_rate": 0.00017250896914613727,
      "loss": 0.0645,
      "step": 9389
    },
    {
      "epoch": 1.3160476524176594,
      "grad_norm": 0.6319864988327026,
      "learning_rate": 0.00017249461851231762,
      "loss": 0.0483,
      "step": 9390
    },
    {
      "epoch": 1.316187806587246,
      "grad_norm": 0.34077081084251404,
      "learning_rate": 0.00017248026787849795,
      "loss": 0.028,
      "step": 9391
    },
    {
      "epoch": 1.3163279607568326,
      "grad_norm": 0.6036697030067444,
      "learning_rate": 0.00017246591724467828,
      "loss": 0.0897,
      "step": 9392
    },
    {
      "epoch": 1.316468114926419,
      "grad_norm": 0.197063609957695,
      "learning_rate": 0.0001724515666108586,
      "loss": 0.0399,
      "step": 9393
    },
    {
      "epoch": 1.3166082690960055,
      "grad_norm": 0.30906596779823303,
      "learning_rate": 0.000172437215977039,
      "loss": 0.0508,
      "step": 9394
    },
    {
      "epoch": 1.3167484232655922,
      "grad_norm": 0.32832279801368713,
      "learning_rate": 0.00017242286534321932,
      "loss": 0.0637,
      "step": 9395
    },
    {
      "epoch": 1.3168885774351786,
      "grad_norm": 0.4837797284126282,
      "learning_rate": 0.00017240851470939965,
      "loss": 0.1044,
      "step": 9396
    },
    {
      "epoch": 1.3170287316047653,
      "grad_norm": 0.7901841998100281,
      "learning_rate": 0.00017239416407558,
      "loss": 0.1214,
      "step": 9397
    },
    {
      "epoch": 1.3171688857743518,
      "grad_norm": 0.6718924641609192,
      "learning_rate": 0.00017237981344176033,
      "loss": 0.0555,
      "step": 9398
    },
    {
      "epoch": 1.3173090399439382,
      "grad_norm": 0.2511550784111023,
      "learning_rate": 0.00017236546280794066,
      "loss": 0.0536,
      "step": 9399
    },
    {
      "epoch": 1.317449194113525,
      "grad_norm": 0.4304676949977875,
      "learning_rate": 0.00017235111217412102,
      "loss": 0.0559,
      "step": 9400
    },
    {
      "epoch": 1.3175893482831114,
      "grad_norm": 0.2692815661430359,
      "learning_rate": 0.00017233676154030134,
      "loss": 0.0447,
      "step": 9401
    },
    {
      "epoch": 1.317729502452698,
      "grad_norm": 0.3264482319355011,
      "learning_rate": 0.00017232241090648167,
      "loss": 0.0345,
      "step": 9402
    },
    {
      "epoch": 1.3178696566222845,
      "grad_norm": 0.19648206233978271,
      "learning_rate": 0.00017230806027266203,
      "loss": 0.0398,
      "step": 9403
    },
    {
      "epoch": 1.318009810791871,
      "grad_norm": 0.3099084198474884,
      "learning_rate": 0.00017229370963884236,
      "loss": 0.0385,
      "step": 9404
    },
    {
      "epoch": 1.3181499649614576,
      "grad_norm": 0.1876903474330902,
      "learning_rate": 0.00017227935900502269,
      "loss": 0.0285,
      "step": 9405
    },
    {
      "epoch": 1.318290119131044,
      "grad_norm": 0.50505131483078,
      "learning_rate": 0.00017226500837120307,
      "loss": 0.0322,
      "step": 9406
    },
    {
      "epoch": 1.3184302733006308,
      "grad_norm": 0.4616032838821411,
      "learning_rate": 0.0001722506577373834,
      "loss": 0.0424,
      "step": 9407
    },
    {
      "epoch": 1.3185704274702172,
      "grad_norm": 0.3054984211921692,
      "learning_rate": 0.00017223630710356373,
      "loss": 0.0521,
      "step": 9408
    },
    {
      "epoch": 1.3187105816398037,
      "grad_norm": 0.45004233717918396,
      "learning_rate": 0.00017222195646974405,
      "loss": 0.0632,
      "step": 9409
    },
    {
      "epoch": 1.3188507358093904,
      "grad_norm": 0.7899589538574219,
      "learning_rate": 0.0001722076058359244,
      "loss": 0.054,
      "step": 9410
    },
    {
      "epoch": 1.3189908899789768,
      "grad_norm": 0.41488322615623474,
      "learning_rate": 0.00017219325520210474,
      "loss": 0.0583,
      "step": 9411
    },
    {
      "epoch": 1.3191310441485635,
      "grad_norm": 0.364869624376297,
      "learning_rate": 0.00017217890456828507,
      "loss": 0.1081,
      "step": 9412
    },
    {
      "epoch": 1.31927119831815,
      "grad_norm": 0.0936540812253952,
      "learning_rate": 0.00017216455393446542,
      "loss": 0.0076,
      "step": 9413
    },
    {
      "epoch": 1.3194113524877364,
      "grad_norm": 0.24324995279312134,
      "learning_rate": 0.00017215020330064575,
      "loss": 0.0191,
      "step": 9414
    },
    {
      "epoch": 1.3195515066573231,
      "grad_norm": 0.5959106087684631,
      "learning_rate": 0.00017213585266682608,
      "loss": 0.0402,
      "step": 9415
    },
    {
      "epoch": 1.3196916608269096,
      "grad_norm": 0.22388191521167755,
      "learning_rate": 0.00017212150203300646,
      "loss": 0.0106,
      "step": 9416
    },
    {
      "epoch": 1.3198318149964963,
      "grad_norm": 0.6405531764030457,
      "learning_rate": 0.0001721071513991868,
      "loss": 0.0582,
      "step": 9417
    },
    {
      "epoch": 1.3199719691660827,
      "grad_norm": 0.33234524726867676,
      "learning_rate": 0.00017209280076536712,
      "loss": 0.0258,
      "step": 9418
    },
    {
      "epoch": 1.3201121233356692,
      "grad_norm": 0.5107434988021851,
      "learning_rate": 0.00017207845013154748,
      "loss": 0.0705,
      "step": 9419
    },
    {
      "epoch": 1.3202522775052556,
      "grad_norm": 0.519616961479187,
      "learning_rate": 0.0001720640994977278,
      "loss": 0.1073,
      "step": 9420
    },
    {
      "epoch": 1.3203924316748423,
      "grad_norm": 0.22811037302017212,
      "learning_rate": 0.00017204974886390813,
      "loss": 0.0211,
      "step": 9421
    },
    {
      "epoch": 1.320532585844429,
      "grad_norm": 0.3887975215911865,
      "learning_rate": 0.0001720353982300885,
      "loss": 0.0743,
      "step": 9422
    },
    {
      "epoch": 1.3206727400140155,
      "grad_norm": 0.36133256554603577,
      "learning_rate": 0.00017202104759626882,
      "loss": 0.0566,
      "step": 9423
    },
    {
      "epoch": 1.320812894183602,
      "grad_norm": 0.1473684459924698,
      "learning_rate": 0.00017200669696244915,
      "loss": 0.0187,
      "step": 9424
    },
    {
      "epoch": 1.3209530483531884,
      "grad_norm": 0.18020454049110413,
      "learning_rate": 0.00017199234632862953,
      "loss": 0.0216,
      "step": 9425
    },
    {
      "epoch": 1.321093202522775,
      "grad_norm": 0.3773222863674164,
      "learning_rate": 0.00017197799569480986,
      "loss": 0.0402,
      "step": 9426
    },
    {
      "epoch": 1.3212333566923615,
      "grad_norm": 0.5045652389526367,
      "learning_rate": 0.00017196364506099018,
      "loss": 0.0258,
      "step": 9427
    },
    {
      "epoch": 1.3213735108619482,
      "grad_norm": 2.0684447288513184,
      "learning_rate": 0.0001719492944271705,
      "loss": 0.0577,
      "step": 9428
    },
    {
      "epoch": 1.3215136650315347,
      "grad_norm": 0.4795096516609192,
      "learning_rate": 0.00017193494379335087,
      "loss": 0.0619,
      "step": 9429
    },
    {
      "epoch": 1.3216538192011211,
      "grad_norm": 0.32334205508232117,
      "learning_rate": 0.0001719205931595312,
      "loss": 0.0442,
      "step": 9430
    },
    {
      "epoch": 1.3217939733707078,
      "grad_norm": 0.5212321281433105,
      "learning_rate": 0.00017190624252571153,
      "loss": 0.0549,
      "step": 9431
    },
    {
      "epoch": 1.3219341275402943,
      "grad_norm": 0.26268699765205383,
      "learning_rate": 0.00017189189189189188,
      "loss": 0.0056,
      "step": 9432
    },
    {
      "epoch": 1.322074281709881,
      "grad_norm": 1.6961874961853027,
      "learning_rate": 0.0001718775412580722,
      "loss": 0.1777,
      "step": 9433
    },
    {
      "epoch": 1.3222144358794674,
      "grad_norm": 6.129197120666504,
      "learning_rate": 0.00017186319062425254,
      "loss": 0.5777,
      "step": 9434
    },
    {
      "epoch": 1.3223545900490539,
      "grad_norm": 1.4504064321517944,
      "learning_rate": 0.0001718488399904329,
      "loss": 0.1988,
      "step": 9435
    },
    {
      "epoch": 1.3224947442186405,
      "grad_norm": 0.37830254435539246,
      "learning_rate": 0.00017183448935661322,
      "loss": 0.0386,
      "step": 9436
    },
    {
      "epoch": 1.322634898388227,
      "grad_norm": 0.29685354232788086,
      "learning_rate": 0.00017182013872279358,
      "loss": 0.0433,
      "step": 9437
    },
    {
      "epoch": 1.3227750525578137,
      "grad_norm": 0.3068963289260864,
      "learning_rate": 0.00017180578808897393,
      "loss": 0.0559,
      "step": 9438
    },
    {
      "epoch": 1.3229152067274002,
      "grad_norm": 0.36144188046455383,
      "learning_rate": 0.00017179143745515426,
      "loss": 0.1003,
      "step": 9439
    },
    {
      "epoch": 1.3230553608969866,
      "grad_norm": 0.23329022526741028,
      "learning_rate": 0.0001717770868213346,
      "loss": 0.0662,
      "step": 9440
    },
    {
      "epoch": 1.3231955150665733,
      "grad_norm": 0.31613385677337646,
      "learning_rate": 0.00017176273618751495,
      "loss": 0.0467,
      "step": 9441
    },
    {
      "epoch": 1.3233356692361598,
      "grad_norm": 0.7425344586372375,
      "learning_rate": 0.00017174838555369528,
      "loss": 0.1065,
      "step": 9442
    },
    {
      "epoch": 1.3234758234057464,
      "grad_norm": 0.2765350639820099,
      "learning_rate": 0.0001717340349198756,
      "loss": 0.0658,
      "step": 9443
    },
    {
      "epoch": 1.323615977575333,
      "grad_norm": 0.35815995931625366,
      "learning_rate": 0.00017171968428605593,
      "loss": 0.0613,
      "step": 9444
    },
    {
      "epoch": 1.3237561317449194,
      "grad_norm": 0.4471622705459595,
      "learning_rate": 0.0001717053336522363,
      "loss": 0.0337,
      "step": 9445
    },
    {
      "epoch": 1.323896285914506,
      "grad_norm": 0.27890393137931824,
      "learning_rate": 0.00017169098301841662,
      "loss": 0.0386,
      "step": 9446
    },
    {
      "epoch": 1.3240364400840925,
      "grad_norm": 0.3145810067653656,
      "learning_rate": 0.00017167663238459695,
      "loss": 0.0507,
      "step": 9447
    },
    {
      "epoch": 1.3241765942536792,
      "grad_norm": 0.11248306185007095,
      "learning_rate": 0.00017166228175077733,
      "loss": 0.0114,
      "step": 9448
    },
    {
      "epoch": 1.3243167484232656,
      "grad_norm": 0.4057770073413849,
      "learning_rate": 0.00017164793111695766,
      "loss": 0.0405,
      "step": 9449
    },
    {
      "epoch": 1.324456902592852,
      "grad_norm": 0.4622826874256134,
      "learning_rate": 0.00017163358048313799,
      "loss": 0.0638,
      "step": 9450
    },
    {
      "epoch": 1.3245970567624386,
      "grad_norm": 0.3345355689525604,
      "learning_rate": 0.00017161922984931834,
      "loss": 0.0568,
      "step": 9451
    },
    {
      "epoch": 1.3247372109320252,
      "grad_norm": 0.11428558826446533,
      "learning_rate": 0.00017160487921549867,
      "loss": 0.014,
      "step": 9452
    },
    {
      "epoch": 1.3248773651016117,
      "grad_norm": 0.218642920255661,
      "learning_rate": 0.000171590528581679,
      "loss": 0.0353,
      "step": 9453
    },
    {
      "epoch": 1.3250175192711984,
      "grad_norm": 0.7660614848136902,
      "learning_rate": 0.00017157617794785935,
      "loss": 0.0524,
      "step": 9454
    },
    {
      "epoch": 1.3251576734407848,
      "grad_norm": 0.38175347447395325,
      "learning_rate": 0.00017156182731403968,
      "loss": 0.0702,
      "step": 9455
    },
    {
      "epoch": 1.3252978276103713,
      "grad_norm": 0.49460795521736145,
      "learning_rate": 0.00017154747668022,
      "loss": 0.0506,
      "step": 9456
    },
    {
      "epoch": 1.325437981779958,
      "grad_norm": 0.3104495108127594,
      "learning_rate": 0.0001715331260464004,
      "loss": 0.062,
      "step": 9457
    },
    {
      "epoch": 1.3255781359495444,
      "grad_norm": 0.241384819149971,
      "learning_rate": 0.00017151877541258072,
      "loss": 0.0708,
      "step": 9458
    },
    {
      "epoch": 1.3257182901191311,
      "grad_norm": 0.6427384614944458,
      "learning_rate": 0.00017150442477876105,
      "loss": 0.0274,
      "step": 9459
    },
    {
      "epoch": 1.3258584442887176,
      "grad_norm": 0.38693928718566895,
      "learning_rate": 0.0001714900741449414,
      "loss": 0.0201,
      "step": 9460
    },
    {
      "epoch": 1.325998598458304,
      "grad_norm": 0.3339323401451111,
      "learning_rate": 0.00017147572351112174,
      "loss": 0.0551,
      "step": 9461
    },
    {
      "epoch": 1.3261387526278907,
      "grad_norm": 0.2760277986526489,
      "learning_rate": 0.00017146137287730206,
      "loss": 0.0368,
      "step": 9462
    },
    {
      "epoch": 1.3262789067974772,
      "grad_norm": 0.8217750191688538,
      "learning_rate": 0.0001714470222434824,
      "loss": 0.0486,
      "step": 9463
    },
    {
      "epoch": 1.3264190609670639,
      "grad_norm": 0.3545800447463989,
      "learning_rate": 0.00017143267160966275,
      "loss": 0.0621,
      "step": 9464
    },
    {
      "epoch": 1.3265592151366503,
      "grad_norm": 0.5317981839179993,
      "learning_rate": 0.00017141832097584308,
      "loss": 0.0639,
      "step": 9465
    },
    {
      "epoch": 1.3266993693062368,
      "grad_norm": 0.10901476442813873,
      "learning_rate": 0.0001714039703420234,
      "loss": 0.0089,
      "step": 9466
    },
    {
      "epoch": 1.3268395234758235,
      "grad_norm": 0.20464099943637848,
      "learning_rate": 0.00017138961970820376,
      "loss": 0.0529,
      "step": 9467
    },
    {
      "epoch": 1.32697967764541,
      "grad_norm": 0.41743746399879456,
      "learning_rate": 0.0001713752690743841,
      "loss": 0.088,
      "step": 9468
    },
    {
      "epoch": 1.3271198318149966,
      "grad_norm": 0.4790174961090088,
      "learning_rate": 0.00017136091844056445,
      "loss": 0.0516,
      "step": 9469
    },
    {
      "epoch": 1.327259985984583,
      "grad_norm": 0.6115158796310425,
      "learning_rate": 0.0001713465678067448,
      "loss": 0.0496,
      "step": 9470
    },
    {
      "epoch": 1.3274001401541695,
      "grad_norm": 0.16216257214546204,
      "learning_rate": 0.00017133221717292513,
      "loss": 0.0192,
      "step": 9471
    },
    {
      "epoch": 1.3275402943237562,
      "grad_norm": 0.33258384466171265,
      "learning_rate": 0.00017131786653910546,
      "loss": 0.0803,
      "step": 9472
    },
    {
      "epoch": 1.3276804484933427,
      "grad_norm": 0.09029435366392136,
      "learning_rate": 0.00017130351590528581,
      "loss": 0.0125,
      "step": 9473
    },
    {
      "epoch": 1.3278206026629293,
      "grad_norm": 0.7412534952163696,
      "learning_rate": 0.00017128916527146614,
      "loss": 0.0242,
      "step": 9474
    },
    {
      "epoch": 1.3279607568325158,
      "grad_norm": 0.20665088295936584,
      "learning_rate": 0.00017127481463764647,
      "loss": 0.0202,
      "step": 9475
    },
    {
      "epoch": 1.3281009110021023,
      "grad_norm": 0.29844722151756287,
      "learning_rate": 0.00017126046400382683,
      "loss": 0.0499,
      "step": 9476
    },
    {
      "epoch": 1.328241065171689,
      "grad_norm": 0.1486787647008896,
      "learning_rate": 0.00017124611337000716,
      "loss": 0.0088,
      "step": 9477
    },
    {
      "epoch": 1.3283812193412754,
      "grad_norm": 0.5192716121673584,
      "learning_rate": 0.00017123176273618748,
      "loss": 0.107,
      "step": 9478
    },
    {
      "epoch": 1.328521373510862,
      "grad_norm": 0.21467429399490356,
      "learning_rate": 0.0001712174121023678,
      "loss": 0.0087,
      "step": 9479
    },
    {
      "epoch": 1.3286615276804485,
      "grad_norm": 0.5761073231697083,
      "learning_rate": 0.0001712030614685482,
      "loss": 0.0783,
      "step": 9480
    },
    {
      "epoch": 1.328801681850035,
      "grad_norm": 0.28764209151268005,
      "learning_rate": 0.00017118871083472852,
      "loss": 0.0407,
      "step": 9481
    },
    {
      "epoch": 1.3289418360196215,
      "grad_norm": 0.5905888676643372,
      "learning_rate": 0.00017117436020090885,
      "loss": 0.0257,
      "step": 9482
    },
    {
      "epoch": 1.3290819901892081,
      "grad_norm": 1.063347578048706,
      "learning_rate": 0.0001711600095670892,
      "loss": 0.1806,
      "step": 9483
    },
    {
      "epoch": 1.3292221443587946,
      "grad_norm": 0.8173499703407288,
      "learning_rate": 0.00017114565893326954,
      "loss": 0.3561,
      "step": 9484
    },
    {
      "epoch": 1.3293622985283813,
      "grad_norm": 2.738908052444458,
      "learning_rate": 0.00017113130829944986,
      "loss": 0.3224,
      "step": 9485
    },
    {
      "epoch": 1.3295024526979677,
      "grad_norm": 0.24104203283786774,
      "learning_rate": 0.00017111695766563022,
      "loss": 0.0421,
      "step": 9486
    },
    {
      "epoch": 1.3296426068675542,
      "grad_norm": 0.3442029058933258,
      "learning_rate": 0.00017110260703181055,
      "loss": 0.1006,
      "step": 9487
    },
    {
      "epoch": 1.3297827610371409,
      "grad_norm": 0.30067411065101624,
      "learning_rate": 0.00017108825639799088,
      "loss": 0.0636,
      "step": 9488
    },
    {
      "epoch": 1.3299229152067273,
      "grad_norm": 0.853289008140564,
      "learning_rate": 0.00017107390576417126,
      "loss": 0.1017,
      "step": 9489
    },
    {
      "epoch": 1.330063069376314,
      "grad_norm": 0.2919696271419525,
      "learning_rate": 0.0001710595551303516,
      "loss": 0.0579,
      "step": 9490
    },
    {
      "epoch": 1.3302032235459005,
      "grad_norm": 0.28185316920280457,
      "learning_rate": 0.00017104520449653192,
      "loss": 0.056,
      "step": 9491
    },
    {
      "epoch": 1.330343377715487,
      "grad_norm": 0.3327508866786957,
      "learning_rate": 0.00017103085386271227,
      "loss": 0.0525,
      "step": 9492
    },
    {
      "epoch": 1.3304835318850736,
      "grad_norm": 0.5602075457572937,
      "learning_rate": 0.0001710165032288926,
      "loss": 0.0495,
      "step": 9493
    },
    {
      "epoch": 1.33062368605466,
      "grad_norm": 0.7398636341094971,
      "learning_rate": 0.00017100215259507293,
      "loss": 0.1476,
      "step": 9494
    },
    {
      "epoch": 1.3307638402242468,
      "grad_norm": 0.1316189467906952,
      "learning_rate": 0.00017098780196125329,
      "loss": 0.0228,
      "step": 9495
    },
    {
      "epoch": 1.3309039943938332,
      "grad_norm": 0.3175063729286194,
      "learning_rate": 0.00017097345132743361,
      "loss": 0.0711,
      "step": 9496
    },
    {
      "epoch": 1.3310441485634197,
      "grad_norm": 0.24527721107006073,
      "learning_rate": 0.00017095910069361394,
      "loss": 0.0232,
      "step": 9497
    },
    {
      "epoch": 1.3311843027330064,
      "grad_norm": 0.21846212446689606,
      "learning_rate": 0.00017094475005979427,
      "loss": 0.0843,
      "step": 9498
    },
    {
      "epoch": 1.3313244569025928,
      "grad_norm": 0.22878167033195496,
      "learning_rate": 0.00017093039942597463,
      "loss": 0.0572,
      "step": 9499
    },
    {
      "epoch": 1.3314646110721795,
      "grad_norm": 0.2170042097568512,
      "learning_rate": 0.00017091604879215496,
      "loss": 0.05,
      "step": 9500
    },
    {
      "epoch": 1.331604765241766,
      "grad_norm": 0.18254008889198303,
      "learning_rate": 0.0001709016981583353,
      "loss": 0.0362,
      "step": 9501
    },
    {
      "epoch": 1.3317449194113524,
      "grad_norm": 0.2711474299430847,
      "learning_rate": 0.00017088734752451567,
      "loss": 0.0601,
      "step": 9502
    },
    {
      "epoch": 1.331885073580939,
      "grad_norm": 0.4067468047142029,
      "learning_rate": 0.000170872996890696,
      "loss": 0.0947,
      "step": 9503
    },
    {
      "epoch": 1.3320252277505256,
      "grad_norm": 0.3171539306640625,
      "learning_rate": 0.00017085864625687632,
      "loss": 0.0742,
      "step": 9504
    },
    {
      "epoch": 1.3321653819201122,
      "grad_norm": 0.15590724349021912,
      "learning_rate": 0.00017084429562305668,
      "loss": 0.0241,
      "step": 9505
    },
    {
      "epoch": 1.3323055360896987,
      "grad_norm": 0.17302237451076508,
      "learning_rate": 0.000170829944989237,
      "loss": 0.0267,
      "step": 9506
    },
    {
      "epoch": 1.3324456902592852,
      "grad_norm": 0.3936539590358734,
      "learning_rate": 0.00017081559435541734,
      "loss": 0.0611,
      "step": 9507
    },
    {
      "epoch": 1.3325858444288716,
      "grad_norm": 0.22893404960632324,
      "learning_rate": 0.0001708012437215977,
      "loss": 0.0368,
      "step": 9508
    },
    {
      "epoch": 1.3327259985984583,
      "grad_norm": 0.3247547447681427,
      "learning_rate": 0.00017078689308777802,
      "loss": 0.0392,
      "step": 9509
    },
    {
      "epoch": 1.332866152768045,
      "grad_norm": 0.17838488519191742,
      "learning_rate": 0.00017077254245395835,
      "loss": 0.0448,
      "step": 9510
    },
    {
      "epoch": 1.3330063069376314,
      "grad_norm": 0.5286068320274353,
      "learning_rate": 0.00017075819182013873,
      "loss": 0.0499,
      "step": 9511
    },
    {
      "epoch": 1.333146461107218,
      "grad_norm": 0.4125949740409851,
      "learning_rate": 0.00017074384118631906,
      "loss": 0.0862,
      "step": 9512
    },
    {
      "epoch": 1.3332866152768044,
      "grad_norm": 0.3173750340938568,
      "learning_rate": 0.0001707294905524994,
      "loss": 0.0523,
      "step": 9513
    },
    {
      "epoch": 1.333426769446391,
      "grad_norm": 0.48030805587768555,
      "learning_rate": 0.00017071513991867972,
      "loss": 0.0916,
      "step": 9514
    },
    {
      "epoch": 1.3335669236159775,
      "grad_norm": 0.38122883439064026,
      "learning_rate": 0.00017070078928486007,
      "loss": 0.0586,
      "step": 9515
    },
    {
      "epoch": 1.3337070777855642,
      "grad_norm": 0.20957478880882263,
      "learning_rate": 0.0001706864386510404,
      "loss": 0.0224,
      "step": 9516
    },
    {
      "epoch": 1.3338472319551506,
      "grad_norm": 0.27692320942878723,
      "learning_rate": 0.00017067208801722073,
      "loss": 0.038,
      "step": 9517
    },
    {
      "epoch": 1.333987386124737,
      "grad_norm": 0.5911213159561157,
      "learning_rate": 0.0001706577373834011,
      "loss": 0.084,
      "step": 9518
    },
    {
      "epoch": 1.3341275402943238,
      "grad_norm": 0.35676783323287964,
      "learning_rate": 0.00017064338674958142,
      "loss": 0.0397,
      "step": 9519
    },
    {
      "epoch": 1.3342676944639102,
      "grad_norm": 0.17341038584709167,
      "learning_rate": 0.00017062903611576174,
      "loss": 0.0556,
      "step": 9520
    },
    {
      "epoch": 1.334407848633497,
      "grad_norm": 0.302238404750824,
      "learning_rate": 0.00017061468548194213,
      "loss": 0.0491,
      "step": 9521
    },
    {
      "epoch": 1.3345480028030834,
      "grad_norm": 0.12353895604610443,
      "learning_rate": 0.00017060033484812246,
      "loss": 0.0214,
      "step": 9522
    },
    {
      "epoch": 1.3346881569726698,
      "grad_norm": 0.3894589841365814,
      "learning_rate": 0.00017058598421430278,
      "loss": 0.0612,
      "step": 9523
    },
    {
      "epoch": 1.3348283111422565,
      "grad_norm": 0.2525026500225067,
      "learning_rate": 0.00017057163358048314,
      "loss": 0.0859,
      "step": 9524
    },
    {
      "epoch": 1.334968465311843,
      "grad_norm": 0.21902169287204742,
      "learning_rate": 0.00017055728294666347,
      "loss": 0.0526,
      "step": 9525
    },
    {
      "epoch": 1.3351086194814297,
      "grad_norm": 0.5129676461219788,
      "learning_rate": 0.0001705429323128438,
      "loss": 0.0578,
      "step": 9526
    },
    {
      "epoch": 1.3352487736510161,
      "grad_norm": 0.678501546382904,
      "learning_rate": 0.00017052858167902415,
      "loss": 0.1178,
      "step": 9527
    },
    {
      "epoch": 1.3353889278206026,
      "grad_norm": 0.07673218101263046,
      "learning_rate": 0.00017051423104520448,
      "loss": 0.0053,
      "step": 9528
    },
    {
      "epoch": 1.3355290819901893,
      "grad_norm": 1.247165560722351,
      "learning_rate": 0.0001704998804113848,
      "loss": 0.0709,
      "step": 9529
    },
    {
      "epoch": 1.3356692361597757,
      "grad_norm": 0.24425484240055084,
      "learning_rate": 0.00017048552977756517,
      "loss": 0.0269,
      "step": 9530
    },
    {
      "epoch": 1.3358093903293624,
      "grad_norm": 0.6372262239456177,
      "learning_rate": 0.0001704711791437455,
      "loss": 0.0474,
      "step": 9531
    },
    {
      "epoch": 1.3359495444989489,
      "grad_norm": 0.1417616456747055,
      "learning_rate": 0.00017045682850992585,
      "loss": 0.0147,
      "step": 9532
    },
    {
      "epoch": 1.3360896986685353,
      "grad_norm": 1.049975872039795,
      "learning_rate": 0.00017044247787610618,
      "loss": 0.1155,
      "step": 9533
    },
    {
      "epoch": 1.336229852838122,
      "grad_norm": 1.076125979423523,
      "learning_rate": 0.00017042812724228653,
      "loss": 0.0353,
      "step": 9534
    },
    {
      "epoch": 1.3363700070077085,
      "grad_norm": 1.0211845636367798,
      "learning_rate": 0.00017041377660846686,
      "loss": 0.0917,
      "step": 9535
    },
    {
      "epoch": 1.3365101611772952,
      "grad_norm": 0.3120881915092468,
      "learning_rate": 0.0001703994259746472,
      "loss": 0.0389,
      "step": 9536
    },
    {
      "epoch": 1.3366503153468816,
      "grad_norm": 0.3107713460922241,
      "learning_rate": 0.00017038507534082755,
      "loss": 0.027,
      "step": 9537
    },
    {
      "epoch": 1.336790469516468,
      "grad_norm": 0.15866436064243317,
      "learning_rate": 0.00017037072470700787,
      "loss": 0.0428,
      "step": 9538
    },
    {
      "epoch": 1.3369306236860545,
      "grad_norm": 0.23351190984249115,
      "learning_rate": 0.0001703563740731882,
      "loss": 0.0182,
      "step": 9539
    },
    {
      "epoch": 1.3370707778556412,
      "grad_norm": 0.09756828844547272,
      "learning_rate": 0.00017034202343936856,
      "loss": 0.0112,
      "step": 9540
    },
    {
      "epoch": 1.3372109320252277,
      "grad_norm": 0.32933470606803894,
      "learning_rate": 0.0001703276728055489,
      "loss": 0.0468,
      "step": 9541
    },
    {
      "epoch": 1.3373510861948144,
      "grad_norm": 0.2580755650997162,
      "learning_rate": 0.00017031332217172922,
      "loss": 0.0423,
      "step": 9542
    },
    {
      "epoch": 1.3374912403644008,
      "grad_norm": 0.26410555839538574,
      "learning_rate": 0.0001702989715379096,
      "loss": 0.0472,
      "step": 9543
    },
    {
      "epoch": 1.3376313945339873,
      "grad_norm": 0.2901782989501953,
      "learning_rate": 0.00017028462090408993,
      "loss": 0.0623,
      "step": 9544
    },
    {
      "epoch": 1.337771548703574,
      "grad_norm": 0.4845621585845947,
      "learning_rate": 0.00017027027027027026,
      "loss": 0.0593,
      "step": 9545
    },
    {
      "epoch": 1.3379117028731604,
      "grad_norm": 0.3676447868347168,
      "learning_rate": 0.0001702559196364506,
      "loss": 0.0767,
      "step": 9546
    },
    {
      "epoch": 1.338051857042747,
      "grad_norm": 0.1000380590558052,
      "learning_rate": 0.00017024156900263094,
      "loss": 0.0067,
      "step": 9547
    },
    {
      "epoch": 1.3381920112123336,
      "grad_norm": 0.11539237946271896,
      "learning_rate": 0.00017022721836881127,
      "loss": 0.0357,
      "step": 9548
    },
    {
      "epoch": 1.33833216538192,
      "grad_norm": 0.22616511583328247,
      "learning_rate": 0.0001702128677349916,
      "loss": 0.0238,
      "step": 9549
    },
    {
      "epoch": 1.3384723195515067,
      "grad_norm": 0.2731891870498657,
      "learning_rate": 0.00017019851710117195,
      "loss": 0.0368,
      "step": 9550
    },
    {
      "epoch": 1.3386124737210932,
      "grad_norm": 0.42878878116607666,
      "learning_rate": 0.00017018416646735228,
      "loss": 0.0822,
      "step": 9551
    },
    {
      "epoch": 1.3387526278906798,
      "grad_norm": 0.2010854333639145,
      "learning_rate": 0.0001701698158335326,
      "loss": 0.0476,
      "step": 9552
    },
    {
      "epoch": 1.3388927820602663,
      "grad_norm": 0.32286015152931213,
      "learning_rate": 0.000170155465199713,
      "loss": 0.0668,
      "step": 9553
    },
    {
      "epoch": 1.3390329362298528,
      "grad_norm": 0.33523309230804443,
      "learning_rate": 0.00017014111456589332,
      "loss": 0.0572,
      "step": 9554
    },
    {
      "epoch": 1.3391730903994394,
      "grad_norm": 0.22790703177452087,
      "learning_rate": 0.00017012676393207365,
      "loss": 0.0397,
      "step": 9555
    },
    {
      "epoch": 1.339313244569026,
      "grad_norm": 0.42198801040649414,
      "learning_rate": 0.000170112413298254,
      "loss": 0.0962,
      "step": 9556
    },
    {
      "epoch": 1.3394533987386126,
      "grad_norm": 0.22848308086395264,
      "learning_rate": 0.00017009806266443433,
      "loss": 0.0249,
      "step": 9557
    },
    {
      "epoch": 1.339593552908199,
      "grad_norm": 0.348146915435791,
      "learning_rate": 0.00017008371203061466,
      "loss": 0.0558,
      "step": 9558
    },
    {
      "epoch": 1.3397337070777855,
      "grad_norm": 0.558443009853363,
      "learning_rate": 0.00017006936139679502,
      "loss": 0.0966,
      "step": 9559
    },
    {
      "epoch": 1.3398738612473722,
      "grad_norm": 0.5223244428634644,
      "learning_rate": 0.00017005501076297535,
      "loss": 0.0451,
      "step": 9560
    },
    {
      "epoch": 1.3400140154169586,
      "grad_norm": 0.6537792086601257,
      "learning_rate": 0.00017004066012915568,
      "loss": 0.0655,
      "step": 9561
    },
    {
      "epoch": 1.3401541695865453,
      "grad_norm": 0.1561972051858902,
      "learning_rate": 0.00017002630949533603,
      "loss": 0.0194,
      "step": 9562
    },
    {
      "epoch": 1.3402943237561318,
      "grad_norm": 0.4724835753440857,
      "learning_rate": 0.00017001195886151636,
      "loss": 0.095,
      "step": 9563
    },
    {
      "epoch": 1.3404344779257182,
      "grad_norm": 0.3880367577075958,
      "learning_rate": 0.00016999760822769672,
      "loss": 0.0566,
      "step": 9564
    },
    {
      "epoch": 1.3405746320953047,
      "grad_norm": 0.1931467354297638,
      "learning_rate": 0.00016998325759387704,
      "loss": 0.0285,
      "step": 9565
    },
    {
      "epoch": 1.3407147862648914,
      "grad_norm": 0.745561420917511,
      "learning_rate": 0.0001699689069600574,
      "loss": 0.0736,
      "step": 9566
    },
    {
      "epoch": 1.340854940434478,
      "grad_norm": 0.14206565916538239,
      "learning_rate": 0.00016995455632623773,
      "loss": 0.0353,
      "step": 9567
    },
    {
      "epoch": 1.3409950946040645,
      "grad_norm": 0.6110050678253174,
      "learning_rate": 0.00016994020569241806,
      "loss": 0.1482,
      "step": 9568
    },
    {
      "epoch": 1.341135248773651,
      "grad_norm": 1.2706602811813354,
      "learning_rate": 0.0001699258550585984,
      "loss": 0.1146,
      "step": 9569
    },
    {
      "epoch": 1.3412754029432374,
      "grad_norm": 0.3216162323951721,
      "learning_rate": 0.00016991150442477874,
      "loss": 0.0456,
      "step": 9570
    },
    {
      "epoch": 1.3414155571128241,
      "grad_norm": 0.2087244987487793,
      "learning_rate": 0.00016989715379095907,
      "loss": 0.0293,
      "step": 9571
    },
    {
      "epoch": 1.3415557112824106,
      "grad_norm": 0.4775845408439636,
      "learning_rate": 0.00016988280315713943,
      "loss": 0.1145,
      "step": 9572
    },
    {
      "epoch": 1.3416958654519973,
      "grad_norm": 0.24230755865573883,
      "learning_rate": 0.00016986845252331975,
      "loss": 0.0869,
      "step": 9573
    },
    {
      "epoch": 1.3418360196215837,
      "grad_norm": 0.12217220664024353,
      "learning_rate": 0.00016985410188950008,
      "loss": 0.018,
      "step": 9574
    },
    {
      "epoch": 1.3419761737911702,
      "grad_norm": 0.2012275755405426,
      "learning_rate": 0.00016983975125568047,
      "loss": 0.072,
      "step": 9575
    },
    {
      "epoch": 1.3421163279607569,
      "grad_norm": 0.2923319935798645,
      "learning_rate": 0.0001698254006218608,
      "loss": 0.0555,
      "step": 9576
    },
    {
      "epoch": 1.3422564821303433,
      "grad_norm": 0.5228692293167114,
      "learning_rate": 0.00016981104998804112,
      "loss": 0.0849,
      "step": 9577
    },
    {
      "epoch": 1.34239663629993,
      "grad_norm": 0.3818853795528412,
      "learning_rate": 0.00016979669935422148,
      "loss": 0.0816,
      "step": 9578
    },
    {
      "epoch": 1.3425367904695165,
      "grad_norm": 0.9247800707817078,
      "learning_rate": 0.0001697823487204018,
      "loss": 0.23,
      "step": 9579
    },
    {
      "epoch": 1.342676944639103,
      "grad_norm": 0.39113649725914,
      "learning_rate": 0.00016976799808658214,
      "loss": 0.0807,
      "step": 9580
    },
    {
      "epoch": 1.3428170988086896,
      "grad_norm": 1.0975922346115112,
      "learning_rate": 0.0001697536474527625,
      "loss": 0.0949,
      "step": 9581
    },
    {
      "epoch": 1.342957252978276,
      "grad_norm": 0.3754643499851227,
      "learning_rate": 0.00016973929681894282,
      "loss": 0.1034,
      "step": 9582
    },
    {
      "epoch": 1.3430974071478627,
      "grad_norm": 0.7964534163475037,
      "learning_rate": 0.00016972494618512315,
      "loss": 0.1745,
      "step": 9583
    },
    {
      "epoch": 1.3432375613174492,
      "grad_norm": 2.2981514930725098,
      "learning_rate": 0.00016971059555130348,
      "loss": 0.069,
      "step": 9584
    },
    {
      "epoch": 1.3433777154870357,
      "grad_norm": 0.1188306137919426,
      "learning_rate": 0.00016969624491748386,
      "loss": 0.011,
      "step": 9585
    },
    {
      "epoch": 1.3435178696566223,
      "grad_norm": 0.32813698053359985,
      "learning_rate": 0.0001696818942836642,
      "loss": 0.0672,
      "step": 9586
    },
    {
      "epoch": 1.3436580238262088,
      "grad_norm": 0.20178063213825226,
      "learning_rate": 0.00016966754364984452,
      "loss": 0.0295,
      "step": 9587
    },
    {
      "epoch": 1.3437981779957955,
      "grad_norm": 0.38333839178085327,
      "learning_rate": 0.00016965319301602487,
      "loss": 0.0726,
      "step": 9588
    },
    {
      "epoch": 1.343938332165382,
      "grad_norm": 0.3554108738899231,
      "learning_rate": 0.0001696388423822052,
      "loss": 0.1224,
      "step": 9589
    },
    {
      "epoch": 1.3440784863349684,
      "grad_norm": 0.3435667157173157,
      "learning_rate": 0.00016962449174838553,
      "loss": 0.1305,
      "step": 9590
    },
    {
      "epoch": 1.344218640504555,
      "grad_norm": 0.20762506127357483,
      "learning_rate": 0.00016961014111456588,
      "loss": 0.0471,
      "step": 9591
    },
    {
      "epoch": 1.3443587946741415,
      "grad_norm": 0.30831727385520935,
      "learning_rate": 0.0001695957904807462,
      "loss": 0.1059,
      "step": 9592
    },
    {
      "epoch": 1.3444989488437282,
      "grad_norm": 0.15234048664569855,
      "learning_rate": 0.00016958143984692654,
      "loss": 0.0501,
      "step": 9593
    },
    {
      "epoch": 1.3446391030133147,
      "grad_norm": 0.24389781057834625,
      "learning_rate": 0.0001695670892131069,
      "loss": 0.0305,
      "step": 9594
    },
    {
      "epoch": 1.3447792571829011,
      "grad_norm": 0.5057260990142822,
      "learning_rate": 0.00016955273857928723,
      "loss": 0.1259,
      "step": 9595
    },
    {
      "epoch": 1.3449194113524876,
      "grad_norm": 0.1339229792356491,
      "learning_rate": 0.00016953838794546758,
      "loss": 0.0202,
      "step": 9596
    },
    {
      "epoch": 1.3450595655220743,
      "grad_norm": 0.40921786427497864,
      "learning_rate": 0.00016952403731164794,
      "loss": 0.0575,
      "step": 9597
    },
    {
      "epoch": 1.3451997196916607,
      "grad_norm": 0.17376172542572021,
      "learning_rate": 0.00016950968667782827,
      "loss": 0.0258,
      "step": 9598
    },
    {
      "epoch": 1.3453398738612474,
      "grad_norm": 0.6082354187965393,
      "learning_rate": 0.0001694953360440086,
      "loss": 0.0709,
      "step": 9599
    },
    {
      "epoch": 1.3454800280308339,
      "grad_norm": 0.22580136358737946,
      "learning_rate": 0.00016948098541018892,
      "loss": 0.0528,
      "step": 9600
    },
    {
      "epoch": 1.3456201822004203,
      "grad_norm": 0.27200090885162354,
      "learning_rate": 0.00016946663477636928,
      "loss": 0.0267,
      "step": 9601
    },
    {
      "epoch": 1.345760336370007,
      "grad_norm": 0.16164040565490723,
      "learning_rate": 0.0001694522841425496,
      "loss": 0.023,
      "step": 9602
    },
    {
      "epoch": 1.3459004905395935,
      "grad_norm": 0.1931738257408142,
      "learning_rate": 0.00016943793350872994,
      "loss": 0.0335,
      "step": 9603
    },
    {
      "epoch": 1.3460406447091802,
      "grad_norm": 0.36990222334861755,
      "learning_rate": 0.0001694235828749103,
      "loss": 0.0676,
      "step": 9604
    },
    {
      "epoch": 1.3461807988787666,
      "grad_norm": 0.24738983809947968,
      "learning_rate": 0.00016940923224109062,
      "loss": 0.0445,
      "step": 9605
    },
    {
      "epoch": 1.346320953048353,
      "grad_norm": 0.24724404513835907,
      "learning_rate": 0.00016939488160727095,
      "loss": 0.0306,
      "step": 9606
    },
    {
      "epoch": 1.3464611072179398,
      "grad_norm": 0.5514512062072754,
      "learning_rate": 0.00016938053097345133,
      "loss": 0.065,
      "step": 9607
    },
    {
      "epoch": 1.3466012613875262,
      "grad_norm": 0.5415422916412354,
      "learning_rate": 0.00016936618033963166,
      "loss": 0.1003,
      "step": 9608
    },
    {
      "epoch": 1.346741415557113,
      "grad_norm": 0.5476313829421997,
      "learning_rate": 0.000169351829705812,
      "loss": 0.0622,
      "step": 9609
    },
    {
      "epoch": 1.3468815697266994,
      "grad_norm": 1.0372077226638794,
      "learning_rate": 0.00016933747907199234,
      "loss": 0.1601,
      "step": 9610
    },
    {
      "epoch": 1.3470217238962858,
      "grad_norm": 0.5090930461883545,
      "learning_rate": 0.00016932312843817267,
      "loss": 0.0855,
      "step": 9611
    },
    {
      "epoch": 1.3471618780658725,
      "grad_norm": 0.5956155061721802,
      "learning_rate": 0.000169308777804353,
      "loss": 0.1916,
      "step": 9612
    },
    {
      "epoch": 1.347302032235459,
      "grad_norm": 0.3450208604335785,
      "learning_rate": 0.00016929442717053336,
      "loss": 0.0636,
      "step": 9613
    },
    {
      "epoch": 1.3474421864050456,
      "grad_norm": 0.4558791518211365,
      "learning_rate": 0.00016928007653671369,
      "loss": 0.0869,
      "step": 9614
    },
    {
      "epoch": 1.347582340574632,
      "grad_norm": 0.0827002078294754,
      "learning_rate": 0.00016926572590289401,
      "loss": 0.0131,
      "step": 9615
    },
    {
      "epoch": 1.3477224947442186,
      "grad_norm": 0.16938596963882446,
      "learning_rate": 0.0001692513752690744,
      "loss": 0.0172,
      "step": 9616
    },
    {
      "epoch": 1.3478626489138053,
      "grad_norm": 0.1471003144979477,
      "learning_rate": 0.00016923702463525473,
      "loss": 0.0142,
      "step": 9617
    },
    {
      "epoch": 1.3480028030833917,
      "grad_norm": 0.2193594127893448,
      "learning_rate": 0.00016922267400143505,
      "loss": 0.0533,
      "step": 9618
    },
    {
      "epoch": 1.3481429572529784,
      "grad_norm": 0.5552753210067749,
      "learning_rate": 0.00016920832336761538,
      "loss": 0.041,
      "step": 9619
    },
    {
      "epoch": 1.3482831114225649,
      "grad_norm": 1.0127137899398804,
      "learning_rate": 0.00016919397273379574,
      "loss": 0.1012,
      "step": 9620
    },
    {
      "epoch": 1.3484232655921513,
      "grad_norm": 0.7789947986602783,
      "learning_rate": 0.00016917962209997607,
      "loss": 0.0374,
      "step": 9621
    },
    {
      "epoch": 1.348563419761738,
      "grad_norm": 0.48137718439102173,
      "learning_rate": 0.0001691652714661564,
      "loss": 0.0399,
      "step": 9622
    },
    {
      "epoch": 1.3487035739313245,
      "grad_norm": 0.2113160938024521,
      "learning_rate": 0.00016915092083233675,
      "loss": 0.0217,
      "step": 9623
    },
    {
      "epoch": 1.3488437281009111,
      "grad_norm": 0.3934256434440613,
      "learning_rate": 0.00016913657019851708,
      "loss": 0.0178,
      "step": 9624
    },
    {
      "epoch": 1.3489838822704976,
      "grad_norm": 0.6783877015113831,
      "learning_rate": 0.0001691222195646974,
      "loss": 0.06,
      "step": 9625
    },
    {
      "epoch": 1.349124036440084,
      "grad_norm": 0.24311049282550812,
      "learning_rate": 0.00016910786893087776,
      "loss": 0.105,
      "step": 9626
    },
    {
      "epoch": 1.3492641906096705,
      "grad_norm": 0.6261190176010132,
      "learning_rate": 0.0001690935182970581,
      "loss": 0.0486,
      "step": 9627
    },
    {
      "epoch": 1.3494043447792572,
      "grad_norm": 0.37981170415878296,
      "learning_rate": 0.00016907916766323845,
      "loss": 0.1636,
      "step": 9628
    },
    {
      "epoch": 1.3495444989488437,
      "grad_norm": 0.5377947688102722,
      "learning_rate": 0.0001690648170294188,
      "loss": 0.1592,
      "step": 9629
    },
    {
      "epoch": 1.3496846531184303,
      "grad_norm": 0.5082780122756958,
      "learning_rate": 0.00016905046639559913,
      "loss": 0.1175,
      "step": 9630
    },
    {
      "epoch": 1.3498248072880168,
      "grad_norm": 0.33813199400901794,
      "learning_rate": 0.00016903611576177946,
      "loss": 0.0213,
      "step": 9631
    },
    {
      "epoch": 1.3499649614576033,
      "grad_norm": 0.9066349864006042,
      "learning_rate": 0.00016902176512795982,
      "loss": 0.1986,
      "step": 9632
    },
    {
      "epoch": 1.35010511562719,
      "grad_norm": 0.3174242377281189,
      "learning_rate": 0.00016900741449414015,
      "loss": 0.0225,
      "step": 9633
    },
    {
      "epoch": 1.3502452697967764,
      "grad_norm": 1.0678372383117676,
      "learning_rate": 0.00016899306386032047,
      "loss": 0.0772,
      "step": 9634
    },
    {
      "epoch": 1.350385423966363,
      "grad_norm": 1.488322138786316,
      "learning_rate": 0.0001689787132265008,
      "loss": 0.0465,
      "step": 9635
    },
    {
      "epoch": 1.3505255781359495,
      "grad_norm": 0.46247369050979614,
      "learning_rate": 0.00016896436259268116,
      "loss": 0.1368,
      "step": 9636
    },
    {
      "epoch": 1.350665732305536,
      "grad_norm": 0.4396011233329773,
      "learning_rate": 0.0001689500119588615,
      "loss": 0.1015,
      "step": 9637
    },
    {
      "epoch": 1.3508058864751227,
      "grad_norm": 0.6034500598907471,
      "learning_rate": 0.00016893566132504182,
      "loss": 0.1036,
      "step": 9638
    },
    {
      "epoch": 1.3509460406447091,
      "grad_norm": 0.15007194876670837,
      "learning_rate": 0.0001689213106912222,
      "loss": 0.0313,
      "step": 9639
    },
    {
      "epoch": 1.3510861948142958,
      "grad_norm": 0.39873605966567993,
      "learning_rate": 0.00016890696005740253,
      "loss": 0.0894,
      "step": 9640
    },
    {
      "epoch": 1.3512263489838823,
      "grad_norm": 0.3135083019733429,
      "learning_rate": 0.00016889260942358285,
      "loss": 0.0641,
      "step": 9641
    },
    {
      "epoch": 1.3513665031534687,
      "grad_norm": 0.7267391681671143,
      "learning_rate": 0.0001688782587897632,
      "loss": 0.1094,
      "step": 9642
    },
    {
      "epoch": 1.3515066573230554,
      "grad_norm": 1.3700371980667114,
      "learning_rate": 0.00016886390815594354,
      "loss": 0.0352,
      "step": 9643
    },
    {
      "epoch": 1.3516468114926419,
      "grad_norm": 0.3135605454444885,
      "learning_rate": 0.00016884955752212387,
      "loss": 0.0435,
      "step": 9644
    },
    {
      "epoch": 1.3517869656622286,
      "grad_norm": 0.3437100946903229,
      "learning_rate": 0.00016883520688830422,
      "loss": 0.0539,
      "step": 9645
    },
    {
      "epoch": 1.351927119831815,
      "grad_norm": 0.4274687170982361,
      "learning_rate": 0.00016882085625448455,
      "loss": 0.0755,
      "step": 9646
    },
    {
      "epoch": 1.3520672740014015,
      "grad_norm": 0.5987633466720581,
      "learning_rate": 0.00016880650562066488,
      "loss": 0.0565,
      "step": 9647
    },
    {
      "epoch": 1.3522074281709882,
      "grad_norm": 0.104233518242836,
      "learning_rate": 0.00016879215498684526,
      "loss": 0.0149,
      "step": 9648
    },
    {
      "epoch": 1.3523475823405746,
      "grad_norm": 0.2899114787578583,
      "learning_rate": 0.0001687778043530256,
      "loss": 0.0314,
      "step": 9649
    },
    {
      "epoch": 1.3524877365101613,
      "grad_norm": 0.19484783709049225,
      "learning_rate": 0.00016876345371920592,
      "loss": 0.0443,
      "step": 9650
    },
    {
      "epoch": 1.3526278906797478,
      "grad_norm": 0.28061410784721375,
      "learning_rate": 0.00016874910308538628,
      "loss": 0.0706,
      "step": 9651
    },
    {
      "epoch": 1.3527680448493342,
      "grad_norm": 0.8848469257354736,
      "learning_rate": 0.0001687347524515666,
      "loss": 0.0875,
      "step": 9652
    },
    {
      "epoch": 1.3529081990189207,
      "grad_norm": 0.37647897005081177,
      "learning_rate": 0.00016872040181774693,
      "loss": 0.0836,
      "step": 9653
    },
    {
      "epoch": 1.3530483531885074,
      "grad_norm": 0.5196483731269836,
      "learning_rate": 0.00016870605118392726,
      "loss": 0.0894,
      "step": 9654
    },
    {
      "epoch": 1.353188507358094,
      "grad_norm": 0.14152616262435913,
      "learning_rate": 0.00016869170055010762,
      "loss": 0.0191,
      "step": 9655
    },
    {
      "epoch": 1.3533286615276805,
      "grad_norm": 0.1502651423215866,
      "learning_rate": 0.00016867734991628795,
      "loss": 0.0092,
      "step": 9656
    },
    {
      "epoch": 1.353468815697267,
      "grad_norm": 0.7941946387290955,
      "learning_rate": 0.00016866299928246827,
      "loss": 0.0511,
      "step": 9657
    },
    {
      "epoch": 1.3536089698668534,
      "grad_norm": 0.30657556653022766,
      "learning_rate": 0.00016864864864864863,
      "loss": 0.0404,
      "step": 9658
    },
    {
      "epoch": 1.35374912403644,
      "grad_norm": 0.717262864112854,
      "learning_rate": 0.00016863429801482899,
      "loss": 0.0424,
      "step": 9659
    },
    {
      "epoch": 1.3538892782060266,
      "grad_norm": 0.18987800180912018,
      "learning_rate": 0.00016861994738100931,
      "loss": 0.0432,
      "step": 9660
    },
    {
      "epoch": 1.3540294323756132,
      "grad_norm": 0.27932873368263245,
      "learning_rate": 0.00016860559674718967,
      "loss": 0.0353,
      "step": 9661
    },
    {
      "epoch": 1.3541695865451997,
      "grad_norm": 0.5111303925514221,
      "learning_rate": 0.00016859124611337,
      "loss": 0.0534,
      "step": 9662
    },
    {
      "epoch": 1.3543097407147862,
      "grad_norm": 0.6268123388290405,
      "learning_rate": 0.00016857689547955033,
      "loss": 0.0836,
      "step": 9663
    },
    {
      "epoch": 1.3544498948843728,
      "grad_norm": 0.6795238256454468,
      "learning_rate": 0.00016856254484573068,
      "loss": 0.1216,
      "step": 9664
    },
    {
      "epoch": 1.3545900490539593,
      "grad_norm": 0.2711315453052521,
      "learning_rate": 0.000168548194211911,
      "loss": 0.0408,
      "step": 9665
    },
    {
      "epoch": 1.354730203223546,
      "grad_norm": 0.24131660163402557,
      "learning_rate": 0.00016853384357809134,
      "loss": 0.0272,
      "step": 9666
    },
    {
      "epoch": 1.3548703573931324,
      "grad_norm": 0.3637286424636841,
      "learning_rate": 0.0001685194929442717,
      "loss": 0.0411,
      "step": 9667
    },
    {
      "epoch": 1.355010511562719,
      "grad_norm": 0.06975194811820984,
      "learning_rate": 0.00016850514231045202,
      "loss": 0.0122,
      "step": 9668
    },
    {
      "epoch": 1.3551506657323056,
      "grad_norm": 0.42182624340057373,
      "learning_rate": 0.00016849079167663235,
      "loss": 0.0614,
      "step": 9669
    },
    {
      "epoch": 1.355290819901892,
      "grad_norm": 0.4486479163169861,
      "learning_rate": 0.00016847644104281268,
      "loss": 0.032,
      "step": 9670
    },
    {
      "epoch": 1.3554309740714787,
      "grad_norm": 1.331338882446289,
      "learning_rate": 0.00016846209040899306,
      "loss": 0.0871,
      "step": 9671
    },
    {
      "epoch": 1.3555711282410652,
      "grad_norm": 0.5530375242233276,
      "learning_rate": 0.0001684477397751734,
      "loss": 0.1335,
      "step": 9672
    },
    {
      "epoch": 1.3557112824106516,
      "grad_norm": 0.4682348966598511,
      "learning_rate": 0.00016843338914135372,
      "loss": 0.0687,
      "step": 9673
    },
    {
      "epoch": 1.3558514365802383,
      "grad_norm": 0.11995410174131393,
      "learning_rate": 0.00016841903850753408,
      "loss": 0.0102,
      "step": 9674
    },
    {
      "epoch": 1.3559915907498248,
      "grad_norm": 0.25365790724754333,
      "learning_rate": 0.0001684046878737144,
      "loss": 0.0423,
      "step": 9675
    },
    {
      "epoch": 1.3561317449194115,
      "grad_norm": 0.20284157991409302,
      "learning_rate": 0.00016839033723989473,
      "loss": 0.0325,
      "step": 9676
    },
    {
      "epoch": 1.356271899088998,
      "grad_norm": 0.280926913022995,
      "learning_rate": 0.0001683759866060751,
      "loss": 0.06,
      "step": 9677
    },
    {
      "epoch": 1.3564120532585844,
      "grad_norm": 0.10474584251642227,
      "learning_rate": 0.00016836163597225542,
      "loss": 0.0074,
      "step": 9678
    },
    {
      "epoch": 1.356552207428171,
      "grad_norm": 0.2583116888999939,
      "learning_rate": 0.00016834728533843575,
      "loss": 0.0372,
      "step": 9679
    },
    {
      "epoch": 1.3566923615977575,
      "grad_norm": 0.2200823277235031,
      "learning_rate": 0.00016833293470461613,
      "loss": 0.0332,
      "step": 9680
    },
    {
      "epoch": 1.3568325157673442,
      "grad_norm": 0.45323115587234497,
      "learning_rate": 0.00016831858407079646,
      "loss": 0.0572,
      "step": 9681
    },
    {
      "epoch": 1.3569726699369307,
      "grad_norm": 0.6593964099884033,
      "learning_rate": 0.0001683042334369768,
      "loss": 0.0793,
      "step": 9682
    },
    {
      "epoch": 1.3571128241065171,
      "grad_norm": 0.8121048808097839,
      "learning_rate": 0.00016828988280315714,
      "loss": 0.0619,
      "step": 9683
    },
    {
      "epoch": 1.3572529782761036,
      "grad_norm": 0.3453994393348694,
      "learning_rate": 0.00016827553216933747,
      "loss": 0.0287,
      "step": 9684
    },
    {
      "epoch": 1.3573931324456903,
      "grad_norm": 0.7168417572975159,
      "learning_rate": 0.0001682611815355178,
      "loss": 0.0737,
      "step": 9685
    },
    {
      "epoch": 1.3575332866152767,
      "grad_norm": 0.31652066111564636,
      "learning_rate": 0.00016824683090169816,
      "loss": 0.0318,
      "step": 9686
    },
    {
      "epoch": 1.3576734407848634,
      "grad_norm": 0.19063116610050201,
      "learning_rate": 0.00016823248026787848,
      "loss": 0.0337,
      "step": 9687
    },
    {
      "epoch": 1.3578135949544499,
      "grad_norm": 0.17053866386413574,
      "learning_rate": 0.0001682181296340588,
      "loss": 0.0387,
      "step": 9688
    },
    {
      "epoch": 1.3579537491240363,
      "grad_norm": 0.41927045583724976,
      "learning_rate": 0.00016820377900023914,
      "loss": 0.0469,
      "step": 9689
    },
    {
      "epoch": 1.358093903293623,
      "grad_norm": 0.29581496119499207,
      "learning_rate": 0.0001681894283664195,
      "loss": 0.0751,
      "step": 9690
    },
    {
      "epoch": 1.3582340574632095,
      "grad_norm": 0.3370482623577118,
      "learning_rate": 0.00016817507773259985,
      "loss": 0.0487,
      "step": 9691
    },
    {
      "epoch": 1.3583742116327961,
      "grad_norm": 0.4582037627696991,
      "learning_rate": 0.00016816072709878018,
      "loss": 0.0708,
      "step": 9692
    },
    {
      "epoch": 1.3585143658023826,
      "grad_norm": 0.5835514664649963,
      "learning_rate": 0.00016814637646496054,
      "loss": 0.0346,
      "step": 9693
    },
    {
      "epoch": 1.358654519971969,
      "grad_norm": 0.6042092442512512,
      "learning_rate": 0.00016813202583114086,
      "loss": 0.0285,
      "step": 9694
    },
    {
      "epoch": 1.3587946741415557,
      "grad_norm": 0.11981189250946045,
      "learning_rate": 0.0001681176751973212,
      "loss": 0.0159,
      "step": 9695
    },
    {
      "epoch": 1.3589348283111422,
      "grad_norm": 0.591228723526001,
      "learning_rate": 0.00016810332456350155,
      "loss": 0.0609,
      "step": 9696
    },
    {
      "epoch": 1.3590749824807289,
      "grad_norm": 0.13521863520145416,
      "learning_rate": 0.00016808897392968188,
      "loss": 0.0286,
      "step": 9697
    },
    {
      "epoch": 1.3592151366503153,
      "grad_norm": 0.1536429226398468,
      "learning_rate": 0.0001680746232958622,
      "loss": 0.0376,
      "step": 9698
    },
    {
      "epoch": 1.3593552908199018,
      "grad_norm": 0.2578083872795105,
      "learning_rate": 0.00016806027266204256,
      "loss": 0.0647,
      "step": 9699
    },
    {
      "epoch": 1.3594954449894885,
      "grad_norm": 0.3452412486076355,
      "learning_rate": 0.0001680459220282229,
      "loss": 0.0451,
      "step": 9700
    },
    {
      "epoch": 1.359635599159075,
      "grad_norm": 0.7062780261039734,
      "learning_rate": 0.00016803157139440322,
      "loss": 0.1317,
      "step": 9701
    },
    {
      "epoch": 1.3597757533286616,
      "grad_norm": 0.17084822058677673,
      "learning_rate": 0.0001680172207605836,
      "loss": 0.0431,
      "step": 9702
    },
    {
      "epoch": 1.359915907498248,
      "grad_norm": 0.4236963391304016,
      "learning_rate": 0.00016800287012676393,
      "loss": 0.0515,
      "step": 9703
    },
    {
      "epoch": 1.3600560616678345,
      "grad_norm": 0.21950465440750122,
      "learning_rate": 0.00016798851949294426,
      "loss": 0.0528,
      "step": 9704
    },
    {
      "epoch": 1.3601962158374212,
      "grad_norm": 0.27367982268333435,
      "learning_rate": 0.0001679741688591246,
      "loss": 0.0519,
      "step": 9705
    },
    {
      "epoch": 1.3603363700070077,
      "grad_norm": 0.15314653515815735,
      "learning_rate": 0.00016795981822530494,
      "loss": 0.0347,
      "step": 9706
    },
    {
      "epoch": 1.3604765241765944,
      "grad_norm": 0.22668534517288208,
      "learning_rate": 0.00016794546759148527,
      "loss": 0.0152,
      "step": 9707
    },
    {
      "epoch": 1.3606166783461808,
      "grad_norm": 0.35102778673171997,
      "learning_rate": 0.0001679311169576656,
      "loss": 0.0439,
      "step": 9708
    },
    {
      "epoch": 1.3607568325157673,
      "grad_norm": 0.35799992084503174,
      "learning_rate": 0.00016791676632384596,
      "loss": 0.0365,
      "step": 9709
    },
    {
      "epoch": 1.360896986685354,
      "grad_norm": 0.5184187889099121,
      "learning_rate": 0.00016790241569002628,
      "loss": 0.0266,
      "step": 9710
    },
    {
      "epoch": 1.3610371408549404,
      "grad_norm": 0.36529093980789185,
      "learning_rate": 0.0001678880650562066,
      "loss": 0.0603,
      "step": 9711
    },
    {
      "epoch": 1.3611772950245271,
      "grad_norm": 0.38146328926086426,
      "learning_rate": 0.000167873714422387,
      "loss": 0.0575,
      "step": 9712
    },
    {
      "epoch": 1.3613174491941136,
      "grad_norm": 0.24646399915218353,
      "learning_rate": 0.00016785936378856732,
      "loss": 0.0218,
      "step": 9713
    },
    {
      "epoch": 1.3614576033637,
      "grad_norm": 0.35222500562667847,
      "learning_rate": 0.00016784501315474765,
      "loss": 0.0281,
      "step": 9714
    },
    {
      "epoch": 1.3615977575332865,
      "grad_norm": 0.22733935713768005,
      "learning_rate": 0.000167830662520928,
      "loss": 0.0369,
      "step": 9715
    },
    {
      "epoch": 1.3617379117028732,
      "grad_norm": 0.1621370017528534,
      "learning_rate": 0.00016781631188710834,
      "loss": 0.018,
      "step": 9716
    },
    {
      "epoch": 1.3618780658724596,
      "grad_norm": 0.38500919938087463,
      "learning_rate": 0.00016780196125328867,
      "loss": 0.0183,
      "step": 9717
    },
    {
      "epoch": 1.3620182200420463,
      "grad_norm": 0.2516532242298126,
      "learning_rate": 0.00016778761061946902,
      "loss": 0.0363,
      "step": 9718
    },
    {
      "epoch": 1.3621583742116328,
      "grad_norm": 0.2872370779514313,
      "learning_rate": 0.00016777325998564935,
      "loss": 0.0326,
      "step": 9719
    },
    {
      "epoch": 1.3622985283812192,
      "grad_norm": 0.1611098349094391,
      "learning_rate": 0.00016775890935182968,
      "loss": 0.0179,
      "step": 9720
    },
    {
      "epoch": 1.362438682550806,
      "grad_norm": 0.28133001923561096,
      "learning_rate": 0.00016774455871801003,
      "loss": 0.051,
      "step": 9721
    },
    {
      "epoch": 1.3625788367203924,
      "grad_norm": 0.29716742038726807,
      "learning_rate": 0.00016773020808419036,
      "loss": 0.0278,
      "step": 9722
    },
    {
      "epoch": 1.362718990889979,
      "grad_norm": 0.43527963757514954,
      "learning_rate": 0.00016771585745037072,
      "loss": 0.1277,
      "step": 9723
    },
    {
      "epoch": 1.3628591450595655,
      "grad_norm": 0.6623085141181946,
      "learning_rate": 0.00016770150681655105,
      "loss": 0.045,
      "step": 9724
    },
    {
      "epoch": 1.362999299229152,
      "grad_norm": 0.13986140489578247,
      "learning_rate": 0.0001676871561827314,
      "loss": 0.0301,
      "step": 9725
    },
    {
      "epoch": 1.3631394533987387,
      "grad_norm": 0.34659212827682495,
      "learning_rate": 0.00016767280554891173,
      "loss": 0.0346,
      "step": 9726
    },
    {
      "epoch": 1.3632796075683251,
      "grad_norm": 0.5105130672454834,
      "learning_rate": 0.00016765845491509206,
      "loss": 0.132,
      "step": 9727
    },
    {
      "epoch": 1.3634197617379118,
      "grad_norm": 0.4852428138256073,
      "learning_rate": 0.00016764410428127242,
      "loss": 0.0433,
      "step": 9728
    },
    {
      "epoch": 1.3635599159074983,
      "grad_norm": 0.8634260296821594,
      "learning_rate": 0.00016762975364745274,
      "loss": 0.0182,
      "step": 9729
    },
    {
      "epoch": 1.3637000700770847,
      "grad_norm": 0.5627794861793518,
      "learning_rate": 0.00016761540301363307,
      "loss": 0.0501,
      "step": 9730
    },
    {
      "epoch": 1.3638402242466714,
      "grad_norm": 0.3760053217411041,
      "learning_rate": 0.00016760105237981343,
      "loss": 0.0509,
      "step": 9731
    },
    {
      "epoch": 1.3639803784162579,
      "grad_norm": 4.839693069458008,
      "learning_rate": 0.00016758670174599376,
      "loss": 0.0624,
      "step": 9732
    },
    {
      "epoch": 1.3641205325858445,
      "grad_norm": 1.5461735725402832,
      "learning_rate": 0.00016757235111217409,
      "loss": 0.109,
      "step": 9733
    },
    {
      "epoch": 1.364260686755431,
      "grad_norm": 2.0265204906463623,
      "learning_rate": 0.00016755800047835447,
      "loss": 0.1984,
      "step": 9734
    },
    {
      "epoch": 1.3644008409250175,
      "grad_norm": 0.41636842489242554,
      "learning_rate": 0.0001675436498445348,
      "loss": 0.0301,
      "step": 9735
    },
    {
      "epoch": 1.3645409950946041,
      "grad_norm": 0.42520642280578613,
      "learning_rate": 0.00016752929921071513,
      "loss": 0.0854,
      "step": 9736
    },
    {
      "epoch": 1.3646811492641906,
      "grad_norm": 0.22899377346038818,
      "learning_rate": 0.00016751494857689548,
      "loss": 0.0684,
      "step": 9737
    },
    {
      "epoch": 1.3648213034337773,
      "grad_norm": 0.498642235994339,
      "learning_rate": 0.0001675005979430758,
      "loss": 0.0625,
      "step": 9738
    },
    {
      "epoch": 1.3649614576033637,
      "grad_norm": 0.2687016725540161,
      "learning_rate": 0.00016748624730925614,
      "loss": 0.0373,
      "step": 9739
    },
    {
      "epoch": 1.3651016117729502,
      "grad_norm": 0.19468122720718384,
      "learning_rate": 0.00016747189667543647,
      "loss": 0.0979,
      "step": 9740
    },
    {
      "epoch": 1.3652417659425367,
      "grad_norm": 0.7452189922332764,
      "learning_rate": 0.00016745754604161682,
      "loss": 0.0324,
      "step": 9741
    },
    {
      "epoch": 1.3653819201121233,
      "grad_norm": 0.16044935584068298,
      "learning_rate": 0.00016744319540779715,
      "loss": 0.0264,
      "step": 9742
    },
    {
      "epoch": 1.36552207428171,
      "grad_norm": 0.3225463628768921,
      "learning_rate": 0.00016742884477397748,
      "loss": 0.03,
      "step": 9743
    },
    {
      "epoch": 1.3656622284512965,
      "grad_norm": 0.3160260021686554,
      "learning_rate": 0.00016741449414015786,
      "loss": 0.0295,
      "step": 9744
    },
    {
      "epoch": 1.365802382620883,
      "grad_norm": 0.390919953584671,
      "learning_rate": 0.0001674001435063382,
      "loss": 0.06,
      "step": 9745
    },
    {
      "epoch": 1.3659425367904694,
      "grad_norm": 0.20790278911590576,
      "learning_rate": 0.00016738579287251852,
      "loss": 0.0275,
      "step": 9746
    },
    {
      "epoch": 1.366082690960056,
      "grad_norm": 0.2458115816116333,
      "learning_rate": 0.00016737144223869887,
      "loss": 0.0697,
      "step": 9747
    },
    {
      "epoch": 1.3662228451296425,
      "grad_norm": 0.2854403555393219,
      "learning_rate": 0.0001673570916048792,
      "loss": 0.0216,
      "step": 9748
    },
    {
      "epoch": 1.3663629992992292,
      "grad_norm": 0.12739400565624237,
      "learning_rate": 0.00016734274097105953,
      "loss": 0.0182,
      "step": 9749
    },
    {
      "epoch": 1.3665031534688157,
      "grad_norm": 0.3343561589717865,
      "learning_rate": 0.0001673283903372399,
      "loss": 0.0578,
      "step": 9750
    },
    {
      "epoch": 1.3666433076384021,
      "grad_norm": 0.24275031685829163,
      "learning_rate": 0.00016731403970342022,
      "loss": 0.0452,
      "step": 9751
    },
    {
      "epoch": 1.3667834618079888,
      "grad_norm": 0.3544938564300537,
      "learning_rate": 0.00016729968906960054,
      "loss": 0.07,
      "step": 9752
    },
    {
      "epoch": 1.3669236159775753,
      "grad_norm": 0.3955884575843811,
      "learning_rate": 0.0001672853384357809,
      "loss": 0.045,
      "step": 9753
    },
    {
      "epoch": 1.367063770147162,
      "grad_norm": 0.3393678069114685,
      "learning_rate": 0.00016727098780196123,
      "loss": 0.0738,
      "step": 9754
    },
    {
      "epoch": 1.3672039243167484,
      "grad_norm": 0.33253100514411926,
      "learning_rate": 0.00016725663716814158,
      "loss": 0.0884,
      "step": 9755
    },
    {
      "epoch": 1.3673440784863349,
      "grad_norm": 0.30154991149902344,
      "learning_rate": 0.00016724228653432194,
      "loss": 0.0529,
      "step": 9756
    },
    {
      "epoch": 1.3674842326559216,
      "grad_norm": 0.40125954151153564,
      "learning_rate": 0.00016722793590050227,
      "loss": 0.048,
      "step": 9757
    },
    {
      "epoch": 1.367624386825508,
      "grad_norm": 0.5775707960128784,
      "learning_rate": 0.0001672135852666826,
      "loss": 0.1117,
      "step": 9758
    },
    {
      "epoch": 1.3677645409950947,
      "grad_norm": 0.14290858805179596,
      "learning_rate": 0.00016719923463286293,
      "loss": 0.0172,
      "step": 9759
    },
    {
      "epoch": 1.3679046951646812,
      "grad_norm": 0.382387638092041,
      "learning_rate": 0.00016718488399904328,
      "loss": 0.0541,
      "step": 9760
    },
    {
      "epoch": 1.3680448493342676,
      "grad_norm": 0.299716055393219,
      "learning_rate": 0.0001671705333652236,
      "loss": 0.0337,
      "step": 9761
    },
    {
      "epoch": 1.3681850035038543,
      "grad_norm": 0.3183436095714569,
      "learning_rate": 0.00016715618273140394,
      "loss": 0.0455,
      "step": 9762
    },
    {
      "epoch": 1.3683251576734408,
      "grad_norm": 0.12458598613739014,
      "learning_rate": 0.0001671418320975843,
      "loss": 0.0093,
      "step": 9763
    },
    {
      "epoch": 1.3684653118430274,
      "grad_norm": 0.23761752247810364,
      "learning_rate": 0.00016712748146376462,
      "loss": 0.0259,
      "step": 9764
    },
    {
      "epoch": 1.368605466012614,
      "grad_norm": 0.28191614151000977,
      "learning_rate": 0.00016711313082994495,
      "loss": 0.0312,
      "step": 9765
    },
    {
      "epoch": 1.3687456201822004,
      "grad_norm": 0.28340035676956177,
      "learning_rate": 0.00016709878019612533,
      "loss": 0.0232,
      "step": 9766
    },
    {
      "epoch": 1.368885774351787,
      "grad_norm": 0.47565537691116333,
      "learning_rate": 0.00016708442956230566,
      "loss": 0.1049,
      "step": 9767
    },
    {
      "epoch": 1.3690259285213735,
      "grad_norm": 0.6011717915534973,
      "learning_rate": 0.000167070078928486,
      "loss": 0.048,
      "step": 9768
    },
    {
      "epoch": 1.3691660826909602,
      "grad_norm": 0.34251531958580017,
      "learning_rate": 0.00016705572829466635,
      "loss": 0.0474,
      "step": 9769
    },
    {
      "epoch": 1.3693062368605466,
      "grad_norm": 0.27362003922462463,
      "learning_rate": 0.00016704137766084668,
      "loss": 0.0451,
      "step": 9770
    },
    {
      "epoch": 1.369446391030133,
      "grad_norm": 0.6736848950386047,
      "learning_rate": 0.000167027027027027,
      "loss": 0.0273,
      "step": 9771
    },
    {
      "epoch": 1.3695865451997196,
      "grad_norm": 0.4624515771865845,
      "learning_rate": 0.00016701267639320736,
      "loss": 0.0355,
      "step": 9772
    },
    {
      "epoch": 1.3697266993693062,
      "grad_norm": 0.19825811684131622,
      "learning_rate": 0.0001669983257593877,
      "loss": 0.0336,
      "step": 9773
    },
    {
      "epoch": 1.3698668535388927,
      "grad_norm": 0.21776919066905975,
      "learning_rate": 0.00016698397512556802,
      "loss": 0.0371,
      "step": 9774
    },
    {
      "epoch": 1.3700070077084794,
      "grad_norm": 0.33590784668922424,
      "learning_rate": 0.00016696962449174835,
      "loss": 0.037,
      "step": 9775
    },
    {
      "epoch": 1.3701471618780658,
      "grad_norm": 0.3645128011703491,
      "learning_rate": 0.00016695527385792873,
      "loss": 0.0422,
      "step": 9776
    },
    {
      "epoch": 1.3702873160476523,
      "grad_norm": 0.13569369912147522,
      "learning_rate": 0.00016694092322410906,
      "loss": 0.0374,
      "step": 9777
    },
    {
      "epoch": 1.370427470217239,
      "grad_norm": 0.4902980625629425,
      "learning_rate": 0.00016692657259028939,
      "loss": 0.0287,
      "step": 9778
    },
    {
      "epoch": 1.3705676243868254,
      "grad_norm": 0.2802170217037201,
      "learning_rate": 0.00016691222195646974,
      "loss": 0.04,
      "step": 9779
    },
    {
      "epoch": 1.3707077785564121,
      "grad_norm": 0.30555084347724915,
      "learning_rate": 0.00016689787132265007,
      "loss": 0.038,
      "step": 9780
    },
    {
      "epoch": 1.3708479327259986,
      "grad_norm": 0.40659022331237793,
      "learning_rate": 0.0001668835206888304,
      "loss": 0.2067,
      "step": 9781
    },
    {
      "epoch": 1.370988086895585,
      "grad_norm": 0.7139325737953186,
      "learning_rate": 0.00016686917005501075,
      "loss": 0.0651,
      "step": 9782
    },
    {
      "epoch": 1.3711282410651717,
      "grad_norm": 0.8820049166679382,
      "learning_rate": 0.00016685481942119108,
      "loss": 0.0697,
      "step": 9783
    },
    {
      "epoch": 1.3712683952347582,
      "grad_norm": 0.4157586097717285,
      "learning_rate": 0.0001668404687873714,
      "loss": 0.084,
      "step": 9784
    },
    {
      "epoch": 1.3714085494043449,
      "grad_norm": 1.549342155456543,
      "learning_rate": 0.00016682611815355177,
      "loss": 0.2265,
      "step": 9785
    },
    {
      "epoch": 1.3715487035739313,
      "grad_norm": 0.35396257042884827,
      "learning_rate": 0.00016681176751973212,
      "loss": 0.0516,
      "step": 9786
    },
    {
      "epoch": 1.3716888577435178,
      "grad_norm": 0.4627709686756134,
      "learning_rate": 0.00016679741688591245,
      "loss": 0.0891,
      "step": 9787
    },
    {
      "epoch": 1.3718290119131045,
      "grad_norm": 0.5053613781929016,
      "learning_rate": 0.0001667830662520928,
      "loss": 0.0904,
      "step": 9788
    },
    {
      "epoch": 1.371969166082691,
      "grad_norm": 1.0574511289596558,
      "learning_rate": 0.00016676871561827314,
      "loss": 0.065,
      "step": 9789
    },
    {
      "epoch": 1.3721093202522776,
      "grad_norm": 0.4142410457134247,
      "learning_rate": 0.00016675436498445346,
      "loss": 0.0352,
      "step": 9790
    },
    {
      "epoch": 1.372249474421864,
      "grad_norm": 0.4920435845851898,
      "learning_rate": 0.00016674001435063382,
      "loss": 0.1189,
      "step": 9791
    },
    {
      "epoch": 1.3723896285914505,
      "grad_norm": 0.3084256052970886,
      "learning_rate": 0.00016672566371681415,
      "loss": 0.0465,
      "step": 9792
    },
    {
      "epoch": 1.3725297827610372,
      "grad_norm": 0.4969826340675354,
      "learning_rate": 0.00016671131308299448,
      "loss": 0.0856,
      "step": 9793
    },
    {
      "epoch": 1.3726699369306237,
      "grad_norm": 0.6154130697250366,
      "learning_rate": 0.0001666969624491748,
      "loss": 0.0839,
      "step": 9794
    },
    {
      "epoch": 1.3728100911002104,
      "grad_norm": 0.20868386328220367,
      "learning_rate": 0.00016668261181535516,
      "loss": 0.0437,
      "step": 9795
    },
    {
      "epoch": 1.3729502452697968,
      "grad_norm": 0.23858319222927094,
      "learning_rate": 0.0001666682611815355,
      "loss": 0.041,
      "step": 9796
    },
    {
      "epoch": 1.3730903994393833,
      "grad_norm": 0.8451023697853088,
      "learning_rate": 0.00016665391054771582,
      "loss": 0.0611,
      "step": 9797
    },
    {
      "epoch": 1.3732305536089697,
      "grad_norm": 0.19176678359508514,
      "learning_rate": 0.0001666395599138962,
      "loss": 0.0279,
      "step": 9798
    },
    {
      "epoch": 1.3733707077785564,
      "grad_norm": 0.4042290449142456,
      "learning_rate": 0.00016662520928007653,
      "loss": 0.16,
      "step": 9799
    },
    {
      "epoch": 1.373510861948143,
      "grad_norm": 0.18294411897659302,
      "learning_rate": 0.00016661085864625686,
      "loss": 0.0354,
      "step": 9800
    },
    {
      "epoch": 1.3736510161177296,
      "grad_norm": 0.28249847888946533,
      "learning_rate": 0.0001665965080124372,
      "loss": 0.1256,
      "step": 9801
    },
    {
      "epoch": 1.373791170287316,
      "grad_norm": 0.46365299820899963,
      "learning_rate": 0.00016658215737861754,
      "loss": 0.057,
      "step": 9802
    },
    {
      "epoch": 1.3739313244569025,
      "grad_norm": 0.3091248571872711,
      "learning_rate": 0.00016656780674479787,
      "loss": 0.0879,
      "step": 9803
    },
    {
      "epoch": 1.3740714786264892,
      "grad_norm": 0.11044400185346603,
      "learning_rate": 0.00016655345611097823,
      "loss": 0.0212,
      "step": 9804
    },
    {
      "epoch": 1.3742116327960756,
      "grad_norm": 0.26344338059425354,
      "learning_rate": 0.00016653910547715855,
      "loss": 0.0507,
      "step": 9805
    },
    {
      "epoch": 1.3743517869656623,
      "grad_norm": 0.16674789786338806,
      "learning_rate": 0.00016652475484333888,
      "loss": 0.0872,
      "step": 9806
    },
    {
      "epoch": 1.3744919411352488,
      "grad_norm": 0.37788036465644836,
      "learning_rate": 0.00016651040420951927,
      "loss": 0.0991,
      "step": 9807
    },
    {
      "epoch": 1.3746320953048352,
      "grad_norm": 0.8262220621109009,
      "learning_rate": 0.0001664960535756996,
      "loss": 0.0394,
      "step": 9808
    },
    {
      "epoch": 1.374772249474422,
      "grad_norm": 0.5292328596115112,
      "learning_rate": 0.00016648170294187992,
      "loss": 0.088,
      "step": 9809
    },
    {
      "epoch": 1.3749124036440084,
      "grad_norm": 0.23596534132957458,
      "learning_rate": 0.00016646735230806025,
      "loss": 0.0852,
      "step": 9810
    },
    {
      "epoch": 1.375052557813595,
      "grad_norm": 0.27616414427757263,
      "learning_rate": 0.0001664530016742406,
      "loss": 0.0427,
      "step": 9811
    },
    {
      "epoch": 1.3751927119831815,
      "grad_norm": 0.25596749782562256,
      "learning_rate": 0.00016643865104042094,
      "loss": 0.0307,
      "step": 9812
    },
    {
      "epoch": 1.375332866152768,
      "grad_norm": 0.6160295605659485,
      "learning_rate": 0.00016642430040660126,
      "loss": 0.101,
      "step": 9813
    },
    {
      "epoch": 1.3754730203223546,
      "grad_norm": 0.3630438446998596,
      "learning_rate": 0.00016640994977278162,
      "loss": 0.0438,
      "step": 9814
    },
    {
      "epoch": 1.375613174491941,
      "grad_norm": 0.15699683129787445,
      "learning_rate": 0.00016639559913896195,
      "loss": 0.0289,
      "step": 9815
    },
    {
      "epoch": 1.3757533286615278,
      "grad_norm": 0.2664378583431244,
      "learning_rate": 0.00016638124850514228,
      "loss": 0.0529,
      "step": 9816
    },
    {
      "epoch": 1.3758934828311142,
      "grad_norm": 0.6827977299690247,
      "learning_rate": 0.00016636689787132263,
      "loss": 0.0913,
      "step": 9817
    },
    {
      "epoch": 1.3760336370007007,
      "grad_norm": 0.1987086832523346,
      "learning_rate": 0.000166352547237503,
      "loss": 0.0255,
      "step": 9818
    },
    {
      "epoch": 1.3761737911702874,
      "grad_norm": 0.20452755689620972,
      "learning_rate": 0.00016633819660368332,
      "loss": 0.0173,
      "step": 9819
    },
    {
      "epoch": 1.3763139453398738,
      "grad_norm": 0.2806962728500366,
      "learning_rate": 0.00016632384596986367,
      "loss": 0.0281,
      "step": 9820
    },
    {
      "epoch": 1.3764540995094605,
      "grad_norm": 0.738853931427002,
      "learning_rate": 0.000166309495336044,
      "loss": 0.0816,
      "step": 9821
    },
    {
      "epoch": 1.376594253679047,
      "grad_norm": 0.3580811023712158,
      "learning_rate": 0.00016629514470222433,
      "loss": 0.077,
      "step": 9822
    },
    {
      "epoch": 1.3767344078486334,
      "grad_norm": 0.273605614900589,
      "learning_rate": 0.00016628079406840469,
      "loss": 0.0192,
      "step": 9823
    },
    {
      "epoch": 1.3768745620182201,
      "grad_norm": 0.3490077257156372,
      "learning_rate": 0.00016626644343458501,
      "loss": 0.0666,
      "step": 9824
    },
    {
      "epoch": 1.3770147161878066,
      "grad_norm": 0.5520818829536438,
      "learning_rate": 0.00016625209280076534,
      "loss": 0.0824,
      "step": 9825
    },
    {
      "epoch": 1.3771548703573933,
      "grad_norm": 0.3373028635978699,
      "learning_rate": 0.0001662377421669457,
      "loss": 0.0363,
      "step": 9826
    },
    {
      "epoch": 1.3772950245269797,
      "grad_norm": 0.3937041759490967,
      "learning_rate": 0.00016622339153312603,
      "loss": 0.0299,
      "step": 9827
    },
    {
      "epoch": 1.3774351786965662,
      "grad_norm": 0.8915595412254333,
      "learning_rate": 0.00016620904089930636,
      "loss": 0.0975,
      "step": 9828
    },
    {
      "epoch": 1.3775753328661526,
      "grad_norm": 0.4106620252132416,
      "learning_rate": 0.00016619469026548668,
      "loss": 0.1457,
      "step": 9829
    },
    {
      "epoch": 1.3777154870357393,
      "grad_norm": 0.5025816559791565,
      "learning_rate": 0.00016618033963166707,
      "loss": 0.0226,
      "step": 9830
    },
    {
      "epoch": 1.3778556412053258,
      "grad_norm": 1.7612367868423462,
      "learning_rate": 0.0001661659889978474,
      "loss": 0.0644,
      "step": 9831
    },
    {
      "epoch": 1.3779957953749125,
      "grad_norm": 1.447753667831421,
      "learning_rate": 0.00016615163836402772,
      "loss": 0.1274,
      "step": 9832
    },
    {
      "epoch": 1.378135949544499,
      "grad_norm": 3.34128737449646,
      "learning_rate": 0.00016613728773020808,
      "loss": 0.2345,
      "step": 9833
    },
    {
      "epoch": 1.3782761037140854,
      "grad_norm": 1.8495852947235107,
      "learning_rate": 0.0001661229370963884,
      "loss": 0.2408,
      "step": 9834
    },
    {
      "epoch": 1.378416257883672,
      "grad_norm": 0.27958783507347107,
      "learning_rate": 0.00016610858646256874,
      "loss": 0.0305,
      "step": 9835
    },
    {
      "epoch": 1.3785564120532585,
      "grad_norm": 0.20330193638801575,
      "learning_rate": 0.0001660942358287491,
      "loss": 0.0303,
      "step": 9836
    },
    {
      "epoch": 1.3786965662228452,
      "grad_norm": 0.18050719797611237,
      "learning_rate": 0.00016607988519492942,
      "loss": 0.0406,
      "step": 9837
    },
    {
      "epoch": 1.3788367203924317,
      "grad_norm": 0.12246309220790863,
      "learning_rate": 0.00016606553456110975,
      "loss": 0.0438,
      "step": 9838
    },
    {
      "epoch": 1.3789768745620181,
      "grad_norm": 0.1493966430425644,
      "learning_rate": 0.00016605118392729013,
      "loss": 0.0226,
      "step": 9839
    },
    {
      "epoch": 1.3791170287316048,
      "grad_norm": 0.3623847961425781,
      "learning_rate": 0.00016603683329347046,
      "loss": 0.0421,
      "step": 9840
    },
    {
      "epoch": 1.3792571829011913,
      "grad_norm": 0.29569578170776367,
      "learning_rate": 0.0001660224826596508,
      "loss": 0.0472,
      "step": 9841
    },
    {
      "epoch": 1.379397337070778,
      "grad_norm": 0.3124988377094269,
      "learning_rate": 0.00016600813202583115,
      "loss": 0.0834,
      "step": 9842
    },
    {
      "epoch": 1.3795374912403644,
      "grad_norm": 0.41684073209762573,
      "learning_rate": 0.00016599378139201147,
      "loss": 0.0954,
      "step": 9843
    },
    {
      "epoch": 1.3796776454099509,
      "grad_norm": 0.3260037899017334,
      "learning_rate": 0.0001659794307581918,
      "loss": 0.0386,
      "step": 9844
    },
    {
      "epoch": 1.3798177995795375,
      "grad_norm": 0.7677415609359741,
      "learning_rate": 0.00016596508012437213,
      "loss": 0.0595,
      "step": 9845
    },
    {
      "epoch": 1.379957953749124,
      "grad_norm": 0.2738877236843109,
      "learning_rate": 0.0001659507294905525,
      "loss": 0.0729,
      "step": 9846
    },
    {
      "epoch": 1.3800981079187107,
      "grad_norm": 0.21391017735004425,
      "learning_rate": 0.00016593637885673282,
      "loss": 0.0178,
      "step": 9847
    },
    {
      "epoch": 1.3802382620882971,
      "grad_norm": 0.3169548213481903,
      "learning_rate": 0.00016592202822291314,
      "loss": 0.0596,
      "step": 9848
    },
    {
      "epoch": 1.3803784162578836,
      "grad_norm": 0.34406572580337524,
      "learning_rate": 0.0001659076775890935,
      "loss": 0.1025,
      "step": 9849
    },
    {
      "epoch": 1.3805185704274703,
      "grad_norm": 0.2814118564128876,
      "learning_rate": 0.00016589332695527386,
      "loss": 0.0707,
      "step": 9850
    },
    {
      "epoch": 1.3806587245970567,
      "grad_norm": 0.2682029902935028,
      "learning_rate": 0.00016587897632145418,
      "loss": 0.0863,
      "step": 9851
    },
    {
      "epoch": 1.3807988787666434,
      "grad_norm": 0.543023943901062,
      "learning_rate": 0.00016586462568763454,
      "loss": 0.0801,
      "step": 9852
    },
    {
      "epoch": 1.3809390329362299,
      "grad_norm": 0.22751758992671967,
      "learning_rate": 0.00016585027505381487,
      "loss": 0.0338,
      "step": 9853
    },
    {
      "epoch": 1.3810791871058163,
      "grad_norm": 0.12478634715080261,
      "learning_rate": 0.0001658359244199952,
      "loss": 0.0157,
      "step": 9854
    },
    {
      "epoch": 1.381219341275403,
      "grad_norm": 0.3831522762775421,
      "learning_rate": 0.00016582157378617555,
      "loss": 0.074,
      "step": 9855
    },
    {
      "epoch": 1.3813594954449895,
      "grad_norm": 0.3365626037120819,
      "learning_rate": 0.00016580722315235588,
      "loss": 0.056,
      "step": 9856
    },
    {
      "epoch": 1.3814996496145762,
      "grad_norm": 0.452978253364563,
      "learning_rate": 0.0001657928725185362,
      "loss": 0.0729,
      "step": 9857
    },
    {
      "epoch": 1.3816398037841626,
      "grad_norm": 0.11168167740106583,
      "learning_rate": 0.00016577852188471656,
      "loss": 0.0219,
      "step": 9858
    },
    {
      "epoch": 1.381779957953749,
      "grad_norm": 0.38828563690185547,
      "learning_rate": 0.0001657641712508969,
      "loss": 0.1007,
      "step": 9859
    },
    {
      "epoch": 1.3819201121233355,
      "grad_norm": 0.49741503596305847,
      "learning_rate": 0.00016574982061707722,
      "loss": 0.0553,
      "step": 9860
    },
    {
      "epoch": 1.3820602662929222,
      "grad_norm": 0.39293739199638367,
      "learning_rate": 0.00016573546998325755,
      "loss": 0.0283,
      "step": 9861
    },
    {
      "epoch": 1.3822004204625087,
      "grad_norm": 0.7633966207504272,
      "learning_rate": 0.00016572111934943793,
      "loss": 0.063,
      "step": 9862
    },
    {
      "epoch": 1.3823405746320954,
      "grad_norm": 0.18341323733329773,
      "learning_rate": 0.00016570676871561826,
      "loss": 0.0399,
      "step": 9863
    },
    {
      "epoch": 1.3824807288016818,
      "grad_norm": 0.16444343328475952,
      "learning_rate": 0.0001656924180817986,
      "loss": 0.059,
      "step": 9864
    },
    {
      "epoch": 1.3826208829712683,
      "grad_norm": 0.6269795298576355,
      "learning_rate": 0.00016567806744797895,
      "loss": 0.1135,
      "step": 9865
    },
    {
      "epoch": 1.382761037140855,
      "grad_norm": 1.1129467487335205,
      "learning_rate": 0.00016566371681415927,
      "loss": 0.069,
      "step": 9866
    },
    {
      "epoch": 1.3829011913104414,
      "grad_norm": 0.3726169168949127,
      "learning_rate": 0.0001656493661803396,
      "loss": 0.0789,
      "step": 9867
    },
    {
      "epoch": 1.383041345480028,
      "grad_norm": 0.3132324814796448,
      "learning_rate": 0.00016563501554651996,
      "loss": 0.1302,
      "step": 9868
    },
    {
      "epoch": 1.3831814996496146,
      "grad_norm": 0.4091932773590088,
      "learning_rate": 0.0001656206649127003,
      "loss": 0.0997,
      "step": 9869
    },
    {
      "epoch": 1.383321653819201,
      "grad_norm": 0.34533199667930603,
      "learning_rate": 0.00016560631427888062,
      "loss": 0.0516,
      "step": 9870
    },
    {
      "epoch": 1.3834618079887877,
      "grad_norm": 0.3048110902309418,
      "learning_rate": 0.000165591963645061,
      "loss": 0.0337,
      "step": 9871
    },
    {
      "epoch": 1.3836019621583742,
      "grad_norm": 0.46481406688690186,
      "learning_rate": 0.00016557761301124133,
      "loss": 0.0624,
      "step": 9872
    },
    {
      "epoch": 1.3837421163279608,
      "grad_norm": 0.3779941201210022,
      "learning_rate": 0.00016556326237742166,
      "loss": 0.0622,
      "step": 9873
    },
    {
      "epoch": 1.3838822704975473,
      "grad_norm": 0.3981720209121704,
      "learning_rate": 0.000165548911743602,
      "loss": 0.0451,
      "step": 9874
    },
    {
      "epoch": 1.3840224246671338,
      "grad_norm": 0.4253818690776825,
      "learning_rate": 0.00016553456110978234,
      "loss": 0.0589,
      "step": 9875
    },
    {
      "epoch": 1.3841625788367204,
      "grad_norm": 0.1879155933856964,
      "learning_rate": 0.00016552021047596267,
      "loss": 0.0242,
      "step": 9876
    },
    {
      "epoch": 1.384302733006307,
      "grad_norm": 0.4387842118740082,
      "learning_rate": 0.00016550585984214302,
      "loss": 0.0472,
      "step": 9877
    },
    {
      "epoch": 1.3844428871758936,
      "grad_norm": 0.6361129879951477,
      "learning_rate": 0.00016549150920832335,
      "loss": 0.0558,
      "step": 9878
    },
    {
      "epoch": 1.38458304134548,
      "grad_norm": 0.6603955626487732,
      "learning_rate": 0.00016547715857450368,
      "loss": 0.0671,
      "step": 9879
    },
    {
      "epoch": 1.3847231955150665,
      "grad_norm": 0.5385135412216187,
      "learning_rate": 0.000165462807940684,
      "loss": 0.1675,
      "step": 9880
    },
    {
      "epoch": 1.3848633496846532,
      "grad_norm": 0.7110384106636047,
      "learning_rate": 0.0001654484573068644,
      "loss": 0.0427,
      "step": 9881
    },
    {
      "epoch": 1.3850035038542396,
      "grad_norm": 0.4004608690738678,
      "learning_rate": 0.00016543410667304472,
      "loss": 0.0692,
      "step": 9882
    },
    {
      "epoch": 1.3851436580238263,
      "grad_norm": 0.5768249034881592,
      "learning_rate": 0.00016541975603922505,
      "loss": 0.0422,
      "step": 9883
    },
    {
      "epoch": 1.3852838121934128,
      "grad_norm": 2.5120275020599365,
      "learning_rate": 0.0001654054054054054,
      "loss": 0.2715,
      "step": 9884
    },
    {
      "epoch": 1.3854239663629992,
      "grad_norm": 1.6612597703933716,
      "learning_rate": 0.00016539105477158573,
      "loss": 0.1563,
      "step": 9885
    },
    {
      "epoch": 1.3855641205325857,
      "grad_norm": 0.13905218243598938,
      "learning_rate": 0.00016537670413776606,
      "loss": 0.0287,
      "step": 9886
    },
    {
      "epoch": 1.3857042747021724,
      "grad_norm": 0.26785966753959656,
      "learning_rate": 0.00016536235350394642,
      "loss": 0.0991,
      "step": 9887
    },
    {
      "epoch": 1.385844428871759,
      "grad_norm": 0.30881965160369873,
      "learning_rate": 0.00016534800287012675,
      "loss": 0.1044,
      "step": 9888
    },
    {
      "epoch": 1.3859845830413455,
      "grad_norm": 0.769124448299408,
      "learning_rate": 0.00016533365223630708,
      "loss": 0.1158,
      "step": 9889
    },
    {
      "epoch": 1.386124737210932,
      "grad_norm": 0.34987568855285645,
      "learning_rate": 0.00016531930160248743,
      "loss": 0.0613,
      "step": 9890
    },
    {
      "epoch": 1.3862648913805184,
      "grad_norm": 0.17850933969020844,
      "learning_rate": 0.00016530495096866776,
      "loss": 0.0396,
      "step": 9891
    },
    {
      "epoch": 1.3864050455501051,
      "grad_norm": 0.13544943928718567,
      "learning_rate": 0.0001652906003348481,
      "loss": 0.0431,
      "step": 9892
    },
    {
      "epoch": 1.3865451997196916,
      "grad_norm": 0.1979440599679947,
      "learning_rate": 0.00016527624970102847,
      "loss": 0.0396,
      "step": 9893
    },
    {
      "epoch": 1.3866853538892783,
      "grad_norm": 0.1560634821653366,
      "learning_rate": 0.0001652618990672088,
      "loss": 0.0227,
      "step": 9894
    },
    {
      "epoch": 1.3868255080588647,
      "grad_norm": 0.1894182562828064,
      "learning_rate": 0.00016524754843338913,
      "loss": 0.0325,
      "step": 9895
    },
    {
      "epoch": 1.3869656622284512,
      "grad_norm": 0.22193019092082977,
      "learning_rate": 0.00016523319779956946,
      "loss": 0.056,
      "step": 9896
    },
    {
      "epoch": 1.3871058163980379,
      "grad_norm": 0.1935570389032364,
      "learning_rate": 0.0001652188471657498,
      "loss": 0.0541,
      "step": 9897
    },
    {
      "epoch": 1.3872459705676243,
      "grad_norm": 0.1711646318435669,
      "learning_rate": 0.00016520449653193014,
      "loss": 0.023,
      "step": 9898
    },
    {
      "epoch": 1.387386124737211,
      "grad_norm": 0.19555114209651947,
      "learning_rate": 0.00016519014589811047,
      "loss": 0.0131,
      "step": 9899
    },
    {
      "epoch": 1.3875262789067975,
      "grad_norm": 0.4711863398551941,
      "learning_rate": 0.00016517579526429083,
      "loss": 0.0373,
      "step": 9900
    },
    {
      "epoch": 1.387666433076384,
      "grad_norm": 0.22323556244373322,
      "learning_rate": 0.00016516144463047115,
      "loss": 0.0476,
      "step": 9901
    },
    {
      "epoch": 1.3878065872459706,
      "grad_norm": 0.4431634247303009,
      "learning_rate": 0.00016514709399665148,
      "loss": 0.0562,
      "step": 9902
    },
    {
      "epoch": 1.387946741415557,
      "grad_norm": 0.26687121391296387,
      "learning_rate": 0.00016513274336283187,
      "loss": 0.1016,
      "step": 9903
    },
    {
      "epoch": 1.3880868955851438,
      "grad_norm": 0.09746091067790985,
      "learning_rate": 0.0001651183927290122,
      "loss": 0.0162,
      "step": 9904
    },
    {
      "epoch": 1.3882270497547302,
      "grad_norm": 0.30017802119255066,
      "learning_rate": 0.00016510404209519252,
      "loss": 0.0566,
      "step": 9905
    },
    {
      "epoch": 1.3883672039243167,
      "grad_norm": 0.2457084208726883,
      "learning_rate": 0.00016508969146137288,
      "loss": 0.0323,
      "step": 9906
    },
    {
      "epoch": 1.3885073580939034,
      "grad_norm": 0.42852237820625305,
      "learning_rate": 0.0001650753408275532,
      "loss": 0.0782,
      "step": 9907
    },
    {
      "epoch": 1.3886475122634898,
      "grad_norm": 0.19700708985328674,
      "learning_rate": 0.00016506099019373353,
      "loss": 0.0125,
      "step": 9908
    },
    {
      "epoch": 1.3887876664330765,
      "grad_norm": 0.3940353989601135,
      "learning_rate": 0.0001650466395599139,
      "loss": 0.0567,
      "step": 9909
    },
    {
      "epoch": 1.388927820602663,
      "grad_norm": 0.5056464076042175,
      "learning_rate": 0.00016503228892609422,
      "loss": 0.0503,
      "step": 9910
    },
    {
      "epoch": 1.3890679747722494,
      "grad_norm": 0.1690114587545395,
      "learning_rate": 0.00016501793829227455,
      "loss": 0.0135,
      "step": 9911
    },
    {
      "epoch": 1.389208128941836,
      "grad_norm": 0.058540597558021545,
      "learning_rate": 0.0001650035876584549,
      "loss": 0.0053,
      "step": 9912
    },
    {
      "epoch": 1.3893482831114226,
      "grad_norm": 0.17253154516220093,
      "learning_rate": 0.00016498923702463526,
      "loss": 0.0354,
      "step": 9913
    },
    {
      "epoch": 1.3894884372810092,
      "grad_norm": 0.6110184192657471,
      "learning_rate": 0.0001649748863908156,
      "loss": 0.0412,
      "step": 9914
    },
    {
      "epoch": 1.3896285914505957,
      "grad_norm": 1.0733479261398315,
      "learning_rate": 0.00016496053575699592,
      "loss": 0.0678,
      "step": 9915
    },
    {
      "epoch": 1.3897687456201822,
      "grad_norm": 0.37653848528862,
      "learning_rate": 0.00016494618512317627,
      "loss": 0.0479,
      "step": 9916
    },
    {
      "epoch": 1.3899088997897686,
      "grad_norm": 0.4403993487358093,
      "learning_rate": 0.0001649318344893566,
      "loss": 0.0755,
      "step": 9917
    },
    {
      "epoch": 1.3900490539593553,
      "grad_norm": 0.11352758854627609,
      "learning_rate": 0.00016491748385553693,
      "loss": 0.048,
      "step": 9918
    },
    {
      "epoch": 1.3901892081289418,
      "grad_norm": 0.3408459424972534,
      "learning_rate": 0.00016490313322171728,
      "loss": 0.0239,
      "step": 9919
    },
    {
      "epoch": 1.3903293622985284,
      "grad_norm": 0.31039145588874817,
      "learning_rate": 0.0001648887825878976,
      "loss": 0.0714,
      "step": 9920
    },
    {
      "epoch": 1.390469516468115,
      "grad_norm": 0.5289487838745117,
      "learning_rate": 0.00016487443195407794,
      "loss": 0.0429,
      "step": 9921
    },
    {
      "epoch": 1.3906096706377014,
      "grad_norm": 0.258137971162796,
      "learning_rate": 0.0001648600813202583,
      "loss": 0.033,
      "step": 9922
    },
    {
      "epoch": 1.390749824807288,
      "grad_norm": 0.3090880215167999,
      "learning_rate": 0.00016484573068643863,
      "loss": 0.0334,
      "step": 9923
    },
    {
      "epoch": 1.3908899789768745,
      "grad_norm": 0.21065670251846313,
      "learning_rate": 0.00016483138005261895,
      "loss": 0.0153,
      "step": 9924
    },
    {
      "epoch": 1.3910301331464612,
      "grad_norm": 1.2045516967773438,
      "learning_rate": 0.00016481702941879934,
      "loss": 0.1021,
      "step": 9925
    },
    {
      "epoch": 1.3911702873160476,
      "grad_norm": 0.22760462760925293,
      "learning_rate": 0.00016480267878497967,
      "loss": 0.0816,
      "step": 9926
    },
    {
      "epoch": 1.391310441485634,
      "grad_norm": 0.8216742873191833,
      "learning_rate": 0.00016478832815116,
      "loss": 0.0496,
      "step": 9927
    },
    {
      "epoch": 1.3914505956552208,
      "grad_norm": 1.1841371059417725,
      "learning_rate": 0.00016477397751734035,
      "loss": 0.1763,
      "step": 9928
    },
    {
      "epoch": 1.3915907498248072,
      "grad_norm": 0.44436272978782654,
      "learning_rate": 0.00016475962688352068,
      "loss": 0.1548,
      "step": 9929
    },
    {
      "epoch": 1.391730903994394,
      "grad_norm": 0.4918588399887085,
      "learning_rate": 0.000164745276249701,
      "loss": 0.0922,
      "step": 9930
    },
    {
      "epoch": 1.3918710581639804,
      "grad_norm": 0.34305915236473083,
      "learning_rate": 0.00016473092561588134,
      "loss": 0.0327,
      "step": 9931
    },
    {
      "epoch": 1.3920112123335668,
      "grad_norm": 0.19560088217258453,
      "learning_rate": 0.0001647165749820617,
      "loss": 0.0179,
      "step": 9932
    },
    {
      "epoch": 1.3921513665031535,
      "grad_norm": 1.6035668849945068,
      "learning_rate": 0.00016470222434824202,
      "loss": 0.1782,
      "step": 9933
    },
    {
      "epoch": 1.39229152067274,
      "grad_norm": 0.9499816298484802,
      "learning_rate": 0.00016468787371442235,
      "loss": 0.0385,
      "step": 9934
    },
    {
      "epoch": 1.3924316748423267,
      "grad_norm": 1.1299548149108887,
      "learning_rate": 0.00016467352308060273,
      "loss": 0.0798,
      "step": 9935
    },
    {
      "epoch": 1.3925718290119131,
      "grad_norm": 0.16624300181865692,
      "learning_rate": 0.00016465917244678306,
      "loss": 0.0426,
      "step": 9936
    },
    {
      "epoch": 1.3927119831814996,
      "grad_norm": 0.2523902356624603,
      "learning_rate": 0.0001646448218129634,
      "loss": 0.0387,
      "step": 9937
    },
    {
      "epoch": 1.3928521373510863,
      "grad_norm": 0.11820819228887558,
      "learning_rate": 0.00016463047117914374,
      "loss": 0.0239,
      "step": 9938
    },
    {
      "epoch": 1.3929922915206727,
      "grad_norm": 0.45123013854026794,
      "learning_rate": 0.00016461612054532407,
      "loss": 0.0516,
      "step": 9939
    },
    {
      "epoch": 1.3931324456902594,
      "grad_norm": 0.3315385580062866,
      "learning_rate": 0.0001646017699115044,
      "loss": 0.0537,
      "step": 9940
    },
    {
      "epoch": 1.3932725998598459,
      "grad_norm": 0.2899155914783478,
      "learning_rate": 0.00016458741927768476,
      "loss": 0.0333,
      "step": 9941
    },
    {
      "epoch": 1.3934127540294323,
      "grad_norm": 0.503903329372406,
      "learning_rate": 0.00016457306864386509,
      "loss": 0.0195,
      "step": 9942
    },
    {
      "epoch": 1.393552908199019,
      "grad_norm": 0.3337000906467438,
      "learning_rate": 0.00016455871801004541,
      "loss": 0.0668,
      "step": 9943
    },
    {
      "epoch": 1.3936930623686055,
      "grad_norm": 0.28747814893722534,
      "learning_rate": 0.00016454436737622577,
      "loss": 0.0257,
      "step": 9944
    },
    {
      "epoch": 1.3938332165381921,
      "grad_norm": 0.3226398527622223,
      "learning_rate": 0.00016453001674240613,
      "loss": 0.053,
      "step": 9945
    },
    {
      "epoch": 1.3939733707077786,
      "grad_norm": 0.08775278180837631,
      "learning_rate": 0.00016451566610858645,
      "loss": 0.0169,
      "step": 9946
    },
    {
      "epoch": 1.394113524877365,
      "grad_norm": 0.40490493178367615,
      "learning_rate": 0.0001645013154747668,
      "loss": 0.0403,
      "step": 9947
    },
    {
      "epoch": 1.3942536790469515,
      "grad_norm": 0.2999856472015381,
      "learning_rate": 0.00016448696484094714,
      "loss": 0.0298,
      "step": 9948
    },
    {
      "epoch": 1.3943938332165382,
      "grad_norm": 0.32002151012420654,
      "learning_rate": 0.00016447261420712747,
      "loss": 0.0349,
      "step": 9949
    },
    {
      "epoch": 1.3945339873861247,
      "grad_norm": 0.3044469356536865,
      "learning_rate": 0.0001644582635733078,
      "loss": 0.0575,
      "step": 9950
    },
    {
      "epoch": 1.3946741415557113,
      "grad_norm": 0.3892817199230194,
      "learning_rate": 0.00016444391293948815,
      "loss": 0.0743,
      "step": 9951
    },
    {
      "epoch": 1.3948142957252978,
      "grad_norm": 0.16460630297660828,
      "learning_rate": 0.00016442956230566848,
      "loss": 0.0112,
      "step": 9952
    },
    {
      "epoch": 1.3949544498948843,
      "grad_norm": 0.2842039167881012,
      "learning_rate": 0.0001644152116718488,
      "loss": 0.0815,
      "step": 9953
    },
    {
      "epoch": 1.395094604064471,
      "grad_norm": 0.29906436800956726,
      "learning_rate": 0.00016440086103802916,
      "loss": 0.0288,
      "step": 9954
    },
    {
      "epoch": 1.3952347582340574,
      "grad_norm": 0.23643511533737183,
      "learning_rate": 0.0001643865104042095,
      "loss": 0.0418,
      "step": 9955
    },
    {
      "epoch": 1.395374912403644,
      "grad_norm": 0.28124886751174927,
      "learning_rate": 0.00016437215977038982,
      "loss": 0.0631,
      "step": 9956
    },
    {
      "epoch": 1.3955150665732305,
      "grad_norm": 0.16888119280338287,
      "learning_rate": 0.0001643578091365702,
      "loss": 0.0201,
      "step": 9957
    },
    {
      "epoch": 1.395655220742817,
      "grad_norm": 0.7826975584030151,
      "learning_rate": 0.00016434345850275053,
      "loss": 0.0539,
      "step": 9958
    },
    {
      "epoch": 1.3957953749124037,
      "grad_norm": 0.7242889404296875,
      "learning_rate": 0.00016432910786893086,
      "loss": 0.1006,
      "step": 9959
    },
    {
      "epoch": 1.3959355290819901,
      "grad_norm": 0.2995014190673828,
      "learning_rate": 0.00016431475723511122,
      "loss": 0.0963,
      "step": 9960
    },
    {
      "epoch": 1.3960756832515768,
      "grad_norm": 0.18630144000053406,
      "learning_rate": 0.00016430040660129154,
      "loss": 0.0276,
      "step": 9961
    },
    {
      "epoch": 1.3962158374211633,
      "grad_norm": 0.9215060472488403,
      "learning_rate": 0.00016428605596747187,
      "loss": 0.0494,
      "step": 9962
    },
    {
      "epoch": 1.3963559915907497,
      "grad_norm": 0.4581790566444397,
      "learning_rate": 0.00016427170533365223,
      "loss": 0.0782,
      "step": 9963
    },
    {
      "epoch": 1.3964961457603364,
      "grad_norm": 0.6915099024772644,
      "learning_rate": 0.00016425735469983256,
      "loss": 0.0374,
      "step": 9964
    },
    {
      "epoch": 1.3966362999299229,
      "grad_norm": 0.4167921245098114,
      "learning_rate": 0.00016424300406601289,
      "loss": 0.0862,
      "step": 9965
    },
    {
      "epoch": 1.3967764540995096,
      "grad_norm": 0.21781854331493378,
      "learning_rate": 0.00016422865343219321,
      "loss": 0.0152,
      "step": 9966
    },
    {
      "epoch": 1.396916608269096,
      "grad_norm": 0.18379028141498566,
      "learning_rate": 0.0001642143027983736,
      "loss": 0.0421,
      "step": 9967
    },
    {
      "epoch": 1.3970567624386825,
      "grad_norm": 0.2734959125518799,
      "learning_rate": 0.00016419995216455393,
      "loss": 0.0414,
      "step": 9968
    },
    {
      "epoch": 1.3971969166082692,
      "grad_norm": 0.5780998468399048,
      "learning_rate": 0.00016418560153073425,
      "loss": 0.0978,
      "step": 9969
    },
    {
      "epoch": 1.3973370707778556,
      "grad_norm": 0.41190165281295776,
      "learning_rate": 0.0001641712508969146,
      "loss": 0.0499,
      "step": 9970
    },
    {
      "epoch": 1.3974772249474423,
      "grad_norm": 0.36172229051589966,
      "learning_rate": 0.00016415690026309494,
      "loss": 0.0414,
      "step": 9971
    },
    {
      "epoch": 1.3976173791170288,
      "grad_norm": 0.2094537913799286,
      "learning_rate": 0.00016414254962927527,
      "loss": 0.0272,
      "step": 9972
    },
    {
      "epoch": 1.3977575332866152,
      "grad_norm": 0.1513061672449112,
      "learning_rate": 0.00016412819899545562,
      "loss": 0.0111,
      "step": 9973
    },
    {
      "epoch": 1.3978976874562017,
      "grad_norm": 0.40582114458084106,
      "learning_rate": 0.00016411384836163595,
      "loss": 0.0763,
      "step": 9974
    },
    {
      "epoch": 1.3980378416257884,
      "grad_norm": 0.456534206867218,
      "learning_rate": 0.00016409949772781628,
      "loss": 0.0761,
      "step": 9975
    },
    {
      "epoch": 1.398177995795375,
      "grad_norm": 0.2686010003089905,
      "learning_rate": 0.00016408514709399664,
      "loss": 0.0342,
      "step": 9976
    },
    {
      "epoch": 1.3983181499649615,
      "grad_norm": 0.4605688452720642,
      "learning_rate": 0.000164070796460177,
      "loss": 0.0509,
      "step": 9977
    },
    {
      "epoch": 1.398458304134548,
      "grad_norm": 0.26086583733558655,
      "learning_rate": 0.00016405644582635732,
      "loss": 0.043,
      "step": 9978
    },
    {
      "epoch": 1.3985984583041344,
      "grad_norm": 0.1798989325761795,
      "learning_rate": 0.00016404209519253768,
      "loss": 0.0291,
      "step": 9979
    },
    {
      "epoch": 1.398738612473721,
      "grad_norm": 0.27406424283981323,
      "learning_rate": 0.000164027744558718,
      "loss": 0.0292,
      "step": 9980
    },
    {
      "epoch": 1.3988787666433076,
      "grad_norm": 0.6544710993766785,
      "learning_rate": 0.00016401339392489833,
      "loss": 0.0131,
      "step": 9981
    },
    {
      "epoch": 1.3990189208128943,
      "grad_norm": 1.0445932149887085,
      "learning_rate": 0.0001639990432910787,
      "loss": 0.1203,
      "step": 9982
    },
    {
      "epoch": 1.3991590749824807,
      "grad_norm": 0.4544623792171478,
      "learning_rate": 0.00016398469265725902,
      "loss": 0.0503,
      "step": 9983
    },
    {
      "epoch": 1.3992992291520672,
      "grad_norm": 0.6019479036331177,
      "learning_rate": 0.00016397034202343935,
      "loss": 0.032,
      "step": 9984
    },
    {
      "epoch": 1.3994393833216539,
      "grad_norm": 1.3350580930709839,
      "learning_rate": 0.00016395599138961967,
      "loss": 0.1664,
      "step": 9985
    },
    {
      "epoch": 1.3995795374912403,
      "grad_norm": 0.1536942422389984,
      "learning_rate": 0.00016394164075580003,
      "loss": 0.0228,
      "step": 9986
    },
    {
      "epoch": 1.399719691660827,
      "grad_norm": 0.35316482186317444,
      "learning_rate": 0.00016392729012198036,
      "loss": 0.0256,
      "step": 9987
    },
    {
      "epoch": 1.3998598458304135,
      "grad_norm": 0.1834089756011963,
      "learning_rate": 0.0001639129394881607,
      "loss": 0.0161,
      "step": 9988
    },
    {
      "epoch": 1.4,
      "grad_norm": 0.4867200553417206,
      "learning_rate": 0.00016389858885434107,
      "loss": 0.0732,
      "step": 9989
    },
    {
      "epoch": 1.4001401541695866,
      "grad_norm": 0.26918044686317444,
      "learning_rate": 0.0001638842382205214,
      "loss": 0.0245,
      "step": 9990
    },
    {
      "epoch": 1.400280308339173,
      "grad_norm": 0.1925928294658661,
      "learning_rate": 0.00016386988758670173,
      "loss": 0.0199,
      "step": 9991
    },
    {
      "epoch": 1.4004204625087597,
      "grad_norm": 0.5353230237960815,
      "learning_rate": 0.00016385553695288208,
      "loss": 0.0599,
      "step": 9992
    },
    {
      "epoch": 1.4005606166783462,
      "grad_norm": 1.0651512145996094,
      "learning_rate": 0.0001638411863190624,
      "loss": 0.053,
      "step": 9993
    },
    {
      "epoch": 1.4007007708479327,
      "grad_norm": 0.31378835439682007,
      "learning_rate": 0.00016382683568524274,
      "loss": 0.0547,
      "step": 9994
    },
    {
      "epoch": 1.4008409250175193,
      "grad_norm": 0.23990264534950256,
      "learning_rate": 0.0001638124850514231,
      "loss": 0.0441,
      "step": 9995
    },
    {
      "epoch": 1.4009810791871058,
      "grad_norm": 0.538273811340332,
      "learning_rate": 0.00016379813441760342,
      "loss": 0.0667,
      "step": 9996
    },
    {
      "epoch": 1.4011212333566925,
      "grad_norm": 0.4067722260951996,
      "learning_rate": 0.00016378378378378375,
      "loss": 0.0829,
      "step": 9997
    },
    {
      "epoch": 1.401261387526279,
      "grad_norm": 0.26510345935821533,
      "learning_rate": 0.00016376943314996414,
      "loss": 0.0667,
      "step": 9998
    },
    {
      "epoch": 1.4014015416958654,
      "grad_norm": 0.3448517918586731,
      "learning_rate": 0.00016375508251614446,
      "loss": 0.0188,
      "step": 9999
    },
    {
      "epoch": 1.401541695865452,
      "grad_norm": 0.6394071578979492,
      "learning_rate": 0.0001637407318823248,
      "loss": 0.0767,
      "step": 10000
    },
    {
      "epoch": 1.4016818500350385,
      "grad_norm": 0.36078816652297974,
      "learning_rate": 0.00016372638124850512,
      "loss": 0.0373,
      "step": 10001
    },
    {
      "epoch": 1.4018220042046252,
      "grad_norm": 0.5789733529090881,
      "learning_rate": 0.00016371203061468548,
      "loss": 0.0424,
      "step": 10002
    },
    {
      "epoch": 1.4019621583742117,
      "grad_norm": 0.308187335729599,
      "learning_rate": 0.0001636976799808658,
      "loss": 0.0769,
      "step": 10003
    },
    {
      "epoch": 1.4021023125437981,
      "grad_norm": 0.21963247656822205,
      "learning_rate": 0.00016368332934704613,
      "loss": 0.0153,
      "step": 10004
    },
    {
      "epoch": 1.4022424667133846,
      "grad_norm": 0.13228927552700043,
      "learning_rate": 0.0001636689787132265,
      "loss": 0.0064,
      "step": 10005
    },
    {
      "epoch": 1.4023826208829713,
      "grad_norm": 0.20588622987270355,
      "learning_rate": 0.00016365462807940682,
      "loss": 0.0135,
      "step": 10006
    },
    {
      "epoch": 1.4025227750525577,
      "grad_norm": 0.49017664790153503,
      "learning_rate": 0.00016364027744558715,
      "loss": 0.0428,
      "step": 10007
    },
    {
      "epoch": 1.4026629292221444,
      "grad_norm": 0.22865651547908783,
      "learning_rate": 0.00016362592681176753,
      "loss": 0.0161,
      "step": 10008
    },
    {
      "epoch": 1.4028030833917309,
      "grad_norm": 0.17678236961364746,
      "learning_rate": 0.00016361157617794786,
      "loss": 0.0231,
      "step": 10009
    },
    {
      "epoch": 1.4029432375613173,
      "grad_norm": 0.24384216964244843,
      "learning_rate": 0.0001635972255441282,
      "loss": 0.0195,
      "step": 10010
    },
    {
      "epoch": 1.403083391730904,
      "grad_norm": 0.3762162923812866,
      "learning_rate": 0.00016358287491030854,
      "loss": 0.0213,
      "step": 10011
    },
    {
      "epoch": 1.4032235459004905,
      "grad_norm": 0.4076460003852844,
      "learning_rate": 0.00016356852427648887,
      "loss": 0.0413,
      "step": 10012
    },
    {
      "epoch": 1.4033637000700772,
      "grad_norm": 0.5227800011634827,
      "learning_rate": 0.0001635541736426692,
      "loss": 0.054,
      "step": 10013
    },
    {
      "epoch": 1.4035038542396636,
      "grad_norm": 0.3644075393676758,
      "learning_rate": 0.00016353982300884955,
      "loss": 0.042,
      "step": 10014
    },
    {
      "epoch": 1.40364400840925,
      "grad_norm": 0.3115498423576355,
      "learning_rate": 0.00016352547237502988,
      "loss": 0.0642,
      "step": 10015
    },
    {
      "epoch": 1.4037841625788368,
      "grad_norm": 0.3390819728374481,
      "learning_rate": 0.0001635111217412102,
      "loss": 0.0405,
      "step": 10016
    },
    {
      "epoch": 1.4039243167484232,
      "grad_norm": 0.3525443971157074,
      "learning_rate": 0.00016349677110739057,
      "loss": 0.0478,
      "step": 10017
    },
    {
      "epoch": 1.40406447091801,
      "grad_norm": 0.3489852547645569,
      "learning_rate": 0.0001634824204735709,
      "loss": 0.0905,
      "step": 10018
    },
    {
      "epoch": 1.4042046250875964,
      "grad_norm": 0.4071124196052551,
      "learning_rate": 0.00016346806983975122,
      "loss": 0.0609,
      "step": 10019
    },
    {
      "epoch": 1.4043447792571828,
      "grad_norm": 0.15567345917224884,
      "learning_rate": 0.00016345371920593155,
      "loss": 0.0509,
      "step": 10020
    },
    {
      "epoch": 1.4044849334267695,
      "grad_norm": 0.062040187418460846,
      "learning_rate": 0.00016343936857211194,
      "loss": 0.0079,
      "step": 10021
    },
    {
      "epoch": 1.404625087596356,
      "grad_norm": 0.3235335350036621,
      "learning_rate": 0.00016342501793829226,
      "loss": 0.0341,
      "step": 10022
    },
    {
      "epoch": 1.4047652417659426,
      "grad_norm": 0.5629802346229553,
      "learning_rate": 0.0001634106673044726,
      "loss": 0.0875,
      "step": 10023
    },
    {
      "epoch": 1.404905395935529,
      "grad_norm": 0.42549726366996765,
      "learning_rate": 0.00016339631667065295,
      "loss": 0.0545,
      "step": 10024
    },
    {
      "epoch": 1.4050455501051156,
      "grad_norm": 0.4296789765357971,
      "learning_rate": 0.00016338196603683328,
      "loss": 0.1332,
      "step": 10025
    },
    {
      "epoch": 1.4051857042747022,
      "grad_norm": 0.4154438376426697,
      "learning_rate": 0.0001633676154030136,
      "loss": 0.066,
      "step": 10026
    },
    {
      "epoch": 1.4053258584442887,
      "grad_norm": 0.8033463358879089,
      "learning_rate": 0.00016335326476919396,
      "loss": 0.0661,
      "step": 10027
    },
    {
      "epoch": 1.4054660126138754,
      "grad_norm": 0.23800277709960938,
      "learning_rate": 0.0001633389141353743,
      "loss": 0.0463,
      "step": 10028
    },
    {
      "epoch": 1.4056061667834618,
      "grad_norm": 1.1812933683395386,
      "learning_rate": 0.00016332456350155462,
      "loss": 0.0678,
      "step": 10029
    },
    {
      "epoch": 1.4057463209530483,
      "grad_norm": 1.805396556854248,
      "learning_rate": 0.000163310212867735,
      "loss": 0.073,
      "step": 10030
    },
    {
      "epoch": 1.4058864751226348,
      "grad_norm": 0.525574266910553,
      "learning_rate": 0.00016329586223391533,
      "loss": 0.0885,
      "step": 10031
    },
    {
      "epoch": 1.4060266292922214,
      "grad_norm": 0.24835549294948578,
      "learning_rate": 0.00016328151160009566,
      "loss": 0.0584,
      "step": 10032
    },
    {
      "epoch": 1.4061667834618081,
      "grad_norm": 0.3491349220275879,
      "learning_rate": 0.00016326716096627601,
      "loss": 0.0255,
      "step": 10033
    },
    {
      "epoch": 1.4063069376313946,
      "grad_norm": 0.9566276669502258,
      "learning_rate": 0.00016325281033245634,
      "loss": 0.1137,
      "step": 10034
    },
    {
      "epoch": 1.406447091800981,
      "grad_norm": 1.8127440214157104,
      "learning_rate": 0.00016323845969863667,
      "loss": 0.1942,
      "step": 10035
    },
    {
      "epoch": 1.4065872459705675,
      "grad_norm": 0.40884366631507874,
      "learning_rate": 0.000163224109064817,
      "loss": 0.0542,
      "step": 10036
    },
    {
      "epoch": 1.4067274001401542,
      "grad_norm": 0.45456331968307495,
      "learning_rate": 0.00016320975843099736,
      "loss": 0.0824,
      "step": 10037
    },
    {
      "epoch": 1.4068675543097406,
      "grad_norm": 0.5281078219413757,
      "learning_rate": 0.00016319540779717768,
      "loss": 0.0659,
      "step": 10038
    },
    {
      "epoch": 1.4070077084793273,
      "grad_norm": 0.07686958461999893,
      "learning_rate": 0.000163181057163358,
      "loss": 0.0109,
      "step": 10039
    },
    {
      "epoch": 1.4071478626489138,
      "grad_norm": 0.2552245855331421,
      "learning_rate": 0.0001631667065295384,
      "loss": 0.0527,
      "step": 10040
    },
    {
      "epoch": 1.4072880168185002,
      "grad_norm": 0.31303641200065613,
      "learning_rate": 0.00016315235589571872,
      "loss": 0.057,
      "step": 10041
    },
    {
      "epoch": 1.407428170988087,
      "grad_norm": 0.26873889565467834,
      "learning_rate": 0.00016313800526189905,
      "loss": 0.0371,
      "step": 10042
    },
    {
      "epoch": 1.4075683251576734,
      "grad_norm": 0.22019919753074646,
      "learning_rate": 0.0001631236546280794,
      "loss": 0.0303,
      "step": 10043
    },
    {
      "epoch": 1.40770847932726,
      "grad_norm": 0.49361923336982727,
      "learning_rate": 0.00016310930399425974,
      "loss": 0.0486,
      "step": 10044
    },
    {
      "epoch": 1.4078486334968465,
      "grad_norm": 0.19308622181415558,
      "learning_rate": 0.00016309495336044007,
      "loss": 0.0395,
      "step": 10045
    },
    {
      "epoch": 1.407988787666433,
      "grad_norm": 0.09779738634824753,
      "learning_rate": 0.00016308060272662042,
      "loss": 0.0134,
      "step": 10046
    },
    {
      "epoch": 1.4081289418360197,
      "grad_norm": 0.3088248372077942,
      "learning_rate": 0.00016306625209280075,
      "loss": 0.0259,
      "step": 10047
    },
    {
      "epoch": 1.4082690960056061,
      "grad_norm": 0.17860445380210876,
      "learning_rate": 0.00016305190145898108,
      "loss": 0.0301,
      "step": 10048
    },
    {
      "epoch": 1.4084092501751928,
      "grad_norm": 0.359048455953598,
      "learning_rate": 0.00016303755082516143,
      "loss": 0.0317,
      "step": 10049
    },
    {
      "epoch": 1.4085494043447793,
      "grad_norm": 0.33246049284935,
      "learning_rate": 0.00016302320019134176,
      "loss": 0.0673,
      "step": 10050
    },
    {
      "epoch": 1.4086895585143657,
      "grad_norm": 0.11317368596792221,
      "learning_rate": 0.0001630088495575221,
      "loss": 0.014,
      "step": 10051
    },
    {
      "epoch": 1.4088297126839524,
      "grad_norm": 0.530755341053009,
      "learning_rate": 0.00016299449892370247,
      "loss": 0.0703,
      "step": 10052
    },
    {
      "epoch": 1.4089698668535389,
      "grad_norm": 0.3518270254135132,
      "learning_rate": 0.0001629801482898828,
      "loss": 0.0581,
      "step": 10053
    },
    {
      "epoch": 1.4091100210231255,
      "grad_norm": 0.13017380237579346,
      "learning_rate": 0.00016296579765606313,
      "loss": 0.0124,
      "step": 10054
    },
    {
      "epoch": 1.409250175192712,
      "grad_norm": 0.3444695472717285,
      "learning_rate": 0.00016295144702224346,
      "loss": 0.0429,
      "step": 10055
    },
    {
      "epoch": 1.4093903293622985,
      "grad_norm": 0.4072650969028473,
      "learning_rate": 0.00016293709638842382,
      "loss": 0.0347,
      "step": 10056
    },
    {
      "epoch": 1.4095304835318851,
      "grad_norm": 0.04853210225701332,
      "learning_rate": 0.00016292274575460414,
      "loss": 0.0043,
      "step": 10057
    },
    {
      "epoch": 1.4096706377014716,
      "grad_norm": 0.1448264718055725,
      "learning_rate": 0.00016290839512078447,
      "loss": 0.0177,
      "step": 10058
    },
    {
      "epoch": 1.4098107918710583,
      "grad_norm": 0.3599625825881958,
      "learning_rate": 0.00016289404448696483,
      "loss": 0.0445,
      "step": 10059
    },
    {
      "epoch": 1.4099509460406447,
      "grad_norm": 0.29080915451049805,
      "learning_rate": 0.00016287969385314516,
      "loss": 0.0864,
      "step": 10060
    },
    {
      "epoch": 1.4100911002102312,
      "grad_norm": 0.1319163590669632,
      "learning_rate": 0.00016286534321932549,
      "loss": 0.0293,
      "step": 10061
    },
    {
      "epoch": 1.4102312543798177,
      "grad_norm": 0.21962696313858032,
      "learning_rate": 0.00016285099258550587,
      "loss": 0.0247,
      "step": 10062
    },
    {
      "epoch": 1.4103714085494043,
      "grad_norm": 0.25393569469451904,
      "learning_rate": 0.0001628366419516862,
      "loss": 0.024,
      "step": 10063
    },
    {
      "epoch": 1.4105115627189908,
      "grad_norm": 0.732363224029541,
      "learning_rate": 0.00016282229131786653,
      "loss": 0.0867,
      "step": 10064
    },
    {
      "epoch": 1.4106517168885775,
      "grad_norm": 0.31279635429382324,
      "learning_rate": 0.00016280794068404688,
      "loss": 0.0456,
      "step": 10065
    },
    {
      "epoch": 1.410791871058164,
      "grad_norm": 0.45657625794410706,
      "learning_rate": 0.0001627935900502272,
      "loss": 0.1078,
      "step": 10066
    },
    {
      "epoch": 1.4109320252277504,
      "grad_norm": 0.48153361678123474,
      "learning_rate": 0.00016277923941640754,
      "loss": 0.1648,
      "step": 10067
    },
    {
      "epoch": 1.411072179397337,
      "grad_norm": 0.38867172598838806,
      "learning_rate": 0.0001627648887825879,
      "loss": 0.0593,
      "step": 10068
    },
    {
      "epoch": 1.4112123335669235,
      "grad_norm": 0.4148240089416504,
      "learning_rate": 0.00016275053814876822,
      "loss": 0.087,
      "step": 10069
    },
    {
      "epoch": 1.4113524877365102,
      "grad_norm": 0.35816872119903564,
      "learning_rate": 0.00016273618751494855,
      "loss": 0.0121,
      "step": 10070
    },
    {
      "epoch": 1.4114926419060967,
      "grad_norm": 0.28633856773376465,
      "learning_rate": 0.00016272183688112888,
      "loss": 0.0375,
      "step": 10071
    },
    {
      "epoch": 1.4116327960756831,
      "grad_norm": 0.4820577800273895,
      "learning_rate": 0.00016270748624730926,
      "loss": 0.0477,
      "step": 10072
    },
    {
      "epoch": 1.4117729502452698,
      "grad_norm": 0.42785048484802246,
      "learning_rate": 0.0001626931356134896,
      "loss": 0.0428,
      "step": 10073
    },
    {
      "epoch": 1.4119131044148563,
      "grad_norm": 0.15436211228370667,
      "learning_rate": 0.00016267878497966992,
      "loss": 0.0342,
      "step": 10074
    },
    {
      "epoch": 1.412053258584443,
      "grad_norm": 0.42595529556274414,
      "learning_rate": 0.00016266443434585027,
      "loss": 0.0294,
      "step": 10075
    },
    {
      "epoch": 1.4121934127540294,
      "grad_norm": 0.13147082924842834,
      "learning_rate": 0.0001626500837120306,
      "loss": 0.0143,
      "step": 10076
    },
    {
      "epoch": 1.412333566923616,
      "grad_norm": 0.35984864830970764,
      "learning_rate": 0.00016263573307821093,
      "loss": 0.0338,
      "step": 10077
    },
    {
      "epoch": 1.4124737210932026,
      "grad_norm": 0.5684127807617188,
      "learning_rate": 0.0001626213824443913,
      "loss": 0.0789,
      "step": 10078
    },
    {
      "epoch": 1.412613875262789,
      "grad_norm": 0.44349196553230286,
      "learning_rate": 0.00016260703181057162,
      "loss": 0.0422,
      "step": 10079
    },
    {
      "epoch": 1.4127540294323757,
      "grad_norm": 0.3814984858036041,
      "learning_rate": 0.00016259268117675194,
      "loss": 0.0343,
      "step": 10080
    },
    {
      "epoch": 1.4128941836019622,
      "grad_norm": 0.45309409499168396,
      "learning_rate": 0.0001625783305429323,
      "loss": 0.0307,
      "step": 10081
    },
    {
      "epoch": 1.4130343377715486,
      "grad_norm": 1.0413120985031128,
      "learning_rate": 0.00016256397990911263,
      "loss": 0.1161,
      "step": 10082
    },
    {
      "epoch": 1.4131744919411353,
      "grad_norm": 0.6609210968017578,
      "learning_rate": 0.00016254962927529296,
      "loss": 0.0809,
      "step": 10083
    },
    {
      "epoch": 1.4133146461107218,
      "grad_norm": 0.49330639839172363,
      "learning_rate": 0.00016253527864147334,
      "loss": 0.03,
      "step": 10084
    },
    {
      "epoch": 1.4134548002803085,
      "grad_norm": 2.1410765647888184,
      "learning_rate": 0.00016252092800765367,
      "loss": 0.2454,
      "step": 10085
    },
    {
      "epoch": 1.413594954449895,
      "grad_norm": 0.5847886204719543,
      "learning_rate": 0.000162506577373834,
      "loss": 0.059,
      "step": 10086
    },
    {
      "epoch": 1.4137351086194814,
      "grad_norm": 0.16128098964691162,
      "learning_rate": 0.00016249222674001435,
      "loss": 0.0109,
      "step": 10087
    },
    {
      "epoch": 1.413875262789068,
      "grad_norm": 0.18059886991977692,
      "learning_rate": 0.00016247787610619468,
      "loss": 0.0153,
      "step": 10088
    },
    {
      "epoch": 1.4140154169586545,
      "grad_norm": 0.17081362009048462,
      "learning_rate": 0.000162463525472375,
      "loss": 0.0429,
      "step": 10089
    },
    {
      "epoch": 1.4141555711282412,
      "grad_norm": 0.546936571598053,
      "learning_rate": 0.00016244917483855534,
      "loss": 0.0825,
      "step": 10090
    },
    {
      "epoch": 1.4142957252978277,
      "grad_norm": 0.3452431261539459,
      "learning_rate": 0.0001624348242047357,
      "loss": 0.0504,
      "step": 10091
    },
    {
      "epoch": 1.4144358794674141,
      "grad_norm": 0.30322957038879395,
      "learning_rate": 0.00016242047357091602,
      "loss": 0.0403,
      "step": 10092
    },
    {
      "epoch": 1.4145760336370006,
      "grad_norm": 0.27789390087127686,
      "learning_rate": 0.00016240612293709635,
      "loss": 0.0334,
      "step": 10093
    },
    {
      "epoch": 1.4147161878065873,
      "grad_norm": 0.23856988549232483,
      "learning_rate": 0.00016239177230327673,
      "loss": 0.0295,
      "step": 10094
    },
    {
      "epoch": 1.4148563419761737,
      "grad_norm": 0.32311302423477173,
      "learning_rate": 0.00016237742166945706,
      "loss": 0.0573,
      "step": 10095
    },
    {
      "epoch": 1.4149964961457604,
      "grad_norm": 0.22626614570617676,
      "learning_rate": 0.0001623630710356374,
      "loss": 0.0451,
      "step": 10096
    },
    {
      "epoch": 1.4151366503153469,
      "grad_norm": 1.0516315698623657,
      "learning_rate": 0.00016234872040181775,
      "loss": 0.0646,
      "step": 10097
    },
    {
      "epoch": 1.4152768044849333,
      "grad_norm": 0.243377223610878,
      "learning_rate": 0.00016233436976799808,
      "loss": 0.0306,
      "step": 10098
    },
    {
      "epoch": 1.41541695865452,
      "grad_norm": 0.2931329011917114,
      "learning_rate": 0.0001623200191341784,
      "loss": 0.0579,
      "step": 10099
    },
    {
      "epoch": 1.4155571128241065,
      "grad_norm": 0.30589163303375244,
      "learning_rate": 0.00016230566850035876,
      "loss": 0.0263,
      "step": 10100
    },
    {
      "epoch": 1.4156972669936931,
      "grad_norm": 0.15806232392787933,
      "learning_rate": 0.0001622913178665391,
      "loss": 0.0233,
      "step": 10101
    },
    {
      "epoch": 1.4158374211632796,
      "grad_norm": 0.4234723746776581,
      "learning_rate": 0.00016227696723271942,
      "loss": 0.0417,
      "step": 10102
    },
    {
      "epoch": 1.415977575332866,
      "grad_norm": 0.5165902376174927,
      "learning_rate": 0.0001622626165988998,
      "loss": 0.0536,
      "step": 10103
    },
    {
      "epoch": 1.4161177295024527,
      "grad_norm": 0.16471479833126068,
      "learning_rate": 0.00016224826596508013,
      "loss": 0.0225,
      "step": 10104
    },
    {
      "epoch": 1.4162578836720392,
      "grad_norm": 0.6931836009025574,
      "learning_rate": 0.00016223391533126046,
      "loss": 0.112,
      "step": 10105
    },
    {
      "epoch": 1.4163980378416259,
      "grad_norm": 0.15529198944568634,
      "learning_rate": 0.00016221956469744079,
      "loss": 0.024,
      "step": 10106
    },
    {
      "epoch": 1.4165381920112123,
      "grad_norm": 0.19974516332149506,
      "learning_rate": 0.00016220521406362114,
      "loss": 0.0262,
      "step": 10107
    },
    {
      "epoch": 1.4166783461807988,
      "grad_norm": 0.13752387464046478,
      "learning_rate": 0.00016219086342980147,
      "loss": 0.0169,
      "step": 10108
    },
    {
      "epoch": 1.4168185003503855,
      "grad_norm": 0.5949627161026001,
      "learning_rate": 0.0001621765127959818,
      "loss": 0.0533,
      "step": 10109
    },
    {
      "epoch": 1.416958654519972,
      "grad_norm": 0.17016081511974335,
      "learning_rate": 0.00016216216216216215,
      "loss": 0.0118,
      "step": 10110
    },
    {
      "epoch": 1.4170988086895586,
      "grad_norm": 0.22535526752471924,
      "learning_rate": 0.00016214781152834248,
      "loss": 0.0573,
      "step": 10111
    },
    {
      "epoch": 1.417238962859145,
      "grad_norm": 0.88917076587677,
      "learning_rate": 0.0001621334608945228,
      "loss": 0.0894,
      "step": 10112
    },
    {
      "epoch": 1.4173791170287315,
      "grad_norm": 0.6522378921508789,
      "learning_rate": 0.00016211911026070317,
      "loss": 0.0809,
      "step": 10113
    },
    {
      "epoch": 1.4175192711983182,
      "grad_norm": 0.4226674437522888,
      "learning_rate": 0.0001621047596268835,
      "loss": 0.0965,
      "step": 10114
    },
    {
      "epoch": 1.4176594253679047,
      "grad_norm": 0.09856314957141876,
      "learning_rate": 0.00016209040899306382,
      "loss": 0.0118,
      "step": 10115
    },
    {
      "epoch": 1.4177995795374914,
      "grad_norm": 0.7640262842178345,
      "learning_rate": 0.0001620760583592442,
      "loss": 0.069,
      "step": 10116
    },
    {
      "epoch": 1.4179397337070778,
      "grad_norm": 0.35214075446128845,
      "learning_rate": 0.00016206170772542454,
      "loss": 0.0344,
      "step": 10117
    },
    {
      "epoch": 1.4180798878766643,
      "grad_norm": 0.6715491414070129,
      "learning_rate": 0.00016204735709160486,
      "loss": 0.0335,
      "step": 10118
    },
    {
      "epoch": 1.4182200420462507,
      "grad_norm": 1.1098538637161255,
      "learning_rate": 0.00016203300645778522,
      "loss": 0.0823,
      "step": 10119
    },
    {
      "epoch": 1.4183601962158374,
      "grad_norm": 0.15266643464565277,
      "learning_rate": 0.00016201865582396555,
      "loss": 0.029,
      "step": 10120
    },
    {
      "epoch": 1.418500350385424,
      "grad_norm": 0.290271133184433,
      "learning_rate": 0.00016200430519014588,
      "loss": 0.0309,
      "step": 10121
    },
    {
      "epoch": 1.4186405045550106,
      "grad_norm": 1.9331448078155518,
      "learning_rate": 0.0001619899545563262,
      "loss": 0.0382,
      "step": 10122
    },
    {
      "epoch": 1.418780658724597,
      "grad_norm": 0.13446784019470215,
      "learning_rate": 0.00016197560392250656,
      "loss": 0.0404,
      "step": 10123
    },
    {
      "epoch": 1.4189208128941835,
      "grad_norm": 0.46477991342544556,
      "learning_rate": 0.0001619612532886869,
      "loss": 0.0375,
      "step": 10124
    },
    {
      "epoch": 1.4190609670637702,
      "grad_norm": 0.23926281929016113,
      "learning_rate": 0.00016194690265486722,
      "loss": 0.0317,
      "step": 10125
    },
    {
      "epoch": 1.4192011212333566,
      "grad_norm": 1.2273597717285156,
      "learning_rate": 0.0001619325520210476,
      "loss": 0.0739,
      "step": 10126
    },
    {
      "epoch": 1.4193412754029433,
      "grad_norm": 0.3223104774951935,
      "learning_rate": 0.00016191820138722793,
      "loss": 0.0295,
      "step": 10127
    },
    {
      "epoch": 1.4194814295725298,
      "grad_norm": 0.7063602805137634,
      "learning_rate": 0.00016190385075340826,
      "loss": 0.1277,
      "step": 10128
    },
    {
      "epoch": 1.4196215837421162,
      "grad_norm": 0.31807786226272583,
      "learning_rate": 0.0001618895001195886,
      "loss": 0.0715,
      "step": 10129
    },
    {
      "epoch": 1.419761737911703,
      "grad_norm": 0.4765188694000244,
      "learning_rate": 0.00016187514948576894,
      "loss": 0.0871,
      "step": 10130
    },
    {
      "epoch": 1.4199018920812894,
      "grad_norm": 0.3904831111431122,
      "learning_rate": 0.00016186079885194927,
      "loss": 0.0461,
      "step": 10131
    },
    {
      "epoch": 1.420042046250876,
      "grad_norm": 0.40957337617874146,
      "learning_rate": 0.00016184644821812963,
      "loss": 0.0222,
      "step": 10132
    },
    {
      "epoch": 1.4201822004204625,
      "grad_norm": 2.021178722381592,
      "learning_rate": 0.00016183209758430995,
      "loss": 0.225,
      "step": 10133
    },
    {
      "epoch": 1.420322354590049,
      "grad_norm": 1.0561689138412476,
      "learning_rate": 0.00016181774695049028,
      "loss": 0.013,
      "step": 10134
    },
    {
      "epoch": 1.4204625087596356,
      "grad_norm": 3.7962841987609863,
      "learning_rate": 0.00016180339631667067,
      "loss": 0.3033,
      "step": 10135
    },
    {
      "epoch": 1.420602662929222,
      "grad_norm": 0.2517479956150055,
      "learning_rate": 0.000161789045682851,
      "loss": 0.0532,
      "step": 10136
    },
    {
      "epoch": 1.4207428170988088,
      "grad_norm": 0.23859789967536926,
      "learning_rate": 0.00016177469504903132,
      "loss": 0.0729,
      "step": 10137
    },
    {
      "epoch": 1.4208829712683952,
      "grad_norm": 0.7044009566307068,
      "learning_rate": 0.00016176034441521168,
      "loss": 0.102,
      "step": 10138
    },
    {
      "epoch": 1.4210231254379817,
      "grad_norm": 0.431848406791687,
      "learning_rate": 0.000161745993781392,
      "loss": 0.0387,
      "step": 10139
    },
    {
      "epoch": 1.4211632796075684,
      "grad_norm": 0.1918434500694275,
      "learning_rate": 0.00016173164314757234,
      "loss": 0.0216,
      "step": 10140
    },
    {
      "epoch": 1.4213034337771548,
      "grad_norm": 0.48974621295928955,
      "learning_rate": 0.00016171729251375266,
      "loss": 0.0809,
      "step": 10141
    },
    {
      "epoch": 1.4214435879467415,
      "grad_norm": 0.54285728931427,
      "learning_rate": 0.00016170294187993302,
      "loss": 0.1183,
      "step": 10142
    },
    {
      "epoch": 1.421583742116328,
      "grad_norm": 0.39793580770492554,
      "learning_rate": 0.00016168859124611335,
      "loss": 0.0733,
      "step": 10143
    },
    {
      "epoch": 1.4217238962859144,
      "grad_norm": 0.30271995067596436,
      "learning_rate": 0.00016167424061229368,
      "loss": 0.0588,
      "step": 10144
    },
    {
      "epoch": 1.4218640504555011,
      "grad_norm": 0.15205909311771393,
      "learning_rate": 0.00016165988997847403,
      "loss": 0.0176,
      "step": 10145
    },
    {
      "epoch": 1.4220042046250876,
      "grad_norm": 0.30136293172836304,
      "learning_rate": 0.00016164553934465436,
      "loss": 0.1326,
      "step": 10146
    },
    {
      "epoch": 1.4221443587946743,
      "grad_norm": 0.1649971604347229,
      "learning_rate": 0.0001616311887108347,
      "loss": 0.0231,
      "step": 10147
    },
    {
      "epoch": 1.4222845129642607,
      "grad_norm": 0.08201868832111359,
      "learning_rate": 0.00016161683807701507,
      "loss": 0.0066,
      "step": 10148
    },
    {
      "epoch": 1.4224246671338472,
      "grad_norm": 0.3914889991283417,
      "learning_rate": 0.0001616024874431954,
      "loss": 0.0584,
      "step": 10149
    },
    {
      "epoch": 1.4225648213034336,
      "grad_norm": 0.5152369737625122,
      "learning_rate": 0.00016158813680937573,
      "loss": 0.0676,
      "step": 10150
    },
    {
      "epoch": 1.4227049754730203,
      "grad_norm": 0.14055362343788147,
      "learning_rate": 0.00016157378617555609,
      "loss": 0.0076,
      "step": 10151
    },
    {
      "epoch": 1.4228451296426068,
      "grad_norm": 0.1619841605424881,
      "learning_rate": 0.00016155943554173641,
      "loss": 0.0176,
      "step": 10152
    },
    {
      "epoch": 1.4229852838121935,
      "grad_norm": 0.5846623778343201,
      "learning_rate": 0.00016154508490791674,
      "loss": 0.0905,
      "step": 10153
    },
    {
      "epoch": 1.42312543798178,
      "grad_norm": 0.30887091159820557,
      "learning_rate": 0.0001615307342740971,
      "loss": 0.0803,
      "step": 10154
    },
    {
      "epoch": 1.4232655921513664,
      "grad_norm": 0.6513183116912842,
      "learning_rate": 0.00016151638364027743,
      "loss": 0.0666,
      "step": 10155
    },
    {
      "epoch": 1.423405746320953,
      "grad_norm": 0.3283507227897644,
      "learning_rate": 0.00016150203300645776,
      "loss": 0.0524,
      "step": 10156
    },
    {
      "epoch": 1.4235459004905395,
      "grad_norm": 0.2082063853740692,
      "learning_rate": 0.00016148768237263808,
      "loss": 0.0385,
      "step": 10157
    },
    {
      "epoch": 1.4236860546601262,
      "grad_norm": 0.32522910833358765,
      "learning_rate": 0.00016147333173881847,
      "loss": 0.0565,
      "step": 10158
    },
    {
      "epoch": 1.4238262088297127,
      "grad_norm": 0.20435641705989838,
      "learning_rate": 0.0001614589811049988,
      "loss": 0.0221,
      "step": 10159
    },
    {
      "epoch": 1.4239663629992991,
      "grad_norm": 0.43393731117248535,
      "learning_rate": 0.00016144463047117912,
      "loss": 0.0561,
      "step": 10160
    },
    {
      "epoch": 1.4241065171688858,
      "grad_norm": 0.24540644884109497,
      "learning_rate": 0.00016143027983735948,
      "loss": 0.0645,
      "step": 10161
    },
    {
      "epoch": 1.4242466713384723,
      "grad_norm": 0.3852110207080841,
      "learning_rate": 0.0001614159292035398,
      "loss": 0.0388,
      "step": 10162
    },
    {
      "epoch": 1.424386825508059,
      "grad_norm": 0.33503320813179016,
      "learning_rate": 0.00016140157856972014,
      "loss": 0.0181,
      "step": 10163
    },
    {
      "epoch": 1.4245269796776454,
      "grad_norm": 0.2334572970867157,
      "learning_rate": 0.0001613872279359005,
      "loss": 0.0445,
      "step": 10164
    },
    {
      "epoch": 1.4246671338472319,
      "grad_norm": 0.6089301109313965,
      "learning_rate": 0.00016137287730208082,
      "loss": 0.0205,
      "step": 10165
    },
    {
      "epoch": 1.4248072880168186,
      "grad_norm": 0.8390827178955078,
      "learning_rate": 0.00016135852666826115,
      "loss": 0.142,
      "step": 10166
    },
    {
      "epoch": 1.424947442186405,
      "grad_norm": 1.045295238494873,
      "learning_rate": 0.00016134417603444153,
      "loss": 0.0549,
      "step": 10167
    },
    {
      "epoch": 1.4250875963559917,
      "grad_norm": 0.4901297986507416,
      "learning_rate": 0.00016132982540062186,
      "loss": 0.0639,
      "step": 10168
    },
    {
      "epoch": 1.4252277505255782,
      "grad_norm": 0.33060768246650696,
      "learning_rate": 0.0001613154747668022,
      "loss": 0.0331,
      "step": 10169
    },
    {
      "epoch": 1.4253679046951646,
      "grad_norm": 0.43889865279197693,
      "learning_rate": 0.00016130112413298255,
      "loss": 0.1722,
      "step": 10170
    },
    {
      "epoch": 1.4255080588647513,
      "grad_norm": 0.15537147223949432,
      "learning_rate": 0.00016128677349916287,
      "loss": 0.0192,
      "step": 10171
    },
    {
      "epoch": 1.4256482130343378,
      "grad_norm": 0.1894560307264328,
      "learning_rate": 0.0001612724228653432,
      "loss": 0.0583,
      "step": 10172
    },
    {
      "epoch": 1.4257883672039244,
      "grad_norm": 0.42838916182518005,
      "learning_rate": 0.00016125807223152356,
      "loss": 0.0956,
      "step": 10173
    },
    {
      "epoch": 1.425928521373511,
      "grad_norm": 0.16536885499954224,
      "learning_rate": 0.00016124372159770389,
      "loss": 0.0147,
      "step": 10174
    },
    {
      "epoch": 1.4260686755430974,
      "grad_norm": 0.4123830795288086,
      "learning_rate": 0.00016122937096388421,
      "loss": 0.0813,
      "step": 10175
    },
    {
      "epoch": 1.426208829712684,
      "grad_norm": 0.29722410440444946,
      "learning_rate": 0.00016121502033006454,
      "loss": 0.0314,
      "step": 10176
    },
    {
      "epoch": 1.4263489838822705,
      "grad_norm": 0.24730896949768066,
      "learning_rate": 0.0001612006696962449,
      "loss": 0.049,
      "step": 10177
    },
    {
      "epoch": 1.4264891380518572,
      "grad_norm": 0.5634830594062805,
      "learning_rate": 0.00016118631906242523,
      "loss": 0.0523,
      "step": 10178
    },
    {
      "epoch": 1.4266292922214436,
      "grad_norm": 0.26038607954978943,
      "learning_rate": 0.00016117196842860556,
      "loss": 0.0206,
      "step": 10179
    },
    {
      "epoch": 1.42676944639103,
      "grad_norm": 0.5175793170928955,
      "learning_rate": 0.00016115761779478594,
      "loss": 0.0481,
      "step": 10180
    },
    {
      "epoch": 1.4269096005606166,
      "grad_norm": 1.546864628791809,
      "learning_rate": 0.00016114326716096627,
      "loss": 0.1191,
      "step": 10181
    },
    {
      "epoch": 1.4270497547302032,
      "grad_norm": 0.6556147933006287,
      "learning_rate": 0.0001611289165271466,
      "loss": 0.1567,
      "step": 10182
    },
    {
      "epoch": 1.4271899088997897,
      "grad_norm": 0.30948296189308167,
      "learning_rate": 0.00016111456589332695,
      "loss": 0.0592,
      "step": 10183
    },
    {
      "epoch": 1.4273300630693764,
      "grad_norm": 0.600836992263794,
      "learning_rate": 0.00016110021525950728,
      "loss": 0.0468,
      "step": 10184
    },
    {
      "epoch": 1.4274702172389628,
      "grad_norm": 3.3685903549194336,
      "learning_rate": 0.0001610858646256876,
      "loss": 0.259,
      "step": 10185
    },
    {
      "epoch": 1.4276103714085493,
      "grad_norm": 0.3439197838306427,
      "learning_rate": 0.00016107151399186796,
      "loss": 0.0749,
      "step": 10186
    },
    {
      "epoch": 1.427750525578136,
      "grad_norm": 0.5156503319740295,
      "learning_rate": 0.0001610571633580483,
      "loss": 0.0469,
      "step": 10187
    },
    {
      "epoch": 1.4278906797477224,
      "grad_norm": 0.3198818266391754,
      "learning_rate": 0.00016104281272422862,
      "loss": 0.0388,
      "step": 10188
    },
    {
      "epoch": 1.4280308339173091,
      "grad_norm": 0.2796226441860199,
      "learning_rate": 0.000161028462090409,
      "loss": 0.0505,
      "step": 10189
    },
    {
      "epoch": 1.4281709880868956,
      "grad_norm": 0.1790735125541687,
      "learning_rate": 0.00016101411145658933,
      "loss": 0.037,
      "step": 10190
    },
    {
      "epoch": 1.428311142256482,
      "grad_norm": 0.5123526453971863,
      "learning_rate": 0.00016099976082276966,
      "loss": 0.0428,
      "step": 10191
    },
    {
      "epoch": 1.4284512964260687,
      "grad_norm": 0.07718978822231293,
      "learning_rate": 0.00016098541018895,
      "loss": 0.0078,
      "step": 10192
    },
    {
      "epoch": 1.4285914505956552,
      "grad_norm": 0.24048765003681183,
      "learning_rate": 0.00016097105955513035,
      "loss": 0.0541,
      "step": 10193
    },
    {
      "epoch": 1.4287316047652419,
      "grad_norm": 0.35560065507888794,
      "learning_rate": 0.00016095670892131067,
      "loss": 0.0487,
      "step": 10194
    },
    {
      "epoch": 1.4288717589348283,
      "grad_norm": 0.32569611072540283,
      "learning_rate": 0.000160942358287491,
      "loss": 0.071,
      "step": 10195
    },
    {
      "epoch": 1.4290119131044148,
      "grad_norm": 0.44907891750335693,
      "learning_rate": 0.00016092800765367136,
      "loss": 0.0574,
      "step": 10196
    },
    {
      "epoch": 1.4291520672740015,
      "grad_norm": 0.1444093883037567,
      "learning_rate": 0.0001609136570198517,
      "loss": 0.03,
      "step": 10197
    },
    {
      "epoch": 1.429292221443588,
      "grad_norm": 0.58797287940979,
      "learning_rate": 0.00016089930638603202,
      "loss": 0.0758,
      "step": 10198
    },
    {
      "epoch": 1.4294323756131746,
      "grad_norm": 0.16892100870609283,
      "learning_rate": 0.0001608849557522124,
      "loss": 0.027,
      "step": 10199
    },
    {
      "epoch": 1.429572529782761,
      "grad_norm": 0.3921225666999817,
      "learning_rate": 0.00016087060511839273,
      "loss": 0.0509,
      "step": 10200
    },
    {
      "epoch": 1.4297126839523475,
      "grad_norm": 0.42925089597702026,
      "learning_rate": 0.00016085625448457306,
      "loss": 0.0476,
      "step": 10201
    },
    {
      "epoch": 1.4298528381219342,
      "grad_norm": 0.27148038148880005,
      "learning_rate": 0.0001608419038507534,
      "loss": 0.0805,
      "step": 10202
    },
    {
      "epoch": 1.4299929922915207,
      "grad_norm": 0.24084679782390594,
      "learning_rate": 0.00016082755321693374,
      "loss": 0.0587,
      "step": 10203
    },
    {
      "epoch": 1.4301331464611073,
      "grad_norm": 0.22835497558116913,
      "learning_rate": 0.00016081320258311407,
      "loss": 0.0357,
      "step": 10204
    },
    {
      "epoch": 1.4302733006306938,
      "grad_norm": 0.14377571642398834,
      "learning_rate": 0.00016079885194929442,
      "loss": 0.0173,
      "step": 10205
    },
    {
      "epoch": 1.4304134548002803,
      "grad_norm": 0.22109854221343994,
      "learning_rate": 0.00016078450131547475,
      "loss": 0.0388,
      "step": 10206
    },
    {
      "epoch": 1.4305536089698667,
      "grad_norm": 0.7799450159072876,
      "learning_rate": 0.00016077015068165508,
      "loss": 0.0655,
      "step": 10207
    },
    {
      "epoch": 1.4306937631394534,
      "grad_norm": 0.4504595696926117,
      "learning_rate": 0.00016075580004783544,
      "loss": 0.0733,
      "step": 10208
    },
    {
      "epoch": 1.43083391730904,
      "grad_norm": 0.3658653497695923,
      "learning_rate": 0.00016074144941401577,
      "loss": 0.0576,
      "step": 10209
    },
    {
      "epoch": 1.4309740714786265,
      "grad_norm": 0.27942484617233276,
      "learning_rate": 0.0001607270987801961,
      "loss": 0.0511,
      "step": 10210
    },
    {
      "epoch": 1.431114225648213,
      "grad_norm": 0.27003490924835205,
      "learning_rate": 0.00016071274814637642,
      "loss": 0.0474,
      "step": 10211
    },
    {
      "epoch": 1.4312543798177995,
      "grad_norm": 0.7827368378639221,
      "learning_rate": 0.0001606983975125568,
      "loss": 0.0849,
      "step": 10212
    },
    {
      "epoch": 1.4313945339873861,
      "grad_norm": 0.5065380930900574,
      "learning_rate": 0.00016068404687873713,
      "loss": 0.0445,
      "step": 10213
    },
    {
      "epoch": 1.4315346881569726,
      "grad_norm": 0.32973483204841614,
      "learning_rate": 0.00016066969624491746,
      "loss": 0.0447,
      "step": 10214
    },
    {
      "epoch": 1.4316748423265593,
      "grad_norm": 0.3289881646633148,
      "learning_rate": 0.00016065534561109782,
      "loss": 0.0523,
      "step": 10215
    },
    {
      "epoch": 1.4318149964961457,
      "grad_norm": 0.15622155368328094,
      "learning_rate": 0.00016064099497727815,
      "loss": 0.0146,
      "step": 10216
    },
    {
      "epoch": 1.4319551506657322,
      "grad_norm": 0.42556485533714294,
      "learning_rate": 0.00016062664434345848,
      "loss": 0.0521,
      "step": 10217
    },
    {
      "epoch": 1.4320953048353189,
      "grad_norm": 0.20520903170108795,
      "learning_rate": 0.00016061229370963883,
      "loss": 0.0499,
      "step": 10218
    },
    {
      "epoch": 1.4322354590049053,
      "grad_norm": 0.6761796474456787,
      "learning_rate": 0.00016059794307581916,
      "loss": 0.0754,
      "step": 10219
    },
    {
      "epoch": 1.432375613174492,
      "grad_norm": 0.40153732895851135,
      "learning_rate": 0.0001605835924419995,
      "loss": 0.1202,
      "step": 10220
    },
    {
      "epoch": 1.4325157673440785,
      "grad_norm": 0.19425968825817108,
      "learning_rate": 0.00016056924180817987,
      "loss": 0.0117,
      "step": 10221
    },
    {
      "epoch": 1.432655921513665,
      "grad_norm": 0.18818245828151703,
      "learning_rate": 0.0001605548911743602,
      "loss": 0.0225,
      "step": 10222
    },
    {
      "epoch": 1.4327960756832516,
      "grad_norm": 0.2861733138561249,
      "learning_rate": 0.00016054054054054053,
      "loss": 0.0745,
      "step": 10223
    },
    {
      "epoch": 1.432936229852838,
      "grad_norm": 0.24111218750476837,
      "learning_rate": 0.00016052618990672088,
      "loss": 0.0581,
      "step": 10224
    },
    {
      "epoch": 1.4330763840224248,
      "grad_norm": 0.19972725212574005,
      "learning_rate": 0.0001605118392729012,
      "loss": 0.0332,
      "step": 10225
    },
    {
      "epoch": 1.4332165381920112,
      "grad_norm": 0.36768510937690735,
      "learning_rate": 0.00016049748863908154,
      "loss": 0.0513,
      "step": 10226
    },
    {
      "epoch": 1.4333566923615977,
      "grad_norm": 0.2722630500793457,
      "learning_rate": 0.00016048313800526187,
      "loss": 0.0424,
      "step": 10227
    },
    {
      "epoch": 1.4334968465311844,
      "grad_norm": 0.47722113132476807,
      "learning_rate": 0.00016046878737144222,
      "loss": 0.0725,
      "step": 10228
    },
    {
      "epoch": 1.4336370007007708,
      "grad_norm": 0.19668526947498322,
      "learning_rate": 0.00016045443673762255,
      "loss": 0.0081,
      "step": 10229
    },
    {
      "epoch": 1.4337771548703575,
      "grad_norm": 1.039320945739746,
      "learning_rate": 0.00016044008610380288,
      "loss": 0.1781,
      "step": 10230
    },
    {
      "epoch": 1.433917309039944,
      "grad_norm": 0.8073946833610535,
      "learning_rate": 0.00016042573546998326,
      "loss": 0.0684,
      "step": 10231
    },
    {
      "epoch": 1.4340574632095304,
      "grad_norm": 0.968450129032135,
      "learning_rate": 0.0001604113848361636,
      "loss": 0.1085,
      "step": 10232
    },
    {
      "epoch": 1.434197617379117,
      "grad_norm": 1.12123703956604,
      "learning_rate": 0.00016039703420234392,
      "loss": 0.0541,
      "step": 10233
    },
    {
      "epoch": 1.4343377715487036,
      "grad_norm": 0.35450249910354614,
      "learning_rate": 0.00016038268356852428,
      "loss": 0.0568,
      "step": 10234
    },
    {
      "epoch": 1.4344779257182902,
      "grad_norm": 3.7186460494995117,
      "learning_rate": 0.0001603683329347046,
      "loss": 0.2854,
      "step": 10235
    },
    {
      "epoch": 1.4346180798878767,
      "grad_norm": 0.21409717202186584,
      "learning_rate": 0.00016035398230088493,
      "loss": 0.0305,
      "step": 10236
    },
    {
      "epoch": 1.4347582340574632,
      "grad_norm": 0.19324368238449097,
      "learning_rate": 0.0001603396316670653,
      "loss": 0.0328,
      "step": 10237
    },
    {
      "epoch": 1.4348983882270496,
      "grad_norm": 0.36743223667144775,
      "learning_rate": 0.00016032528103324562,
      "loss": 0.0417,
      "step": 10238
    },
    {
      "epoch": 1.4350385423966363,
      "grad_norm": 0.13218557834625244,
      "learning_rate": 0.00016031093039942595,
      "loss": 0.0211,
      "step": 10239
    },
    {
      "epoch": 1.4351786965662228,
      "grad_norm": 0.420714795589447,
      "learning_rate": 0.0001602965797656063,
      "loss": 0.0311,
      "step": 10240
    },
    {
      "epoch": 1.4353188507358094,
      "grad_norm": 0.7833425402641296,
      "learning_rate": 0.00016028222913178663,
      "loss": 0.0581,
      "step": 10241
    },
    {
      "epoch": 1.435459004905396,
      "grad_norm": 0.22733062505722046,
      "learning_rate": 0.00016026787849796696,
      "loss": 0.0504,
      "step": 10242
    },
    {
      "epoch": 1.4355991590749824,
      "grad_norm": 0.5046831369400024,
      "learning_rate": 0.00016025352786414734,
      "loss": 0.0469,
      "step": 10243
    },
    {
      "epoch": 1.435739313244569,
      "grad_norm": 0.339307963848114,
      "learning_rate": 0.00016023917723032767,
      "loss": 0.0447,
      "step": 10244
    },
    {
      "epoch": 1.4358794674141555,
      "grad_norm": 0.15975286066532135,
      "learning_rate": 0.000160224826596508,
      "loss": 0.0304,
      "step": 10245
    },
    {
      "epoch": 1.4360196215837422,
      "grad_norm": 0.23918233811855316,
      "learning_rate": 0.00016021047596268833,
      "loss": 0.0392,
      "step": 10246
    },
    {
      "epoch": 1.4361597757533286,
      "grad_norm": 0.3375759720802307,
      "learning_rate": 0.00016019612532886868,
      "loss": 0.0472,
      "step": 10247
    },
    {
      "epoch": 1.436299929922915,
      "grad_norm": 0.2116953283548355,
      "learning_rate": 0.000160181774695049,
      "loss": 0.0705,
      "step": 10248
    },
    {
      "epoch": 1.4364400840925018,
      "grad_norm": 0.4924832582473755,
      "learning_rate": 0.00016016742406122934,
      "loss": 0.059,
      "step": 10249
    },
    {
      "epoch": 1.4365802382620882,
      "grad_norm": 0.36272770166397095,
      "learning_rate": 0.0001601530734274097,
      "loss": 0.0531,
      "step": 10250
    },
    {
      "epoch": 1.436720392431675,
      "grad_norm": 0.4529755413532257,
      "learning_rate": 0.00016013872279359003,
      "loss": 0.0467,
      "step": 10251
    },
    {
      "epoch": 1.4368605466012614,
      "grad_norm": 0.19947658479213715,
      "learning_rate": 0.00016012437215977035,
      "loss": 0.0187,
      "step": 10252
    },
    {
      "epoch": 1.4370007007708478,
      "grad_norm": 0.3391959071159363,
      "learning_rate": 0.00016011002152595074,
      "loss": 0.0426,
      "step": 10253
    },
    {
      "epoch": 1.4371408549404345,
      "grad_norm": 0.15439580380916595,
      "learning_rate": 0.00016009567089213107,
      "loss": 0.024,
      "step": 10254
    },
    {
      "epoch": 1.437281009110021,
      "grad_norm": 0.2740303575992584,
      "learning_rate": 0.0001600813202583114,
      "loss": 0.0407,
      "step": 10255
    },
    {
      "epoch": 1.4374211632796077,
      "grad_norm": 0.7851037979125977,
      "learning_rate": 0.00016006696962449175,
      "loss": 0.0723,
      "step": 10256
    },
    {
      "epoch": 1.4375613174491941,
      "grad_norm": 0.2592664957046509,
      "learning_rate": 0.00016005261899067208,
      "loss": 0.034,
      "step": 10257
    },
    {
      "epoch": 1.4377014716187806,
      "grad_norm": 0.46411001682281494,
      "learning_rate": 0.0001600382683568524,
      "loss": 0.0795,
      "step": 10258
    },
    {
      "epoch": 1.4378416257883673,
      "grad_norm": 0.35372135043144226,
      "learning_rate": 0.00016002391772303276,
      "loss": 0.0381,
      "step": 10259
    },
    {
      "epoch": 1.4379817799579537,
      "grad_norm": 0.8923017382621765,
      "learning_rate": 0.0001600095670892131,
      "loss": 0.0697,
      "step": 10260
    },
    {
      "epoch": 1.4381219341275404,
      "grad_norm": 0.5511233806610107,
      "learning_rate": 0.00015999521645539342,
      "loss": 0.0731,
      "step": 10261
    },
    {
      "epoch": 1.4382620882971269,
      "grad_norm": 0.26875755190849304,
      "learning_rate": 0.00015998086582157375,
      "loss": 0.0505,
      "step": 10262
    },
    {
      "epoch": 1.4384022424667133,
      "grad_norm": 0.18504968285560608,
      "learning_rate": 0.00015996651518775413,
      "loss": 0.0425,
      "step": 10263
    },
    {
      "epoch": 1.4385423966362998,
      "grad_norm": 0.33882051706314087,
      "learning_rate": 0.00015995216455393446,
      "loss": 0.0431,
      "step": 10264
    },
    {
      "epoch": 1.4386825508058865,
      "grad_norm": 0.2804034948348999,
      "learning_rate": 0.0001599378139201148,
      "loss": 0.0667,
      "step": 10265
    },
    {
      "epoch": 1.4388227049754732,
      "grad_norm": 0.464950829744339,
      "learning_rate": 0.00015992346328629514,
      "loss": 0.0471,
      "step": 10266
    },
    {
      "epoch": 1.4389628591450596,
      "grad_norm": 0.3928641676902771,
      "learning_rate": 0.00015990911265247547,
      "loss": 0.0296,
      "step": 10267
    },
    {
      "epoch": 1.439103013314646,
      "grad_norm": 0.4964993894100189,
      "learning_rate": 0.0001598947620186558,
      "loss": 0.0936,
      "step": 10268
    },
    {
      "epoch": 1.4392431674842325,
      "grad_norm": 0.4022435247898102,
      "learning_rate": 0.00015988041138483616,
      "loss": 0.0665,
      "step": 10269
    },
    {
      "epoch": 1.4393833216538192,
      "grad_norm": 0.23466211557388306,
      "learning_rate": 0.00015986606075101649,
      "loss": 0.0175,
      "step": 10270
    },
    {
      "epoch": 1.4395234758234057,
      "grad_norm": 0.6732150912284851,
      "learning_rate": 0.00015985171011719681,
      "loss": 0.093,
      "step": 10271
    },
    {
      "epoch": 1.4396636299929924,
      "grad_norm": 0.3779064118862152,
      "learning_rate": 0.00015983735948337717,
      "loss": 0.0837,
      "step": 10272
    },
    {
      "epoch": 1.4398037841625788,
      "grad_norm": 0.1630546897649765,
      "learning_rate": 0.0001598230088495575,
      "loss": 0.0573,
      "step": 10273
    },
    {
      "epoch": 1.4399439383321653,
      "grad_norm": 0.28557801246643066,
      "learning_rate": 0.00015980865821573783,
      "loss": 0.0527,
      "step": 10274
    },
    {
      "epoch": 1.440084092501752,
      "grad_norm": 0.3272715210914612,
      "learning_rate": 0.0001597943075819182,
      "loss": 0.0572,
      "step": 10275
    },
    {
      "epoch": 1.4402242466713384,
      "grad_norm": 0.22885917127132416,
      "learning_rate": 0.00015977995694809854,
      "loss": 0.0343,
      "step": 10276
    },
    {
      "epoch": 1.440364400840925,
      "grad_norm": 0.6123156547546387,
      "learning_rate": 0.00015976560631427887,
      "loss": 0.0458,
      "step": 10277
    },
    {
      "epoch": 1.4405045550105116,
      "grad_norm": 0.3223572373390198,
      "learning_rate": 0.00015975125568045922,
      "loss": 0.0722,
      "step": 10278
    },
    {
      "epoch": 1.440644709180098,
      "grad_norm": 2.047304391860962,
      "learning_rate": 0.00015973690504663955,
      "loss": 0.0588,
      "step": 10279
    },
    {
      "epoch": 1.4407848633496847,
      "grad_norm": 0.45392248034477234,
      "learning_rate": 0.00015972255441281988,
      "loss": 0.0263,
      "step": 10280
    },
    {
      "epoch": 1.4409250175192712,
      "grad_norm": 0.5961271524429321,
      "learning_rate": 0.0001597082037790002,
      "loss": 0.0662,
      "step": 10281
    },
    {
      "epoch": 1.4410651716888578,
      "grad_norm": 0.09243643283843994,
      "learning_rate": 0.00015969385314518056,
      "loss": 0.0055,
      "step": 10282
    },
    {
      "epoch": 1.4412053258584443,
      "grad_norm": 0.39269953966140747,
      "learning_rate": 0.0001596795025113609,
      "loss": 0.0994,
      "step": 10283
    },
    {
      "epoch": 1.4413454800280308,
      "grad_norm": 1.7446138858795166,
      "learning_rate": 0.00015966515187754122,
      "loss": 0.0865,
      "step": 10284
    },
    {
      "epoch": 1.4414856341976174,
      "grad_norm": 3.0109221935272217,
      "learning_rate": 0.0001596508012437216,
      "loss": 0.1716,
      "step": 10285
    },
    {
      "epoch": 1.441625788367204,
      "grad_norm": 0.37838804721832275,
      "learning_rate": 0.00015963645060990193,
      "loss": 0.0869,
      "step": 10286
    },
    {
      "epoch": 1.4417659425367906,
      "grad_norm": 0.19698014855384827,
      "learning_rate": 0.00015962209997608226,
      "loss": 0.0659,
      "step": 10287
    },
    {
      "epoch": 1.441906096706377,
      "grad_norm": 0.25391724705696106,
      "learning_rate": 0.00015960774934226262,
      "loss": 0.0746,
      "step": 10288
    },
    {
      "epoch": 1.4420462508759635,
      "grad_norm": 0.23755589127540588,
      "learning_rate": 0.00015959339870844294,
      "loss": 0.0994,
      "step": 10289
    },
    {
      "epoch": 1.4421864050455502,
      "grad_norm": 0.34438541531562805,
      "learning_rate": 0.00015957904807462327,
      "loss": 0.0622,
      "step": 10290
    },
    {
      "epoch": 1.4423265592151366,
      "grad_norm": 0.2734167277812958,
      "learning_rate": 0.00015956469744080363,
      "loss": 0.0401,
      "step": 10291
    },
    {
      "epoch": 1.4424667133847233,
      "grad_norm": 0.4822908639907837,
      "learning_rate": 0.00015955034680698396,
      "loss": 0.0699,
      "step": 10292
    },
    {
      "epoch": 1.4426068675543098,
      "grad_norm": 0.2431921809911728,
      "learning_rate": 0.00015953599617316429,
      "loss": 0.1038,
      "step": 10293
    },
    {
      "epoch": 1.4427470217238962,
      "grad_norm": 0.5435600280761719,
      "learning_rate": 0.00015952164553934467,
      "loss": 0.0649,
      "step": 10294
    },
    {
      "epoch": 1.4428871758934827,
      "grad_norm": 0.26899123191833496,
      "learning_rate": 0.000159507294905525,
      "loss": 0.0295,
      "step": 10295
    },
    {
      "epoch": 1.4430273300630694,
      "grad_norm": 0.07756052166223526,
      "learning_rate": 0.00015949294427170533,
      "loss": 0.0074,
      "step": 10296
    },
    {
      "epoch": 1.4431674842326558,
      "grad_norm": 0.3434891700744629,
      "learning_rate": 0.00015947859363788565,
      "loss": 0.0951,
      "step": 10297
    },
    {
      "epoch": 1.4433076384022425,
      "grad_norm": 0.2591734826564789,
      "learning_rate": 0.000159464243004066,
      "loss": 0.033,
      "step": 10298
    },
    {
      "epoch": 1.443447792571829,
      "grad_norm": 0.21404829621315002,
      "learning_rate": 0.00015944989237024634,
      "loss": 0.0431,
      "step": 10299
    },
    {
      "epoch": 1.4435879467414154,
      "grad_norm": 0.2508086860179901,
      "learning_rate": 0.00015943554173642667,
      "loss": 0.0157,
      "step": 10300
    },
    {
      "epoch": 1.4437281009110021,
      "grad_norm": 0.22778648138046265,
      "learning_rate": 0.00015942119110260702,
      "loss": 0.045,
      "step": 10301
    },
    {
      "epoch": 1.4438682550805886,
      "grad_norm": 0.38248202204704285,
      "learning_rate": 0.00015940684046878735,
      "loss": 0.0341,
      "step": 10302
    },
    {
      "epoch": 1.4440084092501753,
      "grad_norm": 0.280750572681427,
      "learning_rate": 0.00015939248983496768,
      "loss": 0.0449,
      "step": 10303
    },
    {
      "epoch": 1.4441485634197617,
      "grad_norm": 0.6564945578575134,
      "learning_rate": 0.00015937813920114804,
      "loss": 0.0844,
      "step": 10304
    },
    {
      "epoch": 1.4442887175893482,
      "grad_norm": 1.5890581607818604,
      "learning_rate": 0.00015936378856732836,
      "loss": 0.1157,
      "step": 10305
    },
    {
      "epoch": 1.4444288717589349,
      "grad_norm": 0.26139146089553833,
      "learning_rate": 0.0001593494379335087,
      "loss": 0.0295,
      "step": 10306
    },
    {
      "epoch": 1.4445690259285213,
      "grad_norm": 0.17264211177825928,
      "learning_rate": 0.00015933508729968908,
      "loss": 0.0106,
      "step": 10307
    },
    {
      "epoch": 1.444709180098108,
      "grad_norm": 0.306117445230484,
      "learning_rate": 0.0001593207366658694,
      "loss": 0.0773,
      "step": 10308
    },
    {
      "epoch": 1.4448493342676945,
      "grad_norm": 0.264789342880249,
      "learning_rate": 0.00015930638603204973,
      "loss": 0.04,
      "step": 10309
    },
    {
      "epoch": 1.444989488437281,
      "grad_norm": 0.2679969370365143,
      "learning_rate": 0.0001592920353982301,
      "loss": 0.0653,
      "step": 10310
    },
    {
      "epoch": 1.4451296426068676,
      "grad_norm": 0.2131221741437912,
      "learning_rate": 0.00015927768476441042,
      "loss": 0.0447,
      "step": 10311
    },
    {
      "epoch": 1.445269796776454,
      "grad_norm": 0.7403803467750549,
      "learning_rate": 0.00015926333413059075,
      "loss": 0.0653,
      "step": 10312
    },
    {
      "epoch": 1.4454099509460407,
      "grad_norm": 0.24347679316997528,
      "learning_rate": 0.0001592489834967711,
      "loss": 0.0387,
      "step": 10313
    },
    {
      "epoch": 1.4455501051156272,
      "grad_norm": 0.41113975644111633,
      "learning_rate": 0.00015923463286295143,
      "loss": 0.0369,
      "step": 10314
    },
    {
      "epoch": 1.4456902592852137,
      "grad_norm": 0.29645198583602905,
      "learning_rate": 0.00015922028222913176,
      "loss": 0.0139,
      "step": 10315
    },
    {
      "epoch": 1.4458304134548003,
      "grad_norm": 0.18170084059238434,
      "learning_rate": 0.0001592059315953121,
      "loss": 0.0214,
      "step": 10316
    },
    {
      "epoch": 1.4459705676243868,
      "grad_norm": 0.33017316460609436,
      "learning_rate": 0.00015919158096149247,
      "loss": 0.1124,
      "step": 10317
    },
    {
      "epoch": 1.4461107217939735,
      "grad_norm": 0.3603140413761139,
      "learning_rate": 0.0001591772303276728,
      "loss": 0.0619,
      "step": 10318
    },
    {
      "epoch": 1.44625087596356,
      "grad_norm": 0.2632472813129425,
      "learning_rate": 0.00015916287969385313,
      "loss": 0.0273,
      "step": 10319
    },
    {
      "epoch": 1.4463910301331464,
      "grad_norm": 0.5321194529533386,
      "learning_rate": 0.00015914852906003348,
      "loss": 0.089,
      "step": 10320
    },
    {
      "epoch": 1.446531184302733,
      "grad_norm": 0.35040658712387085,
      "learning_rate": 0.0001591341784262138,
      "loss": 0.0554,
      "step": 10321
    },
    {
      "epoch": 1.4466713384723195,
      "grad_norm": 0.3102070987224579,
      "learning_rate": 0.00015911982779239414,
      "loss": 0.0221,
      "step": 10322
    },
    {
      "epoch": 1.4468114926419062,
      "grad_norm": 0.22028884291648865,
      "learning_rate": 0.0001591054771585745,
      "loss": 0.0277,
      "step": 10323
    },
    {
      "epoch": 1.4469516468114927,
      "grad_norm": 0.7164686918258667,
      "learning_rate": 0.00015909112652475482,
      "loss": 0.0654,
      "step": 10324
    },
    {
      "epoch": 1.4470918009810791,
      "grad_norm": 0.10332869738340378,
      "learning_rate": 0.00015907677589093515,
      "loss": 0.0177,
      "step": 10325
    },
    {
      "epoch": 1.4472319551506656,
      "grad_norm": 0.27902406454086304,
      "learning_rate": 0.00015906242525711554,
      "loss": 0.088,
      "step": 10326
    },
    {
      "epoch": 1.4473721093202523,
      "grad_norm": 0.32615530490875244,
      "learning_rate": 0.00015904807462329586,
      "loss": 0.0531,
      "step": 10327
    },
    {
      "epoch": 1.4475122634898387,
      "grad_norm": 0.49578773975372314,
      "learning_rate": 0.0001590337239894762,
      "loss": 0.1529,
      "step": 10328
    },
    {
      "epoch": 1.4476524176594254,
      "grad_norm": 0.5705392360687256,
      "learning_rate": 0.00015901937335565655,
      "loss": 0.0756,
      "step": 10329
    },
    {
      "epoch": 1.4477925718290119,
      "grad_norm": 0.4772118031978607,
      "learning_rate": 0.00015900502272183688,
      "loss": 0.0606,
      "step": 10330
    },
    {
      "epoch": 1.4479327259985983,
      "grad_norm": 0.19996708631515503,
      "learning_rate": 0.0001589906720880172,
      "loss": 0.0458,
      "step": 10331
    },
    {
      "epoch": 1.448072880168185,
      "grad_norm": 1.5132197141647339,
      "learning_rate": 0.00015897632145419753,
      "loss": 0.0828,
      "step": 10332
    },
    {
      "epoch": 1.4482130343377715,
      "grad_norm": 2.096529006958008,
      "learning_rate": 0.0001589619708203779,
      "loss": 0.0249,
      "step": 10333
    },
    {
      "epoch": 1.4483531885073582,
      "grad_norm": 1.0235874652862549,
      "learning_rate": 0.00015894762018655822,
      "loss": 0.2314,
      "step": 10334
    },
    {
      "epoch": 1.4484933426769446,
      "grad_norm": 1.823887825012207,
      "learning_rate": 0.00015893326955273855,
      "loss": 0.3277,
      "step": 10335
    },
    {
      "epoch": 1.448633496846531,
      "grad_norm": 0.23627851903438568,
      "learning_rate": 0.0001589189189189189,
      "loss": 0.0189,
      "step": 10336
    },
    {
      "epoch": 1.4487736510161178,
      "grad_norm": 0.22111201286315918,
      "learning_rate": 0.00015890456828509923,
      "loss": 0.0207,
      "step": 10337
    },
    {
      "epoch": 1.4489138051857042,
      "grad_norm": 0.34417077898979187,
      "learning_rate": 0.00015889021765127956,
      "loss": 0.1092,
      "step": 10338
    },
    {
      "epoch": 1.449053959355291,
      "grad_norm": 0.310786634683609,
      "learning_rate": 0.00015887586701745994,
      "loss": 0.0572,
      "step": 10339
    },
    {
      "epoch": 1.4491941135248774,
      "grad_norm": 0.08493835479021072,
      "learning_rate": 0.00015886151638364027,
      "loss": 0.0122,
      "step": 10340
    },
    {
      "epoch": 1.4493342676944638,
      "grad_norm": 0.37680280208587646,
      "learning_rate": 0.0001588471657498206,
      "loss": 0.0364,
      "step": 10341
    },
    {
      "epoch": 1.4494744218640505,
      "grad_norm": 0.3844102621078491,
      "learning_rate": 0.00015883281511600095,
      "loss": 0.059,
      "step": 10342
    },
    {
      "epoch": 1.449614576033637,
      "grad_norm": 0.35041797161102295,
      "learning_rate": 0.00015881846448218128,
      "loss": 0.0681,
      "step": 10343
    },
    {
      "epoch": 1.4497547302032237,
      "grad_norm": 0.36123186349868774,
      "learning_rate": 0.0001588041138483616,
      "loss": 0.0505,
      "step": 10344
    },
    {
      "epoch": 1.44989488437281,
      "grad_norm": 0.3794997036457062,
      "learning_rate": 0.00015878976321454197,
      "loss": 0.1217,
      "step": 10345
    },
    {
      "epoch": 1.4500350385423966,
      "grad_norm": 0.3944834768772125,
      "learning_rate": 0.0001587754125807223,
      "loss": 0.0257,
      "step": 10346
    },
    {
      "epoch": 1.4501751927119833,
      "grad_norm": 0.20971815288066864,
      "learning_rate": 0.00015876106194690262,
      "loss": 0.0204,
      "step": 10347
    },
    {
      "epoch": 1.4503153468815697,
      "grad_norm": 0.2663668394088745,
      "learning_rate": 0.000158746711313083,
      "loss": 0.087,
      "step": 10348
    },
    {
      "epoch": 1.4504555010511564,
      "grad_norm": 0.24102389812469482,
      "learning_rate": 0.00015873236067926334,
      "loss": 0.057,
      "step": 10349
    },
    {
      "epoch": 1.4505956552207429,
      "grad_norm": 0.2945449650287628,
      "learning_rate": 0.00015871801004544366,
      "loss": 0.0267,
      "step": 10350
    },
    {
      "epoch": 1.4507358093903293,
      "grad_norm": 0.33592602610588074,
      "learning_rate": 0.000158703659411624,
      "loss": 0.0756,
      "step": 10351
    },
    {
      "epoch": 1.4508759635599158,
      "grad_norm": 0.40410202741622925,
      "learning_rate": 0.00015868930877780435,
      "loss": 0.0605,
      "step": 10352
    },
    {
      "epoch": 1.4510161177295025,
      "grad_norm": 0.7900621891021729,
      "learning_rate": 0.00015867495814398468,
      "loss": 0.0556,
      "step": 10353
    },
    {
      "epoch": 1.4511562718990891,
      "grad_norm": 0.5548974275588989,
      "learning_rate": 0.000158660607510165,
      "loss": 0.0866,
      "step": 10354
    },
    {
      "epoch": 1.4512964260686756,
      "grad_norm": 0.4954400360584259,
      "learning_rate": 0.00015864625687634536,
      "loss": 0.073,
      "step": 10355
    },
    {
      "epoch": 1.451436580238262,
      "grad_norm": 0.5850778222084045,
      "learning_rate": 0.0001586319062425257,
      "loss": 0.112,
      "step": 10356
    },
    {
      "epoch": 1.4515767344078485,
      "grad_norm": 0.3048843741416931,
      "learning_rate": 0.00015861755560870602,
      "loss": 0.062,
      "step": 10357
    },
    {
      "epoch": 1.4517168885774352,
      "grad_norm": 0.3343328535556793,
      "learning_rate": 0.0001586032049748864,
      "loss": 0.0482,
      "step": 10358
    },
    {
      "epoch": 1.4518570427470217,
      "grad_norm": 0.35425207018852234,
      "learning_rate": 0.00015858885434106673,
      "loss": 0.0632,
      "step": 10359
    },
    {
      "epoch": 1.4519971969166083,
      "grad_norm": 0.1927790492773056,
      "learning_rate": 0.00015857450370724706,
      "loss": 0.0349,
      "step": 10360
    },
    {
      "epoch": 1.4521373510861948,
      "grad_norm": 0.8330217599868774,
      "learning_rate": 0.00015856015307342741,
      "loss": 0.051,
      "step": 10361
    },
    {
      "epoch": 1.4522775052557813,
      "grad_norm": 0.5046206116676331,
      "learning_rate": 0.00015854580243960774,
      "loss": 0.0319,
      "step": 10362
    },
    {
      "epoch": 1.452417659425368,
      "grad_norm": 0.21573689579963684,
      "learning_rate": 0.00015853145180578807,
      "loss": 0.0404,
      "step": 10363
    },
    {
      "epoch": 1.4525578135949544,
      "grad_norm": 0.29211878776550293,
      "learning_rate": 0.00015851710117196843,
      "loss": 0.0507,
      "step": 10364
    },
    {
      "epoch": 1.452697967764541,
      "grad_norm": 0.47367075085639954,
      "learning_rate": 0.00015850275053814876,
      "loss": 0.0756,
      "step": 10365
    },
    {
      "epoch": 1.4528381219341275,
      "grad_norm": 0.659792423248291,
      "learning_rate": 0.00015848839990432908,
      "loss": 0.0919,
      "step": 10366
    },
    {
      "epoch": 1.452978276103714,
      "grad_norm": 0.40301090478897095,
      "learning_rate": 0.0001584740492705094,
      "loss": 0.0542,
      "step": 10367
    },
    {
      "epoch": 1.4531184302733007,
      "grad_norm": 0.244008406996727,
      "learning_rate": 0.00015845969863668977,
      "loss": 0.0225,
      "step": 10368
    },
    {
      "epoch": 1.4532585844428871,
      "grad_norm": 0.2853996455669403,
      "learning_rate": 0.0001584453480028701,
      "loss": 0.0291,
      "step": 10369
    },
    {
      "epoch": 1.4533987386124738,
      "grad_norm": 0.860802948474884,
      "learning_rate": 0.00015843099736905043,
      "loss": 0.0752,
      "step": 10370
    },
    {
      "epoch": 1.4535388927820603,
      "grad_norm": 0.1470966935157776,
      "learning_rate": 0.0001584166467352308,
      "loss": 0.0134,
      "step": 10371
    },
    {
      "epoch": 1.4536790469516467,
      "grad_norm": 0.5490665435791016,
      "learning_rate": 0.00015840229610141114,
      "loss": 0.0723,
      "step": 10372
    },
    {
      "epoch": 1.4538192011212334,
      "grad_norm": 0.14277786016464233,
      "learning_rate": 0.00015838794546759147,
      "loss": 0.022,
      "step": 10373
    },
    {
      "epoch": 1.4539593552908199,
      "grad_norm": 0.49025362730026245,
      "learning_rate": 0.00015837359483377182,
      "loss": 0.0677,
      "step": 10374
    },
    {
      "epoch": 1.4540995094604066,
      "grad_norm": 0.33080458641052246,
      "learning_rate": 0.00015835924419995215,
      "loss": 0.0516,
      "step": 10375
    },
    {
      "epoch": 1.454239663629993,
      "grad_norm": 0.3922242522239685,
      "learning_rate": 0.00015834489356613248,
      "loss": 0.0579,
      "step": 10376
    },
    {
      "epoch": 1.4543798177995795,
      "grad_norm": 0.3718421757221222,
      "learning_rate": 0.00015833054293231283,
      "loss": 0.0363,
      "step": 10377
    },
    {
      "epoch": 1.4545199719691662,
      "grad_norm": 0.4780240058898926,
      "learning_rate": 0.00015831619229849316,
      "loss": 0.0502,
      "step": 10378
    },
    {
      "epoch": 1.4546601261387526,
      "grad_norm": 0.1006823256611824,
      "learning_rate": 0.0001583018416646735,
      "loss": 0.005,
      "step": 10379
    },
    {
      "epoch": 1.4548002803083393,
      "grad_norm": 0.14853745698928833,
      "learning_rate": 0.00015828749103085387,
      "loss": 0.022,
      "step": 10380
    },
    {
      "epoch": 1.4549404344779258,
      "grad_norm": 0.1390381008386612,
      "learning_rate": 0.0001582731403970342,
      "loss": 0.0095,
      "step": 10381
    },
    {
      "epoch": 1.4550805886475122,
      "grad_norm": 0.20819009840488434,
      "learning_rate": 0.00015825878976321453,
      "loss": 0.0088,
      "step": 10382
    },
    {
      "epoch": 1.4552207428170987,
      "grad_norm": 4.472097873687744,
      "learning_rate": 0.0001582444391293949,
      "loss": 0.2471,
      "step": 10383
    },
    {
      "epoch": 1.4553608969866854,
      "grad_norm": 2.834503173828125,
      "learning_rate": 0.00015823008849557522,
      "loss": 0.1322,
      "step": 10384
    },
    {
      "epoch": 1.4555010511562718,
      "grad_norm": 0.6564981341362,
      "learning_rate": 0.00015821573786175554,
      "loss": 0.0367,
      "step": 10385
    },
    {
      "epoch": 1.4556412053258585,
      "grad_norm": 0.24525196850299835,
      "learning_rate": 0.00015820138722793587,
      "loss": 0.0492,
      "step": 10386
    },
    {
      "epoch": 1.455781359495445,
      "grad_norm": 0.15929551422595978,
      "learning_rate": 0.00015818703659411623,
      "loss": 0.019,
      "step": 10387
    },
    {
      "epoch": 1.4559215136650314,
      "grad_norm": 0.3377452790737152,
      "learning_rate": 0.00015817268596029656,
      "loss": 0.0712,
      "step": 10388
    },
    {
      "epoch": 1.456061667834618,
      "grad_norm": 0.33260834217071533,
      "learning_rate": 0.00015815833532647688,
      "loss": 0.0618,
      "step": 10389
    },
    {
      "epoch": 1.4562018220042046,
      "grad_norm": 0.5088858008384705,
      "learning_rate": 0.00015814398469265727,
      "loss": 0.1161,
      "step": 10390
    },
    {
      "epoch": 1.4563419761737912,
      "grad_norm": 0.22829212248325348,
      "learning_rate": 0.0001581296340588376,
      "loss": 0.0555,
      "step": 10391
    },
    {
      "epoch": 1.4564821303433777,
      "grad_norm": 0.4645979106426239,
      "learning_rate": 0.00015811528342501792,
      "loss": 0.0758,
      "step": 10392
    },
    {
      "epoch": 1.4566222845129642,
      "grad_norm": 0.2573792636394501,
      "learning_rate": 0.00015810093279119828,
      "loss": 0.0529,
      "step": 10393
    },
    {
      "epoch": 1.4567624386825508,
      "grad_norm": 0.44291824102401733,
      "learning_rate": 0.0001580865821573786,
      "loss": 0.0286,
      "step": 10394
    },
    {
      "epoch": 1.4569025928521373,
      "grad_norm": 0.5993772745132446,
      "learning_rate": 0.00015807223152355894,
      "loss": 0.0686,
      "step": 10395
    },
    {
      "epoch": 1.457042747021724,
      "grad_norm": 0.21350601315498352,
      "learning_rate": 0.0001580578808897393,
      "loss": 0.0298,
      "step": 10396
    },
    {
      "epoch": 1.4571829011913104,
      "grad_norm": 0.2096581906080246,
      "learning_rate": 0.00015804353025591962,
      "loss": 0.0128,
      "step": 10397
    },
    {
      "epoch": 1.457323055360897,
      "grad_norm": 0.21883641183376312,
      "learning_rate": 0.00015802917962209995,
      "loss": 0.0215,
      "step": 10398
    },
    {
      "epoch": 1.4574632095304836,
      "grad_norm": 0.2923535704612732,
      "learning_rate": 0.0001580148289882803,
      "loss": 0.0213,
      "step": 10399
    },
    {
      "epoch": 1.45760336370007,
      "grad_norm": 0.47698137164115906,
      "learning_rate": 0.00015800047835446063,
      "loss": 0.0337,
      "step": 10400
    },
    {
      "epoch": 1.4577435178696567,
      "grad_norm": 0.44751036167144775,
      "learning_rate": 0.00015798612772064096,
      "loss": 0.0503,
      "step": 10401
    },
    {
      "epoch": 1.4578836720392432,
      "grad_norm": 0.5426173806190491,
      "learning_rate": 0.0001579717770868213,
      "loss": 0.0256,
      "step": 10402
    },
    {
      "epoch": 1.4580238262088296,
      "grad_norm": 0.3416033387184143,
      "learning_rate": 0.00015795742645300167,
      "loss": 0.0426,
      "step": 10403
    },
    {
      "epoch": 1.4581639803784163,
      "grad_norm": 0.3211739957332611,
      "learning_rate": 0.000157943075819182,
      "loss": 0.0505,
      "step": 10404
    },
    {
      "epoch": 1.4583041345480028,
      "grad_norm": 0.594524085521698,
      "learning_rate": 0.00015792872518536233,
      "loss": 0.0409,
      "step": 10405
    },
    {
      "epoch": 1.4584442887175895,
      "grad_norm": 0.6529772281646729,
      "learning_rate": 0.0001579143745515427,
      "loss": 0.0733,
      "step": 10406
    },
    {
      "epoch": 1.458584442887176,
      "grad_norm": 0.24155311286449432,
      "learning_rate": 0.00015790002391772302,
      "loss": 0.0244,
      "step": 10407
    },
    {
      "epoch": 1.4587245970567624,
      "grad_norm": 0.21188224852085114,
      "learning_rate": 0.00015788567328390334,
      "loss": 0.0339,
      "step": 10408
    },
    {
      "epoch": 1.458864751226349,
      "grad_norm": 0.3168196380138397,
      "learning_rate": 0.0001578713226500837,
      "loss": 0.0299,
      "step": 10409
    },
    {
      "epoch": 1.4590049053959355,
      "grad_norm": 0.2256992906332016,
      "learning_rate": 0.00015785697201626403,
      "loss": 0.0503,
      "step": 10410
    },
    {
      "epoch": 1.4591450595655222,
      "grad_norm": 0.29772457480430603,
      "learning_rate": 0.00015784262138244436,
      "loss": 0.0256,
      "step": 10411
    },
    {
      "epoch": 1.4592852137351087,
      "grad_norm": 0.3221682608127594,
      "learning_rate": 0.00015782827074862474,
      "loss": 0.0621,
      "step": 10412
    },
    {
      "epoch": 1.4594253679046951,
      "grad_norm": 0.27890974283218384,
      "learning_rate": 0.00015781392011480507,
      "loss": 0.0379,
      "step": 10413
    },
    {
      "epoch": 1.4595655220742816,
      "grad_norm": 0.42896613478660583,
      "learning_rate": 0.0001577995694809854,
      "loss": 0.044,
      "step": 10414
    },
    {
      "epoch": 1.4597056762438683,
      "grad_norm": 0.3510957360267639,
      "learning_rate": 0.00015778521884716575,
      "loss": 0.0963,
      "step": 10415
    },
    {
      "epoch": 1.4598458304134547,
      "grad_norm": 0.15724126994609833,
      "learning_rate": 0.00015777086821334608,
      "loss": 0.0132,
      "step": 10416
    },
    {
      "epoch": 1.4599859845830414,
      "grad_norm": 0.2976750135421753,
      "learning_rate": 0.0001577565175795264,
      "loss": 0.0678,
      "step": 10417
    },
    {
      "epoch": 1.4601261387526279,
      "grad_norm": 0.2269676774740219,
      "learning_rate": 0.00015774216694570674,
      "loss": 0.02,
      "step": 10418
    },
    {
      "epoch": 1.4602662929222143,
      "grad_norm": 0.5498712658882141,
      "learning_rate": 0.0001577278163118871,
      "loss": 0.0503,
      "step": 10419
    },
    {
      "epoch": 1.460406447091801,
      "grad_norm": 0.343868613243103,
      "learning_rate": 0.00015771346567806742,
      "loss": 0.0541,
      "step": 10420
    },
    {
      "epoch": 1.4605466012613875,
      "grad_norm": 0.62278151512146,
      "learning_rate": 0.00015769911504424775,
      "loss": 0.0407,
      "step": 10421
    },
    {
      "epoch": 1.4606867554309741,
      "grad_norm": 0.310596764087677,
      "learning_rate": 0.00015768476441042813,
      "loss": 0.0705,
      "step": 10422
    },
    {
      "epoch": 1.4608269096005606,
      "grad_norm": 0.3266717791557312,
      "learning_rate": 0.00015767041377660846,
      "loss": 0.0482,
      "step": 10423
    },
    {
      "epoch": 1.460967063770147,
      "grad_norm": 0.3247486352920532,
      "learning_rate": 0.0001576560631427888,
      "loss": 0.0247,
      "step": 10424
    },
    {
      "epoch": 1.4611072179397337,
      "grad_norm": 0.2839023768901825,
      "learning_rate": 0.00015764171250896915,
      "loss": 0.0856,
      "step": 10425
    },
    {
      "epoch": 1.4612473721093202,
      "grad_norm": 0.3268580734729767,
      "learning_rate": 0.00015762736187514948,
      "loss": 0.0378,
      "step": 10426
    },
    {
      "epoch": 1.461387526278907,
      "grad_norm": 0.4395720660686493,
      "learning_rate": 0.0001576130112413298,
      "loss": 0.0887,
      "step": 10427
    },
    {
      "epoch": 1.4615276804484933,
      "grad_norm": 0.6974881291389465,
      "learning_rate": 0.00015759866060751016,
      "loss": 0.039,
      "step": 10428
    },
    {
      "epoch": 1.4616678346180798,
      "grad_norm": 0.35591039061546326,
      "learning_rate": 0.0001575843099736905,
      "loss": 0.0917,
      "step": 10429
    },
    {
      "epoch": 1.4618079887876665,
      "grad_norm": 0.10858973115682602,
      "learning_rate": 0.00015756995933987082,
      "loss": 0.0143,
      "step": 10430
    },
    {
      "epoch": 1.461948142957253,
      "grad_norm": 0.619379460811615,
      "learning_rate": 0.00015755560870605117,
      "loss": 0.0626,
      "step": 10431
    },
    {
      "epoch": 1.4620882971268396,
      "grad_norm": 0.48327958583831787,
      "learning_rate": 0.0001575412580722315,
      "loss": 0.0443,
      "step": 10432
    },
    {
      "epoch": 1.462228451296426,
      "grad_norm": 0.2702605128288269,
      "learning_rate": 0.00015752690743841183,
      "loss": 0.015,
      "step": 10433
    },
    {
      "epoch": 1.4623686054660125,
      "grad_norm": 3.299253463745117,
      "learning_rate": 0.0001575125568045922,
      "loss": 0.0335,
      "step": 10434
    },
    {
      "epoch": 1.4625087596355992,
      "grad_norm": 0.647927463054657,
      "learning_rate": 0.00015749820617077254,
      "loss": 0.07,
      "step": 10435
    },
    {
      "epoch": 1.4626489138051857,
      "grad_norm": 0.15951493382453918,
      "learning_rate": 0.00015748385553695287,
      "loss": 0.054,
      "step": 10436
    },
    {
      "epoch": 1.4627890679747724,
      "grad_norm": 0.3459722697734833,
      "learning_rate": 0.0001574695049031332,
      "loss": 0.042,
      "step": 10437
    },
    {
      "epoch": 1.4629292221443588,
      "grad_norm": 0.6201842427253723,
      "learning_rate": 0.00015745515426931355,
      "loss": 0.0899,
      "step": 10438
    },
    {
      "epoch": 1.4630693763139453,
      "grad_norm": 0.19347669184207916,
      "learning_rate": 0.00015744080363549388,
      "loss": 0.0294,
      "step": 10439
    },
    {
      "epoch": 1.4632095304835318,
      "grad_norm": 0.2756344974040985,
      "learning_rate": 0.0001574264530016742,
      "loss": 0.0197,
      "step": 10440
    },
    {
      "epoch": 1.4633496846531184,
      "grad_norm": 0.3071753680706024,
      "learning_rate": 0.00015741210236785457,
      "loss": 0.0473,
      "step": 10441
    },
    {
      "epoch": 1.4634898388227051,
      "grad_norm": 0.3743177056312561,
      "learning_rate": 0.0001573977517340349,
      "loss": 0.0454,
      "step": 10442
    },
    {
      "epoch": 1.4636299929922916,
      "grad_norm": 0.32004883885383606,
      "learning_rate": 0.00015738340110021522,
      "loss": 0.0682,
      "step": 10443
    },
    {
      "epoch": 1.463770147161878,
      "grad_norm": 0.19337202608585358,
      "learning_rate": 0.0001573690504663956,
      "loss": 0.0245,
      "step": 10444
    },
    {
      "epoch": 1.4639103013314645,
      "grad_norm": 0.31909459829330444,
      "learning_rate": 0.00015735469983257593,
      "loss": 0.0309,
      "step": 10445
    },
    {
      "epoch": 1.4640504555010512,
      "grad_norm": 0.4476185142993927,
      "learning_rate": 0.00015734034919875626,
      "loss": 0.1003,
      "step": 10446
    },
    {
      "epoch": 1.4641906096706376,
      "grad_norm": 0.3821708858013153,
      "learning_rate": 0.00015732599856493662,
      "loss": 0.0352,
      "step": 10447
    },
    {
      "epoch": 1.4643307638402243,
      "grad_norm": 0.2570171058177948,
      "learning_rate": 0.00015731164793111695,
      "loss": 0.0483,
      "step": 10448
    },
    {
      "epoch": 1.4644709180098108,
      "grad_norm": 0.24343939125537872,
      "learning_rate": 0.00015729729729729728,
      "loss": 0.0264,
      "step": 10449
    },
    {
      "epoch": 1.4646110721793972,
      "grad_norm": 0.46883058547973633,
      "learning_rate": 0.00015728294666347763,
      "loss": 0.0583,
      "step": 10450
    },
    {
      "epoch": 1.464751226348984,
      "grad_norm": 0.32669922709465027,
      "learning_rate": 0.00015726859602965796,
      "loss": 0.0705,
      "step": 10451
    },
    {
      "epoch": 1.4648913805185704,
      "grad_norm": 0.40174418687820435,
      "learning_rate": 0.0001572542453958383,
      "loss": 0.0691,
      "step": 10452
    },
    {
      "epoch": 1.465031534688157,
      "grad_norm": 0.5688478946685791,
      "learning_rate": 0.00015723989476201862,
      "loss": 0.0425,
      "step": 10453
    },
    {
      "epoch": 1.4651716888577435,
      "grad_norm": 0.17448224127292633,
      "learning_rate": 0.000157225544128199,
      "loss": 0.0474,
      "step": 10454
    },
    {
      "epoch": 1.46531184302733,
      "grad_norm": 0.1570330262184143,
      "learning_rate": 0.00015721119349437933,
      "loss": 0.0184,
      "step": 10455
    },
    {
      "epoch": 1.4654519971969167,
      "grad_norm": 0.23003444075584412,
      "learning_rate": 0.00015719684286055966,
      "loss": 0.014,
      "step": 10456
    },
    {
      "epoch": 1.4655921513665031,
      "grad_norm": 0.5143041014671326,
      "learning_rate": 0.00015718249222674,
      "loss": 0.0349,
      "step": 10457
    },
    {
      "epoch": 1.4657323055360898,
      "grad_norm": 0.4303840398788452,
      "learning_rate": 0.00015716814159292034,
      "loss": 0.0207,
      "step": 10458
    },
    {
      "epoch": 1.4658724597056763,
      "grad_norm": 0.23060907423496246,
      "learning_rate": 0.00015715379095910067,
      "loss": 0.0208,
      "step": 10459
    },
    {
      "epoch": 1.4660126138752627,
      "grad_norm": 0.1794975996017456,
      "learning_rate": 0.00015713944032528103,
      "loss": 0.0654,
      "step": 10460
    },
    {
      "epoch": 1.4661527680448494,
      "grad_norm": 0.32689523696899414,
      "learning_rate": 0.00015712508969146135,
      "loss": 0.061,
      "step": 10461
    },
    {
      "epoch": 1.4662929222144359,
      "grad_norm": 1.0868775844573975,
      "learning_rate": 0.00015711073905764168,
      "loss": 0.0271,
      "step": 10462
    },
    {
      "epoch": 1.4664330763840225,
      "grad_norm": 0.2191668152809143,
      "learning_rate": 0.00015709638842382204,
      "loss": 0.0294,
      "step": 10463
    },
    {
      "epoch": 1.466573230553609,
      "grad_norm": 0.48042985796928406,
      "learning_rate": 0.00015708203779000237,
      "loss": 0.0621,
      "step": 10464
    },
    {
      "epoch": 1.4667133847231955,
      "grad_norm": 0.26325538754463196,
      "learning_rate": 0.0001570676871561827,
      "loss": 0.037,
      "step": 10465
    },
    {
      "epoch": 1.4668535388927821,
      "grad_norm": 0.335306316614151,
      "learning_rate": 0.00015705333652236308,
      "loss": 0.0897,
      "step": 10466
    },
    {
      "epoch": 1.4669936930623686,
      "grad_norm": 0.10870976746082306,
      "learning_rate": 0.0001570389858885434,
      "loss": 0.0156,
      "step": 10467
    },
    {
      "epoch": 1.4671338472319553,
      "grad_norm": 0.8795617818832397,
      "learning_rate": 0.00015702463525472374,
      "loss": 0.0769,
      "step": 10468
    },
    {
      "epoch": 1.4672740014015417,
      "grad_norm": 0.7768855690956116,
      "learning_rate": 0.0001570102846209041,
      "loss": 0.1468,
      "step": 10469
    },
    {
      "epoch": 1.4674141555711282,
      "grad_norm": 0.14433500170707703,
      "learning_rate": 0.00015699593398708442,
      "loss": 0.0232,
      "step": 10470
    },
    {
      "epoch": 1.4675543097407147,
      "grad_norm": 0.22287726402282715,
      "learning_rate": 0.00015698158335326475,
      "loss": 0.0287,
      "step": 10471
    },
    {
      "epoch": 1.4676944639103013,
      "grad_norm": 0.2655218541622162,
      "learning_rate": 0.00015696723271944508,
      "loss": 0.0282,
      "step": 10472
    },
    {
      "epoch": 1.4678346180798878,
      "grad_norm": 0.22424133121967316,
      "learning_rate": 0.00015695288208562543,
      "loss": 0.0382,
      "step": 10473
    },
    {
      "epoch": 1.4679747722494745,
      "grad_norm": 0.2551825940608978,
      "learning_rate": 0.00015693853145180576,
      "loss": 0.0517,
      "step": 10474
    },
    {
      "epoch": 1.468114926419061,
      "grad_norm": 0.8367336988449097,
      "learning_rate": 0.0001569241808179861,
      "loss": 0.0403,
      "step": 10475
    },
    {
      "epoch": 1.4682550805886474,
      "grad_norm": 0.1715218424797058,
      "learning_rate": 0.00015690983018416647,
      "loss": 0.0274,
      "step": 10476
    },
    {
      "epoch": 1.468395234758234,
      "grad_norm": 0.3328858017921448,
      "learning_rate": 0.0001568954795503468,
      "loss": 0.0433,
      "step": 10477
    },
    {
      "epoch": 1.4685353889278205,
      "grad_norm": 0.47546708583831787,
      "learning_rate": 0.00015688112891652713,
      "loss": 0.1093,
      "step": 10478
    },
    {
      "epoch": 1.4686755430974072,
      "grad_norm": 0.880604088306427,
      "learning_rate": 0.00015686677828270749,
      "loss": 0.1032,
      "step": 10479
    },
    {
      "epoch": 1.4688156972669937,
      "grad_norm": 0.532997727394104,
      "learning_rate": 0.00015685242764888781,
      "loss": 0.0608,
      "step": 10480
    },
    {
      "epoch": 1.4689558514365801,
      "grad_norm": 0.9230180978775024,
      "learning_rate": 0.00015683807701506814,
      "loss": 0.163,
      "step": 10481
    },
    {
      "epoch": 1.4690960056061668,
      "grad_norm": 1.7509791851043701,
      "learning_rate": 0.0001568237263812485,
      "loss": 0.0869,
      "step": 10482
    },
    {
      "epoch": 1.4692361597757533,
      "grad_norm": 1.7276511192321777,
      "learning_rate": 0.00015680937574742883,
      "loss": 0.2058,
      "step": 10483
    },
    {
      "epoch": 1.46937631394534,
      "grad_norm": 0.4504629969596863,
      "learning_rate": 0.00015679502511360916,
      "loss": 0.0833,
      "step": 10484
    },
    {
      "epoch": 1.4695164681149264,
      "grad_norm": 1.767192006111145,
      "learning_rate": 0.00015678067447978954,
      "loss": 0.0561,
      "step": 10485
    },
    {
      "epoch": 1.4696566222845129,
      "grad_norm": 0.046178188174963,
      "learning_rate": 0.00015676632384596987,
      "loss": 0.0058,
      "step": 10486
    },
    {
      "epoch": 1.4697967764540996,
      "grad_norm": 0.3052062392234802,
      "learning_rate": 0.0001567519732121502,
      "loss": 0.0601,
      "step": 10487
    },
    {
      "epoch": 1.469936930623686,
      "grad_norm": 0.2043287605047226,
      "learning_rate": 0.00015673762257833052,
      "loss": 0.0344,
      "step": 10488
    },
    {
      "epoch": 1.4700770847932727,
      "grad_norm": 0.5523789525032043,
      "learning_rate": 0.00015672327194451088,
      "loss": 0.068,
      "step": 10489
    },
    {
      "epoch": 1.4702172389628592,
      "grad_norm": 0.3881550133228302,
      "learning_rate": 0.0001567089213106912,
      "loss": 0.058,
      "step": 10490
    },
    {
      "epoch": 1.4703573931324456,
      "grad_norm": 0.2595837116241455,
      "learning_rate": 0.00015669457067687154,
      "loss": 0.0419,
      "step": 10491
    },
    {
      "epoch": 1.4704975473020323,
      "grad_norm": 0.3189378082752228,
      "learning_rate": 0.0001566802200430519,
      "loss": 0.043,
      "step": 10492
    },
    {
      "epoch": 1.4706377014716188,
      "grad_norm": 0.25390636920928955,
      "learning_rate": 0.00015666586940923222,
      "loss": 0.0901,
      "step": 10493
    },
    {
      "epoch": 1.4707778556412054,
      "grad_norm": 0.5380606651306152,
      "learning_rate": 0.00015665151877541255,
      "loss": 0.04,
      "step": 10494
    },
    {
      "epoch": 1.470918009810792,
      "grad_norm": 0.4755038619041443,
      "learning_rate": 0.0001566371681415929,
      "loss": 0.0411,
      "step": 10495
    },
    {
      "epoch": 1.4710581639803784,
      "grad_norm": 0.568804144859314,
      "learning_rate": 0.00015662281750777323,
      "loss": 0.0364,
      "step": 10496
    },
    {
      "epoch": 1.4711983181499648,
      "grad_norm": 0.2592986226081848,
      "learning_rate": 0.00015660846687395356,
      "loss": 0.0242,
      "step": 10497
    },
    {
      "epoch": 1.4713384723195515,
      "grad_norm": 0.4103192090988159,
      "learning_rate": 0.00015659411624013394,
      "loss": 0.1018,
      "step": 10498
    },
    {
      "epoch": 1.4714786264891382,
      "grad_norm": 0.467492014169693,
      "learning_rate": 0.00015657976560631427,
      "loss": 0.0579,
      "step": 10499
    },
    {
      "epoch": 1.4716187806587246,
      "grad_norm": 0.4619954526424408,
      "learning_rate": 0.0001565654149724946,
      "loss": 0.0639,
      "step": 10500
    },
    {
      "epoch": 1.471758934828311,
      "grad_norm": 0.2719810903072357,
      "learning_rate": 0.00015655106433867496,
      "loss": 0.0702,
      "step": 10501
    },
    {
      "epoch": 1.4718990889978976,
      "grad_norm": 0.5512811541557312,
      "learning_rate": 0.00015653671370485529,
      "loss": 0.1299,
      "step": 10502
    },
    {
      "epoch": 1.4720392431674842,
      "grad_norm": 0.29279544949531555,
      "learning_rate": 0.00015652236307103561,
      "loss": 0.0356,
      "step": 10503
    },
    {
      "epoch": 1.4721793973370707,
      "grad_norm": 0.2552359998226166,
      "learning_rate": 0.00015650801243721597,
      "loss": 0.0255,
      "step": 10504
    },
    {
      "epoch": 1.4723195515066574,
      "grad_norm": 0.5418764352798462,
      "learning_rate": 0.0001564936618033963,
      "loss": 0.0428,
      "step": 10505
    },
    {
      "epoch": 1.4724597056762438,
      "grad_norm": 0.5578295588493347,
      "learning_rate": 0.00015647931116957663,
      "loss": 0.0752,
      "step": 10506
    },
    {
      "epoch": 1.4725998598458303,
      "grad_norm": 0.5077351927757263,
      "learning_rate": 0.00015646496053575696,
      "loss": 0.0532,
      "step": 10507
    },
    {
      "epoch": 1.472740014015417,
      "grad_norm": 0.2052440345287323,
      "learning_rate": 0.00015645060990193734,
      "loss": 0.0491,
      "step": 10508
    },
    {
      "epoch": 1.4728801681850034,
      "grad_norm": 0.24515457451343536,
      "learning_rate": 0.00015643625926811767,
      "loss": 0.0389,
      "step": 10509
    },
    {
      "epoch": 1.4730203223545901,
      "grad_norm": 0.23881974816322327,
      "learning_rate": 0.000156421908634298,
      "loss": 0.0235,
      "step": 10510
    },
    {
      "epoch": 1.4731604765241766,
      "grad_norm": 0.5546131730079651,
      "learning_rate": 0.00015640755800047835,
      "loss": 0.0257,
      "step": 10511
    },
    {
      "epoch": 1.473300630693763,
      "grad_norm": 0.46937084197998047,
      "learning_rate": 0.00015639320736665868,
      "loss": 0.0659,
      "step": 10512
    },
    {
      "epoch": 1.4734407848633497,
      "grad_norm": 1.763956904411316,
      "learning_rate": 0.000156378856732839,
      "loss": 0.0982,
      "step": 10513
    },
    {
      "epoch": 1.4735809390329362,
      "grad_norm": 0.5380129218101501,
      "learning_rate": 0.00015636450609901936,
      "loss": 0.0778,
      "step": 10514
    },
    {
      "epoch": 1.4737210932025229,
      "grad_norm": 0.1430983990430832,
      "learning_rate": 0.0001563501554651997,
      "loss": 0.0276,
      "step": 10515
    },
    {
      "epoch": 1.4738612473721093,
      "grad_norm": 0.21112503111362457,
      "learning_rate": 0.00015633580483138002,
      "loss": 0.0199,
      "step": 10516
    },
    {
      "epoch": 1.4740014015416958,
      "grad_norm": 0.31134650111198425,
      "learning_rate": 0.0001563214541975604,
      "loss": 0.1121,
      "step": 10517
    },
    {
      "epoch": 1.4741415557112825,
      "grad_norm": 0.5994790196418762,
      "learning_rate": 0.00015630710356374073,
      "loss": 0.0658,
      "step": 10518
    },
    {
      "epoch": 1.474281709880869,
      "grad_norm": 0.44840505719184875,
      "learning_rate": 0.00015629275292992106,
      "loss": 0.1078,
      "step": 10519
    },
    {
      "epoch": 1.4744218640504556,
      "grad_norm": 1.0635029077529907,
      "learning_rate": 0.00015627840229610142,
      "loss": 0.0454,
      "step": 10520
    },
    {
      "epoch": 1.474562018220042,
      "grad_norm": 0.43159785866737366,
      "learning_rate": 0.00015626405166228175,
      "loss": 0.1139,
      "step": 10521
    },
    {
      "epoch": 1.4747021723896285,
      "grad_norm": 0.38016262650489807,
      "learning_rate": 0.00015624970102846207,
      "loss": 0.0541,
      "step": 10522
    },
    {
      "epoch": 1.4748423265592152,
      "grad_norm": 0.407825767993927,
      "learning_rate": 0.0001562353503946424,
      "loss": 0.0539,
      "step": 10523
    },
    {
      "epoch": 1.4749824807288017,
      "grad_norm": 0.22972871363162994,
      "learning_rate": 0.00015622099976082276,
      "loss": 0.0633,
      "step": 10524
    },
    {
      "epoch": 1.4751226348983884,
      "grad_norm": 0.5041607618331909,
      "learning_rate": 0.0001562066491270031,
      "loss": 0.0726,
      "step": 10525
    },
    {
      "epoch": 1.4752627890679748,
      "grad_norm": 0.7107272744178772,
      "learning_rate": 0.00015619229849318342,
      "loss": 0.1012,
      "step": 10526
    },
    {
      "epoch": 1.4754029432375613,
      "grad_norm": 0.33229541778564453,
      "learning_rate": 0.00015617794785936377,
      "loss": 0.092,
      "step": 10527
    },
    {
      "epoch": 1.4755430974071477,
      "grad_norm": 0.4425903260707855,
      "learning_rate": 0.0001561635972255441,
      "loss": 0.0369,
      "step": 10528
    },
    {
      "epoch": 1.4756832515767344,
      "grad_norm": 0.2784324884414673,
      "learning_rate": 0.00015614924659172443,
      "loss": 0.0898,
      "step": 10529
    },
    {
      "epoch": 1.4758234057463209,
      "grad_norm": 0.8509828448295593,
      "learning_rate": 0.0001561348959579048,
      "loss": 0.0758,
      "step": 10530
    },
    {
      "epoch": 1.4759635599159076,
      "grad_norm": 1.1231894493103027,
      "learning_rate": 0.00015612054532408514,
      "loss": 0.1082,
      "step": 10531
    },
    {
      "epoch": 1.476103714085494,
      "grad_norm": 2.6428587436676025,
      "learning_rate": 0.00015610619469026547,
      "loss": 0.1552,
      "step": 10532
    },
    {
      "epoch": 1.4762438682550805,
      "grad_norm": 0.774156391620636,
      "learning_rate": 0.00015609184405644582,
      "loss": 0.2044,
      "step": 10533
    },
    {
      "epoch": 1.4763840224246672,
      "grad_norm": 2.5511696338653564,
      "learning_rate": 0.00015607749342262615,
      "loss": 0.1108,
      "step": 10534
    },
    {
      "epoch": 1.4765241765942536,
      "grad_norm": 0.06326726824045181,
      "learning_rate": 0.00015606314278880648,
      "loss": 0.0042,
      "step": 10535
    },
    {
      "epoch": 1.4766643307638403,
      "grad_norm": 0.34364280104637146,
      "learning_rate": 0.00015604879215498684,
      "loss": 0.0563,
      "step": 10536
    },
    {
      "epoch": 1.4768044849334268,
      "grad_norm": 0.4228704273700714,
      "learning_rate": 0.00015603444152116717,
      "loss": 0.145,
      "step": 10537
    },
    {
      "epoch": 1.4769446391030132,
      "grad_norm": 0.189705029129982,
      "learning_rate": 0.0001560200908873475,
      "loss": 0.0464,
      "step": 10538
    },
    {
      "epoch": 1.4770847932726,
      "grad_norm": 0.3275066614151001,
      "learning_rate": 0.00015600574025352788,
      "loss": 0.0474,
      "step": 10539
    },
    {
      "epoch": 1.4772249474421864,
      "grad_norm": 0.270628422498703,
      "learning_rate": 0.0001559913896197082,
      "loss": 0.0459,
      "step": 10540
    },
    {
      "epoch": 1.477365101611773,
      "grad_norm": 0.3875422775745392,
      "learning_rate": 0.00015597703898588853,
      "loss": 0.0499,
      "step": 10541
    },
    {
      "epoch": 1.4775052557813595,
      "grad_norm": 0.47175467014312744,
      "learning_rate": 0.00015596268835206886,
      "loss": 0.079,
      "step": 10542
    },
    {
      "epoch": 1.477645409950946,
      "grad_norm": 0.32115423679351807,
      "learning_rate": 0.00015594833771824922,
      "loss": 0.0659,
      "step": 10543
    },
    {
      "epoch": 1.4777855641205326,
      "grad_norm": 0.27516838908195496,
      "learning_rate": 0.00015593398708442955,
      "loss": 0.0395,
      "step": 10544
    },
    {
      "epoch": 1.477925718290119,
      "grad_norm": 0.20578263700008392,
      "learning_rate": 0.00015591963645060988,
      "loss": 0.0386,
      "step": 10545
    },
    {
      "epoch": 1.4780658724597058,
      "grad_norm": 0.2117847353219986,
      "learning_rate": 0.00015590528581679023,
      "loss": 0.041,
      "step": 10546
    },
    {
      "epoch": 1.4782060266292922,
      "grad_norm": 0.2477148473262787,
      "learning_rate": 0.00015589093518297056,
      "loss": 0.0395,
      "step": 10547
    },
    {
      "epoch": 1.4783461807988787,
      "grad_norm": 0.3784487545490265,
      "learning_rate": 0.0001558765845491509,
      "loss": 0.0628,
      "step": 10548
    },
    {
      "epoch": 1.4784863349684654,
      "grad_norm": 0.4848484992980957,
      "learning_rate": 0.00015586223391533127,
      "loss": 0.0628,
      "step": 10549
    },
    {
      "epoch": 1.4786264891380518,
      "grad_norm": 0.3362005352973938,
      "learning_rate": 0.0001558478832815116,
      "loss": 0.0238,
      "step": 10550
    },
    {
      "epoch": 1.4787666433076385,
      "grad_norm": 0.11578638106584549,
      "learning_rate": 0.00015583353264769193,
      "loss": 0.0105,
      "step": 10551
    },
    {
      "epoch": 1.478906797477225,
      "grad_norm": 0.7384300827980042,
      "learning_rate": 0.00015581918201387228,
      "loss": 0.128,
      "step": 10552
    },
    {
      "epoch": 1.4790469516468114,
      "grad_norm": 0.22978708148002625,
      "learning_rate": 0.0001558048313800526,
      "loss": 0.0636,
      "step": 10553
    },
    {
      "epoch": 1.4791871058163981,
      "grad_norm": 0.22337229549884796,
      "learning_rate": 0.00015579048074623294,
      "loss": 0.0355,
      "step": 10554
    },
    {
      "epoch": 1.4793272599859846,
      "grad_norm": 0.41107049584388733,
      "learning_rate": 0.0001557761301124133,
      "loss": 0.0485,
      "step": 10555
    },
    {
      "epoch": 1.4794674141555713,
      "grad_norm": 0.31415748596191406,
      "learning_rate": 0.00015576177947859362,
      "loss": 0.0492,
      "step": 10556
    },
    {
      "epoch": 1.4796075683251577,
      "grad_norm": 0.260538786649704,
      "learning_rate": 0.00015574742884477395,
      "loss": 0.0405,
      "step": 10557
    },
    {
      "epoch": 1.4797477224947442,
      "grad_norm": 0.5792301893234253,
      "learning_rate": 0.00015573307821095428,
      "loss": 0.0634,
      "step": 10558
    },
    {
      "epoch": 1.4798878766643306,
      "grad_norm": 0.6849362254142761,
      "learning_rate": 0.00015571872757713464,
      "loss": 0.0525,
      "step": 10559
    },
    {
      "epoch": 1.4800280308339173,
      "grad_norm": 0.3748149275779724,
      "learning_rate": 0.00015570437694331497,
      "loss": 0.0718,
      "step": 10560
    },
    {
      "epoch": 1.4801681850035038,
      "grad_norm": 0.2086973637342453,
      "learning_rate": 0.00015569002630949532,
      "loss": 0.0432,
      "step": 10561
    },
    {
      "epoch": 1.4803083391730905,
      "grad_norm": 0.24434258043766022,
      "learning_rate": 0.00015567567567567568,
      "loss": 0.0574,
      "step": 10562
    },
    {
      "epoch": 1.480448493342677,
      "grad_norm": 0.7663661241531372,
      "learning_rate": 0.000155661325041856,
      "loss": 0.1086,
      "step": 10563
    },
    {
      "epoch": 1.4805886475122634,
      "grad_norm": 0.07997685670852661,
      "learning_rate": 0.00015564697440803633,
      "loss": 0.0144,
      "step": 10564
    },
    {
      "epoch": 1.48072880168185,
      "grad_norm": 0.908627986907959,
      "learning_rate": 0.0001556326237742167,
      "loss": 0.0469,
      "step": 10565
    },
    {
      "epoch": 1.4808689558514365,
      "grad_norm": 0.13131190836429596,
      "learning_rate": 0.00015561827314039702,
      "loss": 0.0218,
      "step": 10566
    },
    {
      "epoch": 1.4810091100210232,
      "grad_norm": 0.369503915309906,
      "learning_rate": 0.00015560392250657735,
      "loss": 0.0392,
      "step": 10567
    },
    {
      "epoch": 1.4811492641906097,
      "grad_norm": 0.3099110722541809,
      "learning_rate": 0.0001555895718727577,
      "loss": 0.0371,
      "step": 10568
    },
    {
      "epoch": 1.4812894183601961,
      "grad_norm": 0.2589000165462494,
      "learning_rate": 0.00015557522123893803,
      "loss": 0.0303,
      "step": 10569
    },
    {
      "epoch": 1.4814295725297828,
      "grad_norm": 0.2184293270111084,
      "learning_rate": 0.00015556087060511836,
      "loss": 0.0462,
      "step": 10570
    },
    {
      "epoch": 1.4815697266993693,
      "grad_norm": 0.75166255235672,
      "learning_rate": 0.00015554651997129874,
      "loss": 0.1352,
      "step": 10571
    },
    {
      "epoch": 1.481709880868956,
      "grad_norm": 0.32692375779151917,
      "learning_rate": 0.00015553216933747907,
      "loss": 0.0537,
      "step": 10572
    },
    {
      "epoch": 1.4818500350385424,
      "grad_norm": 0.15704116225242615,
      "learning_rate": 0.0001555178187036594,
      "loss": 0.0399,
      "step": 10573
    },
    {
      "epoch": 1.4819901892081289,
      "grad_norm": 0.10007571429014206,
      "learning_rate": 0.00015550346806983976,
      "loss": 0.0127,
      "step": 10574
    },
    {
      "epoch": 1.4821303433777155,
      "grad_norm": 0.3893049955368042,
      "learning_rate": 0.00015548911743602008,
      "loss": 0.0661,
      "step": 10575
    },
    {
      "epoch": 1.482270497547302,
      "grad_norm": 0.6022841334342957,
      "learning_rate": 0.0001554747668022004,
      "loss": 0.0562,
      "step": 10576
    },
    {
      "epoch": 1.4824106517168887,
      "grad_norm": 0.42999178171157837,
      "learning_rate": 0.00015546041616838074,
      "loss": 0.0729,
      "step": 10577
    },
    {
      "epoch": 1.4825508058864751,
      "grad_norm": 0.3609732389450073,
      "learning_rate": 0.0001554460655345611,
      "loss": 0.0416,
      "step": 10578
    },
    {
      "epoch": 1.4826909600560616,
      "grad_norm": 0.3163759112358093,
      "learning_rate": 0.00015543171490074143,
      "loss": 0.0788,
      "step": 10579
    },
    {
      "epoch": 1.4828311142256483,
      "grad_norm": 1.285809874534607,
      "learning_rate": 0.00015541736426692175,
      "loss": 0.1231,
      "step": 10580
    },
    {
      "epoch": 1.4829712683952347,
      "grad_norm": 0.40848323702812195,
      "learning_rate": 0.00015540301363310214,
      "loss": 0.0812,
      "step": 10581
    },
    {
      "epoch": 1.4831114225648214,
      "grad_norm": 0.8626853227615356,
      "learning_rate": 0.00015538866299928247,
      "loss": 0.1236,
      "step": 10582
    },
    {
      "epoch": 1.4832515767344079,
      "grad_norm": 0.6446532607078552,
      "learning_rate": 0.0001553743123654628,
      "loss": 0.0262,
      "step": 10583
    },
    {
      "epoch": 1.4833917309039943,
      "grad_norm": 1.0703104734420776,
      "learning_rate": 0.00015535996173164315,
      "loss": 0.0841,
      "step": 10584
    },
    {
      "epoch": 1.4835318850735808,
      "grad_norm": 1.4783800840377808,
      "learning_rate": 0.00015534561109782348,
      "loss": 0.1207,
      "step": 10585
    },
    {
      "epoch": 1.4836720392431675,
      "grad_norm": 0.5006522536277771,
      "learning_rate": 0.0001553312604640038,
      "loss": 0.0654,
      "step": 10586
    },
    {
      "epoch": 1.4838121934127542,
      "grad_norm": 0.11923687905073166,
      "learning_rate": 0.00015531690983018416,
      "loss": 0.0365,
      "step": 10587
    },
    {
      "epoch": 1.4839523475823406,
      "grad_norm": 0.2262464463710785,
      "learning_rate": 0.0001553025591963645,
      "loss": 0.0572,
      "step": 10588
    },
    {
      "epoch": 1.484092501751927,
      "grad_norm": 0.2834444046020508,
      "learning_rate": 0.00015528820856254482,
      "loss": 0.0361,
      "step": 10589
    },
    {
      "epoch": 1.4842326559215135,
      "grad_norm": 0.25554704666137695,
      "learning_rate": 0.00015527385792872518,
      "loss": 0.0301,
      "step": 10590
    },
    {
      "epoch": 1.4843728100911002,
      "grad_norm": 0.2724181115627289,
      "learning_rate": 0.0001552595072949055,
      "loss": 0.038,
      "step": 10591
    },
    {
      "epoch": 1.4845129642606867,
      "grad_norm": 0.42190948128700256,
      "learning_rate": 0.00015524515666108583,
      "loss": 0.0221,
      "step": 10592
    },
    {
      "epoch": 1.4846531184302734,
      "grad_norm": 0.35933244228363037,
      "learning_rate": 0.0001552308060272662,
      "loss": 0.0471,
      "step": 10593
    },
    {
      "epoch": 1.4847932725998598,
      "grad_norm": 0.14831778407096863,
      "learning_rate": 0.00015521645539344654,
      "loss": 0.0156,
      "step": 10594
    },
    {
      "epoch": 1.4849334267694463,
      "grad_norm": 0.2693512737751007,
      "learning_rate": 0.00015520210475962687,
      "loss": 0.0338,
      "step": 10595
    },
    {
      "epoch": 1.485073580939033,
      "grad_norm": 0.06814882904291153,
      "learning_rate": 0.0001551877541258072,
      "loss": 0.013,
      "step": 10596
    },
    {
      "epoch": 1.4852137351086194,
      "grad_norm": 0.18754149973392487,
      "learning_rate": 0.00015517340349198756,
      "loss": 0.0462,
      "step": 10597
    },
    {
      "epoch": 1.485353889278206,
      "grad_norm": 0.16206228733062744,
      "learning_rate": 0.00015515905285816789,
      "loss": 0.0342,
      "step": 10598
    },
    {
      "epoch": 1.4854940434477926,
      "grad_norm": 0.3485026955604553,
      "learning_rate": 0.00015514470222434821,
      "loss": 0.0319,
      "step": 10599
    },
    {
      "epoch": 1.485634197617379,
      "grad_norm": 0.37309691309928894,
      "learning_rate": 0.00015513035159052857,
      "loss": 0.0223,
      "step": 10600
    },
    {
      "epoch": 1.4857743517869657,
      "grad_norm": 0.5247063040733337,
      "learning_rate": 0.0001551160009567089,
      "loss": 0.0712,
      "step": 10601
    },
    {
      "epoch": 1.4859145059565522,
      "grad_norm": 0.14952580630779266,
      "learning_rate": 0.00015510165032288923,
      "loss": 0.0207,
      "step": 10602
    },
    {
      "epoch": 1.4860546601261388,
      "grad_norm": 0.24514086544513702,
      "learning_rate": 0.0001550872996890696,
      "loss": 0.0491,
      "step": 10603
    },
    {
      "epoch": 1.4861948142957253,
      "grad_norm": 0.12210065126419067,
      "learning_rate": 0.00015507294905524994,
      "loss": 0.0136,
      "step": 10604
    },
    {
      "epoch": 1.4863349684653118,
      "grad_norm": 0.5551408529281616,
      "learning_rate": 0.00015505859842143027,
      "loss": 0.086,
      "step": 10605
    },
    {
      "epoch": 1.4864751226348984,
      "grad_norm": 0.38685253262519836,
      "learning_rate": 0.00015504424778761062,
      "loss": 0.0317,
      "step": 10606
    },
    {
      "epoch": 1.486615276804485,
      "grad_norm": 0.5873581767082214,
      "learning_rate": 0.00015502989715379095,
      "loss": 0.0603,
      "step": 10607
    },
    {
      "epoch": 1.4867554309740716,
      "grad_norm": 0.5957799553871155,
      "learning_rate": 0.00015501554651997128,
      "loss": 0.0337,
      "step": 10608
    },
    {
      "epoch": 1.486895585143658,
      "grad_norm": 0.6263166666030884,
      "learning_rate": 0.00015500119588615163,
      "loss": 0.0336,
      "step": 10609
    },
    {
      "epoch": 1.4870357393132445,
      "grad_norm": 0.6674920916557312,
      "learning_rate": 0.00015498684525233196,
      "loss": 0.0749,
      "step": 10610
    },
    {
      "epoch": 1.4871758934828312,
      "grad_norm": 0.2550250291824341,
      "learning_rate": 0.0001549724946185123,
      "loss": 0.0387,
      "step": 10611
    },
    {
      "epoch": 1.4873160476524176,
      "grad_norm": 0.4844677448272705,
      "learning_rate": 0.00015495814398469262,
      "loss": 0.0853,
      "step": 10612
    },
    {
      "epoch": 1.4874562018220043,
      "grad_norm": 0.2599497139453888,
      "learning_rate": 0.000154943793350873,
      "loss": 0.0417,
      "step": 10613
    },
    {
      "epoch": 1.4875963559915908,
      "grad_norm": 0.2552608847618103,
      "learning_rate": 0.00015492944271705333,
      "loss": 0.0423,
      "step": 10614
    },
    {
      "epoch": 1.4877365101611773,
      "grad_norm": 0.1310550719499588,
      "learning_rate": 0.00015491509208323366,
      "loss": 0.0112,
      "step": 10615
    },
    {
      "epoch": 1.4878766643307637,
      "grad_norm": 1.0396101474761963,
      "learning_rate": 0.00015490074144941402,
      "loss": 0.0906,
      "step": 10616
    },
    {
      "epoch": 1.4880168185003504,
      "grad_norm": 0.2367522120475769,
      "learning_rate": 0.00015488639081559434,
      "loss": 0.0442,
      "step": 10617
    },
    {
      "epoch": 1.4881569726699369,
      "grad_norm": 0.12786652147769928,
      "learning_rate": 0.00015487204018177467,
      "loss": 0.0171,
      "step": 10618
    },
    {
      "epoch": 1.4882971268395235,
      "grad_norm": 0.6618436574935913,
      "learning_rate": 0.00015485768954795503,
      "loss": 0.1401,
      "step": 10619
    },
    {
      "epoch": 1.48843728100911,
      "grad_norm": 0.16624286770820618,
      "learning_rate": 0.00015484333891413536,
      "loss": 0.0286,
      "step": 10620
    },
    {
      "epoch": 1.4885774351786965,
      "grad_norm": 0.3128873407840729,
      "learning_rate": 0.00015482898828031569,
      "loss": 0.0716,
      "step": 10621
    },
    {
      "epoch": 1.4887175893482831,
      "grad_norm": 0.10659144073724747,
      "learning_rate": 0.00015481463764649604,
      "loss": 0.0108,
      "step": 10622
    },
    {
      "epoch": 1.4888577435178696,
      "grad_norm": 0.43128708004951477,
      "learning_rate": 0.00015480028701267637,
      "loss": 0.1436,
      "step": 10623
    },
    {
      "epoch": 1.4889978976874563,
      "grad_norm": 0.3703378736972809,
      "learning_rate": 0.0001547859363788567,
      "loss": 0.0388,
      "step": 10624
    },
    {
      "epoch": 1.4891380518570427,
      "grad_norm": 0.5118187665939331,
      "learning_rate": 0.00015477158574503708,
      "loss": 0.0707,
      "step": 10625
    },
    {
      "epoch": 1.4892782060266292,
      "grad_norm": 0.3890518546104431,
      "learning_rate": 0.0001547572351112174,
      "loss": 0.0756,
      "step": 10626
    },
    {
      "epoch": 1.4894183601962159,
      "grad_norm": 0.4319985508918762,
      "learning_rate": 0.00015474288447739774,
      "loss": 0.0447,
      "step": 10627
    },
    {
      "epoch": 1.4895585143658023,
      "grad_norm": 0.34239524602890015,
      "learning_rate": 0.00015472853384357807,
      "loss": 0.0637,
      "step": 10628
    },
    {
      "epoch": 1.489698668535389,
      "grad_norm": 0.33493366837501526,
      "learning_rate": 0.00015471418320975842,
      "loss": 0.0213,
      "step": 10629
    },
    {
      "epoch": 1.4898388227049755,
      "grad_norm": 0.7749230265617371,
      "learning_rate": 0.00015469983257593875,
      "loss": 0.1294,
      "step": 10630
    },
    {
      "epoch": 1.489978976874562,
      "grad_norm": 0.31648990511894226,
      "learning_rate": 0.00015468548194211908,
      "loss": 0.0118,
      "step": 10631
    },
    {
      "epoch": 1.4901191310441486,
      "grad_norm": 0.9662306308746338,
      "learning_rate": 0.00015467113130829944,
      "loss": 0.0986,
      "step": 10632
    },
    {
      "epoch": 1.490259285213735,
      "grad_norm": 1.3401962518692017,
      "learning_rate": 0.00015465678067447976,
      "loss": 0.0693,
      "step": 10633
    },
    {
      "epoch": 1.4903994393833218,
      "grad_norm": 0.11344845592975616,
      "learning_rate": 0.0001546424300406601,
      "loss": 0.0053,
      "step": 10634
    },
    {
      "epoch": 1.4905395935529082,
      "grad_norm": 0.4984247386455536,
      "learning_rate": 0.00015462807940684048,
      "loss": 0.0767,
      "step": 10635
    },
    {
      "epoch": 1.4906797477224947,
      "grad_norm": 0.28609466552734375,
      "learning_rate": 0.0001546137287730208,
      "loss": 0.0538,
      "step": 10636
    },
    {
      "epoch": 1.4908199018920814,
      "grad_norm": 0.2780989110469818,
      "learning_rate": 0.00015459937813920113,
      "loss": 0.0409,
      "step": 10637
    },
    {
      "epoch": 1.4909600560616678,
      "grad_norm": 0.18578656017780304,
      "learning_rate": 0.0001545850275053815,
      "loss": 0.028,
      "step": 10638
    },
    {
      "epoch": 1.4911002102312545,
      "grad_norm": 0.3846004903316498,
      "learning_rate": 0.00015457067687156182,
      "loss": 0.0976,
      "step": 10639
    },
    {
      "epoch": 1.491240364400841,
      "grad_norm": 0.34138429164886475,
      "learning_rate": 0.00015455632623774215,
      "loss": 0.0298,
      "step": 10640
    },
    {
      "epoch": 1.4913805185704274,
      "grad_norm": 0.26356932520866394,
      "learning_rate": 0.0001545419756039225,
      "loss": 0.0921,
      "step": 10641
    },
    {
      "epoch": 1.491520672740014,
      "grad_norm": 0.1698404848575592,
      "learning_rate": 0.00015452762497010283,
      "loss": 0.0222,
      "step": 10642
    },
    {
      "epoch": 1.4916608269096006,
      "grad_norm": 0.3921966254711151,
      "learning_rate": 0.00015451327433628316,
      "loss": 0.0329,
      "step": 10643
    },
    {
      "epoch": 1.4918009810791872,
      "grad_norm": 0.21553845703601837,
      "learning_rate": 0.00015449892370246354,
      "loss": 0.0544,
      "step": 10644
    },
    {
      "epoch": 1.4919411352487737,
      "grad_norm": 0.3352382183074951,
      "learning_rate": 0.00015448457306864387,
      "loss": 0.064,
      "step": 10645
    },
    {
      "epoch": 1.4920812894183602,
      "grad_norm": 0.5347132682800293,
      "learning_rate": 0.0001544702224348242,
      "loss": 0.0665,
      "step": 10646
    },
    {
      "epoch": 1.4922214435879466,
      "grad_norm": 0.4259601831436157,
      "learning_rate": 0.00015445587180100453,
      "loss": 0.0677,
      "step": 10647
    },
    {
      "epoch": 1.4923615977575333,
      "grad_norm": 0.2352869212627411,
      "learning_rate": 0.00015444152116718488,
      "loss": 0.0419,
      "step": 10648
    },
    {
      "epoch": 1.4925017519271198,
      "grad_norm": 0.1949128806591034,
      "learning_rate": 0.0001544271705333652,
      "loss": 0.04,
      "step": 10649
    },
    {
      "epoch": 1.4926419060967064,
      "grad_norm": 0.309375137090683,
      "learning_rate": 0.00015441281989954554,
      "loss": 0.046,
      "step": 10650
    },
    {
      "epoch": 1.492782060266293,
      "grad_norm": 0.3392605483531952,
      "learning_rate": 0.0001543984692657259,
      "loss": 0.0211,
      "step": 10651
    },
    {
      "epoch": 1.4929222144358794,
      "grad_norm": 0.45508208870887756,
      "learning_rate": 0.00015438411863190622,
      "loss": 0.0603,
      "step": 10652
    },
    {
      "epoch": 1.493062368605466,
      "grad_norm": 0.3773292303085327,
      "learning_rate": 0.00015436976799808655,
      "loss": 0.0283,
      "step": 10653
    },
    {
      "epoch": 1.4932025227750525,
      "grad_norm": 0.2173190861940384,
      "learning_rate": 0.0001543554173642669,
      "loss": 0.0343,
      "step": 10654
    },
    {
      "epoch": 1.4933426769446392,
      "grad_norm": 0.22467036545276642,
      "learning_rate": 0.00015434106673044724,
      "loss": 0.0396,
      "step": 10655
    },
    {
      "epoch": 1.4934828311142256,
      "grad_norm": 0.1732553392648697,
      "learning_rate": 0.00015432671609662756,
      "loss": 0.0277,
      "step": 10656
    },
    {
      "epoch": 1.493622985283812,
      "grad_norm": 0.6231887936592102,
      "learning_rate": 0.00015431236546280795,
      "loss": 0.0861,
      "step": 10657
    },
    {
      "epoch": 1.4937631394533988,
      "grad_norm": 0.1981246918439865,
      "learning_rate": 0.00015429801482898828,
      "loss": 0.0458,
      "step": 10658
    },
    {
      "epoch": 1.4939032936229852,
      "grad_norm": 0.2730039954185486,
      "learning_rate": 0.0001542836641951686,
      "loss": 0.012,
      "step": 10659
    },
    {
      "epoch": 1.494043447792572,
      "grad_norm": 0.3582146167755127,
      "learning_rate": 0.00015426931356134896,
      "loss": 0.068,
      "step": 10660
    },
    {
      "epoch": 1.4941836019621584,
      "grad_norm": 0.18658442795276642,
      "learning_rate": 0.0001542549629275293,
      "loss": 0.0344,
      "step": 10661
    },
    {
      "epoch": 1.4943237561317448,
      "grad_norm": 0.2621755003929138,
      "learning_rate": 0.00015424061229370962,
      "loss": 0.0432,
      "step": 10662
    },
    {
      "epoch": 1.4944639103013315,
      "grad_norm": 0.2550070881843567,
      "learning_rate": 0.00015422626165988995,
      "loss": 0.0263,
      "step": 10663
    },
    {
      "epoch": 1.494604064470918,
      "grad_norm": 0.7072445750236511,
      "learning_rate": 0.0001542119110260703,
      "loss": 0.0659,
      "step": 10664
    },
    {
      "epoch": 1.4947442186405047,
      "grad_norm": 0.2643151581287384,
      "learning_rate": 0.00015419756039225063,
      "loss": 0.0316,
      "step": 10665
    },
    {
      "epoch": 1.4948843728100911,
      "grad_norm": 0.20512066781520844,
      "learning_rate": 0.00015418320975843096,
      "loss": 0.03,
      "step": 10666
    },
    {
      "epoch": 1.4950245269796776,
      "grad_norm": 0.5069049596786499,
      "learning_rate": 0.00015416885912461134,
      "loss": 0.0823,
      "step": 10667
    },
    {
      "epoch": 1.4951646811492643,
      "grad_norm": 0.1641375720500946,
      "learning_rate": 0.00015415450849079167,
      "loss": 0.0158,
      "step": 10668
    },
    {
      "epoch": 1.4953048353188507,
      "grad_norm": 0.4637381434440613,
      "learning_rate": 0.000154140157856972,
      "loss": 0.043,
      "step": 10669
    },
    {
      "epoch": 1.4954449894884374,
      "grad_norm": 0.15042731165885925,
      "learning_rate": 0.00015412580722315235,
      "loss": 0.0128,
      "step": 10670
    },
    {
      "epoch": 1.4955851436580239,
      "grad_norm": 0.2220517247915268,
      "learning_rate": 0.00015411145658933268,
      "loss": 0.0565,
      "step": 10671
    },
    {
      "epoch": 1.4957252978276103,
      "grad_norm": 0.45739394426345825,
      "learning_rate": 0.000154097105955513,
      "loss": 0.0992,
      "step": 10672
    },
    {
      "epoch": 1.4958654519971968,
      "grad_norm": 0.310108482837677,
      "learning_rate": 0.00015408275532169337,
      "loss": 0.0532,
      "step": 10673
    },
    {
      "epoch": 1.4960056061667835,
      "grad_norm": 0.8512653112411499,
      "learning_rate": 0.0001540684046878737,
      "loss": 0.1773,
      "step": 10674
    },
    {
      "epoch": 1.4961457603363701,
      "grad_norm": 0.6950824856758118,
      "learning_rate": 0.00015405405405405402,
      "loss": 0.0884,
      "step": 10675
    },
    {
      "epoch": 1.4962859145059566,
      "grad_norm": 0.2746245265007019,
      "learning_rate": 0.0001540397034202344,
      "loss": 0.0432,
      "step": 10676
    },
    {
      "epoch": 1.496426068675543,
      "grad_norm": 0.3589267432689667,
      "learning_rate": 0.00015402535278641474,
      "loss": 0.0726,
      "step": 10677
    },
    {
      "epoch": 1.4965662228451295,
      "grad_norm": 0.39547809958457947,
      "learning_rate": 0.00015401100215259506,
      "loss": 0.033,
      "step": 10678
    },
    {
      "epoch": 1.4967063770147162,
      "grad_norm": 0.34382662177085876,
      "learning_rate": 0.0001539966515187754,
      "loss": 0.0869,
      "step": 10679
    },
    {
      "epoch": 1.4968465311843027,
      "grad_norm": 1.4440655708312988,
      "learning_rate": 0.00015398230088495575,
      "loss": 0.0987,
      "step": 10680
    },
    {
      "epoch": 1.4969866853538893,
      "grad_norm": 0.47064104676246643,
      "learning_rate": 0.00015396795025113608,
      "loss": 0.056,
      "step": 10681
    },
    {
      "epoch": 1.4971268395234758,
      "grad_norm": 0.26009801030158997,
      "learning_rate": 0.0001539535996173164,
      "loss": 0.0278,
      "step": 10682
    },
    {
      "epoch": 1.4972669936930623,
      "grad_norm": 0.20467530190944672,
      "learning_rate": 0.00015393924898349676,
      "loss": 0.038,
      "step": 10683
    },
    {
      "epoch": 1.497407147862649,
      "grad_norm": 0.24109293520450592,
      "learning_rate": 0.0001539248983496771,
      "loss": 0.0267,
      "step": 10684
    },
    {
      "epoch": 1.4975473020322354,
      "grad_norm": 1.0723930597305298,
      "learning_rate": 0.00015391054771585742,
      "loss": 0.1705,
      "step": 10685
    },
    {
      "epoch": 1.497687456201822,
      "grad_norm": 0.5197061896324158,
      "learning_rate": 0.00015389619708203777,
      "loss": 0.1189,
      "step": 10686
    },
    {
      "epoch": 1.4978276103714085,
      "grad_norm": 0.3153019845485687,
      "learning_rate": 0.0001538818464482181,
      "loss": 0.0434,
      "step": 10687
    },
    {
      "epoch": 1.497967764540995,
      "grad_norm": 0.20601604878902435,
      "learning_rate": 0.00015386749581439846,
      "loss": 0.0264,
      "step": 10688
    },
    {
      "epoch": 1.4981079187105817,
      "grad_norm": 0.21115687489509583,
      "learning_rate": 0.00015385314518057881,
      "loss": 0.1023,
      "step": 10689
    },
    {
      "epoch": 1.4982480728801681,
      "grad_norm": 0.4875928461551666,
      "learning_rate": 0.00015383879454675914,
      "loss": 0.0847,
      "step": 10690
    },
    {
      "epoch": 1.4983882270497548,
      "grad_norm": 0.4417114555835724,
      "learning_rate": 0.00015382444391293947,
      "loss": 0.0699,
      "step": 10691
    },
    {
      "epoch": 1.4985283812193413,
      "grad_norm": 0.5027158856391907,
      "learning_rate": 0.00015381009327911983,
      "loss": 0.0909,
      "step": 10692
    },
    {
      "epoch": 1.4986685353889277,
      "grad_norm": 0.4453561305999756,
      "learning_rate": 0.00015379574264530016,
      "loss": 0.0523,
      "step": 10693
    },
    {
      "epoch": 1.4988086895585144,
      "grad_norm": 0.19902105629444122,
      "learning_rate": 0.00015378139201148048,
      "loss": 0.0413,
      "step": 10694
    },
    {
      "epoch": 1.4989488437281009,
      "grad_norm": 0.36840540170669556,
      "learning_rate": 0.00015376704137766084,
      "loss": 0.0525,
      "step": 10695
    },
    {
      "epoch": 1.4990889978976876,
      "grad_norm": 0.32745835185050964,
      "learning_rate": 0.00015375269074384117,
      "loss": 0.059,
      "step": 10696
    },
    {
      "epoch": 1.499229152067274,
      "grad_norm": 0.6460821032524109,
      "learning_rate": 0.0001537383401100215,
      "loss": 0.0995,
      "step": 10697
    },
    {
      "epoch": 1.4993693062368605,
      "grad_norm": 0.4269315302371979,
      "learning_rate": 0.00015372398947620183,
      "loss": 0.0592,
      "step": 10698
    },
    {
      "epoch": 1.4995094604064472,
      "grad_norm": 0.20190581679344177,
      "learning_rate": 0.0001537096388423822,
      "loss": 0.0204,
      "step": 10699
    },
    {
      "epoch": 1.4996496145760336,
      "grad_norm": 0.40271300077438354,
      "learning_rate": 0.00015369528820856254,
      "loss": 0.0768,
      "step": 10700
    },
    {
      "epoch": 1.4997897687456203,
      "grad_norm": 0.25591036677360535,
      "learning_rate": 0.00015368093757474287,
      "loss": 0.0386,
      "step": 10701
    },
    {
      "epoch": 1.4999299229152068,
      "grad_norm": 0.2654736042022705,
      "learning_rate": 0.00015366658694092322,
      "loss": 0.0404,
      "step": 10702
    },
    {
      "epoch": 1.5000700770847932,
      "grad_norm": 0.17650796473026276,
      "learning_rate": 0.00015365223630710355,
      "loss": 0.0421,
      "step": 10703
    },
    {
      "epoch": 1.5002102312543797,
      "grad_norm": 0.26553329825401306,
      "learning_rate": 0.00015363788567328388,
      "loss": 0.0405,
      "step": 10704
    },
    {
      "epoch": 1.5003503854239664,
      "grad_norm": 0.15579751133918762,
      "learning_rate": 0.00015362353503946423,
      "loss": 0.0829,
      "step": 10705
    },
    {
      "epoch": 1.500490539593553,
      "grad_norm": 0.3489481806755066,
      "learning_rate": 0.00015360918440564456,
      "loss": 0.0312,
      "step": 10706
    },
    {
      "epoch": 1.5006306937631395,
      "grad_norm": 0.299667090177536,
      "learning_rate": 0.0001535948337718249,
      "loss": 0.0478,
      "step": 10707
    },
    {
      "epoch": 1.500770847932726,
      "grad_norm": 0.09261930733919144,
      "learning_rate": 0.00015358048313800527,
      "loss": 0.0141,
      "step": 10708
    },
    {
      "epoch": 1.5009110021023124,
      "grad_norm": 0.5096018314361572,
      "learning_rate": 0.0001535661325041856,
      "loss": 0.0773,
      "step": 10709
    },
    {
      "epoch": 1.5010511562718991,
      "grad_norm": 0.09015260636806488,
      "learning_rate": 0.00015355178187036593,
      "loss": 0.0113,
      "step": 10710
    },
    {
      "epoch": 1.5011913104414858,
      "grad_norm": 0.8212284445762634,
      "learning_rate": 0.00015353743123654629,
      "loss": 0.0666,
      "step": 10711
    },
    {
      "epoch": 1.5013314646110723,
      "grad_norm": 0.519081711769104,
      "learning_rate": 0.00015352308060272661,
      "loss": 0.1139,
      "step": 10712
    },
    {
      "epoch": 1.5014716187806587,
      "grad_norm": 0.2797171473503113,
      "learning_rate": 0.00015350872996890694,
      "loss": 0.0302,
      "step": 10713
    },
    {
      "epoch": 1.5016117729502452,
      "grad_norm": 0.3629531264305115,
      "learning_rate": 0.00015349437933508727,
      "loss": 0.0469,
      "step": 10714
    },
    {
      "epoch": 1.5017519271198319,
      "grad_norm": 0.28467485308647156,
      "learning_rate": 0.00015348002870126763,
      "loss": 0.06,
      "step": 10715
    },
    {
      "epoch": 1.5018920812894183,
      "grad_norm": 0.28747886419296265,
      "learning_rate": 0.00015346567806744796,
      "loss": 0.0233,
      "step": 10716
    },
    {
      "epoch": 1.502032235459005,
      "grad_norm": 0.4266621172428131,
      "learning_rate": 0.00015345132743362828,
      "loss": 0.0316,
      "step": 10717
    },
    {
      "epoch": 1.5021723896285915,
      "grad_norm": 0.13075195252895355,
      "learning_rate": 0.00015343697679980864,
      "loss": 0.0147,
      "step": 10718
    },
    {
      "epoch": 1.502312543798178,
      "grad_norm": 0.12602153420448303,
      "learning_rate": 0.00015342262616598897,
      "loss": 0.0114,
      "step": 10719
    },
    {
      "epoch": 1.5024526979677646,
      "grad_norm": 0.1261730194091797,
      "learning_rate": 0.00015340827553216932,
      "loss": 0.0088,
      "step": 10720
    },
    {
      "epoch": 1.502592852137351,
      "grad_norm": 1.1034353971481323,
      "learning_rate": 0.00015339392489834968,
      "loss": 0.0737,
      "step": 10721
    },
    {
      "epoch": 1.5027330063069377,
      "grad_norm": 0.32052773237228394,
      "learning_rate": 0.00015337957426453,
      "loss": 0.0283,
      "step": 10722
    },
    {
      "epoch": 1.5028731604765242,
      "grad_norm": 0.4729958772659302,
      "learning_rate": 0.00015336522363071034,
      "loss": 0.0362,
      "step": 10723
    },
    {
      "epoch": 1.5030133146461107,
      "grad_norm": 0.4405868649482727,
      "learning_rate": 0.0001533508729968907,
      "loss": 0.0416,
      "step": 10724
    },
    {
      "epoch": 1.5031534688156971,
      "grad_norm": 0.9098767042160034,
      "learning_rate": 0.00015333652236307102,
      "loss": 0.1684,
      "step": 10725
    },
    {
      "epoch": 1.5032936229852838,
      "grad_norm": 0.16408763825893402,
      "learning_rate": 0.00015332217172925135,
      "loss": 0.0327,
      "step": 10726
    },
    {
      "epoch": 1.5034337771548705,
      "grad_norm": 0.7781208753585815,
      "learning_rate": 0.0001533078210954317,
      "loss": 0.1479,
      "step": 10727
    },
    {
      "epoch": 1.503573931324457,
      "grad_norm": 0.22220692038536072,
      "learning_rate": 0.00015329347046161203,
      "loss": 0.024,
      "step": 10728
    },
    {
      "epoch": 1.5037140854940434,
      "grad_norm": 0.16805599629878998,
      "learning_rate": 0.00015327911982779236,
      "loss": 0.0081,
      "step": 10729
    },
    {
      "epoch": 1.5038542396636299,
      "grad_norm": 0.23774494230747223,
      "learning_rate": 0.00015326476919397275,
      "loss": 0.025,
      "step": 10730
    },
    {
      "epoch": 1.5039943938332165,
      "grad_norm": 1.0608748197555542,
      "learning_rate": 0.00015325041856015307,
      "loss": 0.0874,
      "step": 10731
    },
    {
      "epoch": 1.5041345480028032,
      "grad_norm": 1.797458291053772,
      "learning_rate": 0.0001532360679263334,
      "loss": 0.2008,
      "step": 10732
    },
    {
      "epoch": 1.5042747021723897,
      "grad_norm": 1.239945411682129,
      "learning_rate": 0.00015322171729251373,
      "loss": 0.0383,
      "step": 10733
    },
    {
      "epoch": 1.5044148563419761,
      "grad_norm": 1.1340250968933105,
      "learning_rate": 0.0001532073666586941,
      "loss": 0.0418,
      "step": 10734
    },
    {
      "epoch": 1.5045550105115626,
      "grad_norm": 0.3491586744785309,
      "learning_rate": 0.00015319301602487442,
      "loss": 0.0213,
      "step": 10735
    },
    {
      "epoch": 1.5046951646811493,
      "grad_norm": 0.32779473066329956,
      "learning_rate": 0.00015317866539105474,
      "loss": 0.0553,
      "step": 10736
    },
    {
      "epoch": 1.504835318850736,
      "grad_norm": 1.9793720245361328,
      "learning_rate": 0.0001531643147572351,
      "loss": 0.0548,
      "step": 10737
    },
    {
      "epoch": 1.5049754730203224,
      "grad_norm": 0.9499906301498413,
      "learning_rate": 0.00015314996412341543,
      "loss": 0.0463,
      "step": 10738
    },
    {
      "epoch": 1.5051156271899089,
      "grad_norm": 0.2471480667591095,
      "learning_rate": 0.00015313561348959576,
      "loss": 0.0105,
      "step": 10739
    },
    {
      "epoch": 1.5052557813594953,
      "grad_norm": 0.621599018573761,
      "learning_rate": 0.00015312126285577614,
      "loss": 0.0712,
      "step": 10740
    },
    {
      "epoch": 1.505395935529082,
      "grad_norm": 0.15189777314662933,
      "learning_rate": 0.00015310691222195647,
      "loss": 0.0313,
      "step": 10741
    },
    {
      "epoch": 1.5055360896986687,
      "grad_norm": 0.32547056674957275,
      "learning_rate": 0.0001530925615881368,
      "loss": 0.0655,
      "step": 10742
    },
    {
      "epoch": 1.5056762438682552,
      "grad_norm": 0.31570181250572205,
      "learning_rate": 0.00015307821095431715,
      "loss": 0.0808,
      "step": 10743
    },
    {
      "epoch": 1.5058163980378416,
      "grad_norm": 0.35404089093208313,
      "learning_rate": 0.00015306386032049748,
      "loss": 0.0281,
      "step": 10744
    },
    {
      "epoch": 1.505956552207428,
      "grad_norm": 0.16628332436084747,
      "learning_rate": 0.0001530495096866778,
      "loss": 0.0346,
      "step": 10745
    },
    {
      "epoch": 1.5060967063770148,
      "grad_norm": 0.36109432578086853,
      "learning_rate": 0.00015303515905285817,
      "loss": 0.0804,
      "step": 10746
    },
    {
      "epoch": 1.5062368605466012,
      "grad_norm": 0.5275200009346008,
      "learning_rate": 0.0001530208084190385,
      "loss": 0.088,
      "step": 10747
    },
    {
      "epoch": 1.506377014716188,
      "grad_norm": 0.06977388262748718,
      "learning_rate": 0.00015300645778521882,
      "loss": 0.0089,
      "step": 10748
    },
    {
      "epoch": 1.5065171688857744,
      "grad_norm": 0.352504163980484,
      "learning_rate": 0.00015299210715139915,
      "loss": 0.0482,
      "step": 10749
    },
    {
      "epoch": 1.5066573230553608,
      "grad_norm": 0.3667183518409729,
      "learning_rate": 0.0001529777565175795,
      "loss": 0.0917,
      "step": 10750
    },
    {
      "epoch": 1.5067974772249473,
      "grad_norm": 0.14459437131881714,
      "learning_rate": 0.00015296340588375984,
      "loss": 0.0097,
      "step": 10751
    },
    {
      "epoch": 1.506937631394534,
      "grad_norm": 0.308759868144989,
      "learning_rate": 0.0001529490552499402,
      "loss": 0.0657,
      "step": 10752
    },
    {
      "epoch": 1.5070777855641206,
      "grad_norm": 0.29015445709228516,
      "learning_rate": 0.00015293470461612055,
      "loss": 0.0647,
      "step": 10753
    },
    {
      "epoch": 1.507217939733707,
      "grad_norm": 0.3704620599746704,
      "learning_rate": 0.00015292035398230088,
      "loss": 0.0681,
      "step": 10754
    },
    {
      "epoch": 1.5073580939032936,
      "grad_norm": 0.2279195338487625,
      "learning_rate": 0.0001529060033484812,
      "loss": 0.0319,
      "step": 10755
    },
    {
      "epoch": 1.50749824807288,
      "grad_norm": 0.12060660868883133,
      "learning_rate": 0.00015289165271466156,
      "loss": 0.0147,
      "step": 10756
    },
    {
      "epoch": 1.5076384022424667,
      "grad_norm": 0.5635783076286316,
      "learning_rate": 0.0001528773020808419,
      "loss": 0.0523,
      "step": 10757
    },
    {
      "epoch": 1.5077785564120534,
      "grad_norm": 0.2604323923587799,
      "learning_rate": 0.00015286295144702222,
      "loss": 0.0332,
      "step": 10758
    },
    {
      "epoch": 1.5079187105816398,
      "grad_norm": 0.29043710231781006,
      "learning_rate": 0.00015284860081320257,
      "loss": 0.0611,
      "step": 10759
    },
    {
      "epoch": 1.5080588647512263,
      "grad_norm": 0.28768956661224365,
      "learning_rate": 0.0001528342501793829,
      "loss": 0.0515,
      "step": 10760
    },
    {
      "epoch": 1.5081990189208128,
      "grad_norm": 0.21576617658138275,
      "learning_rate": 0.00015281989954556323,
      "loss": 0.0358,
      "step": 10761
    },
    {
      "epoch": 1.5083391730903994,
      "grad_norm": 0.08806242793798447,
      "learning_rate": 0.0001528055489117436,
      "loss": 0.0063,
      "step": 10762
    },
    {
      "epoch": 1.5084793272599861,
      "grad_norm": 0.25208035111427307,
      "learning_rate": 0.00015279119827792394,
      "loss": 0.0182,
      "step": 10763
    },
    {
      "epoch": 1.5086194814295726,
      "grad_norm": 0.6288411617279053,
      "learning_rate": 0.00015277684764410427,
      "loss": 0.0524,
      "step": 10764
    },
    {
      "epoch": 1.508759635599159,
      "grad_norm": 0.5118399858474731,
      "learning_rate": 0.00015276249701028462,
      "loss": 0.0639,
      "step": 10765
    },
    {
      "epoch": 1.5088997897687455,
      "grad_norm": 0.5515400171279907,
      "learning_rate": 0.00015274814637646495,
      "loss": 0.1634,
      "step": 10766
    },
    {
      "epoch": 1.5090399439383322,
      "grad_norm": 0.4751434028148651,
      "learning_rate": 0.00015273379574264528,
      "loss": 0.0782,
      "step": 10767
    },
    {
      "epoch": 1.5091800981079189,
      "grad_norm": 0.5392279624938965,
      "learning_rate": 0.0001527194451088256,
      "loss": 0.1296,
      "step": 10768
    },
    {
      "epoch": 1.5093202522775053,
      "grad_norm": 0.40058988332748413,
      "learning_rate": 0.00015270509447500597,
      "loss": 0.0742,
      "step": 10769
    },
    {
      "epoch": 1.5094604064470918,
      "grad_norm": 0.36171796917915344,
      "learning_rate": 0.0001526907438411863,
      "loss": 0.0234,
      "step": 10770
    },
    {
      "epoch": 1.5096005606166782,
      "grad_norm": 0.2082325667142868,
      "learning_rate": 0.00015267639320736662,
      "loss": 0.0382,
      "step": 10771
    },
    {
      "epoch": 1.509740714786265,
      "grad_norm": 0.18789510428905487,
      "learning_rate": 0.000152662042573547,
      "loss": 0.0295,
      "step": 10772
    },
    {
      "epoch": 1.5098808689558514,
      "grad_norm": 0.5805644392967224,
      "learning_rate": 0.00015264769193972733,
      "loss": 0.0232,
      "step": 10773
    },
    {
      "epoch": 1.510021023125438,
      "grad_norm": 0.4467480182647705,
      "learning_rate": 0.00015263334130590766,
      "loss": 0.0176,
      "step": 10774
    },
    {
      "epoch": 1.5101611772950245,
      "grad_norm": 0.3863131105899811,
      "learning_rate": 0.00015261899067208802,
      "loss": 0.0422,
      "step": 10775
    },
    {
      "epoch": 1.510301331464611,
      "grad_norm": 0.0412088967859745,
      "learning_rate": 0.00015260464003826835,
      "loss": 0.0041,
      "step": 10776
    },
    {
      "epoch": 1.5104414856341977,
      "grad_norm": 0.4299469292163849,
      "learning_rate": 0.00015259028940444868,
      "loss": 0.1029,
      "step": 10777
    },
    {
      "epoch": 1.5105816398037841,
      "grad_norm": 0.6618829965591431,
      "learning_rate": 0.00015257593877062903,
      "loss": 0.0983,
      "step": 10778
    },
    {
      "epoch": 1.5107217939733708,
      "grad_norm": 0.6282081007957458,
      "learning_rate": 0.00015256158813680936,
      "loss": 0.0961,
      "step": 10779
    },
    {
      "epoch": 1.5108619481429573,
      "grad_norm": 0.028919318690896034,
      "learning_rate": 0.0001525472375029897,
      "loss": 0.0023,
      "step": 10780
    },
    {
      "epoch": 1.5110021023125437,
      "grad_norm": 0.24769586324691772,
      "learning_rate": 0.00015253288686917004,
      "loss": 0.0297,
      "step": 10781
    },
    {
      "epoch": 1.5111422564821302,
      "grad_norm": 0.1989944875240326,
      "learning_rate": 0.00015251853623535037,
      "loss": 0.0208,
      "step": 10782
    },
    {
      "epoch": 1.5112824106517169,
      "grad_norm": 0.196117103099823,
      "learning_rate": 0.0001525041856015307,
      "loss": 0.0237,
      "step": 10783
    },
    {
      "epoch": 1.5114225648213035,
      "grad_norm": 1.185658574104309,
      "learning_rate": 0.00015248983496771106,
      "loss": 0.0879,
      "step": 10784
    },
    {
      "epoch": 1.51156271899089,
      "grad_norm": 2.8151087760925293,
      "learning_rate": 0.0001524754843338914,
      "loss": 0.4963,
      "step": 10785
    },
    {
      "epoch": 1.5117028731604765,
      "grad_norm": 0.08544868230819702,
      "learning_rate": 0.00015246113370007174,
      "loss": 0.0125,
      "step": 10786
    },
    {
      "epoch": 1.511843027330063,
      "grad_norm": 0.25065869092941284,
      "learning_rate": 0.00015244678306625207,
      "loss": 0.0456,
      "step": 10787
    },
    {
      "epoch": 1.5119831814996496,
      "grad_norm": 0.32535651326179504,
      "learning_rate": 0.00015243243243243243,
      "loss": 0.0737,
      "step": 10788
    },
    {
      "epoch": 1.5121233356692363,
      "grad_norm": 0.11083528399467468,
      "learning_rate": 0.00015241808179861275,
      "loss": 0.0047,
      "step": 10789
    },
    {
      "epoch": 1.5122634898388227,
      "grad_norm": 0.5659446716308594,
      "learning_rate": 0.00015240373116479308,
      "loss": 0.1077,
      "step": 10790
    },
    {
      "epoch": 1.5124036440084092,
      "grad_norm": 0.2669200897216797,
      "learning_rate": 0.00015238938053097344,
      "loss": 0.0359,
      "step": 10791
    },
    {
      "epoch": 1.5125437981779957,
      "grad_norm": 0.3054511249065399,
      "learning_rate": 0.00015237502989715377,
      "loss": 0.0374,
      "step": 10792
    },
    {
      "epoch": 1.5126839523475824,
      "grad_norm": 0.1364801675081253,
      "learning_rate": 0.0001523606792633341,
      "loss": 0.0271,
      "step": 10793
    },
    {
      "epoch": 1.512824106517169,
      "grad_norm": 0.2876091003417969,
      "learning_rate": 0.00015234632862951448,
      "loss": 0.0248,
      "step": 10794
    },
    {
      "epoch": 1.5129642606867555,
      "grad_norm": 1.205754280090332,
      "learning_rate": 0.0001523319779956948,
      "loss": 0.054,
      "step": 10795
    },
    {
      "epoch": 1.513104414856342,
      "grad_norm": 0.3665226101875305,
      "learning_rate": 0.00015231762736187514,
      "loss": 0.069,
      "step": 10796
    },
    {
      "epoch": 1.5132445690259284,
      "grad_norm": 0.4154714047908783,
      "learning_rate": 0.0001523032767280555,
      "loss": 0.0285,
      "step": 10797
    },
    {
      "epoch": 1.513384723195515,
      "grad_norm": 0.9619696140289307,
      "learning_rate": 0.00015228892609423582,
      "loss": 0.0872,
      "step": 10798
    },
    {
      "epoch": 1.5135248773651018,
      "grad_norm": 0.28615206480026245,
      "learning_rate": 0.00015227457546041615,
      "loss": 0.0306,
      "step": 10799
    },
    {
      "epoch": 1.5136650315346882,
      "grad_norm": 0.33452072739601135,
      "learning_rate": 0.0001522602248265965,
      "loss": 0.0382,
      "step": 10800
    },
    {
      "epoch": 1.5138051857042747,
      "grad_norm": 0.26759278774261475,
      "learning_rate": 0.00015224587419277683,
      "loss": 0.0381,
      "step": 10801
    },
    {
      "epoch": 1.5139453398738612,
      "grad_norm": 0.2677653729915619,
      "learning_rate": 0.00015223152355895716,
      "loss": 0.0263,
      "step": 10802
    },
    {
      "epoch": 1.5140854940434478,
      "grad_norm": 0.6581940650939941,
      "learning_rate": 0.0001522171729251375,
      "loss": 0.0385,
      "step": 10803
    },
    {
      "epoch": 1.5142256482130343,
      "grad_norm": 0.31521183252334595,
      "learning_rate": 0.00015220282229131787,
      "loss": 0.0744,
      "step": 10804
    },
    {
      "epoch": 1.514365802382621,
      "grad_norm": 0.21869181096553802,
      "learning_rate": 0.0001521884716574982,
      "loss": 0.0468,
      "step": 10805
    },
    {
      "epoch": 1.5145059565522074,
      "grad_norm": 0.5995174050331116,
      "learning_rate": 0.00015217412102367853,
      "loss": 0.0663,
      "step": 10806
    },
    {
      "epoch": 1.514646110721794,
      "grad_norm": 0.3601396977901459,
      "learning_rate": 0.00015215977038985889,
      "loss": 0.0319,
      "step": 10807
    },
    {
      "epoch": 1.5147862648913806,
      "grad_norm": 0.368913471698761,
      "learning_rate": 0.00015214541975603921,
      "loss": 0.0474,
      "step": 10808
    },
    {
      "epoch": 1.514926419060967,
      "grad_norm": 0.44042468070983887,
      "learning_rate": 0.00015213106912221954,
      "loss": 0.0466,
      "step": 10809
    },
    {
      "epoch": 1.5150665732305537,
      "grad_norm": 0.24894589185714722,
      "learning_rate": 0.0001521167184883999,
      "loss": 0.0533,
      "step": 10810
    },
    {
      "epoch": 1.5152067274001402,
      "grad_norm": 0.3458297550678253,
      "learning_rate": 0.00015210236785458023,
      "loss": 0.0454,
      "step": 10811
    },
    {
      "epoch": 1.5153468815697266,
      "grad_norm": 0.5747541189193726,
      "learning_rate": 0.00015208801722076056,
      "loss": 0.0553,
      "step": 10812
    },
    {
      "epoch": 1.515487035739313,
      "grad_norm": 0.6571207046508789,
      "learning_rate": 0.0001520736665869409,
      "loss": 0.1397,
      "step": 10813
    },
    {
      "epoch": 1.5156271899088998,
      "grad_norm": 0.16182446479797363,
      "learning_rate": 0.00015205931595312124,
      "loss": 0.0122,
      "step": 10814
    },
    {
      "epoch": 1.5157673440784865,
      "grad_norm": 0.7133028507232666,
      "learning_rate": 0.0001520449653193016,
      "loss": 0.0396,
      "step": 10815
    },
    {
      "epoch": 1.515907498248073,
      "grad_norm": 0.468172162771225,
      "learning_rate": 0.00015203061468548195,
      "loss": 0.0347,
      "step": 10816
    },
    {
      "epoch": 1.5160476524176594,
      "grad_norm": 0.19602705538272858,
      "learning_rate": 0.00015201626405166228,
      "loss": 0.0264,
      "step": 10817
    },
    {
      "epoch": 1.5161878065872458,
      "grad_norm": 0.25425300002098083,
      "learning_rate": 0.0001520019134178426,
      "loss": 0.0159,
      "step": 10818
    },
    {
      "epoch": 1.5163279607568325,
      "grad_norm": 0.30720216035842896,
      "learning_rate": 0.00015198756278402294,
      "loss": 0.0195,
      "step": 10819
    },
    {
      "epoch": 1.5164681149264192,
      "grad_norm": 0.47727319598197937,
      "learning_rate": 0.0001519732121502033,
      "loss": 0.0581,
      "step": 10820
    },
    {
      "epoch": 1.5166082690960057,
      "grad_norm": 0.2693765163421631,
      "learning_rate": 0.00015195886151638362,
      "loss": 0.0435,
      "step": 10821
    },
    {
      "epoch": 1.5167484232655921,
      "grad_norm": 0.20326267182826996,
      "learning_rate": 0.00015194451088256395,
      "loss": 0.018,
      "step": 10822
    },
    {
      "epoch": 1.5168885774351786,
      "grad_norm": 0.13905519247055054,
      "learning_rate": 0.0001519301602487443,
      "loss": 0.025,
      "step": 10823
    },
    {
      "epoch": 1.5170287316047653,
      "grad_norm": 0.32441219687461853,
      "learning_rate": 0.00015191580961492463,
      "loss": 0.062,
      "step": 10824
    },
    {
      "epoch": 1.517168885774352,
      "grad_norm": 0.25833946466445923,
      "learning_rate": 0.00015190145898110496,
      "loss": 0.0551,
      "step": 10825
    },
    {
      "epoch": 1.5173090399439384,
      "grad_norm": 0.38717103004455566,
      "learning_rate": 0.00015188710834728534,
      "loss": 0.0379,
      "step": 10826
    },
    {
      "epoch": 1.5174491941135249,
      "grad_norm": 0.17563162744045258,
      "learning_rate": 0.00015187275771346567,
      "loss": 0.0163,
      "step": 10827
    },
    {
      "epoch": 1.5175893482831113,
      "grad_norm": 1.8284509181976318,
      "learning_rate": 0.000151858407079646,
      "loss": 0.0683,
      "step": 10828
    },
    {
      "epoch": 1.517729502452698,
      "grad_norm": 0.7358977794647217,
      "learning_rate": 0.00015184405644582636,
      "loss": 0.0851,
      "step": 10829
    },
    {
      "epoch": 1.5178696566222845,
      "grad_norm": 0.20001107454299927,
      "learning_rate": 0.00015182970581200669,
      "loss": 0.0427,
      "step": 10830
    },
    {
      "epoch": 1.5180098107918711,
      "grad_norm": 0.22446341812610626,
      "learning_rate": 0.00015181535517818701,
      "loss": 0.0157,
      "step": 10831
    },
    {
      "epoch": 1.5181499649614576,
      "grad_norm": 0.9096955060958862,
      "learning_rate": 0.00015180100454436737,
      "loss": 0.2369,
      "step": 10832
    },
    {
      "epoch": 1.518290119131044,
      "grad_norm": 1.0665274858474731,
      "learning_rate": 0.0001517866539105477,
      "loss": 0.1144,
      "step": 10833
    },
    {
      "epoch": 1.5184302733006307,
      "grad_norm": 2.253868341445923,
      "learning_rate": 0.00015177230327672803,
      "loss": 0.308,
      "step": 10834
    },
    {
      "epoch": 1.5185704274702172,
      "grad_norm": 1.3804970979690552,
      "learning_rate": 0.0001517579526429084,
      "loss": 0.3864,
      "step": 10835
    },
    {
      "epoch": 1.5187105816398039,
      "grad_norm": 0.4434795081615448,
      "learning_rate": 0.00015174360200908874,
      "loss": 0.0527,
      "step": 10836
    },
    {
      "epoch": 1.5188507358093903,
      "grad_norm": 0.27656427025794983,
      "learning_rate": 0.00015172925137526907,
      "loss": 0.0571,
      "step": 10837
    },
    {
      "epoch": 1.5189908899789768,
      "grad_norm": 0.2262948751449585,
      "learning_rate": 0.0001517149007414494,
      "loss": 0.0714,
      "step": 10838
    },
    {
      "epoch": 1.5191310441485633,
      "grad_norm": 0.15831328928470612,
      "learning_rate": 0.00015170055010762975,
      "loss": 0.0553,
      "step": 10839
    },
    {
      "epoch": 1.51927119831815,
      "grad_norm": 0.3244056701660156,
      "learning_rate": 0.00015168619947381008,
      "loss": 0.1105,
      "step": 10840
    },
    {
      "epoch": 1.5194113524877366,
      "grad_norm": 0.2566559910774231,
      "learning_rate": 0.0001516718488399904,
      "loss": 0.0942,
      "step": 10841
    },
    {
      "epoch": 1.519551506657323,
      "grad_norm": 0.29002147912979126,
      "learning_rate": 0.00015165749820617076,
      "loss": 0.0483,
      "step": 10842
    },
    {
      "epoch": 1.5196916608269095,
      "grad_norm": 0.7608566284179688,
      "learning_rate": 0.0001516431475723511,
      "loss": 0.0785,
      "step": 10843
    },
    {
      "epoch": 1.519831814996496,
      "grad_norm": 0.24261343479156494,
      "learning_rate": 0.00015162879693853142,
      "loss": 0.0563,
      "step": 10844
    },
    {
      "epoch": 1.5199719691660827,
      "grad_norm": 0.11511098593473434,
      "learning_rate": 0.00015161444630471178,
      "loss": 0.0286,
      "step": 10845
    },
    {
      "epoch": 1.5201121233356694,
      "grad_norm": 0.30866873264312744,
      "learning_rate": 0.0001516000956708921,
      "loss": 0.051,
      "step": 10846
    },
    {
      "epoch": 1.5202522775052558,
      "grad_norm": 0.23170363903045654,
      "learning_rate": 0.00015158574503707246,
      "loss": 0.0397,
      "step": 10847
    },
    {
      "epoch": 1.5203924316748423,
      "grad_norm": 0.21092519164085388,
      "learning_rate": 0.00015157139440325282,
      "loss": 0.0439,
      "step": 10848
    },
    {
      "epoch": 1.5205325858444287,
      "grad_norm": 0.14955882728099823,
      "learning_rate": 0.00015155704376943315,
      "loss": 0.0403,
      "step": 10849
    },
    {
      "epoch": 1.5206727400140154,
      "grad_norm": 0.3115289807319641,
      "learning_rate": 0.00015154269313561347,
      "loss": 0.0484,
      "step": 10850
    },
    {
      "epoch": 1.520812894183602,
      "grad_norm": 0.46107035875320435,
      "learning_rate": 0.00015152834250179383,
      "loss": 0.0468,
      "step": 10851
    },
    {
      "epoch": 1.5209530483531886,
      "grad_norm": 0.13584738969802856,
      "learning_rate": 0.00015151399186797416,
      "loss": 0.0282,
      "step": 10852
    },
    {
      "epoch": 1.521093202522775,
      "grad_norm": 0.3119492828845978,
      "learning_rate": 0.0001514996412341545,
      "loss": 0.0657,
      "step": 10853
    },
    {
      "epoch": 1.5212333566923615,
      "grad_norm": 0.35627949237823486,
      "learning_rate": 0.00015148529060033482,
      "loss": 0.0848,
      "step": 10854
    },
    {
      "epoch": 1.5213735108619482,
      "grad_norm": 0.10402132570743561,
      "learning_rate": 0.00015147093996651517,
      "loss": 0.0234,
      "step": 10855
    },
    {
      "epoch": 1.5215136650315348,
      "grad_norm": 0.42414355278015137,
      "learning_rate": 0.0001514565893326955,
      "loss": 0.042,
      "step": 10856
    },
    {
      "epoch": 1.5216538192011213,
      "grad_norm": 0.3917308747768402,
      "learning_rate": 0.00015144223869887583,
      "loss": 0.1049,
      "step": 10857
    },
    {
      "epoch": 1.5217939733707078,
      "grad_norm": 0.2178390771150589,
      "learning_rate": 0.0001514278880650562,
      "loss": 0.016,
      "step": 10858
    },
    {
      "epoch": 1.5219341275402942,
      "grad_norm": 0.1455674022436142,
      "learning_rate": 0.00015141353743123654,
      "loss": 0.0617,
      "step": 10859
    },
    {
      "epoch": 1.522074281709881,
      "grad_norm": 0.2771393358707428,
      "learning_rate": 0.00015139918679741687,
      "loss": 0.0659,
      "step": 10860
    },
    {
      "epoch": 1.5222144358794674,
      "grad_norm": 0.1687132567167282,
      "learning_rate": 0.00015138483616359722,
      "loss": 0.0249,
      "step": 10861
    },
    {
      "epoch": 1.522354590049054,
      "grad_norm": 0.5420262217521667,
      "learning_rate": 0.00015137048552977755,
      "loss": 0.0626,
      "step": 10862
    },
    {
      "epoch": 1.5224947442186405,
      "grad_norm": 0.20648440718650818,
      "learning_rate": 0.00015135613489595788,
      "loss": 0.0362,
      "step": 10863
    },
    {
      "epoch": 1.522634898388227,
      "grad_norm": 0.5218943953514099,
      "learning_rate": 0.00015134178426213824,
      "loss": 0.1662,
      "step": 10864
    },
    {
      "epoch": 1.5227750525578136,
      "grad_norm": 0.12385623902082443,
      "learning_rate": 0.00015132743362831857,
      "loss": 0.014,
      "step": 10865
    },
    {
      "epoch": 1.5229152067274,
      "grad_norm": 0.9721292853355408,
      "learning_rate": 0.0001513130829944989,
      "loss": 0.1214,
      "step": 10866
    },
    {
      "epoch": 1.5230553608969868,
      "grad_norm": 0.4468240439891815,
      "learning_rate": 0.00015129873236067928,
      "loss": 0.1219,
      "step": 10867
    },
    {
      "epoch": 1.5231955150665732,
      "grad_norm": 0.7129539847373962,
      "learning_rate": 0.0001512843817268596,
      "loss": 0.0506,
      "step": 10868
    },
    {
      "epoch": 1.5233356692361597,
      "grad_norm": 0.256916344165802,
      "learning_rate": 0.00015127003109303993,
      "loss": 0.0275,
      "step": 10869
    },
    {
      "epoch": 1.5234758234057462,
      "grad_norm": 0.434985876083374,
      "learning_rate": 0.0001512556804592203,
      "loss": 0.0929,
      "step": 10870
    },
    {
      "epoch": 1.5236159775753328,
      "grad_norm": 0.40868932008743286,
      "learning_rate": 0.00015124132982540062,
      "loss": 0.0319,
      "step": 10871
    },
    {
      "epoch": 1.5237561317449195,
      "grad_norm": 0.3004898130893707,
      "learning_rate": 0.00015122697919158095,
      "loss": 0.0213,
      "step": 10872
    },
    {
      "epoch": 1.523896285914506,
      "grad_norm": 0.5182176828384399,
      "learning_rate": 0.00015121262855776127,
      "loss": 0.0371,
      "step": 10873
    },
    {
      "epoch": 1.5240364400840924,
      "grad_norm": 0.10452108085155487,
      "learning_rate": 0.00015119827792394163,
      "loss": 0.0145,
      "step": 10874
    },
    {
      "epoch": 1.524176594253679,
      "grad_norm": 0.2919504642486572,
      "learning_rate": 0.00015118392729012196,
      "loss": 0.0318,
      "step": 10875
    },
    {
      "epoch": 1.5243167484232656,
      "grad_norm": 0.2737484276294708,
      "learning_rate": 0.0001511695766563023,
      "loss": 0.0642,
      "step": 10876
    },
    {
      "epoch": 1.5244569025928523,
      "grad_norm": 0.8553444147109985,
      "learning_rate": 0.00015115522602248264,
      "loss": 0.133,
      "step": 10877
    },
    {
      "epoch": 1.5245970567624387,
      "grad_norm": 0.3205641806125641,
      "learning_rate": 0.00015114087538866297,
      "loss": 0.0666,
      "step": 10878
    },
    {
      "epoch": 1.5247372109320252,
      "grad_norm": 0.5432232618331909,
      "learning_rate": 0.00015112652475484333,
      "loss": 0.0667,
      "step": 10879
    },
    {
      "epoch": 1.5248773651016116,
      "grad_norm": 0.12649619579315186,
      "learning_rate": 0.00015111217412102368,
      "loss": 0.0124,
      "step": 10880
    },
    {
      "epoch": 1.5250175192711983,
      "grad_norm": 0.1093301922082901,
      "learning_rate": 0.000151097823487204,
      "loss": 0.0086,
      "step": 10881
    },
    {
      "epoch": 1.525157673440785,
      "grad_norm": 0.6656155586242676,
      "learning_rate": 0.00015108347285338434,
      "loss": 0.1079,
      "step": 10882
    },
    {
      "epoch": 1.5252978276103715,
      "grad_norm": 0.6874668598175049,
      "learning_rate": 0.0001510691222195647,
      "loss": 0.1198,
      "step": 10883
    },
    {
      "epoch": 1.525437981779958,
      "grad_norm": 1.210294485092163,
      "learning_rate": 0.00015105477158574502,
      "loss": 0.1503,
      "step": 10884
    },
    {
      "epoch": 1.5255781359495444,
      "grad_norm": 3.4931607246398926,
      "learning_rate": 0.00015104042095192535,
      "loss": 0.7081,
      "step": 10885
    },
    {
      "epoch": 1.525718290119131,
      "grad_norm": 0.17720280587673187,
      "learning_rate": 0.0001510260703181057,
      "loss": 0.032,
      "step": 10886
    },
    {
      "epoch": 1.5258584442887178,
      "grad_norm": 0.2815784811973572,
      "learning_rate": 0.00015101171968428604,
      "loss": 0.0438,
      "step": 10887
    },
    {
      "epoch": 1.5259985984583042,
      "grad_norm": 0.2656289339065552,
      "learning_rate": 0.00015099736905046637,
      "loss": 0.0493,
      "step": 10888
    },
    {
      "epoch": 1.5261387526278907,
      "grad_norm": 0.21136623620986938,
      "learning_rate": 0.0001509830184166467,
      "loss": 0.0575,
      "step": 10889
    },
    {
      "epoch": 1.5262789067974771,
      "grad_norm": 0.14760293066501617,
      "learning_rate": 0.00015096866778282708,
      "loss": 0.0136,
      "step": 10890
    },
    {
      "epoch": 1.5264190609670638,
      "grad_norm": 0.19249001145362854,
      "learning_rate": 0.0001509543171490074,
      "loss": 0.0212,
      "step": 10891
    },
    {
      "epoch": 1.5265592151366503,
      "grad_norm": 0.29600653052330017,
      "learning_rate": 0.00015093996651518773,
      "loss": 0.0235,
      "step": 10892
    },
    {
      "epoch": 1.526699369306237,
      "grad_norm": 0.17708593606948853,
      "learning_rate": 0.0001509256158813681,
      "loss": 0.0265,
      "step": 10893
    },
    {
      "epoch": 1.5268395234758234,
      "grad_norm": 1.0646584033966064,
      "learning_rate": 0.00015091126524754842,
      "loss": 0.0537,
      "step": 10894
    },
    {
      "epoch": 1.5269796776454099,
      "grad_norm": 0.5252156257629395,
      "learning_rate": 0.00015089691461372875,
      "loss": 0.062,
      "step": 10895
    },
    {
      "epoch": 1.5271198318149963,
      "grad_norm": 0.4272652864456177,
      "learning_rate": 0.0001508825639799091,
      "loss": 0.0485,
      "step": 10896
    },
    {
      "epoch": 1.527259985984583,
      "grad_norm": 0.3775762915611267,
      "learning_rate": 0.00015086821334608943,
      "loss": 0.0617,
      "step": 10897
    },
    {
      "epoch": 1.5274001401541697,
      "grad_norm": 0.9619895219802856,
      "learning_rate": 0.00015085386271226976,
      "loss": 0.0674,
      "step": 10898
    },
    {
      "epoch": 1.5275402943237562,
      "grad_norm": 0.18873095512390137,
      "learning_rate": 0.00015083951207845014,
      "loss": 0.0202,
      "step": 10899
    },
    {
      "epoch": 1.5276804484933426,
      "grad_norm": 0.4533403813838959,
      "learning_rate": 0.00015082516144463047,
      "loss": 0.1591,
      "step": 10900
    },
    {
      "epoch": 1.527820602662929,
      "grad_norm": 0.47293156385421753,
      "learning_rate": 0.0001508108108108108,
      "loss": 0.0539,
      "step": 10901
    },
    {
      "epoch": 1.5279607568325158,
      "grad_norm": 0.5269839763641357,
      "learning_rate": 0.00015079646017699116,
      "loss": 0.108,
      "step": 10902
    },
    {
      "epoch": 1.5281009110021024,
      "grad_norm": 0.18391455709934235,
      "learning_rate": 0.00015078210954317148,
      "loss": 0.03,
      "step": 10903
    },
    {
      "epoch": 1.528241065171689,
      "grad_norm": 1.3996319770812988,
      "learning_rate": 0.0001507677589093518,
      "loss": 0.1065,
      "step": 10904
    },
    {
      "epoch": 1.5283812193412754,
      "grad_norm": 0.37039241194725037,
      "learning_rate": 0.00015075340827553217,
      "loss": 0.0438,
      "step": 10905
    },
    {
      "epoch": 1.5285213735108618,
      "grad_norm": 0.20693868398666382,
      "learning_rate": 0.0001507390576417125,
      "loss": 0.0159,
      "step": 10906
    },
    {
      "epoch": 1.5286615276804485,
      "grad_norm": 0.16598059237003326,
      "learning_rate": 0.00015072470700789283,
      "loss": 0.0344,
      "step": 10907
    },
    {
      "epoch": 1.5288016818500352,
      "grad_norm": 0.9544166326522827,
      "learning_rate": 0.00015071035637407315,
      "loss": 0.1074,
      "step": 10908
    },
    {
      "epoch": 1.5289418360196216,
      "grad_norm": 0.3196631669998169,
      "learning_rate": 0.0001506960057402535,
      "loss": 0.0422,
      "step": 10909
    },
    {
      "epoch": 1.529081990189208,
      "grad_norm": 0.34100693464279175,
      "learning_rate": 0.00015068165510643387,
      "loss": 0.0142,
      "step": 10910
    },
    {
      "epoch": 1.5292221443587946,
      "grad_norm": 0.6050891280174255,
      "learning_rate": 0.0001506673044726142,
      "loss": 0.0949,
      "step": 10911
    },
    {
      "epoch": 1.5293622985283812,
      "grad_norm": 0.19235040247440338,
      "learning_rate": 0.00015065295383879455,
      "loss": 0.0389,
      "step": 10912
    },
    {
      "epoch": 1.529502452697968,
      "grad_norm": 0.23279502987861633,
      "learning_rate": 0.00015063860320497488,
      "loss": 0.0352,
      "step": 10913
    },
    {
      "epoch": 1.5296426068675544,
      "grad_norm": 0.18740299344062805,
      "learning_rate": 0.0001506242525711552,
      "loss": 0.0273,
      "step": 10914
    },
    {
      "epoch": 1.5297827610371408,
      "grad_norm": 0.15140323340892792,
      "learning_rate": 0.00015060990193733556,
      "loss": 0.0201,
      "step": 10915
    },
    {
      "epoch": 1.5299229152067273,
      "grad_norm": 0.21403227746486664,
      "learning_rate": 0.0001505955513035159,
      "loss": 0.0355,
      "step": 10916
    },
    {
      "epoch": 1.530063069376314,
      "grad_norm": 0.23515480756759644,
      "learning_rate": 0.00015058120066969622,
      "loss": 0.0498,
      "step": 10917
    },
    {
      "epoch": 1.5302032235459004,
      "grad_norm": 0.24206171929836273,
      "learning_rate": 0.00015056685003587658,
      "loss": 0.0551,
      "step": 10918
    },
    {
      "epoch": 1.5303433777154871,
      "grad_norm": 0.1398821622133255,
      "learning_rate": 0.0001505524994020569,
      "loss": 0.0338,
      "step": 10919
    },
    {
      "epoch": 1.5304835318850736,
      "grad_norm": 0.4756773114204407,
      "learning_rate": 0.00015053814876823723,
      "loss": 0.072,
      "step": 10920
    },
    {
      "epoch": 1.53062368605466,
      "grad_norm": 0.08770030736923218,
      "learning_rate": 0.00015052379813441761,
      "loss": 0.0072,
      "step": 10921
    },
    {
      "epoch": 1.5307638402242467,
      "grad_norm": 0.4261038899421692,
      "learning_rate": 0.00015050944750059794,
      "loss": 0.0904,
      "step": 10922
    },
    {
      "epoch": 1.5309039943938332,
      "grad_norm": 0.48148220777511597,
      "learning_rate": 0.00015049509686677827,
      "loss": 0.0984,
      "step": 10923
    },
    {
      "epoch": 1.5310441485634199,
      "grad_norm": 0.5076977014541626,
      "learning_rate": 0.0001504807462329586,
      "loss": 0.0799,
      "step": 10924
    },
    {
      "epoch": 1.5311843027330063,
      "grad_norm": 0.7440010905265808,
      "learning_rate": 0.00015046639559913896,
      "loss": 0.0977,
      "step": 10925
    },
    {
      "epoch": 1.5313244569025928,
      "grad_norm": 1.4425071477890015,
      "learning_rate": 0.00015045204496531928,
      "loss": 0.0555,
      "step": 10926
    },
    {
      "epoch": 1.5314646110721792,
      "grad_norm": 0.30451399087905884,
      "learning_rate": 0.0001504376943314996,
      "loss": 0.0227,
      "step": 10927
    },
    {
      "epoch": 1.531604765241766,
      "grad_norm": 0.5178651809692383,
      "learning_rate": 0.00015042334369767997,
      "loss": 0.0256,
      "step": 10928
    },
    {
      "epoch": 1.5317449194113526,
      "grad_norm": 0.26351237297058105,
      "learning_rate": 0.0001504089930638603,
      "loss": 0.015,
      "step": 10929
    },
    {
      "epoch": 1.531885073580939,
      "grad_norm": 0.7813514471054077,
      "learning_rate": 0.00015039464243004063,
      "loss": 0.0981,
      "step": 10930
    },
    {
      "epoch": 1.5320252277505255,
      "grad_norm": 0.5308046936988831,
      "learning_rate": 0.000150380291796221,
      "loss": 0.1238,
      "step": 10931
    },
    {
      "epoch": 1.532165381920112,
      "grad_norm": 0.8562813997268677,
      "learning_rate": 0.00015036594116240134,
      "loss": 0.0528,
      "step": 10932
    },
    {
      "epoch": 1.5323055360896987,
      "grad_norm": 0.872998833656311,
      "learning_rate": 0.00015035159052858167,
      "loss": 0.0439,
      "step": 10933
    },
    {
      "epoch": 1.5324456902592853,
      "grad_norm": 1.37801992893219,
      "learning_rate": 0.00015033723989476202,
      "loss": 0.1312,
      "step": 10934
    },
    {
      "epoch": 1.5325858444288718,
      "grad_norm": 0.14941856265068054,
      "learning_rate": 0.00015032288926094235,
      "loss": 0.0056,
      "step": 10935
    },
    {
      "epoch": 1.5327259985984583,
      "grad_norm": 0.20218883454799652,
      "learning_rate": 0.00015030853862712268,
      "loss": 0.0459,
      "step": 10936
    },
    {
      "epoch": 1.5328661527680447,
      "grad_norm": 0.2454679012298584,
      "learning_rate": 0.00015029418799330303,
      "loss": 0.0426,
      "step": 10937
    },
    {
      "epoch": 1.5330063069376314,
      "grad_norm": 0.14818303287029266,
      "learning_rate": 0.00015027983735948336,
      "loss": 0.0232,
      "step": 10938
    },
    {
      "epoch": 1.533146461107218,
      "grad_norm": 0.14970046281814575,
      "learning_rate": 0.0001502654867256637,
      "loss": 0.0354,
      "step": 10939
    },
    {
      "epoch": 1.5332866152768045,
      "grad_norm": 0.04809822142124176,
      "learning_rate": 0.00015025113609184405,
      "loss": 0.0047,
      "step": 10940
    },
    {
      "epoch": 1.533426769446391,
      "grad_norm": 0.20709531009197235,
      "learning_rate": 0.00015023678545802438,
      "loss": 0.0189,
      "step": 10941
    },
    {
      "epoch": 1.5335669236159775,
      "grad_norm": 0.29231974482536316,
      "learning_rate": 0.00015022243482420473,
      "loss": 0.0268,
      "step": 10942
    },
    {
      "epoch": 1.5337070777855641,
      "grad_norm": 0.12852610647678375,
      "learning_rate": 0.00015020808419038506,
      "loss": 0.0411,
      "step": 10943
    },
    {
      "epoch": 1.5338472319551508,
      "grad_norm": 0.18194787204265594,
      "learning_rate": 0.00015019373355656542,
      "loss": 0.0241,
      "step": 10944
    },
    {
      "epoch": 1.5339873861247373,
      "grad_norm": 0.42633524537086487,
      "learning_rate": 0.00015017938292274574,
      "loss": 0.0329,
      "step": 10945
    },
    {
      "epoch": 1.5341275402943237,
      "grad_norm": 0.2647063434123993,
      "learning_rate": 0.00015016503228892607,
      "loss": 0.0262,
      "step": 10946
    },
    {
      "epoch": 1.5342676944639102,
      "grad_norm": 0.5387910604476929,
      "learning_rate": 0.00015015068165510643,
      "loss": 0.0938,
      "step": 10947
    },
    {
      "epoch": 1.5344078486334969,
      "grad_norm": 0.06391759216785431,
      "learning_rate": 0.00015013633102128676,
      "loss": 0.0049,
      "step": 10948
    },
    {
      "epoch": 1.5345480028030833,
      "grad_norm": 0.5705273747444153,
      "learning_rate": 0.00015012198038746709,
      "loss": 0.0952,
      "step": 10949
    },
    {
      "epoch": 1.53468815697267,
      "grad_norm": 0.20019106566905975,
      "learning_rate": 0.00015010762975364744,
      "loss": 0.0556,
      "step": 10950
    },
    {
      "epoch": 1.5348283111422565,
      "grad_norm": 0.23995529115200043,
      "learning_rate": 0.00015009327911982777,
      "loss": 0.0201,
      "step": 10951
    },
    {
      "epoch": 1.534968465311843,
      "grad_norm": 0.4369611144065857,
      "learning_rate": 0.0001500789284860081,
      "loss": 0.032,
      "step": 10952
    },
    {
      "epoch": 1.5351086194814296,
      "grad_norm": 0.3101497292518616,
      "learning_rate": 0.00015006457785218848,
      "loss": 0.0276,
      "step": 10953
    },
    {
      "epoch": 1.535248773651016,
      "grad_norm": 0.31244391202926636,
      "learning_rate": 0.0001500502272183688,
      "loss": 0.0565,
      "step": 10954
    },
    {
      "epoch": 1.5353889278206028,
      "grad_norm": 0.36449599266052246,
      "learning_rate": 0.00015003587658454914,
      "loss": 0.0419,
      "step": 10955
    },
    {
      "epoch": 1.5355290819901892,
      "grad_norm": 0.45624756813049316,
      "learning_rate": 0.0001500215259507295,
      "loss": 0.0983,
      "step": 10956
    },
    {
      "epoch": 1.5356692361597757,
      "grad_norm": 0.25939854979515076,
      "learning_rate": 0.00015000717531690982,
      "loss": 0.0284,
      "step": 10957
    },
    {
      "epoch": 1.5358093903293621,
      "grad_norm": 0.403261661529541,
      "learning_rate": 0.00014999282468309015,
      "loss": 0.0428,
      "step": 10958
    },
    {
      "epoch": 1.5359495444989488,
      "grad_norm": 0.23813696205615997,
      "learning_rate": 0.0001499784740492705,
      "loss": 0.0249,
      "step": 10959
    },
    {
      "epoch": 1.5360896986685355,
      "grad_norm": 0.4732477366924286,
      "learning_rate": 0.00014996412341545084,
      "loss": 0.0726,
      "step": 10960
    },
    {
      "epoch": 1.536229852838122,
      "grad_norm": 0.3605789542198181,
      "learning_rate": 0.00014994977278163116,
      "loss": 0.0336,
      "step": 10961
    },
    {
      "epoch": 1.5363700070077084,
      "grad_norm": 0.2642771899700165,
      "learning_rate": 0.00014993542214781152,
      "loss": 0.0246,
      "step": 10962
    },
    {
      "epoch": 1.5365101611772949,
      "grad_norm": 0.29140397906303406,
      "learning_rate": 0.00014992107151399185,
      "loss": 0.0484,
      "step": 10963
    },
    {
      "epoch": 1.5366503153468816,
      "grad_norm": 0.5289062261581421,
      "learning_rate": 0.0001499067208801722,
      "loss": 0.0566,
      "step": 10964
    },
    {
      "epoch": 1.5367904695164682,
      "grad_norm": 0.3317778706550598,
      "learning_rate": 0.00014989237024635253,
      "loss": 0.0438,
      "step": 10965
    },
    {
      "epoch": 1.5369306236860547,
      "grad_norm": 0.47614800930023193,
      "learning_rate": 0.00014987801961253286,
      "loss": 0.0643,
      "step": 10966
    },
    {
      "epoch": 1.5370707778556412,
      "grad_norm": 1.2505192756652832,
      "learning_rate": 0.00014986366897871322,
      "loss": 0.0503,
      "step": 10967
    },
    {
      "epoch": 1.5372109320252276,
      "grad_norm": 0.45235052704811096,
      "learning_rate": 0.00014984931834489357,
      "loss": 0.0564,
      "step": 10968
    },
    {
      "epoch": 1.5373510861948143,
      "grad_norm": 0.9001206159591675,
      "learning_rate": 0.0001498349677110739,
      "loss": 0.0311,
      "step": 10969
    },
    {
      "epoch": 1.537491240364401,
      "grad_norm": 1.357242226600647,
      "learning_rate": 0.00014982061707725423,
      "loss": 0.0363,
      "step": 10970
    },
    {
      "epoch": 1.5376313945339875,
      "grad_norm": 0.2559105455875397,
      "learning_rate": 0.00014980626644343456,
      "loss": 0.0981,
      "step": 10971
    },
    {
      "epoch": 1.537771548703574,
      "grad_norm": 0.5213108062744141,
      "learning_rate": 0.00014979191580961491,
      "loss": 0.0379,
      "step": 10972
    },
    {
      "epoch": 1.5379117028731604,
      "grad_norm": 0.45825639367103577,
      "learning_rate": 0.00014977756517579524,
      "loss": 0.0381,
      "step": 10973
    },
    {
      "epoch": 1.538051857042747,
      "grad_norm": 0.20507089793682098,
      "learning_rate": 0.0001497632145419756,
      "loss": 0.0519,
      "step": 10974
    },
    {
      "epoch": 1.5381920112123337,
      "grad_norm": 0.3508005738258362,
      "learning_rate": 0.00014974886390815593,
      "loss": 0.0742,
      "step": 10975
    },
    {
      "epoch": 1.5383321653819202,
      "grad_norm": 0.44060322642326355,
      "learning_rate": 0.00014973451327433628,
      "loss": 0.0541,
      "step": 10976
    },
    {
      "epoch": 1.5384723195515067,
      "grad_norm": 0.36605215072631836,
      "learning_rate": 0.0001497201626405166,
      "loss": 0.0798,
      "step": 10977
    },
    {
      "epoch": 1.538612473721093,
      "grad_norm": 0.6941869854927063,
      "learning_rate": 0.00014970581200669694,
      "loss": 0.0341,
      "step": 10978
    },
    {
      "epoch": 1.5387526278906798,
      "grad_norm": 0.3290439248085022,
      "learning_rate": 0.0001496914613728773,
      "loss": 0.0454,
      "step": 10979
    },
    {
      "epoch": 1.5388927820602663,
      "grad_norm": 0.5251160264015198,
      "learning_rate": 0.00014967711073905762,
      "loss": 0.0303,
      "step": 10980
    },
    {
      "epoch": 1.539032936229853,
      "grad_norm": 0.3014562427997589,
      "learning_rate": 0.00014966276010523798,
      "loss": 0.0547,
      "step": 10981
    },
    {
      "epoch": 1.5391730903994394,
      "grad_norm": 0.35790368914604187,
      "learning_rate": 0.0001496484094714183,
      "loss": 0.0514,
      "step": 10982
    },
    {
      "epoch": 1.5393132445690259,
      "grad_norm": 0.8156243562698364,
      "learning_rate": 0.00014963405883759864,
      "loss": 0.0527,
      "step": 10983
    },
    {
      "epoch": 1.5394533987386123,
      "grad_norm": 0.16634903848171234,
      "learning_rate": 0.000149619708203779,
      "loss": 0.0087,
      "step": 10984
    },
    {
      "epoch": 1.539593552908199,
      "grad_norm": 2.464573860168457,
      "learning_rate": 0.00014960535756995932,
      "loss": 0.1041,
      "step": 10985
    },
    {
      "epoch": 1.5397337070777857,
      "grad_norm": 0.3118284344673157,
      "learning_rate": 0.00014959100693613968,
      "loss": 0.0843,
      "step": 10986
    },
    {
      "epoch": 1.5398738612473721,
      "grad_norm": 0.4030275046825409,
      "learning_rate": 0.00014957665630232,
      "loss": 0.0821,
      "step": 10987
    },
    {
      "epoch": 1.5400140154169586,
      "grad_norm": 0.39984509348869324,
      "learning_rate": 0.00014956230566850033,
      "loss": 0.0694,
      "step": 10988
    },
    {
      "epoch": 1.540154169586545,
      "grad_norm": 0.6093841195106506,
      "learning_rate": 0.0001495479550346807,
      "loss": 0.1557,
      "step": 10989
    },
    {
      "epoch": 1.5402943237561317,
      "grad_norm": 0.24070964753627777,
      "learning_rate": 0.00014953360440086102,
      "loss": 0.0324,
      "step": 10990
    },
    {
      "epoch": 1.5404344779257184,
      "grad_norm": 0.5339551568031311,
      "learning_rate": 0.00014951925376704137,
      "loss": 0.1013,
      "step": 10991
    },
    {
      "epoch": 1.5405746320953049,
      "grad_norm": 0.2245163917541504,
      "learning_rate": 0.0001495049031332217,
      "loss": 0.0719,
      "step": 10992
    },
    {
      "epoch": 1.5407147862648913,
      "grad_norm": 0.4870762228965759,
      "learning_rate": 0.00014949055249940203,
      "loss": 0.0248,
      "step": 10993
    },
    {
      "epoch": 1.5408549404344778,
      "grad_norm": 0.8021231293678284,
      "learning_rate": 0.00014947620186558239,
      "loss": 0.1202,
      "step": 10994
    },
    {
      "epoch": 1.5409950946040645,
      "grad_norm": 0.36878955364227295,
      "learning_rate": 0.00014946185123176274,
      "loss": 0.0681,
      "step": 10995
    },
    {
      "epoch": 1.5411352487736512,
      "grad_norm": 0.5719016194343567,
      "learning_rate": 0.00014944750059794307,
      "loss": 0.0737,
      "step": 10996
    },
    {
      "epoch": 1.5412754029432376,
      "grad_norm": 0.7299607992172241,
      "learning_rate": 0.0001494331499641234,
      "loss": 0.066,
      "step": 10997
    },
    {
      "epoch": 1.541415557112824,
      "grad_norm": 0.3183037042617798,
      "learning_rate": 0.00014941879933030373,
      "loss": 0.0471,
      "step": 10998
    },
    {
      "epoch": 1.5415557112824105,
      "grad_norm": 0.2559284567832947,
      "learning_rate": 0.00014940444869648408,
      "loss": 0.0462,
      "step": 10999
    },
    {
      "epoch": 1.5416958654519972,
      "grad_norm": 0.3757286071777344,
      "learning_rate": 0.00014939009806266444,
      "loss": 0.0456,
      "step": 11000
    },
    {
      "epoch": 1.541836019621584,
      "grad_norm": 0.2546945810317993,
      "learning_rate": 0.00014937574742884477,
      "loss": 0.0394,
      "step": 11001
    },
    {
      "epoch": 1.5419761737911704,
      "grad_norm": 0.2045392394065857,
      "learning_rate": 0.0001493613967950251,
      "loss": 0.0487,
      "step": 11002
    },
    {
      "epoch": 1.5421163279607568,
      "grad_norm": 0.19313909113407135,
      "learning_rate": 0.00014934704616120545,
      "loss": 0.0505,
      "step": 11003
    },
    {
      "epoch": 1.5422564821303433,
      "grad_norm": 0.20614683628082275,
      "learning_rate": 0.00014933269552738578,
      "loss": 0.0261,
      "step": 11004
    },
    {
      "epoch": 1.54239663629993,
      "grad_norm": 0.20345093309879303,
      "learning_rate": 0.0001493183448935661,
      "loss": 0.05,
      "step": 11005
    },
    {
      "epoch": 1.5425367904695164,
      "grad_norm": 0.5298324227333069,
      "learning_rate": 0.00014930399425974646,
      "loss": 0.0238,
      "step": 11006
    },
    {
      "epoch": 1.542676944639103,
      "grad_norm": 0.11062458157539368,
      "learning_rate": 0.0001492896436259268,
      "loss": 0.0194,
      "step": 11007
    },
    {
      "epoch": 1.5428170988086896,
      "grad_norm": 0.14331156015396118,
      "learning_rate": 0.00014927529299210715,
      "loss": 0.0201,
      "step": 11008
    },
    {
      "epoch": 1.542957252978276,
      "grad_norm": 0.3429943919181824,
      "learning_rate": 0.00014926094235828748,
      "loss": 0.0498,
      "step": 11009
    },
    {
      "epoch": 1.5430974071478627,
      "grad_norm": 0.5337877869606018,
      "learning_rate": 0.0001492465917244678,
      "loss": 0.0872,
      "step": 11010
    },
    {
      "epoch": 1.5432375613174492,
      "grad_norm": 0.13501450419425964,
      "learning_rate": 0.00014923224109064816,
      "loss": 0.0256,
      "step": 11011
    },
    {
      "epoch": 1.5433777154870358,
      "grad_norm": 0.1723991483449936,
      "learning_rate": 0.0001492178904568285,
      "loss": 0.0253,
      "step": 11012
    },
    {
      "epoch": 1.5435178696566223,
      "grad_norm": 0.18672430515289307,
      "learning_rate": 0.00014920353982300885,
      "loss": 0.0677,
      "step": 11013
    },
    {
      "epoch": 1.5436580238262088,
      "grad_norm": 0.12765507400035858,
      "learning_rate": 0.00014918918918918917,
      "loss": 0.0415,
      "step": 11014
    },
    {
      "epoch": 1.5437981779957952,
      "grad_norm": 0.3994615972042084,
      "learning_rate": 0.0001491748385553695,
      "loss": 0.0277,
      "step": 11015
    },
    {
      "epoch": 1.543938332165382,
      "grad_norm": 0.1665220558643341,
      "learning_rate": 0.00014916048792154986,
      "loss": 0.0321,
      "step": 11016
    },
    {
      "epoch": 1.5440784863349686,
      "grad_norm": 0.19501011073589325,
      "learning_rate": 0.0001491461372877302,
      "loss": 0.0291,
      "step": 11017
    },
    {
      "epoch": 1.544218640504555,
      "grad_norm": 0.17355221509933472,
      "learning_rate": 0.00014913178665391054,
      "loss": 0.0289,
      "step": 11018
    },
    {
      "epoch": 1.5443587946741415,
      "grad_norm": 0.2866831421852112,
      "learning_rate": 0.00014911743602009087,
      "loss": 0.0527,
      "step": 11019
    },
    {
      "epoch": 1.544498948843728,
      "grad_norm": 0.17746466398239136,
      "learning_rate": 0.0001491030853862712,
      "loss": 0.0517,
      "step": 11020
    },
    {
      "epoch": 1.5446391030133146,
      "grad_norm": 0.8315578699111938,
      "learning_rate": 0.00014908873475245156,
      "loss": 0.0588,
      "step": 11021
    },
    {
      "epoch": 1.5447792571829013,
      "grad_norm": 0.17520473897457123,
      "learning_rate": 0.0001490743841186319,
      "loss": 0.0337,
      "step": 11022
    },
    {
      "epoch": 1.5449194113524878,
      "grad_norm": 0.24548864364624023,
      "learning_rate": 0.00014906003348481224,
      "loss": 0.0441,
      "step": 11023
    },
    {
      "epoch": 1.5450595655220742,
      "grad_norm": 0.4432048201560974,
      "learning_rate": 0.00014904568285099257,
      "loss": 0.0945,
      "step": 11024
    },
    {
      "epoch": 1.5451997196916607,
      "grad_norm": 0.17676718533039093,
      "learning_rate": 0.0001490313322171729,
      "loss": 0.0748,
      "step": 11025
    },
    {
      "epoch": 1.5453398738612474,
      "grad_norm": 0.22608895599842072,
      "learning_rate": 0.00014901698158335325,
      "loss": 0.0166,
      "step": 11026
    },
    {
      "epoch": 1.545480028030834,
      "grad_norm": 0.7016151547431946,
      "learning_rate": 0.0001490026309495336,
      "loss": 0.0454,
      "step": 11027
    },
    {
      "epoch": 1.5456201822004205,
      "grad_norm": 0.3259376883506775,
      "learning_rate": 0.00014898828031571394,
      "loss": 0.0618,
      "step": 11028
    },
    {
      "epoch": 1.545760336370007,
      "grad_norm": 0.33096277713775635,
      "learning_rate": 0.00014897392968189426,
      "loss": 0.0718,
      "step": 11029
    },
    {
      "epoch": 1.5459004905395934,
      "grad_norm": 0.4058026373386383,
      "learning_rate": 0.00014895957904807462,
      "loss": 0.0383,
      "step": 11030
    },
    {
      "epoch": 1.5460406447091801,
      "grad_norm": 0.5410714149475098,
      "learning_rate": 0.00014894522841425495,
      "loss": 0.0787,
      "step": 11031
    },
    {
      "epoch": 1.5461807988787668,
      "grad_norm": 0.7713353037834167,
      "learning_rate": 0.0001489308777804353,
      "loss": 0.1656,
      "step": 11032
    },
    {
      "epoch": 1.5463209530483533,
      "grad_norm": 0.3726988732814789,
      "learning_rate": 0.00014891652714661563,
      "loss": 0.1695,
      "step": 11033
    },
    {
      "epoch": 1.5464611072179397,
      "grad_norm": 0.7880122661590576,
      "learning_rate": 0.00014890217651279596,
      "loss": 0.0568,
      "step": 11034
    },
    {
      "epoch": 1.5466012613875262,
      "grad_norm": 1.9528509378433228,
      "learning_rate": 0.00014888782587897632,
      "loss": 0.1263,
      "step": 11035
    },
    {
      "epoch": 1.5467414155571129,
      "grad_norm": 0.2528485953807831,
      "learning_rate": 0.00014887347524515665,
      "loss": 0.0463,
      "step": 11036
    },
    {
      "epoch": 1.5468815697266993,
      "grad_norm": 0.38555899262428284,
      "learning_rate": 0.000148859124611337,
      "loss": 0.0791,
      "step": 11037
    },
    {
      "epoch": 1.547021723896286,
      "grad_norm": 0.49913421273231506,
      "learning_rate": 0.00014884477397751733,
      "loss": 0.0519,
      "step": 11038
    },
    {
      "epoch": 1.5471618780658725,
      "grad_norm": 0.11820407956838608,
      "learning_rate": 0.00014883042334369766,
      "loss": 0.0167,
      "step": 11039
    },
    {
      "epoch": 1.547302032235459,
      "grad_norm": 0.25541189312934875,
      "learning_rate": 0.00014881607270987801,
      "loss": 0.0362,
      "step": 11040
    },
    {
      "epoch": 1.5474421864050456,
      "grad_norm": 0.31139102578163147,
      "learning_rate": 0.00014880172207605834,
      "loss": 0.0366,
      "step": 11041
    },
    {
      "epoch": 1.547582340574632,
      "grad_norm": 0.3352653980255127,
      "learning_rate": 0.00014878737144223867,
      "loss": 0.0451,
      "step": 11042
    },
    {
      "epoch": 1.5477224947442187,
      "grad_norm": 0.32105931639671326,
      "learning_rate": 0.00014877302080841903,
      "loss": 0.1157,
      "step": 11043
    },
    {
      "epoch": 1.5478626489138052,
      "grad_norm": 0.6846283078193665,
      "learning_rate": 0.00014875867017459936,
      "loss": 0.1519,
      "step": 11044
    },
    {
      "epoch": 1.5480028030833917,
      "grad_norm": 0.33372262120246887,
      "learning_rate": 0.0001487443195407797,
      "loss": 0.0615,
      "step": 11045
    },
    {
      "epoch": 1.5481429572529781,
      "grad_norm": 0.32960978150367737,
      "learning_rate": 0.00014872996890696004,
      "loss": 0.0341,
      "step": 11046
    },
    {
      "epoch": 1.5482831114225648,
      "grad_norm": 0.22216135263442993,
      "learning_rate": 0.00014871561827314037,
      "loss": 0.0439,
      "step": 11047
    },
    {
      "epoch": 1.5484232655921515,
      "grad_norm": 0.4431505799293518,
      "learning_rate": 0.00014870126763932072,
      "loss": 0.0633,
      "step": 11048
    },
    {
      "epoch": 1.548563419761738,
      "grad_norm": 0.10634178668260574,
      "learning_rate": 0.00014868691700550105,
      "loss": 0.0128,
      "step": 11049
    },
    {
      "epoch": 1.5487035739313244,
      "grad_norm": 0.9513567686080933,
      "learning_rate": 0.0001486725663716814,
      "loss": 0.0531,
      "step": 11050
    },
    {
      "epoch": 1.5488437281009109,
      "grad_norm": 0.44933417439460754,
      "learning_rate": 0.00014865821573786174,
      "loss": 0.1208,
      "step": 11051
    },
    {
      "epoch": 1.5489838822704975,
      "grad_norm": 0.2294304519891739,
      "learning_rate": 0.00014864386510404207,
      "loss": 0.0387,
      "step": 11052
    },
    {
      "epoch": 1.5491240364400842,
      "grad_norm": 0.11340650916099548,
      "learning_rate": 0.00014862951447022242,
      "loss": 0.0179,
      "step": 11053
    },
    {
      "epoch": 1.5492641906096707,
      "grad_norm": 0.3369068503379822,
      "learning_rate": 0.00014861516383640278,
      "loss": 0.035,
      "step": 11054
    },
    {
      "epoch": 1.5494043447792571,
      "grad_norm": 0.3367752432823181,
      "learning_rate": 0.0001486008132025831,
      "loss": 0.068,
      "step": 11055
    },
    {
      "epoch": 1.5495444989488436,
      "grad_norm": 0.36260631680488586,
      "learning_rate": 0.00014858646256876343,
      "loss": 0.074,
      "step": 11056
    },
    {
      "epoch": 1.5496846531184303,
      "grad_norm": 0.05152274668216705,
      "learning_rate": 0.0001485721119349438,
      "loss": 0.0042,
      "step": 11057
    },
    {
      "epoch": 1.549824807288017,
      "grad_norm": 0.15546537935733795,
      "learning_rate": 0.00014855776130112412,
      "loss": 0.0362,
      "step": 11058
    },
    {
      "epoch": 1.5499649614576034,
      "grad_norm": 0.16330179572105408,
      "learning_rate": 0.00014854341066730447,
      "loss": 0.0298,
      "step": 11059
    },
    {
      "epoch": 1.55010511562719,
      "grad_norm": 0.18959923088550568,
      "learning_rate": 0.0001485290600334848,
      "loss": 0.0612,
      "step": 11060
    },
    {
      "epoch": 1.5502452697967763,
      "grad_norm": 0.16318963468074799,
      "learning_rate": 0.00014851470939966513,
      "loss": 0.0226,
      "step": 11061
    },
    {
      "epoch": 1.550385423966363,
      "grad_norm": 0.07888984680175781,
      "learning_rate": 0.0001485003587658455,
      "loss": 0.012,
      "step": 11062
    },
    {
      "epoch": 1.5505255781359495,
      "grad_norm": 0.19933873414993286,
      "learning_rate": 0.00014848600813202582,
      "loss": 0.0184,
      "step": 11063
    },
    {
      "epoch": 1.5506657323055362,
      "grad_norm": 0.4456675350666046,
      "learning_rate": 0.00014847165749820617,
      "loss": 0.0342,
      "step": 11064
    },
    {
      "epoch": 1.5508058864751226,
      "grad_norm": 0.6584606766700745,
      "learning_rate": 0.0001484573068643865,
      "loss": 0.1309,
      "step": 11065
    },
    {
      "epoch": 1.550946040644709,
      "grad_norm": 0.27750495076179504,
      "learning_rate": 0.00014844295623056683,
      "loss": 0.0299,
      "step": 11066
    },
    {
      "epoch": 1.5510861948142958,
      "grad_norm": 0.06212341785430908,
      "learning_rate": 0.00014842860559674718,
      "loss": 0.0063,
      "step": 11067
    },
    {
      "epoch": 1.5512263489838822,
      "grad_norm": 0.31536439061164856,
      "learning_rate": 0.0001484142549629275,
      "loss": 0.0298,
      "step": 11068
    },
    {
      "epoch": 1.551366503153469,
      "grad_norm": 0.559956431388855,
      "learning_rate": 0.00014839990432910787,
      "loss": 0.0525,
      "step": 11069
    },
    {
      "epoch": 1.5515066573230554,
      "grad_norm": 0.2821565568447113,
      "learning_rate": 0.0001483855536952882,
      "loss": 0.0393,
      "step": 11070
    },
    {
      "epoch": 1.5516468114926418,
      "grad_norm": 0.6767759919166565,
      "learning_rate": 0.00014837120306146853,
      "loss": 0.102,
      "step": 11071
    },
    {
      "epoch": 1.5517869656622283,
      "grad_norm": 0.37384364008903503,
      "learning_rate": 0.00014835685242764888,
      "loss": 0.0377,
      "step": 11072
    },
    {
      "epoch": 1.551927119831815,
      "grad_norm": 0.6912876963615417,
      "learning_rate": 0.0001483425017938292,
      "loss": 0.0718,
      "step": 11073
    },
    {
      "epoch": 1.5520672740014017,
      "grad_norm": 0.8117472529411316,
      "learning_rate": 0.00014832815116000954,
      "loss": 0.0694,
      "step": 11074
    },
    {
      "epoch": 1.5522074281709881,
      "grad_norm": 1.1233464479446411,
      "learning_rate": 0.0001483138005261899,
      "loss": 0.0599,
      "step": 11075
    },
    {
      "epoch": 1.5523475823405746,
      "grad_norm": 0.8551127910614014,
      "learning_rate": 0.00014829944989237022,
      "loss": 0.075,
      "step": 11076
    },
    {
      "epoch": 1.552487736510161,
      "grad_norm": 0.026999225839972496,
      "learning_rate": 0.00014828509925855058,
      "loss": 0.0029,
      "step": 11077
    },
    {
      "epoch": 1.5526278906797477,
      "grad_norm": 0.3826160728931427,
      "learning_rate": 0.0001482707486247309,
      "loss": 0.1069,
      "step": 11078
    },
    {
      "epoch": 1.5527680448493344,
      "grad_norm": 0.24017146229743958,
      "learning_rate": 0.00014825639799091124,
      "loss": 0.0721,
      "step": 11079
    },
    {
      "epoch": 1.5529081990189209,
      "grad_norm": 0.8382569551467896,
      "learning_rate": 0.0001482420473570916,
      "loss": 0.1573,
      "step": 11080
    },
    {
      "epoch": 1.5530483531885073,
      "grad_norm": 0.09319008141756058,
      "learning_rate": 0.00014822769672327195,
      "loss": 0.0122,
      "step": 11081
    },
    {
      "epoch": 1.5531885073580938,
      "grad_norm": 0.42329511046409607,
      "learning_rate": 0.00014821334608945227,
      "loss": 0.0565,
      "step": 11082
    },
    {
      "epoch": 1.5533286615276805,
      "grad_norm": 0.17139148712158203,
      "learning_rate": 0.0001481989954556326,
      "loss": 0.0169,
      "step": 11083
    },
    {
      "epoch": 1.5534688156972671,
      "grad_norm": 0.8504591584205627,
      "learning_rate": 0.00014818464482181293,
      "loss": 0.0654,
      "step": 11084
    },
    {
      "epoch": 1.5536089698668536,
      "grad_norm": 0.12272142618894577,
      "learning_rate": 0.0001481702941879933,
      "loss": 0.0042,
      "step": 11085
    },
    {
      "epoch": 1.55374912403644,
      "grad_norm": 0.5930334329605103,
      "learning_rate": 0.00014815594355417364,
      "loss": 0.0992,
      "step": 11086
    },
    {
      "epoch": 1.5538892782060265,
      "grad_norm": 0.15107546746730804,
      "learning_rate": 0.00014814159292035397,
      "loss": 0.0402,
      "step": 11087
    },
    {
      "epoch": 1.5540294323756132,
      "grad_norm": 0.23581773042678833,
      "learning_rate": 0.0001481272422865343,
      "loss": 0.061,
      "step": 11088
    },
    {
      "epoch": 1.5541695865451999,
      "grad_norm": 0.24607306718826294,
      "learning_rate": 0.00014811289165271466,
      "loss": 0.0416,
      "step": 11089
    },
    {
      "epoch": 1.5543097407147863,
      "grad_norm": 1.1874651908874512,
      "learning_rate": 0.00014809854101889498,
      "loss": 0.0517,
      "step": 11090
    },
    {
      "epoch": 1.5544498948843728,
      "grad_norm": 0.18721534311771393,
      "learning_rate": 0.00014808419038507534,
      "loss": 0.0199,
      "step": 11091
    },
    {
      "epoch": 1.5545900490539593,
      "grad_norm": 0.26073938608169556,
      "learning_rate": 0.00014806983975125567,
      "loss": 0.0665,
      "step": 11092
    },
    {
      "epoch": 1.554730203223546,
      "grad_norm": 0.3165334165096283,
      "learning_rate": 0.000148055489117436,
      "loss": 0.0423,
      "step": 11093
    },
    {
      "epoch": 1.5548703573931324,
      "grad_norm": 0.17674729228019714,
      "learning_rate": 0.00014804113848361635,
      "loss": 0.0228,
      "step": 11094
    },
    {
      "epoch": 1.555010511562719,
      "grad_norm": 0.291576087474823,
      "learning_rate": 0.00014802678784979668,
      "loss": 0.0356,
      "step": 11095
    },
    {
      "epoch": 1.5551506657323055,
      "grad_norm": 0.3110606372356415,
      "learning_rate": 0.00014801243721597704,
      "loss": 0.0242,
      "step": 11096
    },
    {
      "epoch": 1.555290819901892,
      "grad_norm": 0.2618960440158844,
      "learning_rate": 0.00014799808658215737,
      "loss": 0.0326,
      "step": 11097
    },
    {
      "epoch": 1.5554309740714787,
      "grad_norm": 0.239740788936615,
      "learning_rate": 0.0001479837359483377,
      "loss": 0.0216,
      "step": 11098
    },
    {
      "epoch": 1.5555711282410651,
      "grad_norm": 0.5276740193367004,
      "learning_rate": 0.00014796938531451805,
      "loss": 0.0317,
      "step": 11099
    },
    {
      "epoch": 1.5557112824106518,
      "grad_norm": 0.14626654982566833,
      "learning_rate": 0.00014795503468069838,
      "loss": 0.0187,
      "step": 11100
    },
    {
      "epoch": 1.5558514365802383,
      "grad_norm": 0.191325843334198,
      "learning_rate": 0.00014794068404687873,
      "loss": 0.049,
      "step": 11101
    },
    {
      "epoch": 1.5559915907498247,
      "grad_norm": 0.3403986990451813,
      "learning_rate": 0.00014792633341305906,
      "loss": 0.0657,
      "step": 11102
    },
    {
      "epoch": 1.5561317449194112,
      "grad_norm": 0.2034078687429428,
      "learning_rate": 0.0001479119827792394,
      "loss": 0.0199,
      "step": 11103
    },
    {
      "epoch": 1.5562718990889979,
      "grad_norm": 0.14513526856899261,
      "learning_rate": 0.00014789763214541975,
      "loss": 0.0315,
      "step": 11104
    },
    {
      "epoch": 1.5564120532585846,
      "grad_norm": 0.40946337580680847,
      "learning_rate": 0.00014788328151160008,
      "loss": 0.0247,
      "step": 11105
    },
    {
      "epoch": 1.556552207428171,
      "grad_norm": 0.11619977653026581,
      "learning_rate": 0.0001478689308777804,
      "loss": 0.0177,
      "step": 11106
    },
    {
      "epoch": 1.5566923615977575,
      "grad_norm": 0.4205186069011688,
      "learning_rate": 0.00014785458024396076,
      "loss": 0.0508,
      "step": 11107
    },
    {
      "epoch": 1.556832515767344,
      "grad_norm": 0.25226473808288574,
      "learning_rate": 0.00014784022961014112,
      "loss": 0.0748,
      "step": 11108
    },
    {
      "epoch": 1.5569726699369306,
      "grad_norm": 0.7032549977302551,
      "learning_rate": 0.00014782587897632144,
      "loss": 0.0992,
      "step": 11109
    },
    {
      "epoch": 1.5571128241065173,
      "grad_norm": 0.48030897974967957,
      "learning_rate": 0.00014781152834250177,
      "loss": 0.0727,
      "step": 11110
    },
    {
      "epoch": 1.5572529782761038,
      "grad_norm": 0.257094144821167,
      "learning_rate": 0.0001477971777086821,
      "loss": 0.0565,
      "step": 11111
    },
    {
      "epoch": 1.5573931324456902,
      "grad_norm": 0.21146482229232788,
      "learning_rate": 0.00014778282707486246,
      "loss": 0.0477,
      "step": 11112
    },
    {
      "epoch": 1.5575332866152767,
      "grad_norm": 0.1778813898563385,
      "learning_rate": 0.0001477684764410428,
      "loss": 0.0205,
      "step": 11113
    },
    {
      "epoch": 1.5576734407848634,
      "grad_norm": 0.21597367525100708,
      "learning_rate": 0.00014775412580722314,
      "loss": 0.0189,
      "step": 11114
    },
    {
      "epoch": 1.55781359495445,
      "grad_norm": 0.3993750810623169,
      "learning_rate": 0.00014773977517340347,
      "loss": 0.0398,
      "step": 11115
    },
    {
      "epoch": 1.5579537491240365,
      "grad_norm": 0.19708694517612457,
      "learning_rate": 0.00014772542453958383,
      "loss": 0.0112,
      "step": 11116
    },
    {
      "epoch": 1.558093903293623,
      "grad_norm": 0.3637322187423706,
      "learning_rate": 0.00014771107390576415,
      "loss": 0.0437,
      "step": 11117
    },
    {
      "epoch": 1.5582340574632094,
      "grad_norm": 0.2887438237667084,
      "learning_rate": 0.0001476967232719445,
      "loss": 0.026,
      "step": 11118
    },
    {
      "epoch": 1.558374211632796,
      "grad_norm": 0.32681554555892944,
      "learning_rate": 0.00014768237263812484,
      "loss": 0.0196,
      "step": 11119
    },
    {
      "epoch": 1.5585143658023828,
      "grad_norm": 0.528824508190155,
      "learning_rate": 0.00014766802200430517,
      "loss": 0.0178,
      "step": 11120
    },
    {
      "epoch": 1.5586545199719692,
      "grad_norm": 0.211766317486763,
      "learning_rate": 0.00014765367137048552,
      "loss": 0.0308,
      "step": 11121
    },
    {
      "epoch": 1.5587946741415557,
      "grad_norm": 0.5120553970336914,
      "learning_rate": 0.00014763932073666585,
      "loss": 0.0844,
      "step": 11122
    },
    {
      "epoch": 1.5589348283111422,
      "grad_norm": 0.18291109800338745,
      "learning_rate": 0.0001476249701028462,
      "loss": 0.0474,
      "step": 11123
    },
    {
      "epoch": 1.5590749824807288,
      "grad_norm": 0.3133188486099243,
      "learning_rate": 0.00014761061946902654,
      "loss": 0.0164,
      "step": 11124
    },
    {
      "epoch": 1.5592151366503153,
      "grad_norm": 0.2677336633205414,
      "learning_rate": 0.00014759626883520686,
      "loss": 0.0247,
      "step": 11125
    },
    {
      "epoch": 1.559355290819902,
      "grad_norm": 0.3338314890861511,
      "learning_rate": 0.00014758191820138722,
      "loss": 0.0296,
      "step": 11126
    },
    {
      "epoch": 1.5594954449894884,
      "grad_norm": 0.5702020525932312,
      "learning_rate": 0.00014756756756756758,
      "loss": 0.0666,
      "step": 11127
    },
    {
      "epoch": 1.559635599159075,
      "grad_norm": 0.3197702467441559,
      "learning_rate": 0.0001475532169337479,
      "loss": 0.0347,
      "step": 11128
    },
    {
      "epoch": 1.5597757533286614,
      "grad_norm": 0.6559661626815796,
      "learning_rate": 0.00014753886629992823,
      "loss": 0.0724,
      "step": 11129
    },
    {
      "epoch": 1.559915907498248,
      "grad_norm": 1.1053879261016846,
      "learning_rate": 0.00014752451566610856,
      "loss": 0.0729,
      "step": 11130
    },
    {
      "epoch": 1.5600560616678347,
      "grad_norm": 1.476270318031311,
      "learning_rate": 0.00014751016503228892,
      "loss": 0.196,
      "step": 11131
    },
    {
      "epoch": 1.5601962158374212,
      "grad_norm": 0.46585288643836975,
      "learning_rate": 0.00014749581439846927,
      "loss": 0.0949,
      "step": 11132
    },
    {
      "epoch": 1.5603363700070076,
      "grad_norm": 1.2716703414916992,
      "learning_rate": 0.0001474814637646496,
      "loss": 0.1641,
      "step": 11133
    },
    {
      "epoch": 1.560476524176594,
      "grad_norm": 0.8445846438407898,
      "learning_rate": 0.00014746711313082993,
      "loss": 0.1595,
      "step": 11134
    },
    {
      "epoch": 1.5606166783461808,
      "grad_norm": 0.9498838186264038,
      "learning_rate": 0.00014745276249701028,
      "loss": 0.1249,
      "step": 11135
    },
    {
      "epoch": 1.5607568325157675,
      "grad_norm": 0.6306179165840149,
      "learning_rate": 0.0001474384118631906,
      "loss": 0.0813,
      "step": 11136
    },
    {
      "epoch": 1.560896986685354,
      "grad_norm": 0.16381900012493134,
      "learning_rate": 0.00014742406122937094,
      "loss": 0.0303,
      "step": 11137
    },
    {
      "epoch": 1.5610371408549404,
      "grad_norm": 0.22249996662139893,
      "learning_rate": 0.00014740971059555127,
      "loss": 0.0302,
      "step": 11138
    },
    {
      "epoch": 1.5611772950245268,
      "grad_norm": 0.2671750783920288,
      "learning_rate": 0.00014739535996173163,
      "loss": 0.0264,
      "step": 11139
    },
    {
      "epoch": 1.5613174491941135,
      "grad_norm": 0.5948111414909363,
      "learning_rate": 0.00014738100932791198,
      "loss": 0.0585,
      "step": 11140
    },
    {
      "epoch": 1.5614576033637002,
      "grad_norm": 0.44250333309173584,
      "learning_rate": 0.0001473666586940923,
      "loss": 0.0634,
      "step": 11141
    },
    {
      "epoch": 1.5615977575332867,
      "grad_norm": 0.2616373598575592,
      "learning_rate": 0.00014735230806027264,
      "loss": 0.0485,
      "step": 11142
    },
    {
      "epoch": 1.5617379117028731,
      "grad_norm": 0.41670432686805725,
      "learning_rate": 0.000147337957426453,
      "loss": 0.0741,
      "step": 11143
    },
    {
      "epoch": 1.5618780658724596,
      "grad_norm": 0.2678121030330658,
      "learning_rate": 0.00014732360679263332,
      "loss": 0.0564,
      "step": 11144
    },
    {
      "epoch": 1.5620182200420463,
      "grad_norm": 0.3389662206172943,
      "learning_rate": 0.00014730925615881368,
      "loss": 0.0316,
      "step": 11145
    },
    {
      "epoch": 1.562158374211633,
      "grad_norm": 0.2639074921607971,
      "learning_rate": 0.000147294905524994,
      "loss": 0.1061,
      "step": 11146
    },
    {
      "epoch": 1.5622985283812194,
      "grad_norm": 0.2882052958011627,
      "learning_rate": 0.00014728055489117434,
      "loss": 0.0311,
      "step": 11147
    },
    {
      "epoch": 1.5624386825508059,
      "grad_norm": 0.07320276647806168,
      "learning_rate": 0.0001472662042573547,
      "loss": 0.0073,
      "step": 11148
    },
    {
      "epoch": 1.5625788367203923,
      "grad_norm": 0.3554665148258209,
      "learning_rate": 0.00014725185362353502,
      "loss": 0.0559,
      "step": 11149
    },
    {
      "epoch": 1.562718990889979,
      "grad_norm": 0.772337794303894,
      "learning_rate": 0.00014723750298971538,
      "loss": 0.0684,
      "step": 11150
    },
    {
      "epoch": 1.5628591450595655,
      "grad_norm": 0.31310799717903137,
      "learning_rate": 0.0001472231523558957,
      "loss": 0.036,
      "step": 11151
    },
    {
      "epoch": 1.5629992992291522,
      "grad_norm": 0.3384800851345062,
      "learning_rate": 0.00014720880172207603,
      "loss": 0.0979,
      "step": 11152
    },
    {
      "epoch": 1.5631394533987386,
      "grad_norm": 0.15126965939998627,
      "learning_rate": 0.0001471944510882564,
      "loss": 0.0207,
      "step": 11153
    },
    {
      "epoch": 1.563279607568325,
      "grad_norm": 0.31631508469581604,
      "learning_rate": 0.00014718010045443672,
      "loss": 0.0315,
      "step": 11154
    },
    {
      "epoch": 1.5634197617379118,
      "grad_norm": 0.17821283638477325,
      "learning_rate": 0.00014716574982061707,
      "loss": 0.017,
      "step": 11155
    },
    {
      "epoch": 1.5635599159074982,
      "grad_norm": 0.42374420166015625,
      "learning_rate": 0.0001471513991867974,
      "loss": 0.0701,
      "step": 11156
    },
    {
      "epoch": 1.563700070077085,
      "grad_norm": 0.1802091896533966,
      "learning_rate": 0.00014713704855297773,
      "loss": 0.0241,
      "step": 11157
    },
    {
      "epoch": 1.5638402242466714,
      "grad_norm": 0.2635543942451477,
      "learning_rate": 0.00014712269791915809,
      "loss": 0.0346,
      "step": 11158
    },
    {
      "epoch": 1.5639803784162578,
      "grad_norm": 0.5130125284194946,
      "learning_rate": 0.00014710834728533844,
      "loss": 0.0417,
      "step": 11159
    },
    {
      "epoch": 1.5641205325858443,
      "grad_norm": 0.19770728051662445,
      "learning_rate": 0.00014709399665151877,
      "loss": 0.035,
      "step": 11160
    },
    {
      "epoch": 1.564260686755431,
      "grad_norm": 0.09157402813434601,
      "learning_rate": 0.0001470796460176991,
      "loss": 0.007,
      "step": 11161
    },
    {
      "epoch": 1.5644008409250176,
      "grad_norm": 0.7800874710083008,
      "learning_rate": 0.00014706529538387945,
      "loss": 0.0495,
      "step": 11162
    },
    {
      "epoch": 1.564540995094604,
      "grad_norm": 0.1194981038570404,
      "learning_rate": 0.00014705094475005978,
      "loss": 0.0317,
      "step": 11163
    },
    {
      "epoch": 1.5646811492641906,
      "grad_norm": 0.08925904333591461,
      "learning_rate": 0.00014703659411624014,
      "loss": 0.0145,
      "step": 11164
    },
    {
      "epoch": 1.564821303433777,
      "grad_norm": 0.1607254594564438,
      "learning_rate": 0.00014702224348242047,
      "loss": 0.0222,
      "step": 11165
    },
    {
      "epoch": 1.5649614576033637,
      "grad_norm": 0.573932945728302,
      "learning_rate": 0.0001470078928486008,
      "loss": 0.1133,
      "step": 11166
    },
    {
      "epoch": 1.5651016117729504,
      "grad_norm": 0.16129440069198608,
      "learning_rate": 0.00014699354221478115,
      "loss": 0.0234,
      "step": 11167
    },
    {
      "epoch": 1.5652417659425368,
      "grad_norm": 0.1657620072364807,
      "learning_rate": 0.00014697919158096148,
      "loss": 0.0471,
      "step": 11168
    },
    {
      "epoch": 1.5653819201121233,
      "grad_norm": 0.1637999564409256,
      "learning_rate": 0.0001469648409471418,
      "loss": 0.0278,
      "step": 11169
    },
    {
      "epoch": 1.5655220742817098,
      "grad_norm": 0.13079184293746948,
      "learning_rate": 0.00014695049031332216,
      "loss": 0.0267,
      "step": 11170
    },
    {
      "epoch": 1.5656622284512964,
      "grad_norm": 0.24356278777122498,
      "learning_rate": 0.0001469361396795025,
      "loss": 0.0265,
      "step": 11171
    },
    {
      "epoch": 1.5658023826208831,
      "grad_norm": 0.1065502017736435,
      "learning_rate": 0.00014692178904568285,
      "loss": 0.0092,
      "step": 11172
    },
    {
      "epoch": 1.5659425367904696,
      "grad_norm": 0.14290915429592133,
      "learning_rate": 0.00014690743841186318,
      "loss": 0.0216,
      "step": 11173
    },
    {
      "epoch": 1.566082690960056,
      "grad_norm": 0.4905205965042114,
      "learning_rate": 0.0001468930877780435,
      "loss": 0.0779,
      "step": 11174
    },
    {
      "epoch": 1.5662228451296425,
      "grad_norm": 0.5665708780288696,
      "learning_rate": 0.00014687873714422386,
      "loss": 0.022,
      "step": 11175
    },
    {
      "epoch": 1.5663629992992292,
      "grad_norm": 0.23964370787143707,
      "learning_rate": 0.0001468643865104042,
      "loss": 0.0225,
      "step": 11176
    },
    {
      "epoch": 1.5665031534688159,
      "grad_norm": 0.5382786989212036,
      "learning_rate": 0.00014685003587658455,
      "loss": 0.1165,
      "step": 11177
    },
    {
      "epoch": 1.5666433076384023,
      "grad_norm": 0.06651629507541656,
      "learning_rate": 0.00014683568524276487,
      "loss": 0.0085,
      "step": 11178
    },
    {
      "epoch": 1.5667834618079888,
      "grad_norm": 0.5134016871452332,
      "learning_rate": 0.0001468213346089452,
      "loss": 0.1353,
      "step": 11179
    },
    {
      "epoch": 1.5669236159775752,
      "grad_norm": 0.8455452919006348,
      "learning_rate": 0.00014680698397512556,
      "loss": 0.0917,
      "step": 11180
    },
    {
      "epoch": 1.567063770147162,
      "grad_norm": 0.5323025584220886,
      "learning_rate": 0.0001467926333413059,
      "loss": 0.077,
      "step": 11181
    },
    {
      "epoch": 1.5672039243167484,
      "grad_norm": 0.21210433542728424,
      "learning_rate": 0.00014677828270748624,
      "loss": 0.0134,
      "step": 11182
    },
    {
      "epoch": 1.567344078486335,
      "grad_norm": 0.1852567344903946,
      "learning_rate": 0.00014676393207366657,
      "loss": 0.0109,
      "step": 11183
    },
    {
      "epoch": 1.5674842326559215,
      "grad_norm": 1.3256570100784302,
      "learning_rate": 0.0001467495814398469,
      "loss": 0.0751,
      "step": 11184
    },
    {
      "epoch": 1.567624386825508,
      "grad_norm": 2.4334514141082764,
      "learning_rate": 0.00014673523080602726,
      "loss": 0.2014,
      "step": 11185
    },
    {
      "epoch": 1.5677645409950947,
      "grad_norm": 0.35006874799728394,
      "learning_rate": 0.0001467208801722076,
      "loss": 0.081,
      "step": 11186
    },
    {
      "epoch": 1.5679046951646811,
      "grad_norm": 0.37955722212791443,
      "learning_rate": 0.00014670652953838794,
      "loss": 0.0444,
      "step": 11187
    },
    {
      "epoch": 1.5680448493342678,
      "grad_norm": 0.08182496577501297,
      "learning_rate": 0.00014669217890456827,
      "loss": 0.0077,
      "step": 11188
    },
    {
      "epoch": 1.5681850035038543,
      "grad_norm": 0.1904602348804474,
      "learning_rate": 0.0001466778282707486,
      "loss": 0.0463,
      "step": 11189
    },
    {
      "epoch": 1.5683251576734407,
      "grad_norm": 0.5992399454116821,
      "learning_rate": 0.00014666347763692895,
      "loss": 0.0424,
      "step": 11190
    },
    {
      "epoch": 1.5684653118430272,
      "grad_norm": 0.528991162776947,
      "learning_rate": 0.0001466491270031093,
      "loss": 0.0868,
      "step": 11191
    },
    {
      "epoch": 1.5686054660126139,
      "grad_norm": 0.2003084421157837,
      "learning_rate": 0.00014663477636928964,
      "loss": 0.016,
      "step": 11192
    },
    {
      "epoch": 1.5687456201822005,
      "grad_norm": 0.27936655282974243,
      "learning_rate": 0.00014662042573546996,
      "loss": 0.0697,
      "step": 11193
    },
    {
      "epoch": 1.568885774351787,
      "grad_norm": 0.39925622940063477,
      "learning_rate": 0.00014660607510165032,
      "loss": 0.0488,
      "step": 11194
    },
    {
      "epoch": 1.5690259285213735,
      "grad_norm": 0.37023505568504333,
      "learning_rate": 0.00014659172446783065,
      "loss": 0.0271,
      "step": 11195
    },
    {
      "epoch": 1.56916608269096,
      "grad_norm": 0.19889254868030548,
      "learning_rate": 0.000146577373834011,
      "loss": 0.0321,
      "step": 11196
    },
    {
      "epoch": 1.5693062368605466,
      "grad_norm": 0.18486392498016357,
      "learning_rate": 0.00014656302320019133,
      "loss": 0.0143,
      "step": 11197
    },
    {
      "epoch": 1.5694463910301333,
      "grad_norm": 0.1314917802810669,
      "learning_rate": 0.00014654867256637166,
      "loss": 0.0249,
      "step": 11198
    },
    {
      "epoch": 1.5695865451997197,
      "grad_norm": 0.24066641926765442,
      "learning_rate": 0.00014653432193255202,
      "loss": 0.0448,
      "step": 11199
    },
    {
      "epoch": 1.5697266993693062,
      "grad_norm": 0.20847086608409882,
      "learning_rate": 0.00014651997129873235,
      "loss": 0.0629,
      "step": 11200
    },
    {
      "epoch": 1.5698668535388927,
      "grad_norm": 0.18620292842388153,
      "learning_rate": 0.00014650562066491267,
      "loss": 0.0353,
      "step": 11201
    },
    {
      "epoch": 1.5700070077084793,
      "grad_norm": 0.2205834835767746,
      "learning_rate": 0.00014649127003109303,
      "loss": 0.0463,
      "step": 11202
    },
    {
      "epoch": 1.570147161878066,
      "grad_norm": 0.25545647740364075,
      "learning_rate": 0.00014647691939727336,
      "loss": 0.0771,
      "step": 11203
    },
    {
      "epoch": 1.5702873160476525,
      "grad_norm": 0.352230429649353,
      "learning_rate": 0.00014646256876345371,
      "loss": 0.0379,
      "step": 11204
    },
    {
      "epoch": 1.570427470217239,
      "grad_norm": 0.47847968339920044,
      "learning_rate": 0.00014644821812963404,
      "loss": 0.0801,
      "step": 11205
    },
    {
      "epoch": 1.5705676243868254,
      "grad_norm": 0.24556659162044525,
      "learning_rate": 0.00014643386749581437,
      "loss": 0.0526,
      "step": 11206
    },
    {
      "epoch": 1.570707778556412,
      "grad_norm": 0.42434048652648926,
      "learning_rate": 0.00014641951686199473,
      "loss": 0.101,
      "step": 11207
    },
    {
      "epoch": 1.5708479327259988,
      "grad_norm": 0.238752543926239,
      "learning_rate": 0.00014640516622817506,
      "loss": 0.0565,
      "step": 11208
    },
    {
      "epoch": 1.5709880868955852,
      "grad_norm": 0.1340286135673523,
      "learning_rate": 0.0001463908155943554,
      "loss": 0.0114,
      "step": 11209
    },
    {
      "epoch": 1.5711282410651717,
      "grad_norm": 0.6875956654548645,
      "learning_rate": 0.00014637646496053574,
      "loss": 0.0261,
      "step": 11210
    },
    {
      "epoch": 1.5712683952347581,
      "grad_norm": 0.3364204466342926,
      "learning_rate": 0.00014636211432671607,
      "loss": 0.0884,
      "step": 11211
    },
    {
      "epoch": 1.5714085494043448,
      "grad_norm": 0.09406965970993042,
      "learning_rate": 0.00014634776369289642,
      "loss": 0.0156,
      "step": 11212
    },
    {
      "epoch": 1.5715487035739313,
      "grad_norm": 0.06484595686197281,
      "learning_rate": 0.00014633341305907678,
      "loss": 0.0062,
      "step": 11213
    },
    {
      "epoch": 1.571688857743518,
      "grad_norm": 0.5709916353225708,
      "learning_rate": 0.0001463190624252571,
      "loss": 0.0534,
      "step": 11214
    },
    {
      "epoch": 1.5718290119131044,
      "grad_norm": 0.3590030074119568,
      "learning_rate": 0.00014630471179143744,
      "loss": 0.0435,
      "step": 11215
    },
    {
      "epoch": 1.5719691660826909,
      "grad_norm": 0.2900840938091278,
      "learning_rate": 0.00014629036115761777,
      "loss": 0.0423,
      "step": 11216
    },
    {
      "epoch": 1.5721093202522773,
      "grad_norm": 0.5910401940345764,
      "learning_rate": 0.00014627601052379812,
      "loss": 0.0852,
      "step": 11217
    },
    {
      "epoch": 1.572249474421864,
      "grad_norm": 2.503844976425171,
      "learning_rate": 0.00014626165988997848,
      "loss": 0.0583,
      "step": 11218
    },
    {
      "epoch": 1.5723896285914507,
      "grad_norm": 0.07830420136451721,
      "learning_rate": 0.0001462473092561588,
      "loss": 0.006,
      "step": 11219
    },
    {
      "epoch": 1.5725297827610372,
      "grad_norm": 0.2498377561569214,
      "learning_rate": 0.00014623295862233913,
      "loss": 0.0825,
      "step": 11220
    },
    {
      "epoch": 1.5726699369306236,
      "grad_norm": 0.7405567765235901,
      "learning_rate": 0.0001462186079885195,
      "loss": 0.0192,
      "step": 11221
    },
    {
      "epoch": 1.57281009110021,
      "grad_norm": 0.18399973213672638,
      "learning_rate": 0.00014620425735469982,
      "loss": 0.032,
      "step": 11222
    },
    {
      "epoch": 1.5729502452697968,
      "grad_norm": 0.33848488330841064,
      "learning_rate": 0.00014618990672088017,
      "loss": 0.0429,
      "step": 11223
    },
    {
      "epoch": 1.5730903994393834,
      "grad_norm": 0.5522069334983826,
      "learning_rate": 0.0001461755560870605,
      "loss": 0.0577,
      "step": 11224
    },
    {
      "epoch": 1.57323055360897,
      "grad_norm": 0.5384527444839478,
      "learning_rate": 0.00014616120545324083,
      "loss": 0.1057,
      "step": 11225
    },
    {
      "epoch": 1.5733707077785564,
      "grad_norm": 0.4417308270931244,
      "learning_rate": 0.0001461468548194212,
      "loss": 0.017,
      "step": 11226
    },
    {
      "epoch": 1.5735108619481428,
      "grad_norm": 0.3732277452945709,
      "learning_rate": 0.00014613250418560152,
      "loss": 0.0544,
      "step": 11227
    },
    {
      "epoch": 1.5736510161177295,
      "grad_norm": 1.7595444917678833,
      "learning_rate": 0.00014611815355178187,
      "loss": 0.2268,
      "step": 11228
    },
    {
      "epoch": 1.5737911702873162,
      "grad_norm": 0.24302317202091217,
      "learning_rate": 0.0001461038029179622,
      "loss": 0.0359,
      "step": 11229
    },
    {
      "epoch": 1.5739313244569026,
      "grad_norm": 0.24950161576271057,
      "learning_rate": 0.00014608945228414253,
      "loss": 0.0433,
      "step": 11230
    },
    {
      "epoch": 1.574071478626489,
      "grad_norm": 1.941673994064331,
      "learning_rate": 0.00014607510165032288,
      "loss": 0.0538,
      "step": 11231
    },
    {
      "epoch": 1.5742116327960756,
      "grad_norm": 1.4476665258407593,
      "learning_rate": 0.0001460607510165032,
      "loss": 0.12,
      "step": 11232
    },
    {
      "epoch": 1.5743517869656622,
      "grad_norm": 0.214619442820549,
      "learning_rate": 0.00014604640038268354,
      "loss": 0.0121,
      "step": 11233
    },
    {
      "epoch": 1.574491941135249,
      "grad_norm": 0.3862488567829132,
      "learning_rate": 0.0001460320497488639,
      "loss": 0.0177,
      "step": 11234
    },
    {
      "epoch": 1.5746320953048354,
      "grad_norm": 1.9196295738220215,
      "learning_rate": 0.00014601769911504423,
      "loss": 0.2218,
      "step": 11235
    },
    {
      "epoch": 1.5747722494744218,
      "grad_norm": 0.2490377426147461,
      "learning_rate": 0.00014600334848122458,
      "loss": 0.0575,
      "step": 11236
    },
    {
      "epoch": 1.5749124036440083,
      "grad_norm": 0.14667293429374695,
      "learning_rate": 0.0001459889978474049,
      "loss": 0.043,
      "step": 11237
    },
    {
      "epoch": 1.575052557813595,
      "grad_norm": 0.20396338403224945,
      "learning_rate": 0.00014597464721358524,
      "loss": 0.056,
      "step": 11238
    },
    {
      "epoch": 1.5751927119831814,
      "grad_norm": 0.35358285903930664,
      "learning_rate": 0.0001459602965797656,
      "loss": 0.0539,
      "step": 11239
    },
    {
      "epoch": 1.5753328661527681,
      "grad_norm": 0.3170239329338074,
      "learning_rate": 0.00014594594594594595,
      "loss": 0.0527,
      "step": 11240
    },
    {
      "epoch": 1.5754730203223546,
      "grad_norm": 0.6362709403038025,
      "learning_rate": 0.00014593159531212628,
      "loss": 0.1129,
      "step": 11241
    },
    {
      "epoch": 1.575613174491941,
      "grad_norm": 0.22644935548305511,
      "learning_rate": 0.0001459172446783066,
      "loss": 0.0312,
      "step": 11242
    },
    {
      "epoch": 1.5757533286615277,
      "grad_norm": 0.3139689266681671,
      "learning_rate": 0.00014590289404448693,
      "loss": 0.0433,
      "step": 11243
    },
    {
      "epoch": 1.5758934828311142,
      "grad_norm": 0.4308244287967682,
      "learning_rate": 0.0001458885434106673,
      "loss": 0.0526,
      "step": 11244
    },
    {
      "epoch": 1.5760336370007009,
      "grad_norm": 0.2084711641073227,
      "learning_rate": 0.00014587419277684765,
      "loss": 0.0602,
      "step": 11245
    },
    {
      "epoch": 1.5761737911702873,
      "grad_norm": 0.32235977053642273,
      "learning_rate": 0.00014585984214302797,
      "loss": 0.0602,
      "step": 11246
    },
    {
      "epoch": 1.5763139453398738,
      "grad_norm": 0.4344325065612793,
      "learning_rate": 0.0001458454915092083,
      "loss": 0.0296,
      "step": 11247
    },
    {
      "epoch": 1.5764540995094602,
      "grad_norm": 0.39458155632019043,
      "learning_rate": 0.00014583114087538866,
      "loss": 0.0616,
      "step": 11248
    },
    {
      "epoch": 1.576594253679047,
      "grad_norm": 0.5368797183036804,
      "learning_rate": 0.000145816790241569,
      "loss": 0.0771,
      "step": 11249
    },
    {
      "epoch": 1.5767344078486336,
      "grad_norm": 0.3535921573638916,
      "learning_rate": 0.00014580243960774934,
      "loss": 0.0309,
      "step": 11250
    },
    {
      "epoch": 1.57687456201822,
      "grad_norm": 0.12445615231990814,
      "learning_rate": 0.00014578808897392967,
      "loss": 0.0183,
      "step": 11251
    },
    {
      "epoch": 1.5770147161878065,
      "grad_norm": 0.11979249119758606,
      "learning_rate": 0.00014577373834011,
      "loss": 0.0185,
      "step": 11252
    },
    {
      "epoch": 1.577154870357393,
      "grad_norm": 0.38393333554267883,
      "learning_rate": 0.00014575938770629036,
      "loss": 0.0462,
      "step": 11253
    },
    {
      "epoch": 1.5772950245269797,
      "grad_norm": 0.5327111482620239,
      "learning_rate": 0.00014574503707247068,
      "loss": 0.0544,
      "step": 11254
    },
    {
      "epoch": 1.5774351786965664,
      "grad_norm": 0.5200344324111938,
      "learning_rate": 0.00014573068643865104,
      "loss": 0.0904,
      "step": 11255
    },
    {
      "epoch": 1.5775753328661528,
      "grad_norm": 0.4005781412124634,
      "learning_rate": 0.00014571633580483137,
      "loss": 0.0426,
      "step": 11256
    },
    {
      "epoch": 1.5777154870357393,
      "grad_norm": 0.4025375247001648,
      "learning_rate": 0.0001457019851710117,
      "loss": 0.0424,
      "step": 11257
    },
    {
      "epoch": 1.5778556412053257,
      "grad_norm": 0.34456416964530945,
      "learning_rate": 0.00014568763453719205,
      "loss": 0.0383,
      "step": 11258
    },
    {
      "epoch": 1.5779957953749124,
      "grad_norm": 0.2475968450307846,
      "learning_rate": 0.00014567328390337238,
      "loss": 0.0354,
      "step": 11259
    },
    {
      "epoch": 1.578135949544499,
      "grad_norm": 0.07748182862997055,
      "learning_rate": 0.00014565893326955274,
      "loss": 0.0098,
      "step": 11260
    },
    {
      "epoch": 1.5782761037140856,
      "grad_norm": 0.36885011196136475,
      "learning_rate": 0.00014564458263573307,
      "loss": 0.0618,
      "step": 11261
    },
    {
      "epoch": 1.578416257883672,
      "grad_norm": 0.258219450712204,
      "learning_rate": 0.0001456302320019134,
      "loss": 0.0794,
      "step": 11262
    },
    {
      "epoch": 1.5785564120532585,
      "grad_norm": 0.26139187812805176,
      "learning_rate": 0.00014561588136809375,
      "loss": 0.0181,
      "step": 11263
    },
    {
      "epoch": 1.5786965662228452,
      "grad_norm": 0.2537481486797333,
      "learning_rate": 0.00014560153073427408,
      "loss": 0.0884,
      "step": 11264
    },
    {
      "epoch": 1.5788367203924318,
      "grad_norm": 0.12628164887428284,
      "learning_rate": 0.0001455871801004544,
      "loss": 0.025,
      "step": 11265
    },
    {
      "epoch": 1.5789768745620183,
      "grad_norm": 0.34642693400382996,
      "learning_rate": 0.00014557282946663476,
      "loss": 0.0447,
      "step": 11266
    },
    {
      "epoch": 1.5791170287316048,
      "grad_norm": 0.32561546564102173,
      "learning_rate": 0.0001455584788328151,
      "loss": 0.0825,
      "step": 11267
    },
    {
      "epoch": 1.5792571829011912,
      "grad_norm": 0.42550963163375854,
      "learning_rate": 0.00014554412819899545,
      "loss": 0.0422,
      "step": 11268
    },
    {
      "epoch": 1.579397337070778,
      "grad_norm": 0.4909414052963257,
      "learning_rate": 0.00014552977756517578,
      "loss": 0.0591,
      "step": 11269
    },
    {
      "epoch": 1.5795374912403644,
      "grad_norm": 0.5342494249343872,
      "learning_rate": 0.0001455154269313561,
      "loss": 0.0544,
      "step": 11270
    },
    {
      "epoch": 1.579677645409951,
      "grad_norm": 0.44271159172058105,
      "learning_rate": 0.00014550107629753646,
      "loss": 0.0579,
      "step": 11271
    },
    {
      "epoch": 1.5798177995795375,
      "grad_norm": 0.678312361240387,
      "learning_rate": 0.00014548672566371682,
      "loss": 0.0333,
      "step": 11272
    },
    {
      "epoch": 1.579957953749124,
      "grad_norm": 0.32278138399124146,
      "learning_rate": 0.00014547237502989714,
      "loss": 0.0422,
      "step": 11273
    },
    {
      "epoch": 1.5800981079187106,
      "grad_norm": 0.551562488079071,
      "learning_rate": 0.00014545802439607747,
      "loss": 0.028,
      "step": 11274
    },
    {
      "epoch": 1.580238262088297,
      "grad_norm": 1.3201539516448975,
      "learning_rate": 0.00014544367376225783,
      "loss": 0.0868,
      "step": 11275
    },
    {
      "epoch": 1.5803784162578838,
      "grad_norm": 0.2706214189529419,
      "learning_rate": 0.00014542932312843816,
      "loss": 0.0395,
      "step": 11276
    },
    {
      "epoch": 1.5805185704274702,
      "grad_norm": 0.2803541123867035,
      "learning_rate": 0.0001454149724946185,
      "loss": 0.0357,
      "step": 11277
    },
    {
      "epoch": 1.5806587245970567,
      "grad_norm": 0.9173793792724609,
      "learning_rate": 0.00014540062186079884,
      "loss": 0.1996,
      "step": 11278
    },
    {
      "epoch": 1.5807988787666432,
      "grad_norm": 0.2926436960697174,
      "learning_rate": 0.00014538627122697917,
      "loss": 0.0626,
      "step": 11279
    },
    {
      "epoch": 1.5809390329362298,
      "grad_norm": 0.9522509574890137,
      "learning_rate": 0.00014537192059315953,
      "loss": 0.0841,
      "step": 11280
    },
    {
      "epoch": 1.5810791871058165,
      "grad_norm": 0.5323994755744934,
      "learning_rate": 0.00014535756995933985,
      "loss": 0.0709,
      "step": 11281
    },
    {
      "epoch": 1.581219341275403,
      "grad_norm": 0.4189034700393677,
      "learning_rate": 0.0001453432193255202,
      "loss": 0.0764,
      "step": 11282
    },
    {
      "epoch": 1.5813594954449894,
      "grad_norm": 0.1715243011713028,
      "learning_rate": 0.00014532886869170054,
      "loss": 0.0139,
      "step": 11283
    },
    {
      "epoch": 1.581499649614576,
      "grad_norm": 0.7849209904670715,
      "learning_rate": 0.00014531451805788087,
      "loss": 0.0433,
      "step": 11284
    },
    {
      "epoch": 1.5816398037841626,
      "grad_norm": 0.7326869368553162,
      "learning_rate": 0.00014530016742406122,
      "loss": 0.0217,
      "step": 11285
    },
    {
      "epoch": 1.5817799579537493,
      "grad_norm": 0.20545487105846405,
      "learning_rate": 0.00014528581679024155,
      "loss": 0.0177,
      "step": 11286
    },
    {
      "epoch": 1.5819201121233357,
      "grad_norm": 0.28836357593536377,
      "learning_rate": 0.0001452714661564219,
      "loss": 0.0418,
      "step": 11287
    },
    {
      "epoch": 1.5820602662929222,
      "grad_norm": 0.38304904103279114,
      "learning_rate": 0.00014525711552260224,
      "loss": 0.0495,
      "step": 11288
    },
    {
      "epoch": 1.5822004204625086,
      "grad_norm": 0.3287407457828522,
      "learning_rate": 0.00014524276488878256,
      "loss": 0.0342,
      "step": 11289
    },
    {
      "epoch": 1.5823405746320953,
      "grad_norm": 0.358939528465271,
      "learning_rate": 0.00014522841425496292,
      "loss": 0.0388,
      "step": 11290
    },
    {
      "epoch": 1.582480728801682,
      "grad_norm": 0.42045775055885315,
      "learning_rate": 0.00014521406362114328,
      "loss": 0.0647,
      "step": 11291
    },
    {
      "epoch": 1.5826208829712685,
      "grad_norm": 0.6764708161354065,
      "learning_rate": 0.0001451997129873236,
      "loss": 0.1026,
      "step": 11292
    },
    {
      "epoch": 1.582761037140855,
      "grad_norm": 0.6854547262191772,
      "learning_rate": 0.00014518536235350393,
      "loss": 0.0509,
      "step": 11293
    },
    {
      "epoch": 1.5829011913104414,
      "grad_norm": 0.4909321665763855,
      "learning_rate": 0.00014517101171968426,
      "loss": 0.0379,
      "step": 11294
    },
    {
      "epoch": 1.583041345480028,
      "grad_norm": 0.4555673599243164,
      "learning_rate": 0.00014515666108586462,
      "loss": 0.0872,
      "step": 11295
    },
    {
      "epoch": 1.5831814996496145,
      "grad_norm": 0.08482465893030167,
      "learning_rate": 0.00014514231045204494,
      "loss": 0.0153,
      "step": 11296
    },
    {
      "epoch": 1.5833216538192012,
      "grad_norm": 0.5087714791297913,
      "learning_rate": 0.00014512795981822527,
      "loss": 0.1572,
      "step": 11297
    },
    {
      "epoch": 1.5834618079887877,
      "grad_norm": 0.5906250476837158,
      "learning_rate": 0.00014511360918440563,
      "loss": 0.0872,
      "step": 11298
    },
    {
      "epoch": 1.5836019621583741,
      "grad_norm": 0.2855149209499359,
      "learning_rate": 0.00014509925855058598,
      "loss": 0.031,
      "step": 11299
    },
    {
      "epoch": 1.5837421163279608,
      "grad_norm": 0.27233362197875977,
      "learning_rate": 0.0001450849079167663,
      "loss": 0.0327,
      "step": 11300
    },
    {
      "epoch": 1.5838822704975473,
      "grad_norm": 0.9900158643722534,
      "learning_rate": 0.00014507055728294664,
      "loss": 0.1225,
      "step": 11301
    },
    {
      "epoch": 1.584022424667134,
      "grad_norm": 0.27581602334976196,
      "learning_rate": 0.00014505620664912697,
      "loss": 0.0717,
      "step": 11302
    },
    {
      "epoch": 1.5841625788367204,
      "grad_norm": 0.4488271176815033,
      "learning_rate": 0.00014504185601530733,
      "loss": 0.1049,
      "step": 11303
    },
    {
      "epoch": 1.5843027330063069,
      "grad_norm": 0.24682290852069855,
      "learning_rate": 0.00014502750538148768,
      "loss": 0.0366,
      "step": 11304
    },
    {
      "epoch": 1.5844428871758933,
      "grad_norm": 0.46133553981781006,
      "learning_rate": 0.000145013154747668,
      "loss": 0.0565,
      "step": 11305
    },
    {
      "epoch": 1.58458304134548,
      "grad_norm": 0.3794057071208954,
      "learning_rate": 0.00014499880411384834,
      "loss": 0.0938,
      "step": 11306
    },
    {
      "epoch": 1.5847231955150667,
      "grad_norm": 0.2560756504535675,
      "learning_rate": 0.0001449844534800287,
      "loss": 0.0549,
      "step": 11307
    },
    {
      "epoch": 1.5848633496846531,
      "grad_norm": 0.5859915614128113,
      "learning_rate": 0.00014497010284620902,
      "loss": 0.0743,
      "step": 11308
    },
    {
      "epoch": 1.5850035038542396,
      "grad_norm": 0.21086852252483368,
      "learning_rate": 0.00014495575221238938,
      "loss": 0.0447,
      "step": 11309
    },
    {
      "epoch": 1.585143658023826,
      "grad_norm": 0.36471033096313477,
      "learning_rate": 0.0001449414015785697,
      "loss": 0.0466,
      "step": 11310
    },
    {
      "epoch": 1.5852838121934127,
      "grad_norm": 0.20248322188854218,
      "learning_rate": 0.00014492705094475004,
      "loss": 0.0298,
      "step": 11311
    },
    {
      "epoch": 1.5854239663629994,
      "grad_norm": 0.42613497376441956,
      "learning_rate": 0.0001449127003109304,
      "loss": 0.0233,
      "step": 11312
    },
    {
      "epoch": 1.5855641205325859,
      "grad_norm": 0.1068175882101059,
      "learning_rate": 0.00014489834967711072,
      "loss": 0.0081,
      "step": 11313
    },
    {
      "epoch": 1.5857042747021723,
      "grad_norm": 0.3122786581516266,
      "learning_rate": 0.00014488399904329108,
      "loss": 0.0509,
      "step": 11314
    },
    {
      "epoch": 1.5858444288717588,
      "grad_norm": 0.2238869071006775,
      "learning_rate": 0.0001448696484094714,
      "loss": 0.0258,
      "step": 11315
    },
    {
      "epoch": 1.5859845830413455,
      "grad_norm": 0.307205468416214,
      "learning_rate": 0.00014485529777565173,
      "loss": 0.0514,
      "step": 11316
    },
    {
      "epoch": 1.5861247372109322,
      "grad_norm": 0.30395668745040894,
      "learning_rate": 0.0001448409471418321,
      "loss": 0.1037,
      "step": 11317
    },
    {
      "epoch": 1.5862648913805186,
      "grad_norm": 0.20317557454109192,
      "learning_rate": 0.00014482659650801244,
      "loss": 0.0464,
      "step": 11318
    },
    {
      "epoch": 1.586405045550105,
      "grad_norm": 0.47860780358314514,
      "learning_rate": 0.00014481224587419277,
      "loss": 0.0232,
      "step": 11319
    },
    {
      "epoch": 1.5865451997196915,
      "grad_norm": 0.05480014905333519,
      "learning_rate": 0.0001447978952403731,
      "loss": 0.025,
      "step": 11320
    },
    {
      "epoch": 1.5866853538892782,
      "grad_norm": 0.3540691137313843,
      "learning_rate": 0.00014478354460655343,
      "loss": 0.0494,
      "step": 11321
    },
    {
      "epoch": 1.586825508058865,
      "grad_norm": 0.7096140384674072,
      "learning_rate": 0.00014476919397273379,
      "loss": 0.0641,
      "step": 11322
    },
    {
      "epoch": 1.5869656622284514,
      "grad_norm": 0.2946372330188751,
      "learning_rate": 0.00014475484333891414,
      "loss": 0.0548,
      "step": 11323
    },
    {
      "epoch": 1.5871058163980378,
      "grad_norm": 0.08551746606826782,
      "learning_rate": 0.00014474049270509447,
      "loss": 0.0155,
      "step": 11324
    },
    {
      "epoch": 1.5872459705676243,
      "grad_norm": 0.1938301920890808,
      "learning_rate": 0.0001447261420712748,
      "loss": 0.0152,
      "step": 11325
    },
    {
      "epoch": 1.587386124737211,
      "grad_norm": 0.2958170771598816,
      "learning_rate": 0.00014471179143745515,
      "loss": 0.0535,
      "step": 11326
    },
    {
      "epoch": 1.5875262789067974,
      "grad_norm": 0.24976089596748352,
      "learning_rate": 0.00014469744080363548,
      "loss": 0.0353,
      "step": 11327
    },
    {
      "epoch": 1.587666433076384,
      "grad_norm": 0.485668420791626,
      "learning_rate": 0.0001446830901698158,
      "loss": 0.0589,
      "step": 11328
    },
    {
      "epoch": 1.5878065872459706,
      "grad_norm": 0.20511524379253387,
      "learning_rate": 0.00014466873953599614,
      "loss": 0.0521,
      "step": 11329
    },
    {
      "epoch": 1.587946741415557,
      "grad_norm": 0.18602831661701202,
      "learning_rate": 0.0001446543889021765,
      "loss": 0.0199,
      "step": 11330
    },
    {
      "epoch": 1.5880868955851437,
      "grad_norm": 0.1648767590522766,
      "learning_rate": 0.00014464003826835685,
      "loss": 0.011,
      "step": 11331
    },
    {
      "epoch": 1.5882270497547302,
      "grad_norm": 0.3389259874820709,
      "learning_rate": 0.00014462568763453718,
      "loss": 0.0772,
      "step": 11332
    },
    {
      "epoch": 1.5883672039243169,
      "grad_norm": 0.9060108065605164,
      "learning_rate": 0.0001446113370007175,
      "loss": 0.0892,
      "step": 11333
    },
    {
      "epoch": 1.5885073580939033,
      "grad_norm": 1.392243504524231,
      "learning_rate": 0.00014459698636689786,
      "loss": 0.3211,
      "step": 11334
    },
    {
      "epoch": 1.5886475122634898,
      "grad_norm": 0.4251784384250641,
      "learning_rate": 0.0001445826357330782,
      "loss": 0.0475,
      "step": 11335
    },
    {
      "epoch": 1.5887876664330762,
      "grad_norm": 0.13818101584911346,
      "learning_rate": 0.00014456828509925855,
      "loss": 0.0147,
      "step": 11336
    },
    {
      "epoch": 1.588927820602663,
      "grad_norm": 0.16225974261760712,
      "learning_rate": 0.00014455393446543888,
      "loss": 0.0272,
      "step": 11337
    },
    {
      "epoch": 1.5890679747722496,
      "grad_norm": 0.43955108523368835,
      "learning_rate": 0.0001445395838316192,
      "loss": 0.0828,
      "step": 11338
    },
    {
      "epoch": 1.589208128941836,
      "grad_norm": 0.2266266793012619,
      "learning_rate": 0.00014452523319779956,
      "loss": 0.0594,
      "step": 11339
    },
    {
      "epoch": 1.5893482831114225,
      "grad_norm": 0.27316924929618835,
      "learning_rate": 0.0001445108825639799,
      "loss": 0.0506,
      "step": 11340
    },
    {
      "epoch": 1.589488437281009,
      "grad_norm": 0.689016580581665,
      "learning_rate": 0.00014449653193016025,
      "loss": 0.0749,
      "step": 11341
    },
    {
      "epoch": 1.5896285914505957,
      "grad_norm": 0.21097469329833984,
      "learning_rate": 0.00014448218129634057,
      "loss": 0.0211,
      "step": 11342
    },
    {
      "epoch": 1.5897687456201823,
      "grad_norm": 0.1962958723306656,
      "learning_rate": 0.0001444678306625209,
      "loss": 0.0234,
      "step": 11343
    },
    {
      "epoch": 1.5899088997897688,
      "grad_norm": 0.22709308564662933,
      "learning_rate": 0.00014445348002870126,
      "loss": 0.0238,
      "step": 11344
    },
    {
      "epoch": 1.5900490539593553,
      "grad_norm": 0.14744724333286285,
      "learning_rate": 0.0001444391293948816,
      "loss": 0.0165,
      "step": 11345
    },
    {
      "epoch": 1.5901892081289417,
      "grad_norm": 0.22451026737689972,
      "learning_rate": 0.00014442477876106194,
      "loss": 0.0453,
      "step": 11346
    },
    {
      "epoch": 1.5903293622985284,
      "grad_norm": 0.39369678497314453,
      "learning_rate": 0.00014441042812724227,
      "loss": 0.147,
      "step": 11347
    },
    {
      "epoch": 1.590469516468115,
      "grad_norm": 0.24836881458759308,
      "learning_rate": 0.0001443960774934226,
      "loss": 0.0387,
      "step": 11348
    },
    {
      "epoch": 1.5906096706377015,
      "grad_norm": 0.21216589212417603,
      "learning_rate": 0.00014438172685960295,
      "loss": 0.0453,
      "step": 11349
    },
    {
      "epoch": 1.590749824807288,
      "grad_norm": 0.4266994893550873,
      "learning_rate": 0.0001443673762257833,
      "loss": 0.0256,
      "step": 11350
    },
    {
      "epoch": 1.5908899789768745,
      "grad_norm": 0.23873865604400635,
      "learning_rate": 0.00014435302559196364,
      "loss": 0.0622,
      "step": 11351
    },
    {
      "epoch": 1.5910301331464611,
      "grad_norm": 0.45349276065826416,
      "learning_rate": 0.00014433867495814397,
      "loss": 0.0346,
      "step": 11352
    },
    {
      "epoch": 1.5911702873160478,
      "grad_norm": 0.2463388890028,
      "learning_rate": 0.00014432432432432432,
      "loss": 0.0357,
      "step": 11353
    },
    {
      "epoch": 1.5913104414856343,
      "grad_norm": 0.3385111391544342,
      "learning_rate": 0.00014430997369050465,
      "loss": 0.0581,
      "step": 11354
    },
    {
      "epoch": 1.5914505956552207,
      "grad_norm": 0.2506190836429596,
      "learning_rate": 0.000144295623056685,
      "loss": 0.0685,
      "step": 11355
    },
    {
      "epoch": 1.5915907498248072,
      "grad_norm": 0.33793550729751587,
      "learning_rate": 0.00014428127242286534,
      "loss": 0.0494,
      "step": 11356
    },
    {
      "epoch": 1.5917309039943939,
      "grad_norm": 0.5122058987617493,
      "learning_rate": 0.00014426692178904566,
      "loss": 0.0508,
      "step": 11357
    },
    {
      "epoch": 1.5918710581639803,
      "grad_norm": 0.6740719079971313,
      "learning_rate": 0.00014425257115522602,
      "loss": 0.0599,
      "step": 11358
    },
    {
      "epoch": 1.592011212333567,
      "grad_norm": 0.1437353640794754,
      "learning_rate": 0.00014423822052140635,
      "loss": 0.0238,
      "step": 11359
    },
    {
      "epoch": 1.5921513665031535,
      "grad_norm": 0.16703295707702637,
      "learning_rate": 0.00014422386988758668,
      "loss": 0.0181,
      "step": 11360
    },
    {
      "epoch": 1.59229152067274,
      "grad_norm": 0.2900306284427643,
      "learning_rate": 0.00014420951925376703,
      "loss": 0.0286,
      "step": 11361
    },
    {
      "epoch": 1.5924316748423264,
      "grad_norm": 0.1914912313222885,
      "learning_rate": 0.00014419516861994736,
      "loss": 0.067,
      "step": 11362
    },
    {
      "epoch": 1.592571829011913,
      "grad_norm": 0.2010813057422638,
      "learning_rate": 0.00014418081798612772,
      "loss": 0.0431,
      "step": 11363
    },
    {
      "epoch": 1.5927119831814998,
      "grad_norm": 0.2772805392742157,
      "learning_rate": 0.00014416646735230805,
      "loss": 0.0377,
      "step": 11364
    },
    {
      "epoch": 1.5928521373510862,
      "grad_norm": 0.4188619554042816,
      "learning_rate": 0.00014415211671848837,
      "loss": 0.06,
      "step": 11365
    },
    {
      "epoch": 1.5929922915206727,
      "grad_norm": 0.05867023766040802,
      "learning_rate": 0.00014413776608466873,
      "loss": 0.009,
      "step": 11366
    },
    {
      "epoch": 1.5931324456902591,
      "grad_norm": 0.4740632474422455,
      "learning_rate": 0.00014412341545084906,
      "loss": 0.0425,
      "step": 11367
    },
    {
      "epoch": 1.5932725998598458,
      "grad_norm": 0.19765959680080414,
      "learning_rate": 0.00014410906481702941,
      "loss": 0.0319,
      "step": 11368
    },
    {
      "epoch": 1.5934127540294325,
      "grad_norm": 0.399566113948822,
      "learning_rate": 0.00014409471418320974,
      "loss": 0.0519,
      "step": 11369
    },
    {
      "epoch": 1.593552908199019,
      "grad_norm": 0.7658405900001526,
      "learning_rate": 0.00014408036354939007,
      "loss": 0.1309,
      "step": 11370
    },
    {
      "epoch": 1.5936930623686054,
      "grad_norm": 0.10469689965248108,
      "learning_rate": 0.00014406601291557043,
      "loss": 0.0142,
      "step": 11371
    },
    {
      "epoch": 1.5938332165381919,
      "grad_norm": 0.4584612250328064,
      "learning_rate": 0.00014405166228175076,
      "loss": 0.0664,
      "step": 11372
    },
    {
      "epoch": 1.5939733707077786,
      "grad_norm": 0.5687464475631714,
      "learning_rate": 0.0001440373116479311,
      "loss": 0.0527,
      "step": 11373
    },
    {
      "epoch": 1.5941135248773652,
      "grad_norm": 0.5980580449104309,
      "learning_rate": 0.00014402296101411144,
      "loss": 0.0738,
      "step": 11374
    },
    {
      "epoch": 1.5942536790469517,
      "grad_norm": 0.525931179523468,
      "learning_rate": 0.00014400861038029177,
      "loss": 0.0695,
      "step": 11375
    },
    {
      "epoch": 1.5943938332165382,
      "grad_norm": 0.5981613993644714,
      "learning_rate": 0.00014399425974647212,
      "loss": 0.0297,
      "step": 11376
    },
    {
      "epoch": 1.5945339873861246,
      "grad_norm": 0.32686084508895874,
      "learning_rate": 0.00014397990911265248,
      "loss": 0.0537,
      "step": 11377
    },
    {
      "epoch": 1.5946741415557113,
      "grad_norm": 0.36781010031700134,
      "learning_rate": 0.0001439655584788328,
      "loss": 0.1101,
      "step": 11378
    },
    {
      "epoch": 1.594814295725298,
      "grad_norm": 0.33023348450660706,
      "learning_rate": 0.00014395120784501314,
      "loss": 0.0192,
      "step": 11379
    },
    {
      "epoch": 1.5949544498948844,
      "grad_norm": 0.2836973965167999,
      "learning_rate": 0.00014393685721119347,
      "loss": 0.0745,
      "step": 11380
    },
    {
      "epoch": 1.595094604064471,
      "grad_norm": 0.08225195854902267,
      "learning_rate": 0.00014392250657737382,
      "loss": 0.005,
      "step": 11381
    },
    {
      "epoch": 1.5952347582340574,
      "grad_norm": 0.5242540836334229,
      "learning_rate": 0.00014390815594355418,
      "loss": 0.0197,
      "step": 11382
    },
    {
      "epoch": 1.595374912403644,
      "grad_norm": 0.1624547392129898,
      "learning_rate": 0.0001438938053097345,
      "loss": 0.01,
      "step": 11383
    },
    {
      "epoch": 1.5955150665732305,
      "grad_norm": 0.4167322814464569,
      "learning_rate": 0.00014387945467591483,
      "loss": 0.0788,
      "step": 11384
    },
    {
      "epoch": 1.5956552207428172,
      "grad_norm": 3.375951051712036,
      "learning_rate": 0.0001438651040420952,
      "loss": 0.1986,
      "step": 11385
    },
    {
      "epoch": 1.5957953749124036,
      "grad_norm": 0.23198918998241425,
      "learning_rate": 0.00014385075340827552,
      "loss": 0.022,
      "step": 11386
    },
    {
      "epoch": 1.59593552908199,
      "grad_norm": 0.26288193464279175,
      "learning_rate": 0.00014383640277445587,
      "loss": 0.0257,
      "step": 11387
    },
    {
      "epoch": 1.5960756832515768,
      "grad_norm": 0.19199809432029724,
      "learning_rate": 0.0001438220521406362,
      "loss": 0.038,
      "step": 11388
    },
    {
      "epoch": 1.5962158374211632,
      "grad_norm": 0.25046664476394653,
      "learning_rate": 0.00014380770150681653,
      "loss": 0.0625,
      "step": 11389
    },
    {
      "epoch": 1.59635599159075,
      "grad_norm": 0.3365858495235443,
      "learning_rate": 0.0001437933508729969,
      "loss": 0.0897,
      "step": 11390
    },
    {
      "epoch": 1.5964961457603364,
      "grad_norm": 0.13023974001407623,
      "learning_rate": 0.00014377900023917722,
      "loss": 0.031,
      "step": 11391
    },
    {
      "epoch": 1.5966362999299228,
      "grad_norm": 0.356695294380188,
      "learning_rate": 0.00014376464960535754,
      "loss": 0.0335,
      "step": 11392
    },
    {
      "epoch": 1.5967764540995093,
      "grad_norm": 0.2053287923336029,
      "learning_rate": 0.0001437502989715379,
      "loss": 0.0578,
      "step": 11393
    },
    {
      "epoch": 1.596916608269096,
      "grad_norm": 0.45393800735473633,
      "learning_rate": 0.00014373594833771823,
      "loss": 0.0796,
      "step": 11394
    },
    {
      "epoch": 1.5970567624386827,
      "grad_norm": 0.2853446304798126,
      "learning_rate": 0.00014372159770389858,
      "loss": 0.0357,
      "step": 11395
    },
    {
      "epoch": 1.5971969166082691,
      "grad_norm": 0.25550398230552673,
      "learning_rate": 0.0001437072470700789,
      "loss": 0.0839,
      "step": 11396
    },
    {
      "epoch": 1.5973370707778556,
      "grad_norm": 0.13601426780223846,
      "learning_rate": 0.00014369289643625924,
      "loss": 0.0149,
      "step": 11397
    },
    {
      "epoch": 1.597477224947442,
      "grad_norm": 0.1922236680984497,
      "learning_rate": 0.0001436785458024396,
      "loss": 0.028,
      "step": 11398
    },
    {
      "epoch": 1.5976173791170287,
      "grad_norm": 0.3007862865924835,
      "learning_rate": 0.00014366419516861993,
      "loss": 0.0397,
      "step": 11399
    },
    {
      "epoch": 1.5977575332866154,
      "grad_norm": 0.20555178821086884,
      "learning_rate": 0.00014364984453480028,
      "loss": 0.0096,
      "step": 11400
    },
    {
      "epoch": 1.5978976874562019,
      "grad_norm": 0.40918028354644775,
      "learning_rate": 0.0001436354939009806,
      "loss": 0.0468,
      "step": 11401
    },
    {
      "epoch": 1.5980378416257883,
      "grad_norm": 0.24805596470832825,
      "learning_rate": 0.00014362114326716094,
      "loss": 0.056,
      "step": 11402
    },
    {
      "epoch": 1.5981779957953748,
      "grad_norm": 0.2547355592250824,
      "learning_rate": 0.0001436067926333413,
      "loss": 0.0548,
      "step": 11403
    },
    {
      "epoch": 1.5983181499649615,
      "grad_norm": 0.47048670053482056,
      "learning_rate": 0.00014359244199952165,
      "loss": 0.0667,
      "step": 11404
    },
    {
      "epoch": 1.5984583041345481,
      "grad_norm": 0.48704493045806885,
      "learning_rate": 0.00014357809136570198,
      "loss": 0.0786,
      "step": 11405
    },
    {
      "epoch": 1.5985984583041346,
      "grad_norm": 0.14439354836940765,
      "learning_rate": 0.0001435637407318823,
      "loss": 0.042,
      "step": 11406
    },
    {
      "epoch": 1.598738612473721,
      "grad_norm": 0.13267837464809418,
      "learning_rate": 0.00014354939009806263,
      "loss": 0.0127,
      "step": 11407
    },
    {
      "epoch": 1.5988787666433075,
      "grad_norm": 0.772402286529541,
      "learning_rate": 0.000143535039464243,
      "loss": 0.0622,
      "step": 11408
    },
    {
      "epoch": 1.5990189208128942,
      "grad_norm": 0.12240627408027649,
      "learning_rate": 0.00014352068883042335,
      "loss": 0.0073,
      "step": 11409
    },
    {
      "epoch": 1.5991590749824809,
      "grad_norm": 0.32743585109710693,
      "learning_rate": 0.00014350633819660367,
      "loss": 0.0305,
      "step": 11410
    },
    {
      "epoch": 1.5992992291520673,
      "grad_norm": 0.19536246359348297,
      "learning_rate": 0.000143491987562784,
      "loss": 0.0206,
      "step": 11411
    },
    {
      "epoch": 1.5994393833216538,
      "grad_norm": 0.16025663912296295,
      "learning_rate": 0.00014347763692896436,
      "loss": 0.0292,
      "step": 11412
    },
    {
      "epoch": 1.5995795374912403,
      "grad_norm": 0.32042425870895386,
      "learning_rate": 0.0001434632862951447,
      "loss": 0.0394,
      "step": 11413
    },
    {
      "epoch": 1.599719691660827,
      "grad_norm": 0.2431601583957672,
      "learning_rate": 0.00014344893566132504,
      "loss": 0.0357,
      "step": 11414
    },
    {
      "epoch": 1.5998598458304134,
      "grad_norm": 0.4907341003417969,
      "learning_rate": 0.00014343458502750537,
      "loss": 0.042,
      "step": 11415
    },
    {
      "epoch": 1.6,
      "grad_norm": 0.7053388953208923,
      "learning_rate": 0.0001434202343936857,
      "loss": 0.0561,
      "step": 11416
    },
    {
      "epoch": 1.6001401541695865,
      "grad_norm": 0.3977697193622589,
      "learning_rate": 0.00014340588375986606,
      "loss": 0.0364,
      "step": 11417
    },
    {
      "epoch": 1.600280308339173,
      "grad_norm": 0.3205070495605469,
      "learning_rate": 0.00014339153312604638,
      "loss": 0.0907,
      "step": 11418
    },
    {
      "epoch": 1.6004204625087597,
      "grad_norm": 0.39762935042381287,
      "learning_rate": 0.00014337718249222674,
      "loss": 0.1301,
      "step": 11419
    },
    {
      "epoch": 1.6005606166783461,
      "grad_norm": 0.29576027393341064,
      "learning_rate": 0.00014336283185840707,
      "loss": 0.0572,
      "step": 11420
    },
    {
      "epoch": 1.6007007708479328,
      "grad_norm": 0.47650256752967834,
      "learning_rate": 0.0001433484812245874,
      "loss": 0.0958,
      "step": 11421
    },
    {
      "epoch": 1.6008409250175193,
      "grad_norm": 0.3434876501560211,
      "learning_rate": 0.00014333413059076775,
      "loss": 0.0174,
      "step": 11422
    },
    {
      "epoch": 1.6009810791871057,
      "grad_norm": 0.8974648118019104,
      "learning_rate": 0.00014331977995694808,
      "loss": 0.0631,
      "step": 11423
    },
    {
      "epoch": 1.6011212333566922,
      "grad_norm": 0.5391984581947327,
      "learning_rate": 0.0001433054293231284,
      "loss": 0.0246,
      "step": 11424
    },
    {
      "epoch": 1.601261387526279,
      "grad_norm": 0.3260977864265442,
      "learning_rate": 0.00014329107868930877,
      "loss": 0.0336,
      "step": 11425
    },
    {
      "epoch": 1.6014015416958656,
      "grad_norm": 0.4213988780975342,
      "learning_rate": 0.0001432767280554891,
      "loss": 0.0524,
      "step": 11426
    },
    {
      "epoch": 1.601541695865452,
      "grad_norm": 0.4234372675418854,
      "learning_rate": 0.00014326237742166945,
      "loss": 0.1597,
      "step": 11427
    },
    {
      "epoch": 1.6016818500350385,
      "grad_norm": 0.30518388748168945,
      "learning_rate": 0.00014324802678784978,
      "loss": 0.0128,
      "step": 11428
    },
    {
      "epoch": 1.601822004204625,
      "grad_norm": 0.2911543548107147,
      "learning_rate": 0.0001432336761540301,
      "loss": 0.0417,
      "step": 11429
    },
    {
      "epoch": 1.6019621583742116,
      "grad_norm": 0.47299671173095703,
      "learning_rate": 0.00014321932552021046,
      "loss": 0.0709,
      "step": 11430
    },
    {
      "epoch": 1.6021023125437983,
      "grad_norm": 0.47777125239372253,
      "learning_rate": 0.00014320497488639082,
      "loss": 0.0754,
      "step": 11431
    },
    {
      "epoch": 1.6022424667133848,
      "grad_norm": 0.37924832105636597,
      "learning_rate": 0.00014319062425257115,
      "loss": 0.0497,
      "step": 11432
    },
    {
      "epoch": 1.6023826208829712,
      "grad_norm": 2.9140608310699463,
      "learning_rate": 0.00014317627361875148,
      "loss": 0.2088,
      "step": 11433
    },
    {
      "epoch": 1.6025227750525577,
      "grad_norm": 0.6861683130264282,
      "learning_rate": 0.0001431619229849318,
      "loss": 0.0866,
      "step": 11434
    },
    {
      "epoch": 1.6026629292221444,
      "grad_norm": 1.7493194341659546,
      "learning_rate": 0.00014314757235111216,
      "loss": 0.4043,
      "step": 11435
    },
    {
      "epoch": 1.602803083391731,
      "grad_norm": 0.43702423572540283,
      "learning_rate": 0.00014313322171729252,
      "loss": 0.0553,
      "step": 11436
    },
    {
      "epoch": 1.6029432375613175,
      "grad_norm": 0.1277390569448471,
      "learning_rate": 0.00014311887108347284,
      "loss": 0.0148,
      "step": 11437
    },
    {
      "epoch": 1.603083391730904,
      "grad_norm": 0.28713348507881165,
      "learning_rate": 0.00014310452044965317,
      "loss": 0.066,
      "step": 11438
    },
    {
      "epoch": 1.6032235459004904,
      "grad_norm": 0.2675716280937195,
      "learning_rate": 0.00014309016981583353,
      "loss": 0.0452,
      "step": 11439
    },
    {
      "epoch": 1.6033637000700771,
      "grad_norm": 0.19364652037620544,
      "learning_rate": 0.00014307581918201386,
      "loss": 0.0447,
      "step": 11440
    },
    {
      "epoch": 1.6035038542396638,
      "grad_norm": 0.5300313234329224,
      "learning_rate": 0.0001430614685481942,
      "loss": 0.0489,
      "step": 11441
    },
    {
      "epoch": 1.6036440084092503,
      "grad_norm": 0.25991445779800415,
      "learning_rate": 0.00014304711791437454,
      "loss": 0.0835,
      "step": 11442
    },
    {
      "epoch": 1.6037841625788367,
      "grad_norm": 0.4901083707809448,
      "learning_rate": 0.00014303276728055487,
      "loss": 0.0352,
      "step": 11443
    },
    {
      "epoch": 1.6039243167484232,
      "grad_norm": 0.37696927785873413,
      "learning_rate": 0.00014301841664673523,
      "loss": 0.0397,
      "step": 11444
    },
    {
      "epoch": 1.6040644709180099,
      "grad_norm": 0.215261310338974,
      "learning_rate": 0.00014300406601291555,
      "loss": 0.0434,
      "step": 11445
    },
    {
      "epoch": 1.6042046250875963,
      "grad_norm": 0.5213586091995239,
      "learning_rate": 0.0001429897153790959,
      "loss": 0.0566,
      "step": 11446
    },
    {
      "epoch": 1.604344779257183,
      "grad_norm": 0.35481783747673035,
      "learning_rate": 0.00014297536474527624,
      "loss": 0.0821,
      "step": 11447
    },
    {
      "epoch": 1.6044849334267695,
      "grad_norm": 0.2919999659061432,
      "learning_rate": 0.00014296101411145657,
      "loss": 0.0557,
      "step": 11448
    },
    {
      "epoch": 1.604625087596356,
      "grad_norm": 0.18999779224395752,
      "learning_rate": 0.00014294666347763692,
      "loss": 0.0695,
      "step": 11449
    },
    {
      "epoch": 1.6047652417659424,
      "grad_norm": 0.5571599006652832,
      "learning_rate": 0.00014293231284381725,
      "loss": 0.1317,
      "step": 11450
    },
    {
      "epoch": 1.604905395935529,
      "grad_norm": 0.13731858134269714,
      "learning_rate": 0.0001429179622099976,
      "loss": 0.0322,
      "step": 11451
    },
    {
      "epoch": 1.6050455501051157,
      "grad_norm": 0.4020020663738251,
      "learning_rate": 0.00014290361157617794,
      "loss": 0.0509,
      "step": 11452
    },
    {
      "epoch": 1.6051857042747022,
      "grad_norm": 0.7551411986351013,
      "learning_rate": 0.00014288926094235826,
      "loss": 0.1063,
      "step": 11453
    },
    {
      "epoch": 1.6053258584442887,
      "grad_norm": 0.4349111318588257,
      "learning_rate": 0.00014287491030853862,
      "loss": 0.1069,
      "step": 11454
    },
    {
      "epoch": 1.6054660126138751,
      "grad_norm": 0.23612400889396667,
      "learning_rate": 0.00014286055967471895,
      "loss": 0.0439,
      "step": 11455
    },
    {
      "epoch": 1.6056061667834618,
      "grad_norm": 0.2968144118785858,
      "learning_rate": 0.00014284620904089928,
      "loss": 0.0682,
      "step": 11456
    },
    {
      "epoch": 1.6057463209530485,
      "grad_norm": 0.37891146540641785,
      "learning_rate": 0.00014283185840707963,
      "loss": 0.0408,
      "step": 11457
    },
    {
      "epoch": 1.605886475122635,
      "grad_norm": 0.33037495613098145,
      "learning_rate": 0.00014281750777325996,
      "loss": 0.0387,
      "step": 11458
    },
    {
      "epoch": 1.6060266292922214,
      "grad_norm": 0.25307202339172363,
      "learning_rate": 0.00014280315713944032,
      "loss": 0.0655,
      "step": 11459
    },
    {
      "epoch": 1.6061667834618079,
      "grad_norm": 0.4955904483795166,
      "learning_rate": 0.00014278880650562064,
      "loss": 0.0944,
      "step": 11460
    },
    {
      "epoch": 1.6063069376313945,
      "grad_norm": 0.30982741713523865,
      "learning_rate": 0.00014277445587180097,
      "loss": 0.0551,
      "step": 11461
    },
    {
      "epoch": 1.6064470918009812,
      "grad_norm": 0.21471638977527618,
      "learning_rate": 0.00014276010523798133,
      "loss": 0.0248,
      "step": 11462
    },
    {
      "epoch": 1.6065872459705677,
      "grad_norm": 0.29256123304367065,
      "learning_rate": 0.00014274575460416168,
      "loss": 0.0197,
      "step": 11463
    },
    {
      "epoch": 1.6067274001401541,
      "grad_norm": 0.17885033786296844,
      "learning_rate": 0.000142731403970342,
      "loss": 0.0367,
      "step": 11464
    },
    {
      "epoch": 1.6068675543097406,
      "grad_norm": 0.3832833170890808,
      "learning_rate": 0.00014271705333652234,
      "loss": 0.1491,
      "step": 11465
    },
    {
      "epoch": 1.6070077084793273,
      "grad_norm": 0.12220731377601624,
      "learning_rate": 0.0001427027027027027,
      "loss": 0.0198,
      "step": 11466
    },
    {
      "epoch": 1.607147862648914,
      "grad_norm": 0.08932118117809296,
      "learning_rate": 0.00014268835206888303,
      "loss": 0.0096,
      "step": 11467
    },
    {
      "epoch": 1.6072880168185004,
      "grad_norm": 0.18182271718978882,
      "learning_rate": 0.00014267400143506338,
      "loss": 0.0164,
      "step": 11468
    },
    {
      "epoch": 1.6074281709880869,
      "grad_norm": 1.2724153995513916,
      "learning_rate": 0.0001426596508012437,
      "loss": 0.0993,
      "step": 11469
    },
    {
      "epoch": 1.6075683251576733,
      "grad_norm": 0.23609989881515503,
      "learning_rate": 0.00014264530016742404,
      "loss": 0.0403,
      "step": 11470
    },
    {
      "epoch": 1.60770847932726,
      "grad_norm": 0.3410497307777405,
      "learning_rate": 0.0001426309495336044,
      "loss": 0.0476,
      "step": 11471
    },
    {
      "epoch": 1.6078486334968465,
      "grad_norm": 0.6018911004066467,
      "learning_rate": 0.00014261659889978472,
      "loss": 0.0488,
      "step": 11472
    },
    {
      "epoch": 1.6079887876664332,
      "grad_norm": 0.09402860701084137,
      "learning_rate": 0.00014260224826596508,
      "loss": 0.0188,
      "step": 11473
    },
    {
      "epoch": 1.6081289418360196,
      "grad_norm": 0.17101004719734192,
      "learning_rate": 0.0001425878976321454,
      "loss": 0.0223,
      "step": 11474
    },
    {
      "epoch": 1.608269096005606,
      "grad_norm": 0.2607898712158203,
      "learning_rate": 0.00014257354699832574,
      "loss": 0.066,
      "step": 11475
    },
    {
      "epoch": 1.6084092501751928,
      "grad_norm": 0.16262494027614594,
      "learning_rate": 0.0001425591963645061,
      "loss": 0.0066,
      "step": 11476
    },
    {
      "epoch": 1.6085494043447792,
      "grad_norm": 0.24213643372058868,
      "learning_rate": 0.00014254484573068642,
      "loss": 0.0217,
      "step": 11477
    },
    {
      "epoch": 1.608689558514366,
      "grad_norm": 0.24582268297672272,
      "learning_rate": 0.00014253049509686678,
      "loss": 0.0229,
      "step": 11478
    },
    {
      "epoch": 1.6088297126839524,
      "grad_norm": 0.4782690107822418,
      "learning_rate": 0.0001425161444630471,
      "loss": 0.025,
      "step": 11479
    },
    {
      "epoch": 1.6089698668535388,
      "grad_norm": 0.9680556058883667,
      "learning_rate": 0.00014250179382922743,
      "loss": 0.1242,
      "step": 11480
    },
    {
      "epoch": 1.6091100210231253,
      "grad_norm": 0.5895328521728516,
      "learning_rate": 0.0001424874431954078,
      "loss": 0.0631,
      "step": 11481
    },
    {
      "epoch": 1.609250175192712,
      "grad_norm": 0.6573855876922607,
      "learning_rate": 0.00014247309256158814,
      "loss": 0.1201,
      "step": 11482
    },
    {
      "epoch": 1.6093903293622986,
      "grad_norm": 0.572184681892395,
      "learning_rate": 0.00014245874192776847,
      "loss": 0.0553,
      "step": 11483
    },
    {
      "epoch": 1.609530483531885,
      "grad_norm": 1.0979609489440918,
      "learning_rate": 0.0001424443912939488,
      "loss": 0.0784,
      "step": 11484
    },
    {
      "epoch": 1.6096706377014716,
      "grad_norm": 0.2946031987667084,
      "learning_rate": 0.00014243004066012913,
      "loss": 0.0226,
      "step": 11485
    },
    {
      "epoch": 1.609810791871058,
      "grad_norm": 0.6670780181884766,
      "learning_rate": 0.00014241569002630949,
      "loss": 0.0615,
      "step": 11486
    },
    {
      "epoch": 1.6099509460406447,
      "grad_norm": 0.2716189920902252,
      "learning_rate": 0.00014240133939248981,
      "loss": 0.0078,
      "step": 11487
    },
    {
      "epoch": 1.6100911002102314,
      "grad_norm": 0.39290618896484375,
      "learning_rate": 0.00014238698875867017,
      "loss": 0.0371,
      "step": 11488
    },
    {
      "epoch": 1.6102312543798178,
      "grad_norm": 0.2137584388256073,
      "learning_rate": 0.0001423726381248505,
      "loss": 0.0389,
      "step": 11489
    },
    {
      "epoch": 1.6103714085494043,
      "grad_norm": 0.32838350534439087,
      "learning_rate": 0.00014235828749103085,
      "loss": 0.0337,
      "step": 11490
    },
    {
      "epoch": 1.6105115627189908,
      "grad_norm": 0.154434934258461,
      "learning_rate": 0.00014234393685721118,
      "loss": 0.0285,
      "step": 11491
    },
    {
      "epoch": 1.6106517168885774,
      "grad_norm": 0.2677808701992035,
      "learning_rate": 0.0001423295862233915,
      "loss": 0.0232,
      "step": 11492
    },
    {
      "epoch": 1.6107918710581641,
      "grad_norm": 0.3976641297340393,
      "learning_rate": 0.00014231523558957184,
      "loss": 0.0964,
      "step": 11493
    },
    {
      "epoch": 1.6109320252277506,
      "grad_norm": 0.30432915687561035,
      "learning_rate": 0.0001423008849557522,
      "loss": 0.0344,
      "step": 11494
    },
    {
      "epoch": 1.611072179397337,
      "grad_norm": 0.20437073707580566,
      "learning_rate": 0.00014228653432193255,
      "loss": 0.0462,
      "step": 11495
    },
    {
      "epoch": 1.6112123335669235,
      "grad_norm": 0.36792802810668945,
      "learning_rate": 0.00014227218368811288,
      "loss": 0.0842,
      "step": 11496
    },
    {
      "epoch": 1.6113524877365102,
      "grad_norm": 0.3269313871860504,
      "learning_rate": 0.0001422578330542932,
      "loss": 0.0627,
      "step": 11497
    },
    {
      "epoch": 1.6114926419060969,
      "grad_norm": 0.4229220747947693,
      "learning_rate": 0.00014224348242047356,
      "loss": 0.056,
      "step": 11498
    },
    {
      "epoch": 1.6116327960756833,
      "grad_norm": 0.45769762992858887,
      "learning_rate": 0.0001422291317866539,
      "loss": 0.0223,
      "step": 11499
    },
    {
      "epoch": 1.6117729502452698,
      "grad_norm": 0.4342239797115326,
      "learning_rate": 0.00014221478115283425,
      "loss": 0.0297,
      "step": 11500
    },
    {
      "epoch": 1.6119131044148562,
      "grad_norm": 0.49738943576812744,
      "learning_rate": 0.00014220043051901458,
      "loss": 0.0495,
      "step": 11501
    },
    {
      "epoch": 1.612053258584443,
      "grad_norm": 0.30668190121650696,
      "learning_rate": 0.0001421860798851949,
      "loss": 0.0145,
      "step": 11502
    },
    {
      "epoch": 1.6121934127540294,
      "grad_norm": 0.4818120002746582,
      "learning_rate": 0.00014217172925137526,
      "loss": 0.0381,
      "step": 11503
    },
    {
      "epoch": 1.612333566923616,
      "grad_norm": 0.28638604283332825,
      "learning_rate": 0.0001421573786175556,
      "loss": 0.0814,
      "step": 11504
    },
    {
      "epoch": 1.6124737210932025,
      "grad_norm": 0.12675456702709198,
      "learning_rate": 0.00014214302798373595,
      "loss": 0.0203,
      "step": 11505
    },
    {
      "epoch": 1.612613875262789,
      "grad_norm": 0.11540811508893967,
      "learning_rate": 0.00014212867734991627,
      "loss": 0.0265,
      "step": 11506
    },
    {
      "epoch": 1.6127540294323757,
      "grad_norm": 0.14152872562408447,
      "learning_rate": 0.0001421143267160966,
      "loss": 0.0198,
      "step": 11507
    },
    {
      "epoch": 1.6128941836019621,
      "grad_norm": 0.5622986555099487,
      "learning_rate": 0.00014209997608227696,
      "loss": 0.1093,
      "step": 11508
    },
    {
      "epoch": 1.6130343377715488,
      "grad_norm": 0.1203148365020752,
      "learning_rate": 0.0001420856254484573,
      "loss": 0.0391,
      "step": 11509
    },
    {
      "epoch": 1.6131744919411353,
      "grad_norm": 0.4011740982532501,
      "learning_rate": 0.00014207127481463764,
      "loss": 0.031,
      "step": 11510
    },
    {
      "epoch": 1.6133146461107217,
      "grad_norm": 0.5486384034156799,
      "learning_rate": 0.00014205692418081797,
      "loss": 0.025,
      "step": 11511
    },
    {
      "epoch": 1.6134548002803082,
      "grad_norm": 0.32643720507621765,
      "learning_rate": 0.0001420425735469983,
      "loss": 0.0512,
      "step": 11512
    },
    {
      "epoch": 1.6135949544498949,
      "grad_norm": 0.3651643395423889,
      "learning_rate": 0.00014202822291317865,
      "loss": 0.0365,
      "step": 11513
    },
    {
      "epoch": 1.6137351086194816,
      "grad_norm": 0.06403113156557083,
      "learning_rate": 0.000142013872279359,
      "loss": 0.0043,
      "step": 11514
    },
    {
      "epoch": 1.613875262789068,
      "grad_norm": 0.283630907535553,
      "learning_rate": 0.00014199952164553934,
      "loss": 0.0445,
      "step": 11515
    },
    {
      "epoch": 1.6140154169586545,
      "grad_norm": 0.3547370433807373,
      "learning_rate": 0.00014198517101171967,
      "loss": 0.0424,
      "step": 11516
    },
    {
      "epoch": 1.614155571128241,
      "grad_norm": 0.234574094414711,
      "learning_rate": 0.00014197082037790002,
      "loss": 0.0191,
      "step": 11517
    },
    {
      "epoch": 1.6142957252978276,
      "grad_norm": 0.9759863018989563,
      "learning_rate": 0.00014195646974408035,
      "loss": 0.0529,
      "step": 11518
    },
    {
      "epoch": 1.6144358794674143,
      "grad_norm": 0.3272368907928467,
      "learning_rate": 0.00014194211911026068,
      "loss": 0.029,
      "step": 11519
    },
    {
      "epoch": 1.6145760336370008,
      "grad_norm": 0.3000497817993164,
      "learning_rate": 0.00014192776847644104,
      "loss": 0.0532,
      "step": 11520
    },
    {
      "epoch": 1.6147161878065872,
      "grad_norm": 0.806154727935791,
      "learning_rate": 0.00014191341784262136,
      "loss": 0.0596,
      "step": 11521
    },
    {
      "epoch": 1.6148563419761737,
      "grad_norm": 0.46560943126678467,
      "learning_rate": 0.00014189906720880172,
      "loss": 0.0722,
      "step": 11522
    },
    {
      "epoch": 1.6149964961457604,
      "grad_norm": 0.5738423466682434,
      "learning_rate": 0.00014188471657498205,
      "loss": 0.0774,
      "step": 11523
    },
    {
      "epoch": 1.615136650315347,
      "grad_norm": 0.228576198220253,
      "learning_rate": 0.00014187036594116238,
      "loss": 0.047,
      "step": 11524
    },
    {
      "epoch": 1.6152768044849335,
      "grad_norm": 1.1163264513015747,
      "learning_rate": 0.00014185601530734273,
      "loss": 0.0255,
      "step": 11525
    },
    {
      "epoch": 1.61541695865452,
      "grad_norm": 0.14057192206382751,
      "learning_rate": 0.00014184166467352306,
      "loss": 0.0209,
      "step": 11526
    },
    {
      "epoch": 1.6155571128241064,
      "grad_norm": 0.2490060031414032,
      "learning_rate": 0.00014182731403970342,
      "loss": 0.0436,
      "step": 11527
    },
    {
      "epoch": 1.615697266993693,
      "grad_norm": 0.4295102655887604,
      "learning_rate": 0.00014181296340588375,
      "loss": 0.0375,
      "step": 11528
    },
    {
      "epoch": 1.6158374211632796,
      "grad_norm": 0.22657571732997894,
      "learning_rate": 0.00014179861277206407,
      "loss": 0.0441,
      "step": 11529
    },
    {
      "epoch": 1.6159775753328662,
      "grad_norm": 1.6277847290039062,
      "learning_rate": 0.00014178426213824443,
      "loss": 0.0671,
      "step": 11530
    },
    {
      "epoch": 1.6161177295024527,
      "grad_norm": 0.7943706512451172,
      "learning_rate": 0.00014176991150442476,
      "loss": 0.1968,
      "step": 11531
    },
    {
      "epoch": 1.6162578836720392,
      "grad_norm": 0.3241390883922577,
      "learning_rate": 0.00014175556087060511,
      "loss": 0.0566,
      "step": 11532
    },
    {
      "epoch": 1.6163980378416258,
      "grad_norm": 1.4788482189178467,
      "learning_rate": 0.00014174121023678544,
      "loss": 0.1092,
      "step": 11533
    },
    {
      "epoch": 1.6165381920112123,
      "grad_norm": 0.40441399812698364,
      "learning_rate": 0.00014172685960296577,
      "loss": 0.02,
      "step": 11534
    },
    {
      "epoch": 1.616678346180799,
      "grad_norm": 0.21731466054916382,
      "learning_rate": 0.00014171250896914613,
      "loss": 0.0084,
      "step": 11535
    },
    {
      "epoch": 1.6168185003503854,
      "grad_norm": 0.3413139879703522,
      "learning_rate": 0.00014169815833532648,
      "loss": 0.0595,
      "step": 11536
    },
    {
      "epoch": 1.616958654519972,
      "grad_norm": 0.37197378277778625,
      "learning_rate": 0.0001416838077015068,
      "loss": 0.0697,
      "step": 11537
    },
    {
      "epoch": 1.6170988086895584,
      "grad_norm": 0.1412816196680069,
      "learning_rate": 0.00014166945706768714,
      "loss": 0.022,
      "step": 11538
    },
    {
      "epoch": 1.617238962859145,
      "grad_norm": 0.7602114677429199,
      "learning_rate": 0.00014165510643386747,
      "loss": 0.0542,
      "step": 11539
    },
    {
      "epoch": 1.6173791170287317,
      "grad_norm": 0.7870051860809326,
      "learning_rate": 0.00014164075580004782,
      "loss": 0.1007,
      "step": 11540
    },
    {
      "epoch": 1.6175192711983182,
      "grad_norm": 0.0858910009264946,
      "learning_rate": 0.00014162640516622818,
      "loss": 0.0135,
      "step": 11541
    },
    {
      "epoch": 1.6176594253679046,
      "grad_norm": 0.20803296566009521,
      "learning_rate": 0.0001416120545324085,
      "loss": 0.0415,
      "step": 11542
    },
    {
      "epoch": 1.617799579537491,
      "grad_norm": 0.36449846625328064,
      "learning_rate": 0.00014159770389858884,
      "loss": 0.0302,
      "step": 11543
    },
    {
      "epoch": 1.6179397337070778,
      "grad_norm": 0.8823717832565308,
      "learning_rate": 0.0001415833532647692,
      "loss": 0.1131,
      "step": 11544
    },
    {
      "epoch": 1.6180798878766645,
      "grad_norm": 0.5474041104316711,
      "learning_rate": 0.00014156900263094952,
      "loss": 0.0492,
      "step": 11545
    },
    {
      "epoch": 1.618220042046251,
      "grad_norm": 0.25257402658462524,
      "learning_rate": 0.00014155465199712988,
      "loss": 0.0629,
      "step": 11546
    },
    {
      "epoch": 1.6183601962158374,
      "grad_norm": 0.32781097292900085,
      "learning_rate": 0.0001415403013633102,
      "loss": 0.043,
      "step": 11547
    },
    {
      "epoch": 1.6185003503854238,
      "grad_norm": 0.08208863437175751,
      "learning_rate": 0.00014152595072949053,
      "loss": 0.0084,
      "step": 11548
    },
    {
      "epoch": 1.6186405045550105,
      "grad_norm": 0.1464567482471466,
      "learning_rate": 0.0001415116000956709,
      "loss": 0.0362,
      "step": 11549
    },
    {
      "epoch": 1.6187806587245972,
      "grad_norm": 0.1557217538356781,
      "learning_rate": 0.00014149724946185122,
      "loss": 0.0392,
      "step": 11550
    },
    {
      "epoch": 1.6189208128941837,
      "grad_norm": 0.08954671025276184,
      "learning_rate": 0.00014148289882803155,
      "loss": 0.0144,
      "step": 11551
    },
    {
      "epoch": 1.6190609670637701,
      "grad_norm": 0.31153181195259094,
      "learning_rate": 0.0001414685481942119,
      "loss": 0.053,
      "step": 11552
    },
    {
      "epoch": 1.6192011212333566,
      "grad_norm": 0.26565760374069214,
      "learning_rate": 0.00014145419756039223,
      "loss": 0.04,
      "step": 11553
    },
    {
      "epoch": 1.6193412754029433,
      "grad_norm": 0.2863524258136749,
      "learning_rate": 0.0001414398469265726,
      "loss": 0.031,
      "step": 11554
    },
    {
      "epoch": 1.61948142957253,
      "grad_norm": 0.14655672013759613,
      "learning_rate": 0.00014142549629275292,
      "loss": 0.0374,
      "step": 11555
    },
    {
      "epoch": 1.6196215837421164,
      "grad_norm": 0.7252004742622375,
      "learning_rate": 0.00014141114565893324,
      "loss": 0.1329,
      "step": 11556
    },
    {
      "epoch": 1.6197617379117029,
      "grad_norm": 0.21816428005695343,
      "learning_rate": 0.0001413967950251136,
      "loss": 0.0117,
      "step": 11557
    },
    {
      "epoch": 1.6199018920812893,
      "grad_norm": 0.3231394290924072,
      "learning_rate": 0.00014138244439129393,
      "loss": 0.0427,
      "step": 11558
    },
    {
      "epoch": 1.620042046250876,
      "grad_norm": 0.2110644429922104,
      "learning_rate": 0.00014136809375747428,
      "loss": 0.0232,
      "step": 11559
    },
    {
      "epoch": 1.6201822004204625,
      "grad_norm": 0.3024694621562958,
      "learning_rate": 0.0001413537431236546,
      "loss": 0.0393,
      "step": 11560
    },
    {
      "epoch": 1.6203223545900491,
      "grad_norm": 0.29343223571777344,
      "learning_rate": 0.00014133939248983494,
      "loss": 0.071,
      "step": 11561
    },
    {
      "epoch": 1.6204625087596356,
      "grad_norm": 0.20576244592666626,
      "learning_rate": 0.0001413250418560153,
      "loss": 0.067,
      "step": 11562
    },
    {
      "epoch": 1.620602662929222,
      "grad_norm": 0.2312684804201126,
      "learning_rate": 0.00014131069122219562,
      "loss": 0.0889,
      "step": 11563
    },
    {
      "epoch": 1.6207428170988087,
      "grad_norm": 0.5816258788108826,
      "learning_rate": 0.00014129634058837598,
      "loss": 0.0621,
      "step": 11564
    },
    {
      "epoch": 1.6208829712683952,
      "grad_norm": 0.2639305293560028,
      "learning_rate": 0.0001412819899545563,
      "loss": 0.1137,
      "step": 11565
    },
    {
      "epoch": 1.6210231254379819,
      "grad_norm": 0.32290351390838623,
      "learning_rate": 0.00014126763932073664,
      "loss": 0.0773,
      "step": 11566
    },
    {
      "epoch": 1.6211632796075683,
      "grad_norm": 0.6009404063224792,
      "learning_rate": 0.000141253288686917,
      "loss": 0.0695,
      "step": 11567
    },
    {
      "epoch": 1.6213034337771548,
      "grad_norm": 0.3422173857688904,
      "learning_rate": 0.00014123893805309735,
      "loss": 0.0334,
      "step": 11568
    },
    {
      "epoch": 1.6214435879467413,
      "grad_norm": 0.3206561207771301,
      "learning_rate": 0.00014122458741927768,
      "loss": 0.0354,
      "step": 11569
    },
    {
      "epoch": 1.621583742116328,
      "grad_norm": 0.19543053209781647,
      "learning_rate": 0.000141210236785458,
      "loss": 0.0461,
      "step": 11570
    },
    {
      "epoch": 1.6217238962859146,
      "grad_norm": 0.704021155834198,
      "learning_rate": 0.00014119588615163836,
      "loss": 0.0912,
      "step": 11571
    },
    {
      "epoch": 1.621864050455501,
      "grad_norm": 0.2393180876970291,
      "learning_rate": 0.0001411815355178187,
      "loss": 0.0237,
      "step": 11572
    },
    {
      "epoch": 1.6220042046250875,
      "grad_norm": 0.20762500166893005,
      "learning_rate": 0.00014116718488399905,
      "loss": 0.0304,
      "step": 11573
    },
    {
      "epoch": 1.622144358794674,
      "grad_norm": 0.31861111521720886,
      "learning_rate": 0.00014115283425017937,
      "loss": 0.0595,
      "step": 11574
    },
    {
      "epoch": 1.6222845129642607,
      "grad_norm": 0.5564166903495789,
      "learning_rate": 0.0001411384836163597,
      "loss": 0.0657,
      "step": 11575
    },
    {
      "epoch": 1.6224246671338474,
      "grad_norm": 0.19453097879886627,
      "learning_rate": 0.00014112413298254006,
      "loss": 0.0418,
      "step": 11576
    },
    {
      "epoch": 1.6225648213034338,
      "grad_norm": 0.7942109704017639,
      "learning_rate": 0.0001411097823487204,
      "loss": 0.0209,
      "step": 11577
    },
    {
      "epoch": 1.6227049754730203,
      "grad_norm": 0.1301758885383606,
      "learning_rate": 0.00014109543171490074,
      "loss": 0.0116,
      "step": 11578
    },
    {
      "epoch": 1.6228451296426067,
      "grad_norm": 0.0691453069448471,
      "learning_rate": 0.00014108108108108107,
      "loss": 0.0071,
      "step": 11579
    },
    {
      "epoch": 1.6229852838121934,
      "grad_norm": 0.14758393168449402,
      "learning_rate": 0.0001410667304472614,
      "loss": 0.0136,
      "step": 11580
    },
    {
      "epoch": 1.62312543798178,
      "grad_norm": 0.27677419781684875,
      "learning_rate": 0.00014105237981344176,
      "loss": 0.0124,
      "step": 11581
    },
    {
      "epoch": 1.6232655921513666,
      "grad_norm": 0.3889691233634949,
      "learning_rate": 0.00014103802917962208,
      "loss": 0.0691,
      "step": 11582
    },
    {
      "epoch": 1.623405746320953,
      "grad_norm": 0.7682013511657715,
      "learning_rate": 0.0001410236785458024,
      "loss": 0.0568,
      "step": 11583
    },
    {
      "epoch": 1.6235459004905395,
      "grad_norm": 0.4609513282775879,
      "learning_rate": 0.00014100932791198277,
      "loss": 0.0177,
      "step": 11584
    },
    {
      "epoch": 1.6236860546601262,
      "grad_norm": 1.136196494102478,
      "learning_rate": 0.0001409949772781631,
      "loss": 0.1233,
      "step": 11585
    },
    {
      "epoch": 1.6238262088297128,
      "grad_norm": 0.25201964378356934,
      "learning_rate": 0.00014098062664434345,
      "loss": 0.0879,
      "step": 11586
    },
    {
      "epoch": 1.6239663629992993,
      "grad_norm": 0.5443011522293091,
      "learning_rate": 0.00014096627601052378,
      "loss": 0.0692,
      "step": 11587
    },
    {
      "epoch": 1.6241065171688858,
      "grad_norm": 0.9291744828224182,
      "learning_rate": 0.0001409519253767041,
      "loss": 0.0647,
      "step": 11588
    },
    {
      "epoch": 1.6242466713384722,
      "grad_norm": 0.4512588679790497,
      "learning_rate": 0.00014093757474288447,
      "loss": 0.0721,
      "step": 11589
    },
    {
      "epoch": 1.624386825508059,
      "grad_norm": 0.18535341322422028,
      "learning_rate": 0.0001409232241090648,
      "loss": 0.028,
      "step": 11590
    },
    {
      "epoch": 1.6245269796776454,
      "grad_norm": 0.4352816939353943,
      "learning_rate": 0.00014090887347524515,
      "loss": 0.0831,
      "step": 11591
    },
    {
      "epoch": 1.624667133847232,
      "grad_norm": 0.18647170066833496,
      "learning_rate": 0.00014089452284142548,
      "loss": 0.0391,
      "step": 11592
    },
    {
      "epoch": 1.6248072880168185,
      "grad_norm": 0.22422170639038086,
      "learning_rate": 0.0001408801722076058,
      "loss": 0.0118,
      "step": 11593
    },
    {
      "epoch": 1.624947442186405,
      "grad_norm": 0.2361518293619156,
      "learning_rate": 0.00014086582157378616,
      "loss": 0.0513,
      "step": 11594
    },
    {
      "epoch": 1.6250875963559914,
      "grad_norm": 0.2519594728946686,
      "learning_rate": 0.00014085147093996652,
      "loss": 0.0327,
      "step": 11595
    },
    {
      "epoch": 1.625227750525578,
      "grad_norm": 0.2969636023044586,
      "learning_rate": 0.00014083712030614685,
      "loss": 0.0393,
      "step": 11596
    },
    {
      "epoch": 1.6253679046951648,
      "grad_norm": 0.5165436863899231,
      "learning_rate": 0.00014082276967232718,
      "loss": 0.0577,
      "step": 11597
    },
    {
      "epoch": 1.6255080588647512,
      "grad_norm": 0.47275036573410034,
      "learning_rate": 0.0001408084190385075,
      "loss": 0.051,
      "step": 11598
    },
    {
      "epoch": 1.6256482130343377,
      "grad_norm": 0.5552020072937012,
      "learning_rate": 0.00014079406840468786,
      "loss": 0.0511,
      "step": 11599
    },
    {
      "epoch": 1.6257883672039242,
      "grad_norm": 0.16615641117095947,
      "learning_rate": 0.00014077971777086822,
      "loss": 0.0243,
      "step": 11600
    },
    {
      "epoch": 1.6259285213735108,
      "grad_norm": 0.34352177381515503,
      "learning_rate": 0.00014076536713704854,
      "loss": 0.0217,
      "step": 11601
    },
    {
      "epoch": 1.6260686755430975,
      "grad_norm": 0.3269125521183014,
      "learning_rate": 0.00014075101650322887,
      "loss": 0.0482,
      "step": 11602
    },
    {
      "epoch": 1.626208829712684,
      "grad_norm": 0.5768211483955383,
      "learning_rate": 0.00014073666586940923,
      "loss": 0.0418,
      "step": 11603
    },
    {
      "epoch": 1.6263489838822704,
      "grad_norm": 0.49134424328804016,
      "learning_rate": 0.00014072231523558956,
      "loss": 0.0379,
      "step": 11604
    },
    {
      "epoch": 1.626489138051857,
      "grad_norm": 0.09402913600206375,
      "learning_rate": 0.0001407079646017699,
      "loss": 0.0131,
      "step": 11605
    },
    {
      "epoch": 1.6266292922214436,
      "grad_norm": 0.027240999042987823,
      "learning_rate": 0.00014069361396795024,
      "loss": 0.0022,
      "step": 11606
    },
    {
      "epoch": 1.6267694463910303,
      "grad_norm": 0.2057039737701416,
      "learning_rate": 0.00014067926333413057,
      "loss": 0.0263,
      "step": 11607
    },
    {
      "epoch": 1.6269096005606167,
      "grad_norm": 0.19102413952350616,
      "learning_rate": 0.00014066491270031093,
      "loss": 0.0294,
      "step": 11608
    },
    {
      "epoch": 1.6270497547302032,
      "grad_norm": 0.37462911009788513,
      "learning_rate": 0.00014065056206649125,
      "loss": 0.0914,
      "step": 11609
    },
    {
      "epoch": 1.6271899088997896,
      "grad_norm": 0.5419964790344238,
      "learning_rate": 0.0001406362114326716,
      "loss": 0.0386,
      "step": 11610
    },
    {
      "epoch": 1.6273300630693763,
      "grad_norm": 0.2506133019924164,
      "learning_rate": 0.00014062186079885194,
      "loss": 0.0086,
      "step": 11611
    },
    {
      "epoch": 1.627470217238963,
      "grad_norm": 0.19552595913410187,
      "learning_rate": 0.00014060751016503227,
      "loss": 0.0127,
      "step": 11612
    },
    {
      "epoch": 1.6276103714085495,
      "grad_norm": 0.32846346497535706,
      "learning_rate": 0.00014059315953121262,
      "loss": 0.0564,
      "step": 11613
    },
    {
      "epoch": 1.627750525578136,
      "grad_norm": 0.1583506464958191,
      "learning_rate": 0.00014057880889739295,
      "loss": 0.0267,
      "step": 11614
    },
    {
      "epoch": 1.6278906797477224,
      "grad_norm": 0.34699931740760803,
      "learning_rate": 0.0001405644582635733,
      "loss": 0.0401,
      "step": 11615
    },
    {
      "epoch": 1.628030833917309,
      "grad_norm": 0.23199664056301117,
      "learning_rate": 0.00014055010762975363,
      "loss": 0.0319,
      "step": 11616
    },
    {
      "epoch": 1.6281709880868955,
      "grad_norm": 0.1987570822238922,
      "learning_rate": 0.00014053575699593396,
      "loss": 0.0157,
      "step": 11617
    },
    {
      "epoch": 1.6283111422564822,
      "grad_norm": 0.43920180201530457,
      "learning_rate": 0.00014052140636211432,
      "loss": 0.0389,
      "step": 11618
    },
    {
      "epoch": 1.6284512964260687,
      "grad_norm": 0.2858230471611023,
      "learning_rate": 0.00014050705572829465,
      "loss": 0.0895,
      "step": 11619
    },
    {
      "epoch": 1.6285914505956551,
      "grad_norm": 0.2685343325138092,
      "learning_rate": 0.00014049270509447498,
      "loss": 0.0684,
      "step": 11620
    },
    {
      "epoch": 1.6287316047652418,
      "grad_norm": 0.36698684096336365,
      "learning_rate": 0.00014047835446065533,
      "loss": 0.0892,
      "step": 11621
    },
    {
      "epoch": 1.6288717589348283,
      "grad_norm": 0.18439428508281708,
      "learning_rate": 0.0001404640038268357,
      "loss": 0.0316,
      "step": 11622
    },
    {
      "epoch": 1.629011913104415,
      "grad_norm": 0.6981950402259827,
      "learning_rate": 0.00014044965319301602,
      "loss": 0.0594,
      "step": 11623
    },
    {
      "epoch": 1.6291520672740014,
      "grad_norm": 0.36755767464637756,
      "learning_rate": 0.00014043530255919634,
      "loss": 0.0308,
      "step": 11624
    },
    {
      "epoch": 1.6292922214435879,
      "grad_norm": 0.5525476932525635,
      "learning_rate": 0.00014042095192537667,
      "loss": 0.0145,
      "step": 11625
    },
    {
      "epoch": 1.6294323756131743,
      "grad_norm": 1.8820995092391968,
      "learning_rate": 0.00014040660129155703,
      "loss": 0.0557,
      "step": 11626
    },
    {
      "epoch": 1.629572529782761,
      "grad_norm": 0.4943757951259613,
      "learning_rate": 0.00014039225065773738,
      "loss": 0.0555,
      "step": 11627
    },
    {
      "epoch": 1.6297126839523477,
      "grad_norm": 0.32236361503601074,
      "learning_rate": 0.0001403779000239177,
      "loss": 0.0863,
      "step": 11628
    },
    {
      "epoch": 1.6298528381219342,
      "grad_norm": 0.322846382856369,
      "learning_rate": 0.00014036354939009804,
      "loss": 0.013,
      "step": 11629
    },
    {
      "epoch": 1.6299929922915206,
      "grad_norm": 0.18453797698020935,
      "learning_rate": 0.0001403491987562784,
      "loss": 0.0214,
      "step": 11630
    },
    {
      "epoch": 1.630133146461107,
      "grad_norm": 0.3332267701625824,
      "learning_rate": 0.00014033484812245873,
      "loss": 0.031,
      "step": 11631
    },
    {
      "epoch": 1.6302733006306938,
      "grad_norm": 1.2403481006622314,
      "learning_rate": 0.00014032049748863908,
      "loss": 0.0477,
      "step": 11632
    },
    {
      "epoch": 1.6304134548002804,
      "grad_norm": 0.24606643617153168,
      "learning_rate": 0.0001403061468548194,
      "loss": 0.0931,
      "step": 11633
    },
    {
      "epoch": 1.630553608969867,
      "grad_norm": 1.2664353847503662,
      "learning_rate": 0.00014029179622099974,
      "loss": 0.1896,
      "step": 11634
    },
    {
      "epoch": 1.6306937631394534,
      "grad_norm": 0.3728208541870117,
      "learning_rate": 0.0001402774455871801,
      "loss": 0.0263,
      "step": 11635
    },
    {
      "epoch": 1.6308339173090398,
      "grad_norm": 0.11035755276679993,
      "learning_rate": 0.00014026309495336042,
      "loss": 0.0282,
      "step": 11636
    },
    {
      "epoch": 1.6309740714786265,
      "grad_norm": 0.4083704948425293,
      "learning_rate": 0.00014024874431954078,
      "loss": 0.0656,
      "step": 11637
    },
    {
      "epoch": 1.6311142256482132,
      "grad_norm": 0.4946696162223816,
      "learning_rate": 0.0001402343936857211,
      "loss": 0.0442,
      "step": 11638
    },
    {
      "epoch": 1.6312543798177996,
      "grad_norm": 0.38726767897605896,
      "learning_rate": 0.00014022004305190144,
      "loss": 0.0395,
      "step": 11639
    },
    {
      "epoch": 1.631394533987386,
      "grad_norm": 0.49798622727394104,
      "learning_rate": 0.0001402056924180818,
      "loss": 0.0341,
      "step": 11640
    },
    {
      "epoch": 1.6315346881569726,
      "grad_norm": 0.38486430048942566,
      "learning_rate": 0.00014019134178426212,
      "loss": 0.0531,
      "step": 11641
    },
    {
      "epoch": 1.6316748423265592,
      "grad_norm": 0.18969294428825378,
      "learning_rate": 0.00014017699115044248,
      "loss": 0.033,
      "step": 11642
    },
    {
      "epoch": 1.631814996496146,
      "grad_norm": 0.5594385266304016,
      "learning_rate": 0.0001401626405166228,
      "loss": 0.047,
      "step": 11643
    },
    {
      "epoch": 1.6319551506657324,
      "grad_norm": 0.28359031677246094,
      "learning_rate": 0.00014014828988280313,
      "loss": 0.0912,
      "step": 11644
    },
    {
      "epoch": 1.6320953048353188,
      "grad_norm": 0.21976250410079956,
      "learning_rate": 0.0001401339392489835,
      "loss": 0.0231,
      "step": 11645
    },
    {
      "epoch": 1.6322354590049053,
      "grad_norm": 0.13548928499221802,
      "learning_rate": 0.00014011958861516382,
      "loss": 0.0107,
      "step": 11646
    },
    {
      "epoch": 1.632375613174492,
      "grad_norm": 0.04017278179526329,
      "learning_rate": 0.00014010523798134417,
      "loss": 0.0033,
      "step": 11647
    },
    {
      "epoch": 1.6325157673440784,
      "grad_norm": 1.0490429401397705,
      "learning_rate": 0.0001400908873475245,
      "loss": 0.0202,
      "step": 11648
    },
    {
      "epoch": 1.6326559215136651,
      "grad_norm": 0.20822161436080933,
      "learning_rate": 0.00014007653671370486,
      "loss": 0.0426,
      "step": 11649
    },
    {
      "epoch": 1.6327960756832516,
      "grad_norm": 0.4806745648384094,
      "learning_rate": 0.00014006218607988519,
      "loss": 0.0906,
      "step": 11650
    },
    {
      "epoch": 1.632936229852838,
      "grad_norm": 0.16590049862861633,
      "learning_rate": 0.00014004783544606551,
      "loss": 0.0215,
      "step": 11651
    },
    {
      "epoch": 1.6330763840224247,
      "grad_norm": 0.43973755836486816,
      "learning_rate": 0.00014003348481224584,
      "loss": 0.0339,
      "step": 11652
    },
    {
      "epoch": 1.6332165381920112,
      "grad_norm": 0.23996147513389587,
      "learning_rate": 0.0001400191341784262,
      "loss": 0.0323,
      "step": 11653
    },
    {
      "epoch": 1.6333566923615979,
      "grad_norm": 0.15574686229228973,
      "learning_rate": 0.00014000478354460655,
      "loss": 0.0609,
      "step": 11654
    },
    {
      "epoch": 1.6334968465311843,
      "grad_norm": 0.22125369310379028,
      "learning_rate": 0.00013999043291078688,
      "loss": 0.0497,
      "step": 11655
    },
    {
      "epoch": 1.6336370007007708,
      "grad_norm": 0.2110290676355362,
      "learning_rate": 0.0001399760822769672,
      "loss": 0.0192,
      "step": 11656
    },
    {
      "epoch": 1.6337771548703572,
      "grad_norm": 0.2032098025083542,
      "learning_rate": 0.00013996173164314757,
      "loss": 0.0357,
      "step": 11657
    },
    {
      "epoch": 1.633917309039944,
      "grad_norm": 0.25698280334472656,
      "learning_rate": 0.0001399473810093279,
      "loss": 0.078,
      "step": 11658
    },
    {
      "epoch": 1.6340574632095306,
      "grad_norm": 0.21493308246135712,
      "learning_rate": 0.00013993303037550825,
      "loss": 0.0296,
      "step": 11659
    },
    {
      "epoch": 1.634197617379117,
      "grad_norm": 0.23097802698612213,
      "learning_rate": 0.00013991867974168858,
      "loss": 0.0236,
      "step": 11660
    },
    {
      "epoch": 1.6343377715487035,
      "grad_norm": 0.24762603640556335,
      "learning_rate": 0.0001399043291078689,
      "loss": 0.0837,
      "step": 11661
    },
    {
      "epoch": 1.63447792571829,
      "grad_norm": 0.1455816775560379,
      "learning_rate": 0.00013988997847404926,
      "loss": 0.0156,
      "step": 11662
    },
    {
      "epoch": 1.6346180798878767,
      "grad_norm": 0.28169047832489014,
      "learning_rate": 0.0001398756278402296,
      "loss": 0.039,
      "step": 11663
    },
    {
      "epoch": 1.6347582340574633,
      "grad_norm": 0.2514907121658325,
      "learning_rate": 0.00013986127720640995,
      "loss": 0.0432,
      "step": 11664
    },
    {
      "epoch": 1.6348983882270498,
      "grad_norm": 0.22667475044727325,
      "learning_rate": 0.00013984692657259028,
      "loss": 0.0606,
      "step": 11665
    },
    {
      "epoch": 1.6350385423966363,
      "grad_norm": 0.13793246448040009,
      "learning_rate": 0.0001398325759387706,
      "loss": 0.0167,
      "step": 11666
    },
    {
      "epoch": 1.6351786965662227,
      "grad_norm": 0.45482996106147766,
      "learning_rate": 0.00013981822530495096,
      "loss": 0.0429,
      "step": 11667
    },
    {
      "epoch": 1.6353188507358094,
      "grad_norm": 0.2539718449115753,
      "learning_rate": 0.0001398038746711313,
      "loss": 0.044,
      "step": 11668
    },
    {
      "epoch": 1.635459004905396,
      "grad_norm": 0.2680078148841858,
      "learning_rate": 0.00013978952403731164,
      "loss": 0.0223,
      "step": 11669
    },
    {
      "epoch": 1.6355991590749825,
      "grad_norm": 0.45410051941871643,
      "learning_rate": 0.00013977517340349197,
      "loss": 0.1116,
      "step": 11670
    },
    {
      "epoch": 1.635739313244569,
      "grad_norm": 0.632757842540741,
      "learning_rate": 0.0001397608227696723,
      "loss": 0.0835,
      "step": 11671
    },
    {
      "epoch": 1.6358794674141555,
      "grad_norm": 0.8625863194465637,
      "learning_rate": 0.00013974647213585266,
      "loss": 0.1734,
      "step": 11672
    },
    {
      "epoch": 1.6360196215837421,
      "grad_norm": 0.6289317011833191,
      "learning_rate": 0.000139732121502033,
      "loss": 0.0401,
      "step": 11673
    },
    {
      "epoch": 1.6361597757533288,
      "grad_norm": 0.4812743067741394,
      "learning_rate": 0.00013971777086821334,
      "loss": 0.0576,
      "step": 11674
    },
    {
      "epoch": 1.6362999299229153,
      "grad_norm": 0.1312282383441925,
      "learning_rate": 0.00013970342023439367,
      "loss": 0.0148,
      "step": 11675
    },
    {
      "epoch": 1.6364400840925017,
      "grad_norm": 0.1737457513809204,
      "learning_rate": 0.000139689069600574,
      "loss": 0.0222,
      "step": 11676
    },
    {
      "epoch": 1.6365802382620882,
      "grad_norm": 0.2768307626247406,
      "learning_rate": 0.00013967471896675435,
      "loss": 0.0503,
      "step": 11677
    },
    {
      "epoch": 1.6367203924316749,
      "grad_norm": 0.4666915833950043,
      "learning_rate": 0.00013966036833293468,
      "loss": 0.0482,
      "step": 11678
    },
    {
      "epoch": 1.6368605466012613,
      "grad_norm": 0.07669016718864441,
      "learning_rate": 0.00013964601769911504,
      "loss": 0.0087,
      "step": 11679
    },
    {
      "epoch": 1.637000700770848,
      "grad_norm": 0.7221842408180237,
      "learning_rate": 0.00013963166706529537,
      "loss": 0.1014,
      "step": 11680
    },
    {
      "epoch": 1.6371408549404345,
      "grad_norm": 0.562724769115448,
      "learning_rate": 0.00013961731643147572,
      "loss": 0.0569,
      "step": 11681
    },
    {
      "epoch": 1.637281009110021,
      "grad_norm": 1.501454472541809,
      "learning_rate": 0.00013960296579765605,
      "loss": 0.1748,
      "step": 11682
    },
    {
      "epoch": 1.6374211632796074,
      "grad_norm": 0.23456807434558868,
      "learning_rate": 0.00013958861516383638,
      "loss": 0.0124,
      "step": 11683
    },
    {
      "epoch": 1.637561317449194,
      "grad_norm": 0.37844786047935486,
      "learning_rate": 0.00013957426453001674,
      "loss": 0.0428,
      "step": 11684
    },
    {
      "epoch": 1.6377014716187808,
      "grad_norm": 1.2032986879348755,
      "learning_rate": 0.00013955991389619706,
      "loss": 0.2068,
      "step": 11685
    },
    {
      "epoch": 1.6378416257883672,
      "grad_norm": 0.11660397797822952,
      "learning_rate": 0.00013954556326237742,
      "loss": 0.016,
      "step": 11686
    },
    {
      "epoch": 1.6379817799579537,
      "grad_norm": 0.1083175390958786,
      "learning_rate": 0.00013953121262855775,
      "loss": 0.0191,
      "step": 11687
    },
    {
      "epoch": 1.6381219341275401,
      "grad_norm": 0.14667271077632904,
      "learning_rate": 0.00013951686199473808,
      "loss": 0.0284,
      "step": 11688
    },
    {
      "epoch": 1.6382620882971268,
      "grad_norm": 0.1866927444934845,
      "learning_rate": 0.00013950251136091843,
      "loss": 0.0308,
      "step": 11689
    },
    {
      "epoch": 1.6384022424667135,
      "grad_norm": 0.5327773690223694,
      "learning_rate": 0.00013948816072709876,
      "loss": 0.0612,
      "step": 11690
    },
    {
      "epoch": 1.6385423966363,
      "grad_norm": 0.23347429931163788,
      "learning_rate": 0.00013947381009327912,
      "loss": 0.0255,
      "step": 11691
    },
    {
      "epoch": 1.6386825508058864,
      "grad_norm": 0.22115172445774078,
      "learning_rate": 0.00013945945945945945,
      "loss": 0.0359,
      "step": 11692
    },
    {
      "epoch": 1.6388227049754729,
      "grad_norm": 0.35347816348075867,
      "learning_rate": 0.00013944510882563977,
      "loss": 0.0549,
      "step": 11693
    },
    {
      "epoch": 1.6389628591450596,
      "grad_norm": 0.28366562724113464,
      "learning_rate": 0.00013943075819182013,
      "loss": 0.0617,
      "step": 11694
    },
    {
      "epoch": 1.6391030133146463,
      "grad_norm": 0.26388657093048096,
      "learning_rate": 0.00013941640755800046,
      "loss": 0.0434,
      "step": 11695
    },
    {
      "epoch": 1.6392431674842327,
      "grad_norm": 0.4912094175815582,
      "learning_rate": 0.00013940205692418081,
      "loss": 0.0546,
      "step": 11696
    },
    {
      "epoch": 1.6393833216538192,
      "grad_norm": 0.17095795273780823,
      "learning_rate": 0.00013938770629036114,
      "loss": 0.0315,
      "step": 11697
    },
    {
      "epoch": 1.6395234758234056,
      "grad_norm": 0.11469148099422455,
      "learning_rate": 0.00013937335565654147,
      "loss": 0.0103,
      "step": 11698
    },
    {
      "epoch": 1.6396636299929923,
      "grad_norm": 0.3439951539039612,
      "learning_rate": 0.00013935900502272183,
      "loss": 0.1323,
      "step": 11699
    },
    {
      "epoch": 1.639803784162579,
      "grad_norm": 0.3164527416229248,
      "learning_rate": 0.00013934465438890218,
      "loss": 0.0971,
      "step": 11700
    },
    {
      "epoch": 1.6399439383321655,
      "grad_norm": 0.3256116807460785,
      "learning_rate": 0.0001393303037550825,
      "loss": 0.0913,
      "step": 11701
    },
    {
      "epoch": 1.640084092501752,
      "grad_norm": 0.3805089592933655,
      "learning_rate": 0.00013931595312126284,
      "loss": 0.0242,
      "step": 11702
    },
    {
      "epoch": 1.6402242466713384,
      "grad_norm": 0.2286466658115387,
      "learning_rate": 0.00013930160248744317,
      "loss": 0.0555,
      "step": 11703
    },
    {
      "epoch": 1.640364400840925,
      "grad_norm": 0.7314069867134094,
      "learning_rate": 0.00013928725185362352,
      "loss": 0.0478,
      "step": 11704
    },
    {
      "epoch": 1.6405045550105115,
      "grad_norm": 0.3232010304927826,
      "learning_rate": 0.00013927290121980388,
      "loss": 0.052,
      "step": 11705
    },
    {
      "epoch": 1.6406447091800982,
      "grad_norm": 0.21542169153690338,
      "learning_rate": 0.0001392585505859842,
      "loss": 0.0467,
      "step": 11706
    },
    {
      "epoch": 1.6407848633496847,
      "grad_norm": 0.5921143889427185,
      "learning_rate": 0.00013924419995216454,
      "loss": 0.0877,
      "step": 11707
    },
    {
      "epoch": 1.6409250175192711,
      "grad_norm": 0.15161815285682678,
      "learning_rate": 0.0001392298493183449,
      "loss": 0.026,
      "step": 11708
    },
    {
      "epoch": 1.6410651716888578,
      "grad_norm": 0.12948666512966156,
      "learning_rate": 0.00013921549868452522,
      "loss": 0.0145,
      "step": 11709
    },
    {
      "epoch": 1.6412053258584443,
      "grad_norm": 1.161110520362854,
      "learning_rate": 0.00013920114805070558,
      "loss": 0.0703,
      "step": 11710
    },
    {
      "epoch": 1.641345480028031,
      "grad_norm": 0.5644869208335876,
      "learning_rate": 0.0001391867974168859,
      "loss": 0.1123,
      "step": 11711
    },
    {
      "epoch": 1.6414856341976174,
      "grad_norm": 0.26063624024391174,
      "learning_rate": 0.00013917244678306623,
      "loss": 0.0202,
      "step": 11712
    },
    {
      "epoch": 1.6416257883672039,
      "grad_norm": 0.39383062720298767,
      "learning_rate": 0.0001391580961492466,
      "loss": 0.067,
      "step": 11713
    },
    {
      "epoch": 1.6417659425367903,
      "grad_norm": 0.14393767714500427,
      "learning_rate": 0.00013914374551542692,
      "loss": 0.0134,
      "step": 11714
    },
    {
      "epoch": 1.641906096706377,
      "grad_norm": 0.26287737488746643,
      "learning_rate": 0.00013912939488160725,
      "loss": 0.0734,
      "step": 11715
    },
    {
      "epoch": 1.6420462508759637,
      "grad_norm": 0.12290789932012558,
      "learning_rate": 0.0001391150442477876,
      "loss": 0.0478,
      "step": 11716
    },
    {
      "epoch": 1.6421864050455501,
      "grad_norm": 0.5133820176124573,
      "learning_rate": 0.00013910069361396793,
      "loss": 0.0639,
      "step": 11717
    },
    {
      "epoch": 1.6423265592151366,
      "grad_norm": 0.44815823435783386,
      "learning_rate": 0.0001390863429801483,
      "loss": 0.0964,
      "step": 11718
    },
    {
      "epoch": 1.642466713384723,
      "grad_norm": 0.07294411957263947,
      "learning_rate": 0.00013907199234632862,
      "loss": 0.0066,
      "step": 11719
    },
    {
      "epoch": 1.6426068675543097,
      "grad_norm": 0.333614706993103,
      "learning_rate": 0.00013905764171250894,
      "loss": 0.022,
      "step": 11720
    },
    {
      "epoch": 1.6427470217238964,
      "grad_norm": 0.16638225317001343,
      "learning_rate": 0.0001390432910786893,
      "loss": 0.0249,
      "step": 11721
    },
    {
      "epoch": 1.6428871758934829,
      "grad_norm": 0.4734067916870117,
      "learning_rate": 0.00013902894044486963,
      "loss": 0.0359,
      "step": 11722
    },
    {
      "epoch": 1.6430273300630693,
      "grad_norm": 0.8964606523513794,
      "learning_rate": 0.00013901458981104998,
      "loss": 0.0514,
      "step": 11723
    },
    {
      "epoch": 1.6431674842326558,
      "grad_norm": 0.20593541860580444,
      "learning_rate": 0.0001390002391772303,
      "loss": 0.0296,
      "step": 11724
    },
    {
      "epoch": 1.6433076384022425,
      "grad_norm": 0.28778791427612305,
      "learning_rate": 0.00013898588854341064,
      "loss": 0.0959,
      "step": 11725
    },
    {
      "epoch": 1.6434477925718292,
      "grad_norm": 0.4542524218559265,
      "learning_rate": 0.000138971537909591,
      "loss": 0.0763,
      "step": 11726
    },
    {
      "epoch": 1.6435879467414156,
      "grad_norm": 0.3051491677761078,
      "learning_rate": 0.00013895718727577135,
      "loss": 0.1164,
      "step": 11727
    },
    {
      "epoch": 1.643728100911002,
      "grad_norm": 0.24849773943424225,
      "learning_rate": 0.00013894283664195168,
      "loss": 0.0222,
      "step": 11728
    },
    {
      "epoch": 1.6438682550805885,
      "grad_norm": 0.7094378471374512,
      "learning_rate": 0.000138928486008132,
      "loss": 0.026,
      "step": 11729
    },
    {
      "epoch": 1.6440084092501752,
      "grad_norm": 0.47970616817474365,
      "learning_rate": 0.00013891413537431234,
      "loss": 0.1069,
      "step": 11730
    },
    {
      "epoch": 1.644148563419762,
      "grad_norm": 0.3720422685146332,
      "learning_rate": 0.0001388997847404927,
      "loss": 0.014,
      "step": 11731
    },
    {
      "epoch": 1.6442887175893484,
      "grad_norm": 0.21593424677848816,
      "learning_rate": 0.00013888543410667305,
      "loss": 0.0159,
      "step": 11732
    },
    {
      "epoch": 1.6444288717589348,
      "grad_norm": 0.13539251685142517,
      "learning_rate": 0.00013887108347285338,
      "loss": 0.0128,
      "step": 11733
    },
    {
      "epoch": 1.6445690259285213,
      "grad_norm": 1.8731926679611206,
      "learning_rate": 0.0001388567328390337,
      "loss": 0.2625,
      "step": 11734
    },
    {
      "epoch": 1.644709180098108,
      "grad_norm": 1.3780369758605957,
      "learning_rate": 0.00013884238220521406,
      "loss": 0.1894,
      "step": 11735
    },
    {
      "epoch": 1.6448493342676944,
      "grad_norm": 0.426658034324646,
      "learning_rate": 0.0001388280315713944,
      "loss": 0.0731,
      "step": 11736
    },
    {
      "epoch": 1.644989488437281,
      "grad_norm": 0.30276042222976685,
      "learning_rate": 0.00013881368093757475,
      "loss": 0.1117,
      "step": 11737
    },
    {
      "epoch": 1.6451296426068676,
      "grad_norm": 0.12564964592456818,
      "learning_rate": 0.00013879933030375507,
      "loss": 0.0104,
      "step": 11738
    },
    {
      "epoch": 1.645269796776454,
      "grad_norm": 0.16238072514533997,
      "learning_rate": 0.0001387849796699354,
      "loss": 0.0167,
      "step": 11739
    },
    {
      "epoch": 1.6454099509460405,
      "grad_norm": 0.19371038675308228,
      "learning_rate": 0.00013877062903611576,
      "loss": 0.0573,
      "step": 11740
    },
    {
      "epoch": 1.6455501051156272,
      "grad_norm": 0.36758118867874146,
      "learning_rate": 0.0001387562784022961,
      "loss": 0.0345,
      "step": 11741
    },
    {
      "epoch": 1.6456902592852138,
      "grad_norm": 0.44633328914642334,
      "learning_rate": 0.00013874192776847644,
      "loss": 0.0625,
      "step": 11742
    },
    {
      "epoch": 1.6458304134548003,
      "grad_norm": 0.2560858726501465,
      "learning_rate": 0.00013872757713465677,
      "loss": 0.0948,
      "step": 11743
    },
    {
      "epoch": 1.6459705676243868,
      "grad_norm": 0.17425720393657684,
      "learning_rate": 0.0001387132265008371,
      "loss": 0.0375,
      "step": 11744
    },
    {
      "epoch": 1.6461107217939732,
      "grad_norm": 0.3766055107116699,
      "learning_rate": 0.00013869887586701746,
      "loss": 0.0458,
      "step": 11745
    },
    {
      "epoch": 1.64625087596356,
      "grad_norm": 0.20229826867580414,
      "learning_rate": 0.00013868452523319778,
      "loss": 0.0518,
      "step": 11746
    },
    {
      "epoch": 1.6463910301331466,
      "grad_norm": 0.11625409126281738,
      "learning_rate": 0.0001386701745993781,
      "loss": 0.0187,
      "step": 11747
    },
    {
      "epoch": 1.646531184302733,
      "grad_norm": 0.297679603099823,
      "learning_rate": 0.00013865582396555847,
      "loss": 0.0385,
      "step": 11748
    },
    {
      "epoch": 1.6466713384723195,
      "grad_norm": 1.0538545846939087,
      "learning_rate": 0.0001386414733317388,
      "loss": 0.111,
      "step": 11749
    },
    {
      "epoch": 1.646811492641906,
      "grad_norm": 0.08853999525308609,
      "learning_rate": 0.00013862712269791915,
      "loss": 0.01,
      "step": 11750
    },
    {
      "epoch": 1.6469516468114926,
      "grad_norm": 0.3008106052875519,
      "learning_rate": 0.00013861277206409948,
      "loss": 0.0192,
      "step": 11751
    },
    {
      "epoch": 1.6470918009810793,
      "grad_norm": 0.14477674663066864,
      "learning_rate": 0.0001385984214302798,
      "loss": 0.0178,
      "step": 11752
    },
    {
      "epoch": 1.6472319551506658,
      "grad_norm": 0.13875508308410645,
      "learning_rate": 0.00013858407079646017,
      "loss": 0.042,
      "step": 11753
    },
    {
      "epoch": 1.6473721093202522,
      "grad_norm": 0.2620844542980194,
      "learning_rate": 0.0001385697201626405,
      "loss": 0.047,
      "step": 11754
    },
    {
      "epoch": 1.6475122634898387,
      "grad_norm": 0.282022625207901,
      "learning_rate": 0.00013855536952882085,
      "loss": 0.0736,
      "step": 11755
    },
    {
      "epoch": 1.6476524176594254,
      "grad_norm": 0.33656007051467896,
      "learning_rate": 0.00013854101889500118,
      "loss": 0.1235,
      "step": 11756
    },
    {
      "epoch": 1.647792571829012,
      "grad_norm": 0.4576495289802551,
      "learning_rate": 0.0001385266682611815,
      "loss": 0.0617,
      "step": 11757
    },
    {
      "epoch": 1.6479327259985985,
      "grad_norm": 0.5562580823898315,
      "learning_rate": 0.00013851231762736186,
      "loss": 0.0553,
      "step": 11758
    },
    {
      "epoch": 1.648072880168185,
      "grad_norm": 0.20411553978919983,
      "learning_rate": 0.00013849796699354222,
      "loss": 0.0394,
      "step": 11759
    },
    {
      "epoch": 1.6482130343377714,
      "grad_norm": 0.33069273829460144,
      "learning_rate": 0.00013848361635972255,
      "loss": 0.0474,
      "step": 11760
    },
    {
      "epoch": 1.6483531885073581,
      "grad_norm": 0.2694055736064911,
      "learning_rate": 0.00013846926572590288,
      "loss": 0.0675,
      "step": 11761
    },
    {
      "epoch": 1.6484933426769446,
      "grad_norm": 0.35911455750465393,
      "learning_rate": 0.00013845491509208323,
      "loss": 0.0581,
      "step": 11762
    },
    {
      "epoch": 1.6486334968465313,
      "grad_norm": 0.11525940150022507,
      "learning_rate": 0.00013844056445826356,
      "loss": 0.0086,
      "step": 11763
    },
    {
      "epoch": 1.6487736510161177,
      "grad_norm": 0.422592431306839,
      "learning_rate": 0.00013842621382444392,
      "loss": 0.0921,
      "step": 11764
    },
    {
      "epoch": 1.6489138051857042,
      "grad_norm": 0.11063863337039948,
      "learning_rate": 0.00013841186319062424,
      "loss": 0.0159,
      "step": 11765
    },
    {
      "epoch": 1.6490539593552909,
      "grad_norm": 0.09566666185855865,
      "learning_rate": 0.00013839751255680457,
      "loss": 0.0114,
      "step": 11766
    },
    {
      "epoch": 1.6491941135248773,
      "grad_norm": 0.7951347827911377,
      "learning_rate": 0.00013838316192298493,
      "loss": 0.0443,
      "step": 11767
    },
    {
      "epoch": 1.649334267694464,
      "grad_norm": 0.3839142918586731,
      "learning_rate": 0.00013836881128916526,
      "loss": 0.0532,
      "step": 11768
    },
    {
      "epoch": 1.6494744218640505,
      "grad_norm": 0.2822932004928589,
      "learning_rate": 0.0001383544606553456,
      "loss": 0.0421,
      "step": 11769
    },
    {
      "epoch": 1.649614576033637,
      "grad_norm": 0.4533008635044098,
      "learning_rate": 0.00013834011002152594,
      "loss": 0.0844,
      "step": 11770
    },
    {
      "epoch": 1.6497547302032234,
      "grad_norm": 0.15165278315544128,
      "learning_rate": 0.00013832575938770627,
      "loss": 0.015,
      "step": 11771
    },
    {
      "epoch": 1.64989488437281,
      "grad_norm": 0.9360820651054382,
      "learning_rate": 0.00013831140875388663,
      "loss": 0.1712,
      "step": 11772
    },
    {
      "epoch": 1.6500350385423967,
      "grad_norm": 0.2267729789018631,
      "learning_rate": 0.00013829705812006695,
      "loss": 0.0155,
      "step": 11773
    },
    {
      "epoch": 1.6501751927119832,
      "grad_norm": 0.4546307921409607,
      "learning_rate": 0.0001382827074862473,
      "loss": 0.0392,
      "step": 11774
    },
    {
      "epoch": 1.6503153468815697,
      "grad_norm": 0.9851316213607788,
      "learning_rate": 0.00013826835685242764,
      "loss": 0.0519,
      "step": 11775
    },
    {
      "epoch": 1.6504555010511561,
      "grad_norm": 0.11964554339647293,
      "learning_rate": 0.00013825400621860797,
      "loss": 0.0162,
      "step": 11776
    },
    {
      "epoch": 1.6505956552207428,
      "grad_norm": 0.6148589253425598,
      "learning_rate": 0.00013823965558478832,
      "loss": 0.028,
      "step": 11777
    },
    {
      "epoch": 1.6507358093903295,
      "grad_norm": 0.5039327144622803,
      "learning_rate": 0.00013822530495096865,
      "loss": 0.0574,
      "step": 11778
    },
    {
      "epoch": 1.650875963559916,
      "grad_norm": 0.4020152986049652,
      "learning_rate": 0.00013821095431714898,
      "loss": 0.0416,
      "step": 11779
    },
    {
      "epoch": 1.6510161177295024,
      "grad_norm": 0.405556321144104,
      "learning_rate": 0.00013819660368332933,
      "loss": 0.084,
      "step": 11780
    },
    {
      "epoch": 1.6511562718990889,
      "grad_norm": 0.2820827066898346,
      "learning_rate": 0.00013818225304950966,
      "loss": 0.0692,
      "step": 11781
    },
    {
      "epoch": 1.6512964260686755,
      "grad_norm": 1.0139774084091187,
      "learning_rate": 0.00013816790241569002,
      "loss": 0.2055,
      "step": 11782
    },
    {
      "epoch": 1.6514365802382622,
      "grad_norm": 1.809464454650879,
      "learning_rate": 0.00013815355178187035,
      "loss": 0.2039,
      "step": 11783
    },
    {
      "epoch": 1.6515767344078487,
      "grad_norm": 0.8287193179130554,
      "learning_rate": 0.00013813920114805068,
      "loss": 0.0421,
      "step": 11784
    },
    {
      "epoch": 1.6517168885774351,
      "grad_norm": 0.5953776836395264,
      "learning_rate": 0.00013812485051423103,
      "loss": 0.036,
      "step": 11785
    },
    {
      "epoch": 1.6518570427470216,
      "grad_norm": 0.2768326997756958,
      "learning_rate": 0.0001381104998804114,
      "loss": 0.0897,
      "step": 11786
    },
    {
      "epoch": 1.6519971969166083,
      "grad_norm": 0.14924399554729462,
      "learning_rate": 0.00013809614924659172,
      "loss": 0.014,
      "step": 11787
    },
    {
      "epoch": 1.652137351086195,
      "grad_norm": 0.19737350940704346,
      "learning_rate": 0.00013808179861277204,
      "loss": 0.0482,
      "step": 11788
    },
    {
      "epoch": 1.6522775052557814,
      "grad_norm": 0.184872105717659,
      "learning_rate": 0.00013806744797895237,
      "loss": 0.0499,
      "step": 11789
    },
    {
      "epoch": 1.652417659425368,
      "grad_norm": 0.17982976138591766,
      "learning_rate": 0.00013805309734513273,
      "loss": 0.0286,
      "step": 11790
    },
    {
      "epoch": 1.6525578135949544,
      "grad_norm": 0.3252630829811096,
      "learning_rate": 0.00013803874671131308,
      "loss": 0.0691,
      "step": 11791
    },
    {
      "epoch": 1.652697967764541,
      "grad_norm": 0.18803054094314575,
      "learning_rate": 0.0001380243960774934,
      "loss": 0.0475,
      "step": 11792
    },
    {
      "epoch": 1.6528381219341275,
      "grad_norm": 0.19127315282821655,
      "learning_rate": 0.00013801004544367374,
      "loss": 0.0394,
      "step": 11793
    },
    {
      "epoch": 1.6529782761037142,
      "grad_norm": 0.2496771216392517,
      "learning_rate": 0.0001379956948098541,
      "loss": 0.0535,
      "step": 11794
    },
    {
      "epoch": 1.6531184302733006,
      "grad_norm": 0.2117943912744522,
      "learning_rate": 0.00013798134417603443,
      "loss": 0.0747,
      "step": 11795
    },
    {
      "epoch": 1.653258584442887,
      "grad_norm": 0.2620314061641693,
      "learning_rate": 0.00013796699354221478,
      "loss": 0.0688,
      "step": 11796
    },
    {
      "epoch": 1.6533987386124738,
      "grad_norm": 0.18790596723556519,
      "learning_rate": 0.0001379526429083951,
      "loss": 0.0277,
      "step": 11797
    },
    {
      "epoch": 1.6535388927820602,
      "grad_norm": 0.10476590692996979,
      "learning_rate": 0.00013793829227457544,
      "loss": 0.0215,
      "step": 11798
    },
    {
      "epoch": 1.653679046951647,
      "grad_norm": 0.4203190207481384,
      "learning_rate": 0.0001379239416407558,
      "loss": 0.0523,
      "step": 11799
    },
    {
      "epoch": 1.6538192011212334,
      "grad_norm": 0.320651113986969,
      "learning_rate": 0.00013790959100693612,
      "loss": 0.032,
      "step": 11800
    },
    {
      "epoch": 1.6539593552908198,
      "grad_norm": 0.13263949751853943,
      "learning_rate": 0.00013789524037311648,
      "loss": 0.0393,
      "step": 11801
    },
    {
      "epoch": 1.6540995094604063,
      "grad_norm": 0.1796700805425644,
      "learning_rate": 0.0001378808897392968,
      "loss": 0.031,
      "step": 11802
    },
    {
      "epoch": 1.654239663629993,
      "grad_norm": 0.21623770892620087,
      "learning_rate": 0.00013786653910547714,
      "loss": 0.052,
      "step": 11803
    },
    {
      "epoch": 1.6543798177995797,
      "grad_norm": 0.4184766411781311,
      "learning_rate": 0.0001378521884716575,
      "loss": 0.0553,
      "step": 11804
    },
    {
      "epoch": 1.6545199719691661,
      "grad_norm": 0.29444655776023865,
      "learning_rate": 0.00013783783783783782,
      "loss": 0.0381,
      "step": 11805
    },
    {
      "epoch": 1.6546601261387526,
      "grad_norm": 0.281023234128952,
      "learning_rate": 0.00013782348720401818,
      "loss": 0.0473,
      "step": 11806
    },
    {
      "epoch": 1.654800280308339,
      "grad_norm": 0.22504088282585144,
      "learning_rate": 0.0001378091365701985,
      "loss": 0.0356,
      "step": 11807
    },
    {
      "epoch": 1.6549404344779257,
      "grad_norm": 0.2260611206293106,
      "learning_rate": 0.00013779478593637883,
      "loss": 0.0595,
      "step": 11808
    },
    {
      "epoch": 1.6550805886475124,
      "grad_norm": 0.18638841807842255,
      "learning_rate": 0.0001377804353025592,
      "loss": 0.038,
      "step": 11809
    },
    {
      "epoch": 1.6552207428170989,
      "grad_norm": 0.153292715549469,
      "learning_rate": 0.00013776608466873952,
      "loss": 0.0173,
      "step": 11810
    },
    {
      "epoch": 1.6553608969866853,
      "grad_norm": 0.47503137588500977,
      "learning_rate": 0.00013775173403491985,
      "loss": 0.0236,
      "step": 11811
    },
    {
      "epoch": 1.6555010511562718,
      "grad_norm": 0.7097020745277405,
      "learning_rate": 0.0001377373834011002,
      "loss": 0.1166,
      "step": 11812
    },
    {
      "epoch": 1.6556412053258585,
      "grad_norm": 0.9861078262329102,
      "learning_rate": 0.00013772303276728056,
      "loss": 0.1646,
      "step": 11813
    },
    {
      "epoch": 1.6557813594954451,
      "grad_norm": 0.08831826597452164,
      "learning_rate": 0.00013770868213346089,
      "loss": 0.0246,
      "step": 11814
    },
    {
      "epoch": 1.6559215136650316,
      "grad_norm": 0.3325512707233429,
      "learning_rate": 0.00013769433149964121,
      "loss": 0.0505,
      "step": 11815
    },
    {
      "epoch": 1.656061667834618,
      "grad_norm": 0.2166481912136078,
      "learning_rate": 0.00013767998086582154,
      "loss": 0.0195,
      "step": 11816
    },
    {
      "epoch": 1.6562018220042045,
      "grad_norm": 0.5999817252159119,
      "learning_rate": 0.0001376656302320019,
      "loss": 0.0572,
      "step": 11817
    },
    {
      "epoch": 1.6563419761737912,
      "grad_norm": 0.6211214065551758,
      "learning_rate": 0.00013765127959818225,
      "loss": 0.0394,
      "step": 11818
    },
    {
      "epoch": 1.6564821303433779,
      "grad_norm": 0.6740303039550781,
      "learning_rate": 0.00013763692896436258,
      "loss": 0.0365,
      "step": 11819
    },
    {
      "epoch": 1.6566222845129643,
      "grad_norm": 0.22392001748085022,
      "learning_rate": 0.0001376225783305429,
      "loss": 0.0422,
      "step": 11820
    },
    {
      "epoch": 1.6567624386825508,
      "grad_norm": 0.21644064784049988,
      "learning_rate": 0.00013760822769672327,
      "loss": 0.0176,
      "step": 11821
    },
    {
      "epoch": 1.6569025928521373,
      "grad_norm": 0.3878958523273468,
      "learning_rate": 0.0001375938770629036,
      "loss": 0.1003,
      "step": 11822
    },
    {
      "epoch": 1.657042747021724,
      "grad_norm": 0.0884343609213829,
      "learning_rate": 0.00013757952642908395,
      "loss": 0.0085,
      "step": 11823
    },
    {
      "epoch": 1.6571829011913104,
      "grad_norm": 0.3534504771232605,
      "learning_rate": 0.00013756517579526428,
      "loss": 0.0194,
      "step": 11824
    },
    {
      "epoch": 1.657323055360897,
      "grad_norm": 0.15540985763072968,
      "learning_rate": 0.0001375508251614446,
      "loss": 0.0319,
      "step": 11825
    },
    {
      "epoch": 1.6574632095304835,
      "grad_norm": 0.3599049746990204,
      "learning_rate": 0.00013753647452762496,
      "loss": 0.0392,
      "step": 11826
    },
    {
      "epoch": 1.65760336370007,
      "grad_norm": 0.6477541923522949,
      "learning_rate": 0.0001375221238938053,
      "loss": 0.1121,
      "step": 11827
    },
    {
      "epoch": 1.6577435178696565,
      "grad_norm": 0.4080917239189148,
      "learning_rate": 0.00013750777325998565,
      "loss": 0.0615,
      "step": 11828
    },
    {
      "epoch": 1.6578836720392431,
      "grad_norm": 1.1862410306930542,
      "learning_rate": 0.00013749342262616598,
      "loss": 0.179,
      "step": 11829
    },
    {
      "epoch": 1.6580238262088298,
      "grad_norm": 0.1407335102558136,
      "learning_rate": 0.0001374790719923463,
      "loss": 0.0117,
      "step": 11830
    },
    {
      "epoch": 1.6581639803784163,
      "grad_norm": 0.6689997911453247,
      "learning_rate": 0.00013746472135852666,
      "loss": 0.1023,
      "step": 11831
    },
    {
      "epoch": 1.6583041345480027,
      "grad_norm": 0.37400004267692566,
      "learning_rate": 0.00013745037072470702,
      "loss": 0.0475,
      "step": 11832
    },
    {
      "epoch": 1.6584442887175892,
      "grad_norm": 0.5994638800621033,
      "learning_rate": 0.00013743602009088734,
      "loss": 0.2705,
      "step": 11833
    },
    {
      "epoch": 1.6585844428871759,
      "grad_norm": 3.1800317764282227,
      "learning_rate": 0.00013742166945706767,
      "loss": 0.1777,
      "step": 11834
    },
    {
      "epoch": 1.6587245970567626,
      "grad_norm": 1.0134509801864624,
      "learning_rate": 0.000137407318823248,
      "loss": 0.1518,
      "step": 11835
    },
    {
      "epoch": 1.658864751226349,
      "grad_norm": 0.3975442945957184,
      "learning_rate": 0.00013739296818942836,
      "loss": 0.0707,
      "step": 11836
    },
    {
      "epoch": 1.6590049053959355,
      "grad_norm": 0.286672979593277,
      "learning_rate": 0.0001373786175556087,
      "loss": 0.0592,
      "step": 11837
    },
    {
      "epoch": 1.659145059565522,
      "grad_norm": 0.46812012791633606,
      "learning_rate": 0.00013736426692178904,
      "loss": 0.0426,
      "step": 11838
    },
    {
      "epoch": 1.6592852137351086,
      "grad_norm": 0.42614510655403137,
      "learning_rate": 0.00013734991628796937,
      "loss": 0.0641,
      "step": 11839
    },
    {
      "epoch": 1.6594253679046953,
      "grad_norm": 0.3011021912097931,
      "learning_rate": 0.00013733556565414973,
      "loss": 0.0517,
      "step": 11840
    },
    {
      "epoch": 1.6595655220742818,
      "grad_norm": 0.6219186186790466,
      "learning_rate": 0.00013732121502033005,
      "loss": 0.0863,
      "step": 11841
    },
    {
      "epoch": 1.6597056762438682,
      "grad_norm": 0.3697461187839508,
      "learning_rate": 0.00013730686438651038,
      "loss": 0.0448,
      "step": 11842
    },
    {
      "epoch": 1.6598458304134547,
      "grad_norm": 0.15110893547534943,
      "learning_rate": 0.0001372925137526907,
      "loss": 0.025,
      "step": 11843
    },
    {
      "epoch": 1.6599859845830414,
      "grad_norm": 0.13687993586063385,
      "learning_rate": 0.00013727816311887107,
      "loss": 0.0138,
      "step": 11844
    },
    {
      "epoch": 1.660126138752628,
      "grad_norm": 0.13671039044857025,
      "learning_rate": 0.00013726381248505142,
      "loss": 0.0147,
      "step": 11845
    },
    {
      "epoch": 1.6602662929222145,
      "grad_norm": 0.5719878077507019,
      "learning_rate": 0.00013724946185123175,
      "loss": 0.1206,
      "step": 11846
    },
    {
      "epoch": 1.660406447091801,
      "grad_norm": 0.38528135418891907,
      "learning_rate": 0.00013723511121741208,
      "loss": 0.0852,
      "step": 11847
    },
    {
      "epoch": 1.6605466012613874,
      "grad_norm": 0.5481377243995667,
      "learning_rate": 0.00013722076058359244,
      "loss": 0.0681,
      "step": 11848
    },
    {
      "epoch": 1.660686755430974,
      "grad_norm": 0.3499343693256378,
      "learning_rate": 0.00013720640994977276,
      "loss": 0.0914,
      "step": 11849
    },
    {
      "epoch": 1.6608269096005606,
      "grad_norm": 0.21624954044818878,
      "learning_rate": 0.00013719205931595312,
      "loss": 0.0174,
      "step": 11850
    },
    {
      "epoch": 1.6609670637701472,
      "grad_norm": 0.4076785445213318,
      "learning_rate": 0.00013717770868213345,
      "loss": 0.0426,
      "step": 11851
    },
    {
      "epoch": 1.6611072179397337,
      "grad_norm": 0.5568191409111023,
      "learning_rate": 0.00013716335804831378,
      "loss": 0.0278,
      "step": 11852
    },
    {
      "epoch": 1.6612473721093202,
      "grad_norm": 0.37976574897766113,
      "learning_rate": 0.00013714900741449413,
      "loss": 0.0342,
      "step": 11853
    },
    {
      "epoch": 1.6613875262789068,
      "grad_norm": 0.6761259436607361,
      "learning_rate": 0.00013713465678067446,
      "loss": 0.0534,
      "step": 11854
    },
    {
      "epoch": 1.6615276804484933,
      "grad_norm": 0.2505970597267151,
      "learning_rate": 0.00013712030614685482,
      "loss": 0.0396,
      "step": 11855
    },
    {
      "epoch": 1.66166783461808,
      "grad_norm": 0.26204952597618103,
      "learning_rate": 0.00013710595551303515,
      "loss": 0.0281,
      "step": 11856
    },
    {
      "epoch": 1.6618079887876664,
      "grad_norm": 0.2770780920982361,
      "learning_rate": 0.00013709160487921547,
      "loss": 0.0344,
      "step": 11857
    },
    {
      "epoch": 1.661948142957253,
      "grad_norm": 0.40327659249305725,
      "learning_rate": 0.00013707725424539583,
      "loss": 0.029,
      "step": 11858
    },
    {
      "epoch": 1.6620882971268394,
      "grad_norm": 0.11049145460128784,
      "learning_rate": 0.00013706290361157616,
      "loss": 0.0152,
      "step": 11859
    },
    {
      "epoch": 1.662228451296426,
      "grad_norm": 0.17752285301685333,
      "learning_rate": 0.00013704855297775651,
      "loss": 0.0462,
      "step": 11860
    },
    {
      "epoch": 1.6623686054660127,
      "grad_norm": 0.531157374382019,
      "learning_rate": 0.00013703420234393684,
      "loss": 0.0894,
      "step": 11861
    },
    {
      "epoch": 1.6625087596355992,
      "grad_norm": 0.4597202241420746,
      "learning_rate": 0.00013701985171011717,
      "loss": 0.0238,
      "step": 11862
    },
    {
      "epoch": 1.6626489138051856,
      "grad_norm": 0.38533341884613037,
      "learning_rate": 0.00013700550107629753,
      "loss": 0.0382,
      "step": 11863
    },
    {
      "epoch": 1.662789067974772,
      "grad_norm": 0.2667778730392456,
      "learning_rate": 0.00013699115044247788,
      "loss": 0.0123,
      "step": 11864
    },
    {
      "epoch": 1.6629292221443588,
      "grad_norm": 0.15610401332378387,
      "learning_rate": 0.0001369767998086582,
      "loss": 0.0502,
      "step": 11865
    },
    {
      "epoch": 1.6630693763139455,
      "grad_norm": 0.1264648586511612,
      "learning_rate": 0.00013696244917483854,
      "loss": 0.0272,
      "step": 11866
    },
    {
      "epoch": 1.663209530483532,
      "grad_norm": 0.5182936191558838,
      "learning_rate": 0.00013694809854101887,
      "loss": 0.0867,
      "step": 11867
    },
    {
      "epoch": 1.6633496846531184,
      "grad_norm": 0.2675636112689972,
      "learning_rate": 0.00013693374790719922,
      "loss": 0.037,
      "step": 11868
    },
    {
      "epoch": 1.6634898388227048,
      "grad_norm": 0.2956010699272156,
      "learning_rate": 0.00013691939727337958,
      "loss": 0.0814,
      "step": 11869
    },
    {
      "epoch": 1.6636299929922915,
      "grad_norm": 0.15889495611190796,
      "learning_rate": 0.0001369050466395599,
      "loss": 0.0207,
      "step": 11870
    },
    {
      "epoch": 1.6637701471618782,
      "grad_norm": 0.32948222756385803,
      "learning_rate": 0.00013689069600574024,
      "loss": 0.063,
      "step": 11871
    },
    {
      "epoch": 1.6639103013314647,
      "grad_norm": 0.7254371643066406,
      "learning_rate": 0.0001368763453719206,
      "loss": 0.0651,
      "step": 11872
    },
    {
      "epoch": 1.6640504555010511,
      "grad_norm": 0.49235978722572327,
      "learning_rate": 0.00013686199473810092,
      "loss": 0.1158,
      "step": 11873
    },
    {
      "epoch": 1.6641906096706376,
      "grad_norm": 0.24491435289382935,
      "learning_rate": 0.00013684764410428125,
      "loss": 0.0519,
      "step": 11874
    },
    {
      "epoch": 1.6643307638402243,
      "grad_norm": 0.3322868347167969,
      "learning_rate": 0.0001368332934704616,
      "loss": 0.0933,
      "step": 11875
    },
    {
      "epoch": 1.664470918009811,
      "grad_norm": 0.26144763827323914,
      "learning_rate": 0.00013681894283664193,
      "loss": 0.0459,
      "step": 11876
    },
    {
      "epoch": 1.6646110721793974,
      "grad_norm": 0.18525584042072296,
      "learning_rate": 0.0001368045922028223,
      "loss": 0.0299,
      "step": 11877
    },
    {
      "epoch": 1.6647512263489839,
      "grad_norm": 0.8073279857635498,
      "learning_rate": 0.00013679024156900262,
      "loss": 0.0784,
      "step": 11878
    },
    {
      "epoch": 1.6648913805185703,
      "grad_norm": 0.30658721923828125,
      "learning_rate": 0.00013677589093518295,
      "loss": 0.0316,
      "step": 11879
    },
    {
      "epoch": 1.665031534688157,
      "grad_norm": 0.45898351073265076,
      "learning_rate": 0.0001367615403013633,
      "loss": 0.0467,
      "step": 11880
    },
    {
      "epoch": 1.6651716888577435,
      "grad_norm": 0.45159777998924255,
      "learning_rate": 0.00013674718966754363,
      "loss": 0.0329,
      "step": 11881
    },
    {
      "epoch": 1.6653118430273302,
      "grad_norm": 0.5975531339645386,
      "learning_rate": 0.00013673283903372399,
      "loss": 0.1153,
      "step": 11882
    },
    {
      "epoch": 1.6654519971969166,
      "grad_norm": 0.22012656927108765,
      "learning_rate": 0.00013671848839990431,
      "loss": 0.0084,
      "step": 11883
    },
    {
      "epoch": 1.665592151366503,
      "grad_norm": 0.6469978094100952,
      "learning_rate": 0.00013670413776608464,
      "loss": 0.0548,
      "step": 11884
    },
    {
      "epoch": 1.6657323055360898,
      "grad_norm": 1.1799291372299194,
      "learning_rate": 0.000136689787132265,
      "loss": 0.0779,
      "step": 11885
    },
    {
      "epoch": 1.6658724597056762,
      "grad_norm": 0.2911396324634552,
      "learning_rate": 0.00013667543649844533,
      "loss": 0.0301,
      "step": 11886
    },
    {
      "epoch": 1.666012613875263,
      "grad_norm": 0.932594895362854,
      "learning_rate": 0.00013666108586462568,
      "loss": 0.0368,
      "step": 11887
    },
    {
      "epoch": 1.6661527680448494,
      "grad_norm": 0.28875109553337097,
      "learning_rate": 0.000136646735230806,
      "loss": 0.0762,
      "step": 11888
    },
    {
      "epoch": 1.6662929222144358,
      "grad_norm": 0.2923699915409088,
      "learning_rate": 0.00013663238459698634,
      "loss": 0.0347,
      "step": 11889
    },
    {
      "epoch": 1.6664330763840223,
      "grad_norm": 0.15817172825336456,
      "learning_rate": 0.0001366180339631667,
      "loss": 0.0107,
      "step": 11890
    },
    {
      "epoch": 1.666573230553609,
      "grad_norm": 0.30916720628738403,
      "learning_rate": 0.00013660368332934705,
      "loss": 0.038,
      "step": 11891
    },
    {
      "epoch": 1.6667133847231956,
      "grad_norm": 0.4304237961769104,
      "learning_rate": 0.00013658933269552738,
      "loss": 0.0245,
      "step": 11892
    },
    {
      "epoch": 1.666853538892782,
      "grad_norm": 0.6282362341880798,
      "learning_rate": 0.0001365749820617077,
      "loss": 0.0945,
      "step": 11893
    },
    {
      "epoch": 1.6669936930623686,
      "grad_norm": 0.3297266364097595,
      "learning_rate": 0.00013656063142788804,
      "loss": 0.0446,
      "step": 11894
    },
    {
      "epoch": 1.667133847231955,
      "grad_norm": 0.28222498297691345,
      "learning_rate": 0.0001365462807940684,
      "loss": 0.0443,
      "step": 11895
    },
    {
      "epoch": 1.6672740014015417,
      "grad_norm": 0.26622357964515686,
      "learning_rate": 0.00013653193016024875,
      "loss": 0.061,
      "step": 11896
    },
    {
      "epoch": 1.6674141555711284,
      "grad_norm": 0.9223150610923767,
      "learning_rate": 0.00013651757952642908,
      "loss": 0.1193,
      "step": 11897
    },
    {
      "epoch": 1.6675543097407148,
      "grad_norm": 0.05580474063754082,
      "learning_rate": 0.0001365032288926094,
      "loss": 0.0056,
      "step": 11898
    },
    {
      "epoch": 1.6676944639103013,
      "grad_norm": 0.5558134913444519,
      "learning_rate": 0.00013648887825878976,
      "loss": 0.0852,
      "step": 11899
    },
    {
      "epoch": 1.6678346180798878,
      "grad_norm": 0.1503482460975647,
      "learning_rate": 0.0001364745276249701,
      "loss": 0.0373,
      "step": 11900
    },
    {
      "epoch": 1.6679747722494744,
      "grad_norm": 0.38089436292648315,
      "learning_rate": 0.00013646017699115045,
      "loss": 0.0735,
      "step": 11901
    },
    {
      "epoch": 1.6681149264190611,
      "grad_norm": 0.41297438740730286,
      "learning_rate": 0.00013644582635733077,
      "loss": 0.084,
      "step": 11902
    },
    {
      "epoch": 1.6682550805886476,
      "grad_norm": 0.3086937963962555,
      "learning_rate": 0.0001364314757235111,
      "loss": 0.0422,
      "step": 11903
    },
    {
      "epoch": 1.668395234758234,
      "grad_norm": 0.48585644364356995,
      "learning_rate": 0.00013641712508969146,
      "loss": 0.0449,
      "step": 11904
    },
    {
      "epoch": 1.6685353889278205,
      "grad_norm": 0.24671518802642822,
      "learning_rate": 0.0001364027744558718,
      "loss": 0.0385,
      "step": 11905
    },
    {
      "epoch": 1.6686755430974072,
      "grad_norm": 0.3495645523071289,
      "learning_rate": 0.00013638842382205212,
      "loss": 0.0385,
      "step": 11906
    },
    {
      "epoch": 1.6688156972669939,
      "grad_norm": 0.3558703362941742,
      "learning_rate": 0.00013637407318823247,
      "loss": 0.0238,
      "step": 11907
    },
    {
      "epoch": 1.6689558514365803,
      "grad_norm": 0.07500531524419785,
      "learning_rate": 0.0001363597225544128,
      "loss": 0.0084,
      "step": 11908
    },
    {
      "epoch": 1.6690960056061668,
      "grad_norm": 0.13769300282001495,
      "learning_rate": 0.00013634537192059316,
      "loss": 0.0841,
      "step": 11909
    },
    {
      "epoch": 1.6692361597757532,
      "grad_norm": 0.2148124724626541,
      "learning_rate": 0.00013633102128677348,
      "loss": 0.0503,
      "step": 11910
    },
    {
      "epoch": 1.66937631394534,
      "grad_norm": 0.09973196685314178,
      "learning_rate": 0.0001363166706529538,
      "loss": 0.0182,
      "step": 11911
    },
    {
      "epoch": 1.6695164681149264,
      "grad_norm": 0.31246715784072876,
      "learning_rate": 0.00013630232001913417,
      "loss": 0.0401,
      "step": 11912
    },
    {
      "epoch": 1.669656622284513,
      "grad_norm": 0.36144205927848816,
      "learning_rate": 0.0001362879693853145,
      "loss": 0.0355,
      "step": 11913
    },
    {
      "epoch": 1.6697967764540995,
      "grad_norm": 0.24027907848358154,
      "learning_rate": 0.00013627361875149485,
      "loss": 0.0226,
      "step": 11914
    },
    {
      "epoch": 1.669936930623686,
      "grad_norm": 0.224004328250885,
      "learning_rate": 0.00013625926811767518,
      "loss": 0.0171,
      "step": 11915
    },
    {
      "epoch": 1.6700770847932724,
      "grad_norm": 0.3434467613697052,
      "learning_rate": 0.0001362449174838555,
      "loss": 0.0405,
      "step": 11916
    },
    {
      "epoch": 1.6702172389628591,
      "grad_norm": 0.21136601269245148,
      "learning_rate": 0.00013623056685003587,
      "loss": 0.0368,
      "step": 11917
    },
    {
      "epoch": 1.6703573931324458,
      "grad_norm": 0.24941956996917725,
      "learning_rate": 0.00013621621621621622,
      "loss": 0.064,
      "step": 11918
    },
    {
      "epoch": 1.6704975473020323,
      "grad_norm": 0.4517078995704651,
      "learning_rate": 0.00013620186558239655,
      "loss": 0.0393,
      "step": 11919
    },
    {
      "epoch": 1.6706377014716187,
      "grad_norm": 0.5788573026657104,
      "learning_rate": 0.00013618751494857688,
      "loss": 0.0166,
      "step": 11920
    },
    {
      "epoch": 1.6707778556412052,
      "grad_norm": 0.53787761926651,
      "learning_rate": 0.0001361731643147572,
      "loss": 0.0436,
      "step": 11921
    },
    {
      "epoch": 1.6709180098107919,
      "grad_norm": 0.4707808494567871,
      "learning_rate": 0.00013615881368093756,
      "loss": 0.088,
      "step": 11922
    },
    {
      "epoch": 1.6710581639803785,
      "grad_norm": 0.08876273781061172,
      "learning_rate": 0.00013614446304711792,
      "loss": 0.0138,
      "step": 11923
    },
    {
      "epoch": 1.671198318149965,
      "grad_norm": 0.5747291445732117,
      "learning_rate": 0.00013613011241329825,
      "loss": 0.0347,
      "step": 11924
    },
    {
      "epoch": 1.6713384723195515,
      "grad_norm": 0.433500736951828,
      "learning_rate": 0.00013611576177947858,
      "loss": 0.0539,
      "step": 11925
    },
    {
      "epoch": 1.671478626489138,
      "grad_norm": 0.1903090626001358,
      "learning_rate": 0.00013610141114565893,
      "loss": 0.0279,
      "step": 11926
    },
    {
      "epoch": 1.6716187806587246,
      "grad_norm": 0.19112512469291687,
      "learning_rate": 0.00013608706051183926,
      "loss": 0.0166,
      "step": 11927
    },
    {
      "epoch": 1.6717589348283113,
      "grad_norm": 0.5589933395385742,
      "learning_rate": 0.00013607270987801962,
      "loss": 0.0586,
      "step": 11928
    },
    {
      "epoch": 1.6718990889978977,
      "grad_norm": 0.559594988822937,
      "learning_rate": 0.00013605835924419994,
      "loss": 0.0152,
      "step": 11929
    },
    {
      "epoch": 1.6720392431674842,
      "grad_norm": 0.5386563539505005,
      "learning_rate": 0.00013604400861038027,
      "loss": 0.1003,
      "step": 11930
    },
    {
      "epoch": 1.6721793973370707,
      "grad_norm": 0.9070091843605042,
      "learning_rate": 0.00013602965797656063,
      "loss": 0.1477,
      "step": 11931
    },
    {
      "epoch": 1.6723195515066573,
      "grad_norm": 0.29393908381462097,
      "learning_rate": 0.00013601530734274096,
      "loss": 0.0238,
      "step": 11932
    },
    {
      "epoch": 1.672459705676244,
      "grad_norm": 0.36104804277420044,
      "learning_rate": 0.0001360009567089213,
      "loss": 0.0118,
      "step": 11933
    },
    {
      "epoch": 1.6725998598458305,
      "grad_norm": 0.4152359366416931,
      "learning_rate": 0.00013598660607510164,
      "loss": 0.0233,
      "step": 11934
    },
    {
      "epoch": 1.672740014015417,
      "grad_norm": 2.065208673477173,
      "learning_rate": 0.00013597225544128197,
      "loss": 0.1959,
      "step": 11935
    },
    {
      "epoch": 1.6728801681850034,
      "grad_norm": 0.22712886333465576,
      "learning_rate": 0.00013595790480746232,
      "loss": 0.0317,
      "step": 11936
    },
    {
      "epoch": 1.67302032235459,
      "grad_norm": 0.34828656911849976,
      "learning_rate": 0.00013594355417364265,
      "loss": 0.0311,
      "step": 11937
    },
    {
      "epoch": 1.6731604765241765,
      "grad_norm": 0.3816492259502411,
      "learning_rate": 0.00013592920353982298,
      "loss": 0.0815,
      "step": 11938
    },
    {
      "epoch": 1.6733006306937632,
      "grad_norm": 0.2463635355234146,
      "learning_rate": 0.00013591485290600334,
      "loss": 0.0484,
      "step": 11939
    },
    {
      "epoch": 1.6734407848633497,
      "grad_norm": 0.24040736258029938,
      "learning_rate": 0.00013590050227218367,
      "loss": 0.0465,
      "step": 11940
    },
    {
      "epoch": 1.6735809390329361,
      "grad_norm": 0.3465491235256195,
      "learning_rate": 0.00013588615163836402,
      "loss": 0.1006,
      "step": 11941
    },
    {
      "epoch": 1.6737210932025228,
      "grad_norm": 0.1676328033208847,
      "learning_rate": 0.00013587180100454435,
      "loss": 0.0321,
      "step": 11942
    },
    {
      "epoch": 1.6738612473721093,
      "grad_norm": 0.4115368127822876,
      "learning_rate": 0.00013585745037072468,
      "loss": 0.0318,
      "step": 11943
    },
    {
      "epoch": 1.674001401541696,
      "grad_norm": 0.12745307385921478,
      "learning_rate": 0.00013584309973690503,
      "loss": 0.0162,
      "step": 11944
    },
    {
      "epoch": 1.6741415557112824,
      "grad_norm": 0.35464954376220703,
      "learning_rate": 0.0001358287491030854,
      "loss": 0.0727,
      "step": 11945
    },
    {
      "epoch": 1.6742817098808689,
      "grad_norm": 0.182016059756279,
      "learning_rate": 0.00013581439846926572,
      "loss": 0.0164,
      "step": 11946
    },
    {
      "epoch": 1.6744218640504553,
      "grad_norm": 0.3788032531738281,
      "learning_rate": 0.00013580004783544605,
      "loss": 0.0537,
      "step": 11947
    },
    {
      "epoch": 1.674562018220042,
      "grad_norm": 0.45415472984313965,
      "learning_rate": 0.00013578569720162638,
      "loss": 0.0801,
      "step": 11948
    },
    {
      "epoch": 1.6747021723896287,
      "grad_norm": 0.19531764090061188,
      "learning_rate": 0.00013577134656780673,
      "loss": 0.0308,
      "step": 11949
    },
    {
      "epoch": 1.6748423265592152,
      "grad_norm": 0.220106840133667,
      "learning_rate": 0.0001357569959339871,
      "loss": 0.0343,
      "step": 11950
    },
    {
      "epoch": 1.6749824807288016,
      "grad_norm": 0.17745056748390198,
      "learning_rate": 0.00013574264530016742,
      "loss": 0.0344,
      "step": 11951
    },
    {
      "epoch": 1.675122634898388,
      "grad_norm": 0.3207990229129791,
      "learning_rate": 0.00013572829466634774,
      "loss": 0.0435,
      "step": 11952
    },
    {
      "epoch": 1.6752627890679748,
      "grad_norm": 0.38948920369148254,
      "learning_rate": 0.0001357139440325281,
      "loss": 0.0935,
      "step": 11953
    },
    {
      "epoch": 1.6754029432375614,
      "grad_norm": 0.16411563754081726,
      "learning_rate": 0.00013569959339870843,
      "loss": 0.014,
      "step": 11954
    },
    {
      "epoch": 1.675543097407148,
      "grad_norm": 0.15310940146446228,
      "learning_rate": 0.00013568524276488878,
      "loss": 0.0238,
      "step": 11955
    },
    {
      "epoch": 1.6756832515767344,
      "grad_norm": 0.9212971329689026,
      "learning_rate": 0.0001356708921310691,
      "loss": 0.1334,
      "step": 11956
    },
    {
      "epoch": 1.6758234057463208,
      "grad_norm": 0.13926981389522552,
      "learning_rate": 0.00013565654149724944,
      "loss": 0.0231,
      "step": 11957
    },
    {
      "epoch": 1.6759635599159075,
      "grad_norm": 0.2559824585914612,
      "learning_rate": 0.0001356421908634298,
      "loss": 0.0457,
      "step": 11958
    },
    {
      "epoch": 1.6761037140854942,
      "grad_norm": 0.3078050911426544,
      "learning_rate": 0.00013562784022961013,
      "loss": 0.053,
      "step": 11959
    },
    {
      "epoch": 1.6762438682550806,
      "grad_norm": 0.22574682533740997,
      "learning_rate": 0.00013561348959579048,
      "loss": 0.0371,
      "step": 11960
    },
    {
      "epoch": 1.676384022424667,
      "grad_norm": 0.3852347135543823,
      "learning_rate": 0.0001355991389619708,
      "loss": 0.0914,
      "step": 11961
    },
    {
      "epoch": 1.6765241765942536,
      "grad_norm": 0.21751011908054352,
      "learning_rate": 0.00013558478832815114,
      "loss": 0.0635,
      "step": 11962
    },
    {
      "epoch": 1.6766643307638402,
      "grad_norm": 0.2538410723209381,
      "learning_rate": 0.0001355704376943315,
      "loss": 0.0095,
      "step": 11963
    },
    {
      "epoch": 1.676804484933427,
      "grad_norm": 1.156904935836792,
      "learning_rate": 0.00013555608706051182,
      "loss": 0.102,
      "step": 11964
    },
    {
      "epoch": 1.6769446391030134,
      "grad_norm": 0.21310707926750183,
      "learning_rate": 0.00013554173642669218,
      "loss": 0.0232,
      "step": 11965
    },
    {
      "epoch": 1.6770847932725998,
      "grad_norm": 0.19779279828071594,
      "learning_rate": 0.0001355273857928725,
      "loss": 0.0283,
      "step": 11966
    },
    {
      "epoch": 1.6772249474421863,
      "grad_norm": 0.30429092049598694,
      "learning_rate": 0.00013551303515905284,
      "loss": 0.0277,
      "step": 11967
    },
    {
      "epoch": 1.677365101611773,
      "grad_norm": 0.17025989294052124,
      "learning_rate": 0.0001354986845252332,
      "loss": 0.0144,
      "step": 11968
    },
    {
      "epoch": 1.6775052557813595,
      "grad_norm": 0.3282012641429901,
      "learning_rate": 0.00013548433389141352,
      "loss": 0.0386,
      "step": 11969
    },
    {
      "epoch": 1.6776454099509461,
      "grad_norm": 0.90810227394104,
      "learning_rate": 0.00013546998325759385,
      "loss": 0.1024,
      "step": 11970
    },
    {
      "epoch": 1.6777855641205326,
      "grad_norm": 0.2610308825969696,
      "learning_rate": 0.0001354556326237742,
      "loss": 0.0708,
      "step": 11971
    },
    {
      "epoch": 1.677925718290119,
      "grad_norm": 0.5114941000938416,
      "learning_rate": 0.00013544128198995453,
      "loss": 0.0414,
      "step": 11972
    },
    {
      "epoch": 1.6780658724597055,
      "grad_norm": 0.09413972496986389,
      "learning_rate": 0.0001354269313561349,
      "loss": 0.0164,
      "step": 11973
    },
    {
      "epoch": 1.6782060266292922,
      "grad_norm": 0.19549696147441864,
      "learning_rate": 0.00013541258072231522,
      "loss": 0.0225,
      "step": 11974
    },
    {
      "epoch": 1.6783461807988789,
      "grad_norm": 0.27949437499046326,
      "learning_rate": 0.00013539823008849555,
      "loss": 0.0467,
      "step": 11975
    },
    {
      "epoch": 1.6784863349684653,
      "grad_norm": 0.7189096212387085,
      "learning_rate": 0.0001353838794546759,
      "loss": 0.0456,
      "step": 11976
    },
    {
      "epoch": 1.6786264891380518,
      "grad_norm": 0.3418213129043579,
      "learning_rate": 0.00013536952882085626,
      "loss": 0.0724,
      "step": 11977
    },
    {
      "epoch": 1.6787666433076383,
      "grad_norm": 0.36146309971809387,
      "learning_rate": 0.00013535517818703659,
      "loss": 0.0733,
      "step": 11978
    },
    {
      "epoch": 1.678906797477225,
      "grad_norm": 0.5192902088165283,
      "learning_rate": 0.00013534082755321691,
      "loss": 0.0903,
      "step": 11979
    },
    {
      "epoch": 1.6790469516468116,
      "grad_norm": 0.015255287289619446,
      "learning_rate": 0.00013532647691939727,
      "loss": 0.0018,
      "step": 11980
    },
    {
      "epoch": 1.679187105816398,
      "grad_norm": 0.7141194343566895,
      "learning_rate": 0.0001353121262855776,
      "loss": 0.0434,
      "step": 11981
    },
    {
      "epoch": 1.6793272599859845,
      "grad_norm": 0.345865935087204,
      "learning_rate": 0.00013529777565175795,
      "loss": 0.0654,
      "step": 11982
    },
    {
      "epoch": 1.679467414155571,
      "grad_norm": 0.5288901329040527,
      "learning_rate": 0.00013528342501793828,
      "loss": 0.0869,
      "step": 11983
    },
    {
      "epoch": 1.6796075683251577,
      "grad_norm": 0.980785608291626,
      "learning_rate": 0.0001352690743841186,
      "loss": 0.084,
      "step": 11984
    },
    {
      "epoch": 1.6797477224947444,
      "grad_norm": 0.10756443440914154,
      "learning_rate": 0.00013525472375029897,
      "loss": 0.0123,
      "step": 11985
    },
    {
      "epoch": 1.6798878766643308,
      "grad_norm": 0.08367438614368439,
      "learning_rate": 0.0001352403731164793,
      "loss": 0.0065,
      "step": 11986
    },
    {
      "epoch": 1.6800280308339173,
      "grad_norm": 0.31768864393234253,
      "learning_rate": 0.00013522602248265965,
      "loss": 0.0268,
      "step": 11987
    },
    {
      "epoch": 1.6801681850035037,
      "grad_norm": 0.44531092047691345,
      "learning_rate": 0.00013521167184883998,
      "loss": 0.0426,
      "step": 11988
    },
    {
      "epoch": 1.6803083391730904,
      "grad_norm": 0.12503424286842346,
      "learning_rate": 0.0001351973212150203,
      "loss": 0.0128,
      "step": 11989
    },
    {
      "epoch": 1.680448493342677,
      "grad_norm": 0.5373589992523193,
      "learning_rate": 0.00013518297058120066,
      "loss": 0.0679,
      "step": 11990
    },
    {
      "epoch": 1.6805886475122636,
      "grad_norm": 0.2363787293434143,
      "learning_rate": 0.000135168619947381,
      "loss": 0.027,
      "step": 11991
    },
    {
      "epoch": 1.68072880168185,
      "grad_norm": 0.29088252782821655,
      "learning_rate": 0.00013515426931356135,
      "loss": 0.025,
      "step": 11992
    },
    {
      "epoch": 1.6808689558514365,
      "grad_norm": 0.36560165882110596,
      "learning_rate": 0.00013513991867974168,
      "loss": 0.0442,
      "step": 11993
    },
    {
      "epoch": 1.6810091100210232,
      "grad_norm": 0.1881323754787445,
      "learning_rate": 0.000135125568045922,
      "loss": 0.0537,
      "step": 11994
    },
    {
      "epoch": 1.6811492641906096,
      "grad_norm": 0.2522684335708618,
      "learning_rate": 0.00013511121741210236,
      "loss": 0.0568,
      "step": 11995
    },
    {
      "epoch": 1.6812894183601963,
      "grad_norm": 0.2547159194946289,
      "learning_rate": 0.00013509686677828272,
      "loss": 0.0457,
      "step": 11996
    },
    {
      "epoch": 1.6814295725297828,
      "grad_norm": 0.12966875731945038,
      "learning_rate": 0.00013508251614446304,
      "loss": 0.0122,
      "step": 11997
    },
    {
      "epoch": 1.6815697266993692,
      "grad_norm": 0.22635617852210999,
      "learning_rate": 0.00013506816551064337,
      "loss": 0.0732,
      "step": 11998
    },
    {
      "epoch": 1.681709880868956,
      "grad_norm": 0.3793053925037384,
      "learning_rate": 0.0001350538148768237,
      "loss": 0.071,
      "step": 11999
    },
    {
      "epoch": 1.6818500350385424,
      "grad_norm": 0.17158959805965424,
      "learning_rate": 0.00013503946424300406,
      "loss": 0.0558,
      "step": 12000
    },
    {
      "epoch": 1.681990189208129,
      "grad_norm": 0.4342384934425354,
      "learning_rate": 0.00013502511360918439,
      "loss": 0.1214,
      "step": 12001
    },
    {
      "epoch": 1.6821303433777155,
      "grad_norm": 0.15962554514408112,
      "learning_rate": 0.00013501076297536471,
      "loss": 0.0177,
      "step": 12002
    },
    {
      "epoch": 1.682270497547302,
      "grad_norm": 0.25122955441474915,
      "learning_rate": 0.00013499641234154507,
      "loss": 0.0481,
      "step": 12003
    },
    {
      "epoch": 1.6824106517168884,
      "grad_norm": 0.9374120831489563,
      "learning_rate": 0.00013498206170772543,
      "loss": 0.0802,
      "step": 12004
    },
    {
      "epoch": 1.682550805886475,
      "grad_norm": 0.5195601582527161,
      "learning_rate": 0.00013496771107390575,
      "loss": 0.0588,
      "step": 12005
    },
    {
      "epoch": 1.6826909600560618,
      "grad_norm": 0.232902392745018,
      "learning_rate": 0.00013495336044008608,
      "loss": 0.0627,
      "step": 12006
    },
    {
      "epoch": 1.6828311142256482,
      "grad_norm": 0.3968820571899414,
      "learning_rate": 0.0001349390098062664,
      "loss": 0.0711,
      "step": 12007
    },
    {
      "epoch": 1.6829712683952347,
      "grad_norm": 0.415885865688324,
      "learning_rate": 0.00013492465917244677,
      "loss": 0.0662,
      "step": 12008
    },
    {
      "epoch": 1.6831114225648212,
      "grad_norm": 0.22248360514640808,
      "learning_rate": 0.00013491030853862712,
      "loss": 0.0297,
      "step": 12009
    },
    {
      "epoch": 1.6832515767344078,
      "grad_norm": 0.13106459379196167,
      "learning_rate": 0.00013489595790480745,
      "loss": 0.0233,
      "step": 12010
    },
    {
      "epoch": 1.6833917309039945,
      "grad_norm": 0.4800136685371399,
      "learning_rate": 0.00013488160727098778,
      "loss": 0.0604,
      "step": 12011
    },
    {
      "epoch": 1.683531885073581,
      "grad_norm": 0.5863081812858582,
      "learning_rate": 0.00013486725663716814,
      "loss": 0.0523,
      "step": 12012
    },
    {
      "epoch": 1.6836720392431674,
      "grad_norm": 0.2394528090953827,
      "learning_rate": 0.00013485290600334846,
      "loss": 0.1183,
      "step": 12013
    },
    {
      "epoch": 1.683812193412754,
      "grad_norm": 0.9481812119483948,
      "learning_rate": 0.00013483855536952882,
      "loss": 0.0397,
      "step": 12014
    },
    {
      "epoch": 1.6839523475823406,
      "grad_norm": 0.410428524017334,
      "learning_rate": 0.00013482420473570915,
      "loss": 0.0487,
      "step": 12015
    },
    {
      "epoch": 1.6840925017519273,
      "grad_norm": 0.49864962697029114,
      "learning_rate": 0.00013480985410188948,
      "loss": 0.0247,
      "step": 12016
    },
    {
      "epoch": 1.6842326559215137,
      "grad_norm": 0.1838756948709488,
      "learning_rate": 0.00013479550346806983,
      "loss": 0.0294,
      "step": 12017
    },
    {
      "epoch": 1.6843728100911002,
      "grad_norm": 0.30036959052085876,
      "learning_rate": 0.00013478115283425016,
      "loss": 0.0465,
      "step": 12018
    },
    {
      "epoch": 1.6845129642606866,
      "grad_norm": 0.2414606660604477,
      "learning_rate": 0.00013476680220043052,
      "loss": 0.0271,
      "step": 12019
    },
    {
      "epoch": 1.6846531184302733,
      "grad_norm": 0.08831758052110672,
      "learning_rate": 0.00013475245156661085,
      "loss": 0.0126,
      "step": 12020
    },
    {
      "epoch": 1.68479327259986,
      "grad_norm": 0.6286929845809937,
      "learning_rate": 0.00013473810093279117,
      "loss": 0.0612,
      "step": 12021
    },
    {
      "epoch": 1.6849334267694465,
      "grad_norm": 0.21967561542987823,
      "learning_rate": 0.00013472375029897153,
      "loss": 0.0214,
      "step": 12022
    },
    {
      "epoch": 1.685073580939033,
      "grad_norm": 0.19949324429035187,
      "learning_rate": 0.00013470939966515189,
      "loss": 0.0152,
      "step": 12023
    },
    {
      "epoch": 1.6852137351086194,
      "grad_norm": 0.27890774607658386,
      "learning_rate": 0.00013469504903133221,
      "loss": 0.0303,
      "step": 12024
    },
    {
      "epoch": 1.685353889278206,
      "grad_norm": 0.3173663020133972,
      "learning_rate": 0.00013468069839751254,
      "loss": 0.0398,
      "step": 12025
    },
    {
      "epoch": 1.6854940434477925,
      "grad_norm": 0.23873330652713776,
      "learning_rate": 0.00013466634776369287,
      "loss": 0.0552,
      "step": 12026
    },
    {
      "epoch": 1.6856341976173792,
      "grad_norm": 0.27269285917282104,
      "learning_rate": 0.00013465199712987323,
      "loss": 0.1135,
      "step": 12027
    },
    {
      "epoch": 1.6857743517869657,
      "grad_norm": 0.425577312707901,
      "learning_rate": 0.00013463764649605358,
      "loss": 0.0232,
      "step": 12028
    },
    {
      "epoch": 1.6859145059565521,
      "grad_norm": 0.40970930457115173,
      "learning_rate": 0.0001346232958622339,
      "loss": 0.0219,
      "step": 12029
    },
    {
      "epoch": 1.6860546601261388,
      "grad_norm": 0.5845174789428711,
      "learning_rate": 0.00013460894522841424,
      "loss": 0.0905,
      "step": 12030
    },
    {
      "epoch": 1.6861948142957253,
      "grad_norm": 0.5986014604568481,
      "learning_rate": 0.0001345945945945946,
      "loss": 0.0415,
      "step": 12031
    },
    {
      "epoch": 1.686334968465312,
      "grad_norm": 0.4809368848800659,
      "learning_rate": 0.00013458024396077492,
      "loss": 0.0374,
      "step": 12032
    },
    {
      "epoch": 1.6864751226348984,
      "grad_norm": 0.5704452991485596,
      "learning_rate": 0.00013456589332695525,
      "loss": 0.0962,
      "step": 12033
    },
    {
      "epoch": 1.6866152768044849,
      "grad_norm": 1.1903120279312134,
      "learning_rate": 0.00013455154269313558,
      "loss": 0.2128,
      "step": 12034
    },
    {
      "epoch": 1.6867554309740713,
      "grad_norm": 0.45623716711997986,
      "learning_rate": 0.00013453719205931594,
      "loss": 0.0673,
      "step": 12035
    },
    {
      "epoch": 1.686895585143658,
      "grad_norm": 0.1426125466823578,
      "learning_rate": 0.0001345228414254963,
      "loss": 0.0301,
      "step": 12036
    },
    {
      "epoch": 1.6870357393132447,
      "grad_norm": 0.2537117600440979,
      "learning_rate": 0.00013450849079167662,
      "loss": 0.0449,
      "step": 12037
    },
    {
      "epoch": 1.6871758934828311,
      "grad_norm": 0.365544855594635,
      "learning_rate": 0.00013449414015785695,
      "loss": 0.0585,
      "step": 12038
    },
    {
      "epoch": 1.6873160476524176,
      "grad_norm": 0.13807782530784607,
      "learning_rate": 0.0001344797895240373,
      "loss": 0.0296,
      "step": 12039
    },
    {
      "epoch": 1.687456201822004,
      "grad_norm": 0.9301925897598267,
      "learning_rate": 0.00013446543889021763,
      "loss": 0.0942,
      "step": 12040
    },
    {
      "epoch": 1.6875963559915907,
      "grad_norm": 0.1984456330537796,
      "learning_rate": 0.000134451088256398,
      "loss": 0.0462,
      "step": 12041
    },
    {
      "epoch": 1.6877365101611774,
      "grad_norm": 0.610122799873352,
      "learning_rate": 0.00013443673762257832,
      "loss": 0.0649,
      "step": 12042
    },
    {
      "epoch": 1.6878766643307639,
      "grad_norm": 0.6308504939079285,
      "learning_rate": 0.00013442238698875865,
      "loss": 0.0709,
      "step": 12043
    },
    {
      "epoch": 1.6880168185003503,
      "grad_norm": 0.2242647409439087,
      "learning_rate": 0.000134408036354939,
      "loss": 0.0362,
      "step": 12044
    },
    {
      "epoch": 1.6881569726699368,
      "grad_norm": 0.18173405528068542,
      "learning_rate": 0.00013439368572111933,
      "loss": 0.0323,
      "step": 12045
    },
    {
      "epoch": 1.6882971268395235,
      "grad_norm": 0.33567434549331665,
      "learning_rate": 0.00013437933508729969,
      "loss": 0.0517,
      "step": 12046
    },
    {
      "epoch": 1.6884372810091102,
      "grad_norm": 0.5238831043243408,
      "learning_rate": 0.00013436498445348001,
      "loss": 0.034,
      "step": 12047
    },
    {
      "epoch": 1.6885774351786966,
      "grad_norm": 0.9118971824645996,
      "learning_rate": 0.00013435063381966034,
      "loss": 0.0924,
      "step": 12048
    },
    {
      "epoch": 1.688717589348283,
      "grad_norm": 0.15738297998905182,
      "learning_rate": 0.0001343362831858407,
      "loss": 0.0116,
      "step": 12049
    },
    {
      "epoch": 1.6888577435178695,
      "grad_norm": 0.6067864894866943,
      "learning_rate": 0.00013432193255202103,
      "loss": 0.0649,
      "step": 12050
    },
    {
      "epoch": 1.6889978976874562,
      "grad_norm": 0.5457295179367065,
      "learning_rate": 0.00013430758191820138,
      "loss": 0.0477,
      "step": 12051
    },
    {
      "epoch": 1.689138051857043,
      "grad_norm": 1.9149231910705566,
      "learning_rate": 0.0001342932312843817,
      "loss": 0.0733,
      "step": 12052
    },
    {
      "epoch": 1.6892782060266294,
      "grad_norm": 0.2596793472766876,
      "learning_rate": 0.00013427888065056204,
      "loss": 0.0376,
      "step": 12053
    },
    {
      "epoch": 1.6894183601962158,
      "grad_norm": 0.2406296730041504,
      "learning_rate": 0.0001342645300167424,
      "loss": 0.0341,
      "step": 12054
    },
    {
      "epoch": 1.6895585143658023,
      "grad_norm": 0.14211568236351013,
      "learning_rate": 0.00013425017938292275,
      "loss": 0.0225,
      "step": 12055
    },
    {
      "epoch": 1.689698668535389,
      "grad_norm": 1.1997653245925903,
      "learning_rate": 0.00013423582874910308,
      "loss": 0.0957,
      "step": 12056
    },
    {
      "epoch": 1.6898388227049754,
      "grad_norm": 0.20580078661441803,
      "learning_rate": 0.0001342214781152834,
      "loss": 0.0561,
      "step": 12057
    },
    {
      "epoch": 1.689978976874562,
      "grad_norm": 0.16731882095336914,
      "learning_rate": 0.00013420712748146376,
      "loss": 0.0247,
      "step": 12058
    },
    {
      "epoch": 1.6901191310441486,
      "grad_norm": 0.37701407074928284,
      "learning_rate": 0.0001341927768476441,
      "loss": 0.0487,
      "step": 12059
    },
    {
      "epoch": 1.690259285213735,
      "grad_norm": 0.15563316643238068,
      "learning_rate": 0.00013417842621382445,
      "loss": 0.0283,
      "step": 12060
    },
    {
      "epoch": 1.6903994393833215,
      "grad_norm": 0.3963906168937683,
      "learning_rate": 0.00013416407558000478,
      "loss": 0.0656,
      "step": 12061
    },
    {
      "epoch": 1.6905395935529082,
      "grad_norm": 0.3319525122642517,
      "learning_rate": 0.0001341497249461851,
      "loss": 0.0333,
      "step": 12062
    },
    {
      "epoch": 1.6906797477224949,
      "grad_norm": 0.4549751877784729,
      "learning_rate": 0.00013413537431236546,
      "loss": 0.0458,
      "step": 12063
    },
    {
      "epoch": 1.6908199018920813,
      "grad_norm": 0.16253913938999176,
      "learning_rate": 0.0001341210236785458,
      "loss": 0.0335,
      "step": 12064
    },
    {
      "epoch": 1.6909600560616678,
      "grad_norm": 0.9755932092666626,
      "learning_rate": 0.00013410667304472612,
      "loss": 0.0516,
      "step": 12065
    },
    {
      "epoch": 1.6911002102312542,
      "grad_norm": 0.08407410234212875,
      "learning_rate": 0.00013409232241090647,
      "loss": 0.0117,
      "step": 12066
    },
    {
      "epoch": 1.691240364400841,
      "grad_norm": 0.5107954740524292,
      "learning_rate": 0.0001340779717770868,
      "loss": 0.1058,
      "step": 12067
    },
    {
      "epoch": 1.6913805185704276,
      "grad_norm": 0.12123275548219681,
      "learning_rate": 0.00013406362114326716,
      "loss": 0.0214,
      "step": 12068
    },
    {
      "epoch": 1.691520672740014,
      "grad_norm": 0.2946118712425232,
      "learning_rate": 0.0001340492705094475,
      "loss": 0.0335,
      "step": 12069
    },
    {
      "epoch": 1.6916608269096005,
      "grad_norm": 0.49260231852531433,
      "learning_rate": 0.00013403491987562782,
      "loss": 0.0509,
      "step": 12070
    },
    {
      "epoch": 1.691800981079187,
      "grad_norm": 0.524235725402832,
      "learning_rate": 0.00013402056924180817,
      "loss": 0.1372,
      "step": 12071
    },
    {
      "epoch": 1.6919411352487737,
      "grad_norm": 0.3608382046222687,
      "learning_rate": 0.0001340062186079885,
      "loss": 0.0568,
      "step": 12072
    },
    {
      "epoch": 1.6920812894183603,
      "grad_norm": 0.5260193943977356,
      "learning_rate": 0.00013399186797416886,
      "loss": 0.061,
      "step": 12073
    },
    {
      "epoch": 1.6922214435879468,
      "grad_norm": 0.4607023596763611,
      "learning_rate": 0.00013397751734034918,
      "loss": 0.0478,
      "step": 12074
    },
    {
      "epoch": 1.6923615977575333,
      "grad_norm": 0.19261333346366882,
      "learning_rate": 0.0001339631667065295,
      "loss": 0.0246,
      "step": 12075
    },
    {
      "epoch": 1.6925017519271197,
      "grad_norm": 0.6916095614433289,
      "learning_rate": 0.00013394881607270987,
      "loss": 0.0836,
      "step": 12076
    },
    {
      "epoch": 1.6926419060967064,
      "grad_norm": 0.379151850938797,
      "learning_rate": 0.0001339344654388902,
      "loss": 0.0368,
      "step": 12077
    },
    {
      "epoch": 1.692782060266293,
      "grad_norm": 0.40913838148117065,
      "learning_rate": 0.00013392011480507055,
      "loss": 0.0511,
      "step": 12078
    },
    {
      "epoch": 1.6929222144358795,
      "grad_norm": 0.45786041021347046,
      "learning_rate": 0.00013390576417125088,
      "loss": 0.0673,
      "step": 12079
    },
    {
      "epoch": 1.693062368605466,
      "grad_norm": 1.3063641786575317,
      "learning_rate": 0.0001338914135374312,
      "loss": 0.0423,
      "step": 12080
    },
    {
      "epoch": 1.6932025227750525,
      "grad_norm": 0.31361743807792664,
      "learning_rate": 0.00013387706290361157,
      "loss": 0.0708,
      "step": 12081
    },
    {
      "epoch": 1.6933426769446391,
      "grad_norm": 0.5006605386734009,
      "learning_rate": 0.00013386271226979192,
      "loss": 0.044,
      "step": 12082
    },
    {
      "epoch": 1.6934828311142256,
      "grad_norm": 0.31330978870391846,
      "learning_rate": 0.00013384836163597225,
      "loss": 0.0758,
      "step": 12083
    },
    {
      "epoch": 1.6936229852838123,
      "grad_norm": 0.5071033835411072,
      "learning_rate": 0.00013383401100215258,
      "loss": 0.0746,
      "step": 12084
    },
    {
      "epoch": 1.6937631394533987,
      "grad_norm": 1.7994585037231445,
      "learning_rate": 0.0001338196603683329,
      "loss": 0.1479,
      "step": 12085
    },
    {
      "epoch": 1.6939032936229852,
      "grad_norm": 0.04736752808094025,
      "learning_rate": 0.00013380530973451326,
      "loss": 0.0065,
      "step": 12086
    },
    {
      "epoch": 1.6940434477925719,
      "grad_norm": 0.14732757210731506,
      "learning_rate": 0.00013379095910069362,
      "loss": 0.0176,
      "step": 12087
    },
    {
      "epoch": 1.6941836019621583,
      "grad_norm": 0.11965247243642807,
      "learning_rate": 0.00013377660846687395,
      "loss": 0.0264,
      "step": 12088
    },
    {
      "epoch": 1.694323756131745,
      "grad_norm": 0.3270857632160187,
      "learning_rate": 0.00013376225783305428,
      "loss": 0.0382,
      "step": 12089
    },
    {
      "epoch": 1.6944639103013315,
      "grad_norm": 0.2812128961086273,
      "learning_rate": 0.00013374790719923463,
      "loss": 0.0311,
      "step": 12090
    },
    {
      "epoch": 1.694604064470918,
      "grad_norm": 0.24704235792160034,
      "learning_rate": 0.00013373355656541496,
      "loss": 0.0595,
      "step": 12091
    },
    {
      "epoch": 1.6947442186405044,
      "grad_norm": 0.4344119727611542,
      "learning_rate": 0.00013371920593159532,
      "loss": 0.0698,
      "step": 12092
    },
    {
      "epoch": 1.694884372810091,
      "grad_norm": 0.11858907341957092,
      "learning_rate": 0.00013370485529777564,
      "loss": 0.0173,
      "step": 12093
    },
    {
      "epoch": 1.6950245269796778,
      "grad_norm": 0.14710776507854462,
      "learning_rate": 0.00013369050466395597,
      "loss": 0.0139,
      "step": 12094
    },
    {
      "epoch": 1.6951646811492642,
      "grad_norm": 0.16257426142692566,
      "learning_rate": 0.00013367615403013633,
      "loss": 0.015,
      "step": 12095
    },
    {
      "epoch": 1.6953048353188507,
      "grad_norm": 0.3220033645629883,
      "learning_rate": 0.00013366180339631666,
      "loss": 0.0327,
      "step": 12096
    },
    {
      "epoch": 1.6954449894884371,
      "grad_norm": 0.3824523091316223,
      "learning_rate": 0.00013364745276249698,
      "loss": 0.0514,
      "step": 12097
    },
    {
      "epoch": 1.6955851436580238,
      "grad_norm": 0.5479860305786133,
      "learning_rate": 0.00013363310212867734,
      "loss": 0.0653,
      "step": 12098
    },
    {
      "epoch": 1.6957252978276105,
      "grad_norm": 0.22309648990631104,
      "learning_rate": 0.00013361875149485767,
      "loss": 0.0644,
      "step": 12099
    },
    {
      "epoch": 1.695865451997197,
      "grad_norm": 0.314375638961792,
      "learning_rate": 0.00013360440086103802,
      "loss": 0.0634,
      "step": 12100
    },
    {
      "epoch": 1.6960056061667834,
      "grad_norm": 0.14579297602176666,
      "learning_rate": 0.00013359005022721835,
      "loss": 0.0251,
      "step": 12101
    },
    {
      "epoch": 1.6961457603363699,
      "grad_norm": 0.22480905055999756,
      "learning_rate": 0.00013357569959339868,
      "loss": 0.0156,
      "step": 12102
    },
    {
      "epoch": 1.6962859145059566,
      "grad_norm": 0.21081236004829407,
      "learning_rate": 0.00013356134895957904,
      "loss": 0.0454,
      "step": 12103
    },
    {
      "epoch": 1.6964260686755432,
      "grad_norm": 0.34576883912086487,
      "learning_rate": 0.00013354699832575937,
      "loss": 0.1142,
      "step": 12104
    },
    {
      "epoch": 1.6965662228451297,
      "grad_norm": 0.12170752882957458,
      "learning_rate": 0.00013353264769193972,
      "loss": 0.0082,
      "step": 12105
    },
    {
      "epoch": 1.6967063770147162,
      "grad_norm": 0.5246629118919373,
      "learning_rate": 0.00013351829705812005,
      "loss": 0.0463,
      "step": 12106
    },
    {
      "epoch": 1.6968465311843026,
      "grad_norm": 1.375457525253296,
      "learning_rate": 0.00013350394642430038,
      "loss": 0.0543,
      "step": 12107
    },
    {
      "epoch": 1.6969866853538893,
      "grad_norm": 0.6639530658721924,
      "learning_rate": 0.00013348959579048073,
      "loss": 0.0436,
      "step": 12108
    },
    {
      "epoch": 1.697126839523476,
      "grad_norm": 0.8654889464378357,
      "learning_rate": 0.0001334752451566611,
      "loss": 0.0966,
      "step": 12109
    },
    {
      "epoch": 1.6972669936930624,
      "grad_norm": 0.3038753271102905,
      "learning_rate": 0.00013346089452284142,
      "loss": 0.0818,
      "step": 12110
    },
    {
      "epoch": 1.697407147862649,
      "grad_norm": 0.8238230347633362,
      "learning_rate": 0.00013344654388902175,
      "loss": 0.1143,
      "step": 12111
    },
    {
      "epoch": 1.6975473020322354,
      "grad_norm": 0.30154138803482056,
      "learning_rate": 0.00013343219325520208,
      "loss": 0.0342,
      "step": 12112
    },
    {
      "epoch": 1.697687456201822,
      "grad_norm": 0.18004603683948517,
      "learning_rate": 0.00013341784262138243,
      "loss": 0.0174,
      "step": 12113
    },
    {
      "epoch": 1.6978276103714085,
      "grad_norm": 0.2873094975948334,
      "learning_rate": 0.0001334034919875628,
      "loss": 0.1025,
      "step": 12114
    },
    {
      "epoch": 1.6979677645409952,
      "grad_norm": 0.17505615949630737,
      "learning_rate": 0.00013338914135374312,
      "loss": 0.0705,
      "step": 12115
    },
    {
      "epoch": 1.6981079187105816,
      "grad_norm": 0.37265855073928833,
      "learning_rate": 0.00013337479071992344,
      "loss": 0.0374,
      "step": 12116
    },
    {
      "epoch": 1.698248072880168,
      "grad_norm": 0.2789835035800934,
      "learning_rate": 0.0001333604400861038,
      "loss": 0.0722,
      "step": 12117
    },
    {
      "epoch": 1.6983882270497548,
      "grad_norm": 0.4448922276496887,
      "learning_rate": 0.00013334608945228413,
      "loss": 0.0728,
      "step": 12118
    },
    {
      "epoch": 1.6985283812193412,
      "grad_norm": 0.4655461609363556,
      "learning_rate": 0.00013333173881846448,
      "loss": 0.0247,
      "step": 12119
    },
    {
      "epoch": 1.698668535388928,
      "grad_norm": 0.20748305320739746,
      "learning_rate": 0.0001333173881846448,
      "loss": 0.0115,
      "step": 12120
    },
    {
      "epoch": 1.6988086895585144,
      "grad_norm": 0.32459160685539246,
      "learning_rate": 0.00013330303755082514,
      "loss": 0.0642,
      "step": 12121
    },
    {
      "epoch": 1.6989488437281008,
      "grad_norm": 0.46259671449661255,
      "learning_rate": 0.0001332886869170055,
      "loss": 0.0867,
      "step": 12122
    },
    {
      "epoch": 1.6990889978976873,
      "grad_norm": 0.2355385571718216,
      "learning_rate": 0.00013327433628318583,
      "loss": 0.0339,
      "step": 12123
    },
    {
      "epoch": 1.699229152067274,
      "grad_norm": 0.31749945878982544,
      "learning_rate": 0.00013325998564936618,
      "loss": 0.076,
      "step": 12124
    },
    {
      "epoch": 1.6993693062368607,
      "grad_norm": 0.23552361130714417,
      "learning_rate": 0.0001332456350155465,
      "loss": 0.0266,
      "step": 12125
    },
    {
      "epoch": 1.6995094604064471,
      "grad_norm": 0.19038169085979462,
      "learning_rate": 0.00013323128438172684,
      "loss": 0.0676,
      "step": 12126
    },
    {
      "epoch": 1.6996496145760336,
      "grad_norm": 0.4941211938858032,
      "learning_rate": 0.0001332169337479072,
      "loss": 0.0839,
      "step": 12127
    },
    {
      "epoch": 1.69978976874562,
      "grad_norm": 0.32394328713417053,
      "learning_rate": 0.00013320258311408752,
      "loss": 0.0378,
      "step": 12128
    },
    {
      "epoch": 1.6999299229152067,
      "grad_norm": 0.11059322953224182,
      "learning_rate": 0.00013318823248026785,
      "loss": 0.0058,
      "step": 12129
    },
    {
      "epoch": 1.7000700770847934,
      "grad_norm": 0.8874800801277161,
      "learning_rate": 0.0001331738818464482,
      "loss": 0.0688,
      "step": 12130
    },
    {
      "epoch": 1.7002102312543799,
      "grad_norm": 0.2018873542547226,
      "learning_rate": 0.00013315953121262854,
      "loss": 0.0138,
      "step": 12131
    },
    {
      "epoch": 1.7003503854239663,
      "grad_norm": 0.3651524782180786,
      "learning_rate": 0.0001331451805788089,
      "loss": 0.0537,
      "step": 12132
    },
    {
      "epoch": 1.7004905395935528,
      "grad_norm": 0.4067952036857605,
      "learning_rate": 0.00013313082994498922,
      "loss": 0.0213,
      "step": 12133
    },
    {
      "epoch": 1.7006306937631395,
      "grad_norm": 0.7026558518409729,
      "learning_rate": 0.00013311647931116955,
      "loss": 0.0631,
      "step": 12134
    },
    {
      "epoch": 1.7007708479327261,
      "grad_norm": 1.1000531911849976,
      "learning_rate": 0.0001331021286773499,
      "loss": 0.2094,
      "step": 12135
    },
    {
      "epoch": 1.7009110021023126,
      "grad_norm": 0.17267359793186188,
      "learning_rate": 0.00013308777804353026,
      "loss": 0.0398,
      "step": 12136
    },
    {
      "epoch": 1.701051156271899,
      "grad_norm": 0.18348808586597443,
      "learning_rate": 0.0001330734274097106,
      "loss": 0.0305,
      "step": 12137
    },
    {
      "epoch": 1.7011913104414855,
      "grad_norm": 0.10993392020463943,
      "learning_rate": 0.00013305907677589092,
      "loss": 0.0143,
      "step": 12138
    },
    {
      "epoch": 1.7013314646110722,
      "grad_norm": 0.2789594233036041,
      "learning_rate": 0.00013304472614207125,
      "loss": 0.0598,
      "step": 12139
    },
    {
      "epoch": 1.701471618780659,
      "grad_norm": 0.2597116529941559,
      "learning_rate": 0.0001330303755082516,
      "loss": 0.039,
      "step": 12140
    },
    {
      "epoch": 1.7016117729502453,
      "grad_norm": 0.18684706091880798,
      "learning_rate": 0.00013301602487443196,
      "loss": 0.0336,
      "step": 12141
    },
    {
      "epoch": 1.7017519271198318,
      "grad_norm": 0.3200320899486542,
      "learning_rate": 0.00013300167424061229,
      "loss": 0.0778,
      "step": 12142
    },
    {
      "epoch": 1.7018920812894183,
      "grad_norm": 0.14672788977622986,
      "learning_rate": 0.00013298732360679261,
      "loss": 0.027,
      "step": 12143
    },
    {
      "epoch": 1.702032235459005,
      "grad_norm": 0.5173974633216858,
      "learning_rate": 0.00013297297297297297,
      "loss": 0.1067,
      "step": 12144
    },
    {
      "epoch": 1.7021723896285914,
      "grad_norm": 0.4318787157535553,
      "learning_rate": 0.0001329586223391533,
      "loss": 0.0261,
      "step": 12145
    },
    {
      "epoch": 1.702312543798178,
      "grad_norm": 0.28429049253463745,
      "learning_rate": 0.00013294427170533365,
      "loss": 0.0607,
      "step": 12146
    },
    {
      "epoch": 1.7024526979677645,
      "grad_norm": 0.2931499183177948,
      "learning_rate": 0.00013292992107151398,
      "loss": 0.0574,
      "step": 12147
    },
    {
      "epoch": 1.702592852137351,
      "grad_norm": 0.20604759454727173,
      "learning_rate": 0.0001329155704376943,
      "loss": 0.0437,
      "step": 12148
    },
    {
      "epoch": 1.7027330063069375,
      "grad_norm": 0.372398316860199,
      "learning_rate": 0.00013290121980387467,
      "loss": 0.0277,
      "step": 12149
    },
    {
      "epoch": 1.7028731604765242,
      "grad_norm": 0.0969495102763176,
      "learning_rate": 0.000132886869170055,
      "loss": 0.008,
      "step": 12150
    },
    {
      "epoch": 1.7030133146461108,
      "grad_norm": 0.08814334124326706,
      "learning_rate": 0.00013287251853623535,
      "loss": 0.0124,
      "step": 12151
    },
    {
      "epoch": 1.7031534688156973,
      "grad_norm": 0.25535672903060913,
      "learning_rate": 0.00013285816790241568,
      "loss": 0.0823,
      "step": 12152
    },
    {
      "epoch": 1.7032936229852838,
      "grad_norm": 0.17418822646141052,
      "learning_rate": 0.000132843817268596,
      "loss": 0.0089,
      "step": 12153
    },
    {
      "epoch": 1.7034337771548702,
      "grad_norm": 0.1341678351163864,
      "learning_rate": 0.00013282946663477636,
      "loss": 0.054,
      "step": 12154
    },
    {
      "epoch": 1.703573931324457,
      "grad_norm": 0.24280256032943726,
      "learning_rate": 0.0001328151160009567,
      "loss": 0.0277,
      "step": 12155
    },
    {
      "epoch": 1.7037140854940436,
      "grad_norm": 0.5688652992248535,
      "learning_rate": 0.00013280076536713705,
      "loss": 0.0478,
      "step": 12156
    },
    {
      "epoch": 1.70385423966363,
      "grad_norm": 0.20240440964698792,
      "learning_rate": 0.00013278641473331738,
      "loss": 0.0232,
      "step": 12157
    },
    {
      "epoch": 1.7039943938332165,
      "grad_norm": 0.5226565599441528,
      "learning_rate": 0.0001327720640994977,
      "loss": 0.077,
      "step": 12158
    },
    {
      "epoch": 1.704134548002803,
      "grad_norm": 0.27666181325912476,
      "learning_rate": 0.00013275771346567806,
      "loss": 0.0347,
      "step": 12159
    },
    {
      "epoch": 1.7042747021723896,
      "grad_norm": 0.17397163808345795,
      "learning_rate": 0.0001327433628318584,
      "loss": 0.0748,
      "step": 12160
    },
    {
      "epoch": 1.7044148563419763,
      "grad_norm": 0.2695830464363098,
      "learning_rate": 0.00013272901219803874,
      "loss": 0.0285,
      "step": 12161
    },
    {
      "epoch": 1.7045550105115628,
      "grad_norm": 0.22675687074661255,
      "learning_rate": 0.00013271466156421907,
      "loss": 0.0253,
      "step": 12162
    },
    {
      "epoch": 1.7046951646811492,
      "grad_norm": 0.1300095021724701,
      "learning_rate": 0.0001327003109303994,
      "loss": 0.0138,
      "step": 12163
    },
    {
      "epoch": 1.7048353188507357,
      "grad_norm": 0.12879200279712677,
      "learning_rate": 0.00013268596029657976,
      "loss": 0.0354,
      "step": 12164
    },
    {
      "epoch": 1.7049754730203224,
      "grad_norm": 0.23569846153259277,
      "learning_rate": 0.00013267160966276009,
      "loss": 0.031,
      "step": 12165
    },
    {
      "epoch": 1.705115627189909,
      "grad_norm": 0.07775438576936722,
      "learning_rate": 0.00013265725902894041,
      "loss": 0.0122,
      "step": 12166
    },
    {
      "epoch": 1.7052557813594955,
      "grad_norm": 0.27824512124061584,
      "learning_rate": 0.00013264290839512077,
      "loss": 0.0183,
      "step": 12167
    },
    {
      "epoch": 1.705395935529082,
      "grad_norm": 0.20355655252933502,
      "learning_rate": 0.00013262855776130113,
      "loss": 0.0202,
      "step": 12168
    },
    {
      "epoch": 1.7055360896986684,
      "grad_norm": 0.12762348353862762,
      "learning_rate": 0.00013261420712748145,
      "loss": 0.0131,
      "step": 12169
    },
    {
      "epoch": 1.7056762438682551,
      "grad_norm": 0.33934980630874634,
      "learning_rate": 0.00013259985649366178,
      "loss": 0.0518,
      "step": 12170
    },
    {
      "epoch": 1.7058163980378416,
      "grad_norm": 0.27709197998046875,
      "learning_rate": 0.00013258550585984214,
      "loss": 0.047,
      "step": 12171
    },
    {
      "epoch": 1.7059565522074283,
      "grad_norm": 0.39742711186408997,
      "learning_rate": 0.00013257115522602247,
      "loss": 0.0687,
      "step": 12172
    },
    {
      "epoch": 1.7060967063770147,
      "grad_norm": 0.649774432182312,
      "learning_rate": 0.00013255680459220282,
      "loss": 0.0601,
      "step": 12173
    },
    {
      "epoch": 1.7062368605466012,
      "grad_norm": 0.5390459895133972,
      "learning_rate": 0.00013254245395838315,
      "loss": 0.0225,
      "step": 12174
    },
    {
      "epoch": 1.7063770147161879,
      "grad_norm": 0.5721665024757385,
      "learning_rate": 0.00013252810332456348,
      "loss": 0.1058,
      "step": 12175
    },
    {
      "epoch": 1.7065171688857743,
      "grad_norm": 1.3042230606079102,
      "learning_rate": 0.00013251375269074384,
      "loss": 0.0652,
      "step": 12176
    },
    {
      "epoch": 1.706657323055361,
      "grad_norm": 0.8957727551460266,
      "learning_rate": 0.00013249940205692416,
      "loss": 0.0337,
      "step": 12177
    },
    {
      "epoch": 1.7067974772249475,
      "grad_norm": 0.06076275557279587,
      "learning_rate": 0.00013248505142310452,
      "loss": 0.0046,
      "step": 12178
    },
    {
      "epoch": 1.706937631394534,
      "grad_norm": 0.5878314971923828,
      "learning_rate": 0.00013247070078928485,
      "loss": 0.0282,
      "step": 12179
    },
    {
      "epoch": 1.7070777855641204,
      "grad_norm": 0.22794058918952942,
      "learning_rate": 0.00013245635015546518,
      "loss": 0.0242,
      "step": 12180
    },
    {
      "epoch": 1.707217939733707,
      "grad_norm": 0.3879918158054352,
      "learning_rate": 0.00013244199952164553,
      "loss": 0.0516,
      "step": 12181
    },
    {
      "epoch": 1.7073580939032937,
      "grad_norm": 0.2316022366285324,
      "learning_rate": 0.00013242764888782586,
      "loss": 0.0895,
      "step": 12182
    },
    {
      "epoch": 1.7074982480728802,
      "grad_norm": 0.9732415676116943,
      "learning_rate": 0.00013241329825400622,
      "loss": 0.0567,
      "step": 12183
    },
    {
      "epoch": 1.7076384022424667,
      "grad_norm": 0.1015130802989006,
      "learning_rate": 0.00013239894762018655,
      "loss": 0.005,
      "step": 12184
    },
    {
      "epoch": 1.7077785564120531,
      "grad_norm": 1.2262755632400513,
      "learning_rate": 0.00013238459698636687,
      "loss": 0.0316,
      "step": 12185
    },
    {
      "epoch": 1.7079187105816398,
      "grad_norm": 0.2844727635383606,
      "learning_rate": 0.00013237024635254723,
      "loss": 0.0766,
      "step": 12186
    },
    {
      "epoch": 1.7080588647512265,
      "grad_norm": 0.9894722104072571,
      "learning_rate": 0.00013235589571872759,
      "loss": 0.1161,
      "step": 12187
    },
    {
      "epoch": 1.708199018920813,
      "grad_norm": 0.35616248846054077,
      "learning_rate": 0.00013234154508490791,
      "loss": 0.096,
      "step": 12188
    },
    {
      "epoch": 1.7083391730903994,
      "grad_norm": 0.2448006123304367,
      "learning_rate": 0.00013232719445108824,
      "loss": 0.0161,
      "step": 12189
    },
    {
      "epoch": 1.7084793272599859,
      "grad_norm": 0.41324755549430847,
      "learning_rate": 0.00013231284381726857,
      "loss": 0.09,
      "step": 12190
    },
    {
      "epoch": 1.7086194814295725,
      "grad_norm": 0.4179147481918335,
      "learning_rate": 0.00013229849318344893,
      "loss": 0.0459,
      "step": 12191
    },
    {
      "epoch": 1.7087596355991592,
      "grad_norm": 0.14356225728988647,
      "learning_rate": 0.00013228414254962926,
      "loss": 0.0494,
      "step": 12192
    },
    {
      "epoch": 1.7088997897687457,
      "grad_norm": 0.05739745497703552,
      "learning_rate": 0.0001322697919158096,
      "loss": 0.0096,
      "step": 12193
    },
    {
      "epoch": 1.7090399439383321,
      "grad_norm": 0.1322007030248642,
      "learning_rate": 0.00013225544128198994,
      "loss": 0.0158,
      "step": 12194
    },
    {
      "epoch": 1.7091800981079186,
      "grad_norm": 0.25732842087745667,
      "learning_rate": 0.0001322410906481703,
      "loss": 0.0288,
      "step": 12195
    },
    {
      "epoch": 1.7093202522775053,
      "grad_norm": 0.22450783848762512,
      "learning_rate": 0.00013222674001435062,
      "loss": 0.0429,
      "step": 12196
    },
    {
      "epoch": 1.709460406447092,
      "grad_norm": 0.5321676731109619,
      "learning_rate": 0.00013221238938053095,
      "loss": 0.0947,
      "step": 12197
    },
    {
      "epoch": 1.7096005606166784,
      "grad_norm": 0.5037558674812317,
      "learning_rate": 0.00013219803874671128,
      "loss": 0.0685,
      "step": 12198
    },
    {
      "epoch": 1.7097407147862649,
      "grad_norm": 0.4735731780529022,
      "learning_rate": 0.00013218368811289164,
      "loss": 0.0445,
      "step": 12199
    },
    {
      "epoch": 1.7098808689558513,
      "grad_norm": 0.2632163465023041,
      "learning_rate": 0.000132169337479072,
      "loss": 0.0321,
      "step": 12200
    },
    {
      "epoch": 1.710021023125438,
      "grad_norm": 0.5000625848770142,
      "learning_rate": 0.00013215498684525232,
      "loss": 0.029,
      "step": 12201
    },
    {
      "epoch": 1.7101611772950245,
      "grad_norm": 0.22909273207187653,
      "learning_rate": 0.00013214063621143265,
      "loss": 0.0413,
      "step": 12202
    },
    {
      "epoch": 1.7103013314646112,
      "grad_norm": 0.2071690708398819,
      "learning_rate": 0.000132126285577613,
      "loss": 0.0193,
      "step": 12203
    },
    {
      "epoch": 1.7104414856341976,
      "grad_norm": 0.262101948261261,
      "learning_rate": 0.00013211193494379333,
      "loss": 0.0414,
      "step": 12204
    },
    {
      "epoch": 1.710581639803784,
      "grad_norm": 0.09843871742486954,
      "learning_rate": 0.0001320975843099737,
      "loss": 0.0139,
      "step": 12205
    },
    {
      "epoch": 1.7107217939733705,
      "grad_norm": 0.3094974160194397,
      "learning_rate": 0.00013208323367615402,
      "loss": 0.0611,
      "step": 12206
    },
    {
      "epoch": 1.7108619481429572,
      "grad_norm": 0.28097647428512573,
      "learning_rate": 0.00013206888304233435,
      "loss": 0.0511,
      "step": 12207
    },
    {
      "epoch": 1.711002102312544,
      "grad_norm": 0.366079717874527,
      "learning_rate": 0.0001320545324085147,
      "loss": 0.0388,
      "step": 12208
    },
    {
      "epoch": 1.7111422564821304,
      "grad_norm": 0.24471531808376312,
      "learning_rate": 0.00013204018177469503,
      "loss": 0.0487,
      "step": 12209
    },
    {
      "epoch": 1.7112824106517168,
      "grad_norm": 0.27892395853996277,
      "learning_rate": 0.00013202583114087539,
      "loss": 0.0428,
      "step": 12210
    },
    {
      "epoch": 1.7114225648213033,
      "grad_norm": 0.24249088764190674,
      "learning_rate": 0.00013201148050705571,
      "loss": 0.0806,
      "step": 12211
    },
    {
      "epoch": 1.71156271899089,
      "grad_norm": 0.1949528455734253,
      "learning_rate": 0.00013199712987323604,
      "loss": 0.0327,
      "step": 12212
    },
    {
      "epoch": 1.7117028731604766,
      "grad_norm": 0.5891454219818115,
      "learning_rate": 0.0001319827792394164,
      "loss": 0.0363,
      "step": 12213
    },
    {
      "epoch": 1.711843027330063,
      "grad_norm": 0.34219926595687866,
      "learning_rate": 0.00013196842860559675,
      "loss": 0.0498,
      "step": 12214
    },
    {
      "epoch": 1.7119831814996496,
      "grad_norm": 0.463893860578537,
      "learning_rate": 0.00013195407797177708,
      "loss": 0.039,
      "step": 12215
    },
    {
      "epoch": 1.712123335669236,
      "grad_norm": 0.21816730499267578,
      "learning_rate": 0.0001319397273379574,
      "loss": 0.0299,
      "step": 12216
    },
    {
      "epoch": 1.7122634898388227,
      "grad_norm": 0.2313939779996872,
      "learning_rate": 0.00013192537670413774,
      "loss": 0.0229,
      "step": 12217
    },
    {
      "epoch": 1.7124036440084094,
      "grad_norm": 0.37385743856430054,
      "learning_rate": 0.0001319110260703181,
      "loss": 0.0861,
      "step": 12218
    },
    {
      "epoch": 1.7125437981779958,
      "grad_norm": 0.09055466949939728,
      "learning_rate": 0.00013189667543649845,
      "loss": 0.0094,
      "step": 12219
    },
    {
      "epoch": 1.7126839523475823,
      "grad_norm": 0.39656758308410645,
      "learning_rate": 0.00013188232480267878,
      "loss": 0.0283,
      "step": 12220
    },
    {
      "epoch": 1.7128241065171688,
      "grad_norm": 0.9020304083824158,
      "learning_rate": 0.0001318679741688591,
      "loss": 0.0887,
      "step": 12221
    },
    {
      "epoch": 1.7129642606867554,
      "grad_norm": 0.869001030921936,
      "learning_rate": 0.00013185362353503946,
      "loss": 0.0993,
      "step": 12222
    },
    {
      "epoch": 1.7131044148563421,
      "grad_norm": 0.31396469473838806,
      "learning_rate": 0.0001318392729012198,
      "loss": 0.0577,
      "step": 12223
    },
    {
      "epoch": 1.7132445690259286,
      "grad_norm": 0.28609851002693176,
      "learning_rate": 0.00013182492226740012,
      "loss": 0.0629,
      "step": 12224
    },
    {
      "epoch": 1.713384723195515,
      "grad_norm": 0.18209898471832275,
      "learning_rate": 0.00013181057163358048,
      "loss": 0.0248,
      "step": 12225
    },
    {
      "epoch": 1.7135248773651015,
      "grad_norm": 0.4208248257637024,
      "learning_rate": 0.0001317962209997608,
      "loss": 0.0244,
      "step": 12226
    },
    {
      "epoch": 1.7136650315346882,
      "grad_norm": 0.15248838067054749,
      "learning_rate": 0.00013178187036594116,
      "loss": 0.0122,
      "step": 12227
    },
    {
      "epoch": 1.7138051857042746,
      "grad_norm": 0.16308845579624176,
      "learning_rate": 0.0001317675197321215,
      "loss": 0.0251,
      "step": 12228
    },
    {
      "epoch": 1.7139453398738613,
      "grad_norm": 1.4393203258514404,
      "learning_rate": 0.00013175316909830182,
      "loss": 0.2633,
      "step": 12229
    },
    {
      "epoch": 1.7140854940434478,
      "grad_norm": 0.49943456053733826,
      "learning_rate": 0.00013173881846448217,
      "loss": 0.0958,
      "step": 12230
    },
    {
      "epoch": 1.7142256482130342,
      "grad_norm": 0.3061561584472656,
      "learning_rate": 0.0001317244678306625,
      "loss": 0.036,
      "step": 12231
    },
    {
      "epoch": 1.714365802382621,
      "grad_norm": 0.6713414788246155,
      "learning_rate": 0.00013171011719684286,
      "loss": 0.0952,
      "step": 12232
    },
    {
      "epoch": 1.7145059565522074,
      "grad_norm": 0.06964185833930969,
      "learning_rate": 0.0001316957665630232,
      "loss": 0.0071,
      "step": 12233
    },
    {
      "epoch": 1.714646110721794,
      "grad_norm": 0.7420777678489685,
      "learning_rate": 0.00013168141592920352,
      "loss": 0.1465,
      "step": 12234
    },
    {
      "epoch": 1.7147862648913805,
      "grad_norm": 0.8891483545303345,
      "learning_rate": 0.00013166706529538387,
      "loss": 0.1991,
      "step": 12235
    },
    {
      "epoch": 1.714926419060967,
      "grad_norm": 0.16871501505374908,
      "learning_rate": 0.0001316527146615642,
      "loss": 0.0497,
      "step": 12236
    },
    {
      "epoch": 1.7150665732305534,
      "grad_norm": 0.2156693935394287,
      "learning_rate": 0.00013163836402774456,
      "loss": 0.059,
      "step": 12237
    },
    {
      "epoch": 1.7152067274001401,
      "grad_norm": 0.31691744923591614,
      "learning_rate": 0.00013162401339392488,
      "loss": 0.0239,
      "step": 12238
    },
    {
      "epoch": 1.7153468815697268,
      "grad_norm": 0.45139381289482117,
      "learning_rate": 0.0001316096627601052,
      "loss": 0.0754,
      "step": 12239
    },
    {
      "epoch": 1.7154870357393133,
      "grad_norm": 0.34668833017349243,
      "learning_rate": 0.00013159531212628557,
      "loss": 0.1447,
      "step": 12240
    },
    {
      "epoch": 1.7156271899088997,
      "grad_norm": 0.43707892298698425,
      "learning_rate": 0.00013158096149246592,
      "loss": 0.0425,
      "step": 12241
    },
    {
      "epoch": 1.7157673440784862,
      "grad_norm": 0.3323785662651062,
      "learning_rate": 0.00013156661085864625,
      "loss": 0.0686,
      "step": 12242
    },
    {
      "epoch": 1.7159074982480729,
      "grad_norm": 0.24211536347866058,
      "learning_rate": 0.00013155226022482658,
      "loss": 0.0579,
      "step": 12243
    },
    {
      "epoch": 1.7160476524176596,
      "grad_norm": 0.5556278824806213,
      "learning_rate": 0.0001315379095910069,
      "loss": 0.0423,
      "step": 12244
    },
    {
      "epoch": 1.716187806587246,
      "grad_norm": 0.45714691281318665,
      "learning_rate": 0.00013152355895718727,
      "loss": 0.0294,
      "step": 12245
    },
    {
      "epoch": 1.7163279607568325,
      "grad_norm": 0.39925500750541687,
      "learning_rate": 0.00013150920832336762,
      "loss": 0.0571,
      "step": 12246
    },
    {
      "epoch": 1.716468114926419,
      "grad_norm": 0.15344958007335663,
      "learning_rate": 0.00013149485768954795,
      "loss": 0.0304,
      "step": 12247
    },
    {
      "epoch": 1.7166082690960056,
      "grad_norm": 0.2583089768886566,
      "learning_rate": 0.00013148050705572828,
      "loss": 0.0322,
      "step": 12248
    },
    {
      "epoch": 1.7167484232655923,
      "grad_norm": 0.22716905176639557,
      "learning_rate": 0.00013146615642190863,
      "loss": 0.0525,
      "step": 12249
    },
    {
      "epoch": 1.7168885774351788,
      "grad_norm": 0.3833186626434326,
      "learning_rate": 0.00013145180578808896,
      "loss": 0.0286,
      "step": 12250
    },
    {
      "epoch": 1.7170287316047652,
      "grad_norm": 0.24826520681381226,
      "learning_rate": 0.00013143745515426932,
      "loss": 0.0217,
      "step": 12251
    },
    {
      "epoch": 1.7171688857743517,
      "grad_norm": 0.289717435836792,
      "learning_rate": 0.00013142310452044965,
      "loss": 0.0764,
      "step": 12252
    },
    {
      "epoch": 1.7173090399439384,
      "grad_norm": 0.2958870828151703,
      "learning_rate": 0.00013140875388662998,
      "loss": 0.0631,
      "step": 12253
    },
    {
      "epoch": 1.717449194113525,
      "grad_norm": 0.37041962146759033,
      "learning_rate": 0.00013139440325281033,
      "loss": 0.0559,
      "step": 12254
    },
    {
      "epoch": 1.7175893482831115,
      "grad_norm": 0.18747790157794952,
      "learning_rate": 0.00013138005261899066,
      "loss": 0.0479,
      "step": 12255
    },
    {
      "epoch": 1.717729502452698,
      "grad_norm": 1.0447160005569458,
      "learning_rate": 0.000131365701985171,
      "loss": 0.0373,
      "step": 12256
    },
    {
      "epoch": 1.7178696566222844,
      "grad_norm": 0.4998255968093872,
      "learning_rate": 0.00013135135135135134,
      "loss": 0.0722,
      "step": 12257
    },
    {
      "epoch": 1.718009810791871,
      "grad_norm": 0.2755364775657654,
      "learning_rate": 0.00013133700071753167,
      "loss": 0.0466,
      "step": 12258
    },
    {
      "epoch": 1.7181499649614576,
      "grad_norm": 0.27703437209129333,
      "learning_rate": 0.00013132265008371203,
      "loss": 0.058,
      "step": 12259
    },
    {
      "epoch": 1.7182901191310442,
      "grad_norm": 0.34295231103897095,
      "learning_rate": 0.00013130829944989236,
      "loss": 0.0421,
      "step": 12260
    },
    {
      "epoch": 1.7184302733006307,
      "grad_norm": 0.3623409867286682,
      "learning_rate": 0.00013129394881607268,
      "loss": 0.1082,
      "step": 12261
    },
    {
      "epoch": 1.7185704274702172,
      "grad_norm": 0.16262243688106537,
      "learning_rate": 0.00013127959818225304,
      "loss": 0.0292,
      "step": 12262
    },
    {
      "epoch": 1.7187105816398038,
      "grad_norm": 0.24371591210365295,
      "learning_rate": 0.00013126524754843337,
      "loss": 0.0196,
      "step": 12263
    },
    {
      "epoch": 1.7188507358093903,
      "grad_norm": 0.20839661359786987,
      "learning_rate": 0.00013125089691461372,
      "loss": 0.0124,
      "step": 12264
    },
    {
      "epoch": 1.718990889978977,
      "grad_norm": 0.26452597975730896,
      "learning_rate": 0.00013123654628079405,
      "loss": 0.0393,
      "step": 12265
    },
    {
      "epoch": 1.7191310441485634,
      "grad_norm": 0.22257104516029358,
      "learning_rate": 0.00013122219564697438,
      "loss": 0.0178,
      "step": 12266
    },
    {
      "epoch": 1.71927119831815,
      "grad_norm": 0.4206107258796692,
      "learning_rate": 0.00013120784501315474,
      "loss": 0.0825,
      "step": 12267
    },
    {
      "epoch": 1.7194113524877364,
      "grad_norm": 0.2450208216905594,
      "learning_rate": 0.00013119349437933507,
      "loss": 0.0632,
      "step": 12268
    },
    {
      "epoch": 1.719551506657323,
      "grad_norm": 0.238607719540596,
      "learning_rate": 0.00013117914374551542,
      "loss": 0.0551,
      "step": 12269
    },
    {
      "epoch": 1.7196916608269097,
      "grad_norm": 0.5203726887702942,
      "learning_rate": 0.00013116479311169575,
      "loss": 0.0445,
      "step": 12270
    },
    {
      "epoch": 1.7198318149964962,
      "grad_norm": 0.13038702309131622,
      "learning_rate": 0.00013115044247787608,
      "loss": 0.0114,
      "step": 12271
    },
    {
      "epoch": 1.7199719691660826,
      "grad_norm": 0.2569502890110016,
      "learning_rate": 0.00013113609184405643,
      "loss": 0.0567,
      "step": 12272
    },
    {
      "epoch": 1.720112123335669,
      "grad_norm": 0.6292478442192078,
      "learning_rate": 0.0001311217412102368,
      "loss": 0.0685,
      "step": 12273
    },
    {
      "epoch": 1.7202522775052558,
      "grad_norm": 0.13167506456375122,
      "learning_rate": 0.00013110739057641712,
      "loss": 0.0083,
      "step": 12274
    },
    {
      "epoch": 1.7203924316748425,
      "grad_norm": 0.36134523153305054,
      "learning_rate": 0.00013109303994259745,
      "loss": 0.0549,
      "step": 12275
    },
    {
      "epoch": 1.720532585844429,
      "grad_norm": 0.285233736038208,
      "learning_rate": 0.0001310786893087778,
      "loss": 0.0448,
      "step": 12276
    },
    {
      "epoch": 1.7206727400140154,
      "grad_norm": 0.21198295056819916,
      "learning_rate": 0.00013106433867495813,
      "loss": 0.0136,
      "step": 12277
    },
    {
      "epoch": 1.7208128941836018,
      "grad_norm": 0.12797924876213074,
      "learning_rate": 0.0001310499880411385,
      "loss": 0.0052,
      "step": 12278
    },
    {
      "epoch": 1.7209530483531885,
      "grad_norm": 0.34781551361083984,
      "learning_rate": 0.00013103563740731882,
      "loss": 0.0534,
      "step": 12279
    },
    {
      "epoch": 1.7210932025227752,
      "grad_norm": 2.350677967071533,
      "learning_rate": 0.00013102128677349914,
      "loss": 0.1168,
      "step": 12280
    },
    {
      "epoch": 1.7212333566923617,
      "grad_norm": 0.22225646674633026,
      "learning_rate": 0.0001310069361396795,
      "loss": 0.0242,
      "step": 12281
    },
    {
      "epoch": 1.7213735108619481,
      "grad_norm": 0.9159325957298279,
      "learning_rate": 0.00013099258550585983,
      "loss": 0.0751,
      "step": 12282
    },
    {
      "epoch": 1.7215136650315346,
      "grad_norm": 0.6331193447113037,
      "learning_rate": 0.00013097823487204018,
      "loss": 0.0395,
      "step": 12283
    },
    {
      "epoch": 1.7216538192011213,
      "grad_norm": 1.145039677619934,
      "learning_rate": 0.0001309638842382205,
      "loss": 0.0717,
      "step": 12284
    },
    {
      "epoch": 1.721793973370708,
      "grad_norm": 1.5468385219573975,
      "learning_rate": 0.00013094953360440084,
      "loss": 0.0691,
      "step": 12285
    },
    {
      "epoch": 1.7219341275402944,
      "grad_norm": 0.18451419472694397,
      "learning_rate": 0.0001309351829705812,
      "loss": 0.0217,
      "step": 12286
    },
    {
      "epoch": 1.7220742817098809,
      "grad_norm": 0.3441013693809509,
      "learning_rate": 0.00013092083233676153,
      "loss": 0.0656,
      "step": 12287
    },
    {
      "epoch": 1.7222144358794673,
      "grad_norm": 0.12094729393720627,
      "learning_rate": 0.00013090648170294188,
      "loss": 0.0167,
      "step": 12288
    },
    {
      "epoch": 1.722354590049054,
      "grad_norm": 0.1621893048286438,
      "learning_rate": 0.0001308921310691222,
      "loss": 0.0256,
      "step": 12289
    },
    {
      "epoch": 1.7224947442186405,
      "grad_norm": 0.15836484730243683,
      "learning_rate": 0.00013087778043530254,
      "loss": 0.0198,
      "step": 12290
    },
    {
      "epoch": 1.7226348983882271,
      "grad_norm": 0.22203406691551208,
      "learning_rate": 0.0001308634298014829,
      "loss": 0.0115,
      "step": 12291
    },
    {
      "epoch": 1.7227750525578136,
      "grad_norm": 0.3655462861061096,
      "learning_rate": 0.00013084907916766322,
      "loss": 0.0419,
      "step": 12292
    },
    {
      "epoch": 1.7229152067274,
      "grad_norm": 0.17145375907421112,
      "learning_rate": 0.00013083472853384355,
      "loss": 0.0251,
      "step": 12293
    },
    {
      "epoch": 1.7230553608969865,
      "grad_norm": 0.2717093229293823,
      "learning_rate": 0.0001308203779000239,
      "loss": 0.0469,
      "step": 12294
    },
    {
      "epoch": 1.7231955150665732,
      "grad_norm": 0.05062802881002426,
      "learning_rate": 0.00013080602726620424,
      "loss": 0.0055,
      "step": 12295
    },
    {
      "epoch": 1.7233356692361599,
      "grad_norm": 0.24106605350971222,
      "learning_rate": 0.0001307916766323846,
      "loss": 0.041,
      "step": 12296
    },
    {
      "epoch": 1.7234758234057463,
      "grad_norm": 0.1237282082438469,
      "learning_rate": 0.00013077732599856492,
      "loss": 0.0229,
      "step": 12297
    },
    {
      "epoch": 1.7236159775753328,
      "grad_norm": 0.20759360492229462,
      "learning_rate": 0.00013076297536474525,
      "loss": 0.0456,
      "step": 12298
    },
    {
      "epoch": 1.7237561317449193,
      "grad_norm": 0.5264992117881775,
      "learning_rate": 0.0001307486247309256,
      "loss": 0.0601,
      "step": 12299
    },
    {
      "epoch": 1.723896285914506,
      "grad_norm": 0.4196541905403137,
      "learning_rate": 0.00013073427409710596,
      "loss": 0.0481,
      "step": 12300
    },
    {
      "epoch": 1.7240364400840926,
      "grad_norm": 0.33874571323394775,
      "learning_rate": 0.0001307199234632863,
      "loss": 0.033,
      "step": 12301
    },
    {
      "epoch": 1.724176594253679,
      "grad_norm": 0.30149513483047485,
      "learning_rate": 0.00013070557282946662,
      "loss": 0.0505,
      "step": 12302
    },
    {
      "epoch": 1.7243167484232655,
      "grad_norm": 0.18182025849819183,
      "learning_rate": 0.00013069122219564695,
      "loss": 0.03,
      "step": 12303
    },
    {
      "epoch": 1.724456902592852,
      "grad_norm": 0.45294544100761414,
      "learning_rate": 0.0001306768715618273,
      "loss": 0.0627,
      "step": 12304
    },
    {
      "epoch": 1.7245970567624387,
      "grad_norm": 0.18393060564994812,
      "learning_rate": 0.00013066252092800766,
      "loss": 0.0215,
      "step": 12305
    },
    {
      "epoch": 1.7247372109320254,
      "grad_norm": 0.8840623497962952,
      "learning_rate": 0.00013064817029418799,
      "loss": 0.0389,
      "step": 12306
    },
    {
      "epoch": 1.7248773651016118,
      "grad_norm": 0.4182656705379486,
      "learning_rate": 0.00013063381966036831,
      "loss": 0.0559,
      "step": 12307
    },
    {
      "epoch": 1.7250175192711983,
      "grad_norm": 0.3140980303287506,
      "learning_rate": 0.00013061946902654867,
      "loss": 0.081,
      "step": 12308
    },
    {
      "epoch": 1.7251576734407847,
      "grad_norm": 0.24665966629981995,
      "learning_rate": 0.000130605118392729,
      "loss": 0.0196,
      "step": 12309
    },
    {
      "epoch": 1.7252978276103714,
      "grad_norm": 1.1210683584213257,
      "learning_rate": 0.00013059076775890935,
      "loss": 0.0796,
      "step": 12310
    },
    {
      "epoch": 1.725437981779958,
      "grad_norm": 0.28329533338546753,
      "learning_rate": 0.00013057641712508968,
      "loss": 0.053,
      "step": 12311
    },
    {
      "epoch": 1.7255781359495446,
      "grad_norm": 0.28426188230514526,
      "learning_rate": 0.00013056206649127,
      "loss": 0.0288,
      "step": 12312
    },
    {
      "epoch": 1.725718290119131,
      "grad_norm": 0.4659711420536041,
      "learning_rate": 0.00013054771585745037,
      "loss": 0.0547,
      "step": 12313
    },
    {
      "epoch": 1.7258584442887175,
      "grad_norm": 0.7289787530899048,
      "learning_rate": 0.0001305333652236307,
      "loss": 0.1183,
      "step": 12314
    },
    {
      "epoch": 1.7259985984583042,
      "grad_norm": 0.1230320855975151,
      "learning_rate": 0.00013051901458981105,
      "loss": 0.043,
      "step": 12315
    },
    {
      "epoch": 1.7261387526278906,
      "grad_norm": 0.3219144642353058,
      "learning_rate": 0.00013050466395599138,
      "loss": 0.0472,
      "step": 12316
    },
    {
      "epoch": 1.7262789067974773,
      "grad_norm": 0.20390263199806213,
      "learning_rate": 0.0001304903133221717,
      "loss": 0.0097,
      "step": 12317
    },
    {
      "epoch": 1.7264190609670638,
      "grad_norm": 0.17698785662651062,
      "learning_rate": 0.00013047596268835206,
      "loss": 0.0582,
      "step": 12318
    },
    {
      "epoch": 1.7265592151366502,
      "grad_norm": 0.7207682132720947,
      "learning_rate": 0.0001304616120545324,
      "loss": 0.1119,
      "step": 12319
    },
    {
      "epoch": 1.726699369306237,
      "grad_norm": 0.4833196699619293,
      "learning_rate": 0.00013044726142071275,
      "loss": 0.0427,
      "step": 12320
    },
    {
      "epoch": 1.7268395234758234,
      "grad_norm": 0.3859612047672272,
      "learning_rate": 0.00013043291078689308,
      "loss": 0.0292,
      "step": 12321
    },
    {
      "epoch": 1.72697967764541,
      "grad_norm": 0.8018611073493958,
      "learning_rate": 0.0001304185601530734,
      "loss": 0.0267,
      "step": 12322
    },
    {
      "epoch": 1.7271198318149965,
      "grad_norm": 0.127554252743721,
      "learning_rate": 0.00013040420951925376,
      "loss": 0.0102,
      "step": 12323
    },
    {
      "epoch": 1.727259985984583,
      "grad_norm": 0.09663911908864975,
      "learning_rate": 0.0001303898588854341,
      "loss": 0.0144,
      "step": 12324
    },
    {
      "epoch": 1.7274001401541694,
      "grad_norm": 0.5925981402397156,
      "learning_rate": 0.00013037550825161442,
      "loss": 0.0556,
      "step": 12325
    },
    {
      "epoch": 1.727540294323756,
      "grad_norm": 0.18153047561645508,
      "learning_rate": 0.00013036115761779477,
      "loss": 0.018,
      "step": 12326
    },
    {
      "epoch": 1.7276804484933428,
      "grad_norm": 0.3498193025588989,
      "learning_rate": 0.00013034680698397513,
      "loss": 0.032,
      "step": 12327
    },
    {
      "epoch": 1.7278206026629293,
      "grad_norm": 0.11203636974096298,
      "learning_rate": 0.00013033245635015546,
      "loss": 0.0137,
      "step": 12328
    },
    {
      "epoch": 1.7279607568325157,
      "grad_norm": 0.44035327434539795,
      "learning_rate": 0.00013031810571633579,
      "loss": 0.0518,
      "step": 12329
    },
    {
      "epoch": 1.7281009110021022,
      "grad_norm": 0.42255085706710815,
      "learning_rate": 0.00013030375508251611,
      "loss": 0.0316,
      "step": 12330
    },
    {
      "epoch": 1.7282410651716889,
      "grad_norm": 1.9962173700332642,
      "learning_rate": 0.00013028940444869647,
      "loss": 0.1031,
      "step": 12331
    },
    {
      "epoch": 1.7283812193412755,
      "grad_norm": 0.9048944115638733,
      "learning_rate": 0.00013027505381487683,
      "loss": 0.0503,
      "step": 12332
    },
    {
      "epoch": 1.728521373510862,
      "grad_norm": 0.9001339077949524,
      "learning_rate": 0.00013026070318105715,
      "loss": 0.0656,
      "step": 12333
    },
    {
      "epoch": 1.7286615276804485,
      "grad_norm": 0.5447548031806946,
      "learning_rate": 0.00013024635254723748,
      "loss": 0.0347,
      "step": 12334
    },
    {
      "epoch": 1.728801681850035,
      "grad_norm": 2.1652841567993164,
      "learning_rate": 0.00013023200191341784,
      "loss": 0.1731,
      "step": 12335
    },
    {
      "epoch": 1.7289418360196216,
      "grad_norm": 0.33333322405815125,
      "learning_rate": 0.00013021765127959817,
      "loss": 0.0396,
      "step": 12336
    },
    {
      "epoch": 1.7290819901892083,
      "grad_norm": 0.31905025243759155,
      "learning_rate": 0.00013020330064577852,
      "loss": 0.0305,
      "step": 12337
    },
    {
      "epoch": 1.7292221443587947,
      "grad_norm": 0.3603152632713318,
      "learning_rate": 0.00013018895001195885,
      "loss": 0.0798,
      "step": 12338
    },
    {
      "epoch": 1.7293622985283812,
      "grad_norm": 0.26587095856666565,
      "learning_rate": 0.00013017459937813918,
      "loss": 0.0537,
      "step": 12339
    },
    {
      "epoch": 1.7295024526979677,
      "grad_norm": 0.20560120046138763,
      "learning_rate": 0.00013016024874431954,
      "loss": 0.0293,
      "step": 12340
    },
    {
      "epoch": 1.7296426068675543,
      "grad_norm": 0.21184688806533813,
      "learning_rate": 0.00013014589811049986,
      "loss": 0.0583,
      "step": 12341
    },
    {
      "epoch": 1.729782761037141,
      "grad_norm": 0.2364475429058075,
      "learning_rate": 0.00013013154747668022,
      "loss": 0.0308,
      "step": 12342
    },
    {
      "epoch": 1.7299229152067275,
      "grad_norm": 0.2784612476825714,
      "learning_rate": 0.00013011719684286055,
      "loss": 0.0629,
      "step": 12343
    },
    {
      "epoch": 1.730063069376314,
      "grad_norm": 0.5540226101875305,
      "learning_rate": 0.00013010284620904088,
      "loss": 0.043,
      "step": 12344
    },
    {
      "epoch": 1.7302032235459004,
      "grad_norm": 0.34508460760116577,
      "learning_rate": 0.00013008849557522123,
      "loss": 0.0408,
      "step": 12345
    },
    {
      "epoch": 1.730343377715487,
      "grad_norm": 0.19782060384750366,
      "learning_rate": 0.00013007414494140156,
      "loss": 0.0138,
      "step": 12346
    },
    {
      "epoch": 1.7304835318850735,
      "grad_norm": 0.30429843068122864,
      "learning_rate": 0.00013005979430758192,
      "loss": 0.0624,
      "step": 12347
    },
    {
      "epoch": 1.7306236860546602,
      "grad_norm": 0.13465876877307892,
      "learning_rate": 0.00013004544367376225,
      "loss": 0.015,
      "step": 12348
    },
    {
      "epoch": 1.7307638402242467,
      "grad_norm": 0.4841441214084625,
      "learning_rate": 0.00013003109303994257,
      "loss": 0.096,
      "step": 12349
    },
    {
      "epoch": 1.7309039943938331,
      "grad_norm": 0.2868037521839142,
      "learning_rate": 0.00013001674240612293,
      "loss": 0.0316,
      "step": 12350
    },
    {
      "epoch": 1.7310441485634198,
      "grad_norm": 0.17500494420528412,
      "learning_rate": 0.00013000239177230326,
      "loss": 0.0513,
      "step": 12351
    },
    {
      "epoch": 1.7311843027330063,
      "grad_norm": 0.2766554057598114,
      "learning_rate": 0.00012998804113848361,
      "loss": 0.0488,
      "step": 12352
    },
    {
      "epoch": 1.731324456902593,
      "grad_norm": 0.9096641540527344,
      "learning_rate": 0.00012997369050466394,
      "loss": 0.074,
      "step": 12353
    },
    {
      "epoch": 1.7314646110721794,
      "grad_norm": 0.1750549077987671,
      "learning_rate": 0.0001299593398708443,
      "loss": 0.0231,
      "step": 12354
    },
    {
      "epoch": 1.7316047652417659,
      "grad_norm": 0.3142986297607422,
      "learning_rate": 0.00012994498923702463,
      "loss": 0.0318,
      "step": 12355
    },
    {
      "epoch": 1.7317449194113523,
      "grad_norm": 0.1438191831111908,
      "learning_rate": 0.00012993063860320496,
      "loss": 0.0259,
      "step": 12356
    },
    {
      "epoch": 1.731885073580939,
      "grad_norm": 0.2323543280363083,
      "learning_rate": 0.00012991628796938528,
      "loss": 0.022,
      "step": 12357
    },
    {
      "epoch": 1.7320252277505257,
      "grad_norm": 0.5318880081176758,
      "learning_rate": 0.00012990193733556564,
      "loss": 0.0236,
      "step": 12358
    },
    {
      "epoch": 1.7321653819201122,
      "grad_norm": 0.4395744204521179,
      "learning_rate": 0.000129887586701746,
      "loss": 0.0494,
      "step": 12359
    },
    {
      "epoch": 1.7323055360896986,
      "grad_norm": 0.3908523619174957,
      "learning_rate": 0.00012987323606792632,
      "loss": 0.0524,
      "step": 12360
    },
    {
      "epoch": 1.732445690259285,
      "grad_norm": 0.3383214771747589,
      "learning_rate": 0.00012985888543410665,
      "loss": 0.0327,
      "step": 12361
    },
    {
      "epoch": 1.7325858444288718,
      "grad_norm": 0.24420246481895447,
      "learning_rate": 0.000129844534800287,
      "loss": 0.122,
      "step": 12362
    },
    {
      "epoch": 1.7327259985984584,
      "grad_norm": 0.29437127709388733,
      "learning_rate": 0.00012983018416646734,
      "loss": 0.0965,
      "step": 12363
    },
    {
      "epoch": 1.732866152768045,
      "grad_norm": 0.1582673192024231,
      "learning_rate": 0.0001298158335326477,
      "loss": 0.0381,
      "step": 12364
    },
    {
      "epoch": 1.7330063069376314,
      "grad_norm": 0.490459680557251,
      "learning_rate": 0.00012980148289882802,
      "loss": 0.1023,
      "step": 12365
    },
    {
      "epoch": 1.7331464611072178,
      "grad_norm": 0.3984527289867401,
      "learning_rate": 0.00012978713226500835,
      "loss": 0.0569,
      "step": 12366
    },
    {
      "epoch": 1.7332866152768045,
      "grad_norm": 0.4091145396232605,
      "learning_rate": 0.0001297727816311887,
      "loss": 0.1168,
      "step": 12367
    },
    {
      "epoch": 1.7334267694463912,
      "grad_norm": 0.3573763966560364,
      "learning_rate": 0.00012975843099736903,
      "loss": 0.0488,
      "step": 12368
    },
    {
      "epoch": 1.7335669236159776,
      "grad_norm": 0.1730595976114273,
      "learning_rate": 0.0001297440803635494,
      "loss": 0.0092,
      "step": 12369
    },
    {
      "epoch": 1.733707077785564,
      "grad_norm": 0.26960843801498413,
      "learning_rate": 0.00012972972972972972,
      "loss": 0.0193,
      "step": 12370
    },
    {
      "epoch": 1.7338472319551506,
      "grad_norm": 0.30998438596725464,
      "learning_rate": 0.00012971537909591005,
      "loss": 0.0623,
      "step": 12371
    },
    {
      "epoch": 1.7339873861247372,
      "grad_norm": 0.21065787971019745,
      "learning_rate": 0.0001297010284620904,
      "loss": 0.0435,
      "step": 12372
    },
    {
      "epoch": 1.734127540294324,
      "grad_norm": 0.07457490265369415,
      "learning_rate": 0.00012968667782827073,
      "loss": 0.0076,
      "step": 12373
    },
    {
      "epoch": 1.7342676944639104,
      "grad_norm": 0.21273885667324066,
      "learning_rate": 0.00012967232719445109,
      "loss": 0.0296,
      "step": 12374
    },
    {
      "epoch": 1.7344078486334968,
      "grad_norm": 0.6699028611183167,
      "learning_rate": 0.00012965797656063141,
      "loss": 0.0864,
      "step": 12375
    },
    {
      "epoch": 1.7345480028030833,
      "grad_norm": 0.3473144471645355,
      "learning_rate": 0.00012964362592681174,
      "loss": 0.0371,
      "step": 12376
    },
    {
      "epoch": 1.73468815697267,
      "grad_norm": 0.49465423822402954,
      "learning_rate": 0.0001296292752929921,
      "loss": 0.0852,
      "step": 12377
    },
    {
      "epoch": 1.7348283111422564,
      "grad_norm": 0.4723498821258545,
      "learning_rate": 0.00012961492465917245,
      "loss": 0.012,
      "step": 12378
    },
    {
      "epoch": 1.7349684653118431,
      "grad_norm": 0.5831929445266724,
      "learning_rate": 0.00012960057402535278,
      "loss": 0.1802,
      "step": 12379
    },
    {
      "epoch": 1.7351086194814296,
      "grad_norm": 0.6744077801704407,
      "learning_rate": 0.0001295862233915331,
      "loss": 0.0776,
      "step": 12380
    },
    {
      "epoch": 1.735248773651016,
      "grad_norm": 0.029793143272399902,
      "learning_rate": 0.00012957187275771344,
      "loss": 0.0025,
      "step": 12381
    },
    {
      "epoch": 1.7353889278206025,
      "grad_norm": 0.6070799231529236,
      "learning_rate": 0.0001295575221238938,
      "loss": 0.165,
      "step": 12382
    },
    {
      "epoch": 1.7355290819901892,
      "grad_norm": 0.18198765814304352,
      "learning_rate": 0.00012954317149007412,
      "loss": 0.0123,
      "step": 12383
    },
    {
      "epoch": 1.7356692361597759,
      "grad_norm": 0.28456178307533264,
      "learning_rate": 0.00012952882085625448,
      "loss": 0.0297,
      "step": 12384
    },
    {
      "epoch": 1.7358093903293623,
      "grad_norm": 1.2448192834854126,
      "learning_rate": 0.0001295144702224348,
      "loss": 0.0986,
      "step": 12385
    },
    {
      "epoch": 1.7359495444989488,
      "grad_norm": 0.18204334378242493,
      "learning_rate": 0.00012950011958861516,
      "loss": 0.0382,
      "step": 12386
    },
    {
      "epoch": 1.7360896986685352,
      "grad_norm": 0.2056664228439331,
      "learning_rate": 0.0001294857689547955,
      "loss": 0.0467,
      "step": 12387
    },
    {
      "epoch": 1.736229852838122,
      "grad_norm": 0.5056012272834778,
      "learning_rate": 0.00012947141832097582,
      "loss": 0.0965,
      "step": 12388
    },
    {
      "epoch": 1.7363700070077086,
      "grad_norm": 0.21838964521884918,
      "learning_rate": 0.00012945706768715618,
      "loss": 0.0352,
      "step": 12389
    },
    {
      "epoch": 1.736510161177295,
      "grad_norm": 0.19418999552726746,
      "learning_rate": 0.0001294427170533365,
      "loss": 0.0207,
      "step": 12390
    },
    {
      "epoch": 1.7366503153468815,
      "grad_norm": 0.3644949197769165,
      "learning_rate": 0.00012942836641951686,
      "loss": 0.0532,
      "step": 12391
    },
    {
      "epoch": 1.736790469516468,
      "grad_norm": 0.25886279344558716,
      "learning_rate": 0.0001294140157856972,
      "loss": 0.0347,
      "step": 12392
    },
    {
      "epoch": 1.7369306236860547,
      "grad_norm": 0.2497645914554596,
      "learning_rate": 0.00012939966515187752,
      "loss": 0.0593,
      "step": 12393
    },
    {
      "epoch": 1.7370707778556413,
      "grad_norm": 0.3633016049861908,
      "learning_rate": 0.00012938531451805787,
      "loss": 0.0514,
      "step": 12394
    },
    {
      "epoch": 1.7372109320252278,
      "grad_norm": 0.14847776293754578,
      "learning_rate": 0.0001293709638842382,
      "loss": 0.0262,
      "step": 12395
    },
    {
      "epoch": 1.7373510861948143,
      "grad_norm": 0.5249321460723877,
      "learning_rate": 0.00012935661325041856,
      "loss": 0.0642,
      "step": 12396
    },
    {
      "epoch": 1.7374912403644007,
      "grad_norm": 0.3293091654777527,
      "learning_rate": 0.0001293422626165989,
      "loss": 0.0949,
      "step": 12397
    },
    {
      "epoch": 1.7376313945339874,
      "grad_norm": 0.5687469840049744,
      "learning_rate": 0.00012932791198277922,
      "loss": 0.0358,
      "step": 12398
    },
    {
      "epoch": 1.737771548703574,
      "grad_norm": 0.15011931955814362,
      "learning_rate": 0.00012931356134895957,
      "loss": 0.0195,
      "step": 12399
    },
    {
      "epoch": 1.7379117028731605,
      "grad_norm": 0.07403405010700226,
      "learning_rate": 0.0001292992107151399,
      "loss": 0.01,
      "step": 12400
    },
    {
      "epoch": 1.738051857042747,
      "grad_norm": 0.432157963514328,
      "learning_rate": 0.00012928486008132026,
      "loss": 0.0603,
      "step": 12401
    },
    {
      "epoch": 1.7381920112123335,
      "grad_norm": 0.16148848831653595,
      "learning_rate": 0.00012927050944750058,
      "loss": 0.0218,
      "step": 12402
    },
    {
      "epoch": 1.7383321653819201,
      "grad_norm": 0.5359057188034058,
      "learning_rate": 0.0001292561588136809,
      "loss": 0.0738,
      "step": 12403
    },
    {
      "epoch": 1.7384723195515066,
      "grad_norm": 0.2861936092376709,
      "learning_rate": 0.00012924180817986127,
      "loss": 0.0551,
      "step": 12404
    },
    {
      "epoch": 1.7386124737210933,
      "grad_norm": 0.7421399354934692,
      "learning_rate": 0.00012922745754604162,
      "loss": 0.0626,
      "step": 12405
    },
    {
      "epoch": 1.7387526278906797,
      "grad_norm": 0.1533435434103012,
      "learning_rate": 0.00012921310691222195,
      "loss": 0.0152,
      "step": 12406
    },
    {
      "epoch": 1.7388927820602662,
      "grad_norm": 0.2357248067855835,
      "learning_rate": 0.00012919875627840228,
      "loss": 0.0423,
      "step": 12407
    },
    {
      "epoch": 1.7390329362298529,
      "grad_norm": 0.31651562452316284,
      "learning_rate": 0.0001291844056445826,
      "loss": 0.0935,
      "step": 12408
    },
    {
      "epoch": 1.7391730903994393,
      "grad_norm": 0.4581146538257599,
      "learning_rate": 0.00012917005501076297,
      "loss": 0.0553,
      "step": 12409
    },
    {
      "epoch": 1.739313244569026,
      "grad_norm": 0.24639490246772766,
      "learning_rate": 0.00012915570437694332,
      "loss": 0.0378,
      "step": 12410
    },
    {
      "epoch": 1.7394533987386125,
      "grad_norm": 0.14299839735031128,
      "learning_rate": 0.00012914135374312365,
      "loss": 0.0251,
      "step": 12411
    },
    {
      "epoch": 1.739593552908199,
      "grad_norm": 0.17846910655498505,
      "learning_rate": 0.00012912700310930398,
      "loss": 0.0197,
      "step": 12412
    },
    {
      "epoch": 1.7397337070777854,
      "grad_norm": 0.6507803201675415,
      "learning_rate": 0.00012911265247548433,
      "loss": 0.0966,
      "step": 12413
    },
    {
      "epoch": 1.739873861247372,
      "grad_norm": 0.6710510849952698,
      "learning_rate": 0.00012909830184166466,
      "loss": 0.1416,
      "step": 12414
    },
    {
      "epoch": 1.7400140154169588,
      "grad_norm": 0.1861286610364914,
      "learning_rate": 0.00012908395120784502,
      "loss": 0.0245,
      "step": 12415
    },
    {
      "epoch": 1.7401541695865452,
      "grad_norm": 0.3210747539997101,
      "learning_rate": 0.00012906960057402535,
      "loss": 0.0994,
      "step": 12416
    },
    {
      "epoch": 1.7402943237561317,
      "grad_norm": 0.20362047851085663,
      "learning_rate": 0.00012905524994020567,
      "loss": 0.0204,
      "step": 12417
    },
    {
      "epoch": 1.7404344779257181,
      "grad_norm": 0.22663837671279907,
      "learning_rate": 0.00012904089930638603,
      "loss": 0.0431,
      "step": 12418
    },
    {
      "epoch": 1.7405746320953048,
      "grad_norm": 0.4013076424598694,
      "learning_rate": 0.00012902654867256636,
      "loss": 0.06,
      "step": 12419
    },
    {
      "epoch": 1.7407147862648915,
      "grad_norm": 0.40635499358177185,
      "learning_rate": 0.0001290121980387467,
      "loss": 0.0574,
      "step": 12420
    },
    {
      "epoch": 1.740854940434478,
      "grad_norm": 0.2186608910560608,
      "learning_rate": 0.00012899784740492704,
      "loss": 0.0215,
      "step": 12421
    },
    {
      "epoch": 1.7409950946040644,
      "grad_norm": 0.5750024318695068,
      "learning_rate": 0.00012898349677110737,
      "loss": 0.0252,
      "step": 12422
    },
    {
      "epoch": 1.741135248773651,
      "grad_norm": 0.1944301575422287,
      "learning_rate": 0.00012896914613728773,
      "loss": 0.0235,
      "step": 12423
    },
    {
      "epoch": 1.7412754029432376,
      "grad_norm": 0.6095818877220154,
      "learning_rate": 0.00012895479550346806,
      "loss": 0.0655,
      "step": 12424
    },
    {
      "epoch": 1.7414155571128243,
      "grad_norm": 0.6217092871665955,
      "learning_rate": 0.00012894044486964838,
      "loss": 0.019,
      "step": 12425
    },
    {
      "epoch": 1.7415557112824107,
      "grad_norm": 0.4872361421585083,
      "learning_rate": 0.00012892609423582874,
      "loss": 0.1592,
      "step": 12426
    },
    {
      "epoch": 1.7416958654519972,
      "grad_norm": 0.16807933151721954,
      "learning_rate": 0.00012891174360200907,
      "loss": 0.0339,
      "step": 12427
    },
    {
      "epoch": 1.7418360196215836,
      "grad_norm": 0.20999501645565033,
      "learning_rate": 0.00012889739296818942,
      "loss": 0.0601,
      "step": 12428
    },
    {
      "epoch": 1.7419761737911703,
      "grad_norm": 0.3214479982852936,
      "learning_rate": 0.00012888304233436975,
      "loss": 0.0455,
      "step": 12429
    },
    {
      "epoch": 1.742116327960757,
      "grad_norm": 1.0307008028030396,
      "learning_rate": 0.00012886869170055008,
      "loss": 0.0652,
      "step": 12430
    },
    {
      "epoch": 1.7422564821303435,
      "grad_norm": 0.3579634130001068,
      "learning_rate": 0.00012885434106673044,
      "loss": 0.0578,
      "step": 12431
    },
    {
      "epoch": 1.74239663629993,
      "grad_norm": 0.5004786849021912,
      "learning_rate": 0.0001288399904329108,
      "loss": 0.0156,
      "step": 12432
    },
    {
      "epoch": 1.7425367904695164,
      "grad_norm": 0.2483609914779663,
      "learning_rate": 0.00012882563979909112,
      "loss": 0.0115,
      "step": 12433
    },
    {
      "epoch": 1.742676944639103,
      "grad_norm": 0.5183850526809692,
      "learning_rate": 0.00012881128916527145,
      "loss": 0.0544,
      "step": 12434
    },
    {
      "epoch": 1.7428170988086895,
      "grad_norm": 4.178595542907715,
      "learning_rate": 0.00012879693853145178,
      "loss": 0.2263,
      "step": 12435
    },
    {
      "epoch": 1.7429572529782762,
      "grad_norm": 0.14125262200832367,
      "learning_rate": 0.00012878258789763213,
      "loss": 0.0267,
      "step": 12436
    },
    {
      "epoch": 1.7430974071478627,
      "grad_norm": 0.9474937915802002,
      "learning_rate": 0.0001287682372638125,
      "loss": 0.0879,
      "step": 12437
    },
    {
      "epoch": 1.7432375613174491,
      "grad_norm": 0.4898061454296112,
      "learning_rate": 0.00012875388662999282,
      "loss": 0.066,
      "step": 12438
    },
    {
      "epoch": 1.7433777154870356,
      "grad_norm": 0.22721178829669952,
      "learning_rate": 0.00012873953599617315,
      "loss": 0.0146,
      "step": 12439
    },
    {
      "epoch": 1.7435178696566223,
      "grad_norm": 0.47369828820228577,
      "learning_rate": 0.0001287251853623535,
      "loss": 0.0544,
      "step": 12440
    },
    {
      "epoch": 1.743658023826209,
      "grad_norm": 0.7142036557197571,
      "learning_rate": 0.00012871083472853383,
      "loss": 0.038,
      "step": 12441
    },
    {
      "epoch": 1.7437981779957954,
      "grad_norm": 0.49921169877052307,
      "learning_rate": 0.0001286964840947142,
      "loss": 0.1423,
      "step": 12442
    },
    {
      "epoch": 1.7439383321653819,
      "grad_norm": 0.34828388690948486,
      "learning_rate": 0.00012868213346089452,
      "loss": 0.0498,
      "step": 12443
    },
    {
      "epoch": 1.7440784863349683,
      "grad_norm": 0.48462727665901184,
      "learning_rate": 0.00012866778282707484,
      "loss": 0.0513,
      "step": 12444
    },
    {
      "epoch": 1.744218640504555,
      "grad_norm": 0.24147053062915802,
      "learning_rate": 0.0001286534321932552,
      "loss": 0.0516,
      "step": 12445
    },
    {
      "epoch": 1.7443587946741417,
      "grad_norm": 0.37030845880508423,
      "learning_rate": 0.00012863908155943553,
      "loss": 0.0279,
      "step": 12446
    },
    {
      "epoch": 1.7444989488437281,
      "grad_norm": 0.21156297624111176,
      "learning_rate": 0.00012862473092561588,
      "loss": 0.0402,
      "step": 12447
    },
    {
      "epoch": 1.7446391030133146,
      "grad_norm": 0.26465553045272827,
      "learning_rate": 0.0001286103802917962,
      "loss": 0.0219,
      "step": 12448
    },
    {
      "epoch": 1.744779257182901,
      "grad_norm": 0.43751251697540283,
      "learning_rate": 0.00012859602965797654,
      "loss": 0.0872,
      "step": 12449
    },
    {
      "epoch": 1.7449194113524877,
      "grad_norm": 0.10660535097122192,
      "learning_rate": 0.0001285816790241569,
      "loss": 0.0097,
      "step": 12450
    },
    {
      "epoch": 1.7450595655220744,
      "grad_norm": 0.3799755573272705,
      "learning_rate": 0.00012856732839033723,
      "loss": 0.0468,
      "step": 12451
    },
    {
      "epoch": 1.7451997196916609,
      "grad_norm": 0.20369702577590942,
      "learning_rate": 0.00012855297775651755,
      "loss": 0.0537,
      "step": 12452
    },
    {
      "epoch": 1.7453398738612473,
      "grad_norm": 0.1461147665977478,
      "learning_rate": 0.0001285386271226979,
      "loss": 0.0142,
      "step": 12453
    },
    {
      "epoch": 1.7454800280308338,
      "grad_norm": 0.18199720978736877,
      "learning_rate": 0.00012852427648887824,
      "loss": 0.0504,
      "step": 12454
    },
    {
      "epoch": 1.7456201822004205,
      "grad_norm": 0.5083375573158264,
      "learning_rate": 0.0001285099258550586,
      "loss": 0.088,
      "step": 12455
    },
    {
      "epoch": 1.7457603363700072,
      "grad_norm": 0.17230340838432312,
      "learning_rate": 0.00012849557522123892,
      "loss": 0.0498,
      "step": 12456
    },
    {
      "epoch": 1.7459004905395936,
      "grad_norm": 0.36871016025543213,
      "learning_rate": 0.00012848122458741925,
      "loss": 0.0327,
      "step": 12457
    },
    {
      "epoch": 1.74604064470918,
      "grad_norm": 0.6613603234291077,
      "learning_rate": 0.0001284668739535996,
      "loss": 0.045,
      "step": 12458
    },
    {
      "epoch": 1.7461807988787665,
      "grad_norm": 0.27450990676879883,
      "learning_rate": 0.00012845252331977994,
      "loss": 0.0391,
      "step": 12459
    },
    {
      "epoch": 1.7463209530483532,
      "grad_norm": 0.2846451997756958,
      "learning_rate": 0.0001284381726859603,
      "loss": 0.0205,
      "step": 12460
    },
    {
      "epoch": 1.7464611072179397,
      "grad_norm": 0.22938039898872375,
      "learning_rate": 0.00012842382205214062,
      "loss": 0.0154,
      "step": 12461
    },
    {
      "epoch": 1.7466012613875264,
      "grad_norm": 0.11990543454885483,
      "learning_rate": 0.00012840947141832095,
      "loss": 0.0154,
      "step": 12462
    },
    {
      "epoch": 1.7467414155571128,
      "grad_norm": 0.6128911972045898,
      "learning_rate": 0.0001283951207845013,
      "loss": 0.066,
      "step": 12463
    },
    {
      "epoch": 1.7468815697266993,
      "grad_norm": 0.5303910970687866,
      "learning_rate": 0.00012838077015068166,
      "loss": 0.0696,
      "step": 12464
    },
    {
      "epoch": 1.747021723896286,
      "grad_norm": 0.5912019610404968,
      "learning_rate": 0.000128366419516862,
      "loss": 0.1024,
      "step": 12465
    },
    {
      "epoch": 1.7471618780658724,
      "grad_norm": 0.12200074642896652,
      "learning_rate": 0.00012835206888304232,
      "loss": 0.0113,
      "step": 12466
    },
    {
      "epoch": 1.747302032235459,
      "grad_norm": 0.27069538831710815,
      "learning_rate": 0.00012833771824922267,
      "loss": 0.0235,
      "step": 12467
    },
    {
      "epoch": 1.7474421864050456,
      "grad_norm": 0.1351957768201828,
      "learning_rate": 0.000128323367615403,
      "loss": 0.0069,
      "step": 12468
    },
    {
      "epoch": 1.747582340574632,
      "grad_norm": 0.1599879115819931,
      "learning_rate": 0.00012830901698158336,
      "loss": 0.0342,
      "step": 12469
    },
    {
      "epoch": 1.7477224947442185,
      "grad_norm": 0.6961497664451599,
      "learning_rate": 0.00012829466634776368,
      "loss": 0.1281,
      "step": 12470
    },
    {
      "epoch": 1.7478626489138052,
      "grad_norm": 0.193365678191185,
      "learning_rate": 0.000128280315713944,
      "loss": 0.0128,
      "step": 12471
    },
    {
      "epoch": 1.7480028030833918,
      "grad_norm": 0.5707042217254639,
      "learning_rate": 0.00012826596508012437,
      "loss": 0.1072,
      "step": 12472
    },
    {
      "epoch": 1.7481429572529783,
      "grad_norm": 0.6319882273674011,
      "learning_rate": 0.0001282516144463047,
      "loss": 0.0486,
      "step": 12473
    },
    {
      "epoch": 1.7482831114225648,
      "grad_norm": 0.69849693775177,
      "learning_rate": 0.00012823726381248505,
      "loss": 0.0646,
      "step": 12474
    },
    {
      "epoch": 1.7484232655921512,
      "grad_norm": 0.28756093978881836,
      "learning_rate": 0.00012822291317866538,
      "loss": 0.0273,
      "step": 12475
    },
    {
      "epoch": 1.748563419761738,
      "grad_norm": 0.2201293706893921,
      "learning_rate": 0.0001282085625448457,
      "loss": 0.0484,
      "step": 12476
    },
    {
      "epoch": 1.7487035739313246,
      "grad_norm": 1.8964568376541138,
      "learning_rate": 0.00012819421191102607,
      "loss": 0.1547,
      "step": 12477
    },
    {
      "epoch": 1.748843728100911,
      "grad_norm": 0.2972404360771179,
      "learning_rate": 0.0001281798612772064,
      "loss": 0.023,
      "step": 12478
    },
    {
      "epoch": 1.7489838822704975,
      "grad_norm": 0.26198846101760864,
      "learning_rate": 0.00012816551064338675,
      "loss": 0.0605,
      "step": 12479
    },
    {
      "epoch": 1.749124036440084,
      "grad_norm": 0.29685354232788086,
      "learning_rate": 0.00012815116000956708,
      "loss": 0.0471,
      "step": 12480
    },
    {
      "epoch": 1.7492641906096706,
      "grad_norm": 1.3867894411087036,
      "learning_rate": 0.0001281368093757474,
      "loss": 0.0662,
      "step": 12481
    },
    {
      "epoch": 1.7494043447792573,
      "grad_norm": 0.6764283180236816,
      "learning_rate": 0.00012812245874192776,
      "loss": 0.1135,
      "step": 12482
    },
    {
      "epoch": 1.7495444989488438,
      "grad_norm": 0.7813418507575989,
      "learning_rate": 0.0001281081081081081,
      "loss": 0.0672,
      "step": 12483
    },
    {
      "epoch": 1.7496846531184302,
      "grad_norm": 2.208521842956543,
      "learning_rate": 0.00012809375747428842,
      "loss": 0.2374,
      "step": 12484
    },
    {
      "epoch": 1.7498248072880167,
      "grad_norm": 0.6415435671806335,
      "learning_rate": 0.00012807940684046878,
      "loss": 0.1071,
      "step": 12485
    },
    {
      "epoch": 1.7499649614576034,
      "grad_norm": 0.16769105195999146,
      "learning_rate": 0.0001280650562066491,
      "loss": 0.0225,
      "step": 12486
    },
    {
      "epoch": 1.75010511562719,
      "grad_norm": 0.09176711738109589,
      "learning_rate": 0.00012805070557282946,
      "loss": 0.014,
      "step": 12487
    },
    {
      "epoch": 1.7502452697967765,
      "grad_norm": 0.4044470489025116,
      "learning_rate": 0.0001280363549390098,
      "loss": 0.1298,
      "step": 12488
    },
    {
      "epoch": 1.750385423966363,
      "grad_norm": 0.41471362113952637,
      "learning_rate": 0.00012802200430519012,
      "loss": 0.0418,
      "step": 12489
    },
    {
      "epoch": 1.7505255781359494,
      "grad_norm": 0.36400818824768066,
      "learning_rate": 0.00012800765367137047,
      "loss": 0.0882,
      "step": 12490
    },
    {
      "epoch": 1.7506657323055361,
      "grad_norm": 0.18132257461547852,
      "learning_rate": 0.00012799330303755083,
      "loss": 0.0341,
      "step": 12491
    },
    {
      "epoch": 1.7508058864751226,
      "grad_norm": 0.2690911591053009,
      "learning_rate": 0.00012797895240373116,
      "loss": 0.0218,
      "step": 12492
    },
    {
      "epoch": 1.7509460406447093,
      "grad_norm": 0.4310668706893921,
      "learning_rate": 0.00012796460176991149,
      "loss": 0.0644,
      "step": 12493
    },
    {
      "epoch": 1.7510861948142957,
      "grad_norm": 0.19673354923725128,
      "learning_rate": 0.00012795025113609181,
      "loss": 0.0251,
      "step": 12494
    },
    {
      "epoch": 1.7512263489838822,
      "grad_norm": 0.3696749806404114,
      "learning_rate": 0.00012793590050227217,
      "loss": 0.136,
      "step": 12495
    },
    {
      "epoch": 1.7513665031534689,
      "grad_norm": 0.43675506114959717,
      "learning_rate": 0.00012792154986845253,
      "loss": 0.0272,
      "step": 12496
    },
    {
      "epoch": 1.7515066573230553,
      "grad_norm": 0.1094985380768776,
      "learning_rate": 0.00012790719923463285,
      "loss": 0.0121,
      "step": 12497
    },
    {
      "epoch": 1.751646811492642,
      "grad_norm": 0.43299415707588196,
      "learning_rate": 0.00012789284860081318,
      "loss": 0.0416,
      "step": 12498
    },
    {
      "epoch": 1.7517869656622285,
      "grad_norm": 0.4543834626674652,
      "learning_rate": 0.00012787849796699354,
      "loss": 0.0433,
      "step": 12499
    },
    {
      "epoch": 1.751927119831815,
      "grad_norm": 0.16920851171016693,
      "learning_rate": 0.00012786414733317387,
      "loss": 0.0459,
      "step": 12500
    },
    {
      "epoch": 1.7520672740014014,
      "grad_norm": 0.36703088879585266,
      "learning_rate": 0.00012784979669935422,
      "loss": 0.0481,
      "step": 12501
    },
    {
      "epoch": 1.752207428170988,
      "grad_norm": 0.4576791822910309,
      "learning_rate": 0.00012783544606553455,
      "loss": 0.0442,
      "step": 12502
    },
    {
      "epoch": 1.7523475823405747,
      "grad_norm": 0.13480620086193085,
      "learning_rate": 0.00012782109543171488,
      "loss": 0.0216,
      "step": 12503
    },
    {
      "epoch": 1.7524877365101612,
      "grad_norm": 0.2754122316837311,
      "learning_rate": 0.00012780674479789524,
      "loss": 0.0856,
      "step": 12504
    },
    {
      "epoch": 1.7526278906797477,
      "grad_norm": 0.2122202068567276,
      "learning_rate": 0.00012779239416407556,
      "loss": 0.0199,
      "step": 12505
    },
    {
      "epoch": 1.7527680448493341,
      "grad_norm": 0.1711021065711975,
      "learning_rate": 0.00012777804353025592,
      "loss": 0.0402,
      "step": 12506
    },
    {
      "epoch": 1.7529081990189208,
      "grad_norm": 0.26081499457359314,
      "learning_rate": 0.00012776369289643625,
      "loss": 0.028,
      "step": 12507
    },
    {
      "epoch": 1.7530483531885075,
      "grad_norm": 0.41491949558258057,
      "learning_rate": 0.00012774934226261658,
      "loss": 0.0355,
      "step": 12508
    },
    {
      "epoch": 1.753188507358094,
      "grad_norm": 0.6401037573814392,
      "learning_rate": 0.00012773499162879693,
      "loss": 0.1195,
      "step": 12509
    },
    {
      "epoch": 1.7533286615276804,
      "grad_norm": 0.0715518668293953,
      "learning_rate": 0.0001277206409949773,
      "loss": 0.007,
      "step": 12510
    },
    {
      "epoch": 1.7534688156972669,
      "grad_norm": 0.48080724477767944,
      "learning_rate": 0.00012770629036115762,
      "loss": 0.0434,
      "step": 12511
    },
    {
      "epoch": 1.7536089698668536,
      "grad_norm": 0.1719878613948822,
      "learning_rate": 0.00012769193972733795,
      "loss": 0.0085,
      "step": 12512
    },
    {
      "epoch": 1.7537491240364402,
      "grad_norm": 0.29505372047424316,
      "learning_rate": 0.00012767758909351827,
      "loss": 0.0474,
      "step": 12513
    },
    {
      "epoch": 1.7538892782060267,
      "grad_norm": 0.15355399250984192,
      "learning_rate": 0.00012766323845969863,
      "loss": 0.0184,
      "step": 12514
    },
    {
      "epoch": 1.7540294323756132,
      "grad_norm": 0.5316483378410339,
      "learning_rate": 0.00012764888782587896,
      "loss": 0.0394,
      "step": 12515
    },
    {
      "epoch": 1.7541695865451996,
      "grad_norm": 0.4025880992412567,
      "learning_rate": 0.0001276345371920593,
      "loss": 0.0219,
      "step": 12516
    },
    {
      "epoch": 1.7543097407147863,
      "grad_norm": 0.2650052607059479,
      "learning_rate": 0.00012762018655823964,
      "loss": 0.0648,
      "step": 12517
    },
    {
      "epoch": 1.754449894884373,
      "grad_norm": 0.3519507944583893,
      "learning_rate": 0.00012760583592442,
      "loss": 0.0496,
      "step": 12518
    },
    {
      "epoch": 1.7545900490539594,
      "grad_norm": 0.10431823134422302,
      "learning_rate": 0.00012759148529060033,
      "loss": 0.0104,
      "step": 12519
    },
    {
      "epoch": 1.754730203223546,
      "grad_norm": 1.1775903701782227,
      "learning_rate": 0.00012757713465678066,
      "loss": 0.0481,
      "step": 12520
    },
    {
      "epoch": 1.7548703573931324,
      "grad_norm": 0.2676998972892761,
      "learning_rate": 0.00012756278402296098,
      "loss": 0.0701,
      "step": 12521
    },
    {
      "epoch": 1.755010511562719,
      "grad_norm": 0.4892907738685608,
      "learning_rate": 0.00012754843338914134,
      "loss": 0.0164,
      "step": 12522
    },
    {
      "epoch": 1.7551506657323055,
      "grad_norm": 0.1719372421503067,
      "learning_rate": 0.0001275340827553217,
      "loss": 0.0207,
      "step": 12523
    },
    {
      "epoch": 1.7552908199018922,
      "grad_norm": 0.4776909053325653,
      "learning_rate": 0.00012751973212150202,
      "loss": 0.0542,
      "step": 12524
    },
    {
      "epoch": 1.7554309740714786,
      "grad_norm": 0.5701606273651123,
      "learning_rate": 0.00012750538148768235,
      "loss": 0.1174,
      "step": 12525
    },
    {
      "epoch": 1.755571128241065,
      "grad_norm": 0.3333530128002167,
      "learning_rate": 0.0001274910308538627,
      "loss": 0.0624,
      "step": 12526
    },
    {
      "epoch": 1.7557112824106516,
      "grad_norm": 0.2036724090576172,
      "learning_rate": 0.00012747668022004304,
      "loss": 0.0242,
      "step": 12527
    },
    {
      "epoch": 1.7558514365802382,
      "grad_norm": 0.4787029027938843,
      "learning_rate": 0.0001274623295862234,
      "loss": 0.0878,
      "step": 12528
    },
    {
      "epoch": 1.755991590749825,
      "grad_norm": 0.49031561613082886,
      "learning_rate": 0.00012744797895240372,
      "loss": 0.0734,
      "step": 12529
    },
    {
      "epoch": 1.7561317449194114,
      "grad_norm": 0.6599199771881104,
      "learning_rate": 0.00012743362831858405,
      "loss": 0.0442,
      "step": 12530
    },
    {
      "epoch": 1.7562718990889978,
      "grad_norm": 0.4643518626689911,
      "learning_rate": 0.0001274192776847644,
      "loss": 0.0354,
      "step": 12531
    },
    {
      "epoch": 1.7564120532585843,
      "grad_norm": 0.5022676587104797,
      "learning_rate": 0.00012740492705094473,
      "loss": 0.0571,
      "step": 12532
    },
    {
      "epoch": 1.756552207428171,
      "grad_norm": 0.4180651605129242,
      "learning_rate": 0.0001273905764171251,
      "loss": 0.2197,
      "step": 12533
    },
    {
      "epoch": 1.7566923615977577,
      "grad_norm": 0.8144434094429016,
      "learning_rate": 0.00012737622578330542,
      "loss": 0.116,
      "step": 12534
    },
    {
      "epoch": 1.7568325157673441,
      "grad_norm": 1.8669880628585815,
      "learning_rate": 0.00012736187514948575,
      "loss": 0.353,
      "step": 12535
    },
    {
      "epoch": 1.7569726699369306,
      "grad_norm": 0.14623229205608368,
      "learning_rate": 0.0001273475245156661,
      "loss": 0.0201,
      "step": 12536
    },
    {
      "epoch": 1.757112824106517,
      "grad_norm": 0.23059025406837463,
      "learning_rate": 0.00012733317388184646,
      "loss": 0.0321,
      "step": 12537
    },
    {
      "epoch": 1.7572529782761037,
      "grad_norm": 0.2146468162536621,
      "learning_rate": 0.00012731882324802679,
      "loss": 0.0583,
      "step": 12538
    },
    {
      "epoch": 1.7573931324456904,
      "grad_norm": 0.09678258746862411,
      "learning_rate": 0.00012730447261420711,
      "loss": 0.0231,
      "step": 12539
    },
    {
      "epoch": 1.7575332866152769,
      "grad_norm": 0.34229788184165955,
      "learning_rate": 0.00012729012198038744,
      "loss": 0.0565,
      "step": 12540
    },
    {
      "epoch": 1.7576734407848633,
      "grad_norm": 0.3117468059062958,
      "learning_rate": 0.0001272757713465678,
      "loss": 0.0464,
      "step": 12541
    },
    {
      "epoch": 1.7578135949544498,
      "grad_norm": 0.11006274819374084,
      "learning_rate": 0.00012726142071274815,
      "loss": 0.01,
      "step": 12542
    },
    {
      "epoch": 1.7579537491240365,
      "grad_norm": 0.40979981422424316,
      "learning_rate": 0.00012724707007892848,
      "loss": 0.0505,
      "step": 12543
    },
    {
      "epoch": 1.7580939032936231,
      "grad_norm": 0.22459720075130463,
      "learning_rate": 0.0001272327194451088,
      "loss": 0.0254,
      "step": 12544
    },
    {
      "epoch": 1.7582340574632096,
      "grad_norm": 0.45098814368247986,
      "learning_rate": 0.00012721836881128917,
      "loss": 0.0477,
      "step": 12545
    },
    {
      "epoch": 1.758374211632796,
      "grad_norm": 0.13434384763240814,
      "learning_rate": 0.0001272040181774695,
      "loss": 0.0161,
      "step": 12546
    },
    {
      "epoch": 1.7585143658023825,
      "grad_norm": 0.3728131949901581,
      "learning_rate": 0.00012718966754364982,
      "loss": 0.0577,
      "step": 12547
    },
    {
      "epoch": 1.7586545199719692,
      "grad_norm": 0.43375343084335327,
      "learning_rate": 0.00012717531690983015,
      "loss": 0.0356,
      "step": 12548
    },
    {
      "epoch": 1.7587946741415557,
      "grad_norm": 0.2617349922657013,
      "learning_rate": 0.0001271609662760105,
      "loss": 0.0499,
      "step": 12549
    },
    {
      "epoch": 1.7589348283111423,
      "grad_norm": 0.1481235772371292,
      "learning_rate": 0.00012714661564219086,
      "loss": 0.0566,
      "step": 12550
    },
    {
      "epoch": 1.7590749824807288,
      "grad_norm": 0.4470871388912201,
      "learning_rate": 0.0001271322650083712,
      "loss": 0.0251,
      "step": 12551
    },
    {
      "epoch": 1.7592151366503153,
      "grad_norm": 0.2650703489780426,
      "learning_rate": 0.00012711791437455152,
      "loss": 0.0293,
      "step": 12552
    },
    {
      "epoch": 1.759355290819902,
      "grad_norm": 0.5229439735412598,
      "learning_rate": 0.00012710356374073188,
      "loss": 0.1084,
      "step": 12553
    },
    {
      "epoch": 1.7594954449894884,
      "grad_norm": 0.06834203004837036,
      "learning_rate": 0.0001270892131069122,
      "loss": 0.0108,
      "step": 12554
    },
    {
      "epoch": 1.759635599159075,
      "grad_norm": 0.5743789672851562,
      "learning_rate": 0.00012707486247309256,
      "loss": 0.0539,
      "step": 12555
    },
    {
      "epoch": 1.7597757533286615,
      "grad_norm": 0.24601314961910248,
      "learning_rate": 0.0001270605118392729,
      "loss": 0.0431,
      "step": 12556
    },
    {
      "epoch": 1.759915907498248,
      "grad_norm": 0.24687445163726807,
      "learning_rate": 0.00012704616120545322,
      "loss": 0.0428,
      "step": 12557
    },
    {
      "epoch": 1.7600560616678345,
      "grad_norm": 0.5515462160110474,
      "learning_rate": 0.00012703181057163357,
      "loss": 0.0263,
      "step": 12558
    },
    {
      "epoch": 1.7601962158374211,
      "grad_norm": 0.776067316532135,
      "learning_rate": 0.0001270174599378139,
      "loss": 0.091,
      "step": 12559
    },
    {
      "epoch": 1.7603363700070078,
      "grad_norm": 0.15359188616275787,
      "learning_rate": 0.00012700310930399426,
      "loss": 0.0412,
      "step": 12560
    },
    {
      "epoch": 1.7604765241765943,
      "grad_norm": 0.18060021102428436,
      "learning_rate": 0.0001269887586701746,
      "loss": 0.0277,
      "step": 12561
    },
    {
      "epoch": 1.7606166783461807,
      "grad_norm": 0.4621140956878662,
      "learning_rate": 0.00012697440803635492,
      "loss": 0.0487,
      "step": 12562
    },
    {
      "epoch": 1.7607568325157672,
      "grad_norm": 0.31673869490623474,
      "learning_rate": 0.00012696005740253527,
      "loss": 0.0607,
      "step": 12563
    },
    {
      "epoch": 1.7608969866853539,
      "grad_norm": 0.0712527483701706,
      "learning_rate": 0.0001269457067687156,
      "loss": 0.0134,
      "step": 12564
    },
    {
      "epoch": 1.7610371408549406,
      "grad_norm": 0.12757079303264618,
      "learning_rate": 0.00012693135613489596,
      "loss": 0.013,
      "step": 12565
    },
    {
      "epoch": 1.761177295024527,
      "grad_norm": 0.19716481864452362,
      "learning_rate": 0.00012691700550107628,
      "loss": 0.0343,
      "step": 12566
    },
    {
      "epoch": 1.7613174491941135,
      "grad_norm": 0.431725412607193,
      "learning_rate": 0.0001269026548672566,
      "loss": 0.0711,
      "step": 12567
    },
    {
      "epoch": 1.7614576033637,
      "grad_norm": 0.3966410160064697,
      "learning_rate": 0.00012688830423343697,
      "loss": 0.05,
      "step": 12568
    },
    {
      "epoch": 1.7615977575332866,
      "grad_norm": 0.4786522090435028,
      "learning_rate": 0.00012687395359961732,
      "loss": 0.0343,
      "step": 12569
    },
    {
      "epoch": 1.7617379117028733,
      "grad_norm": 0.4577521085739136,
      "learning_rate": 0.00012685960296579765,
      "loss": 0.0753,
      "step": 12570
    },
    {
      "epoch": 1.7618780658724598,
      "grad_norm": 0.12372224777936935,
      "learning_rate": 0.00012684525233197798,
      "loss": 0.0058,
      "step": 12571
    },
    {
      "epoch": 1.7620182200420462,
      "grad_norm": 0.30933916568756104,
      "learning_rate": 0.0001268309016981583,
      "loss": 0.0389,
      "step": 12572
    },
    {
      "epoch": 1.7621583742116327,
      "grad_norm": 0.25998735427856445,
      "learning_rate": 0.00012681655106433867,
      "loss": 0.0213,
      "step": 12573
    },
    {
      "epoch": 1.7622985283812194,
      "grad_norm": 0.5214083194732666,
      "learning_rate": 0.00012680220043051902,
      "loss": 0.0998,
      "step": 12574
    },
    {
      "epoch": 1.762438682550806,
      "grad_norm": 0.4503547251224518,
      "learning_rate": 0.00012678784979669935,
      "loss": 0.0799,
      "step": 12575
    },
    {
      "epoch": 1.7625788367203925,
      "grad_norm": 0.5093770027160645,
      "learning_rate": 0.00012677349916287968,
      "loss": 0.1242,
      "step": 12576
    },
    {
      "epoch": 1.762718990889979,
      "grad_norm": 0.19426004588603973,
      "learning_rate": 0.00012675914852906003,
      "loss": 0.0219,
      "step": 12577
    },
    {
      "epoch": 1.7628591450595654,
      "grad_norm": 0.43884512782096863,
      "learning_rate": 0.00012674479789524036,
      "loss": 0.0539,
      "step": 12578
    },
    {
      "epoch": 1.762999299229152,
      "grad_norm": 0.7678201794624329,
      "learning_rate": 0.0001267304472614207,
      "loss": 0.0818,
      "step": 12579
    },
    {
      "epoch": 1.7631394533987386,
      "grad_norm": 0.2963252365589142,
      "learning_rate": 0.00012671609662760105,
      "loss": 0.031,
      "step": 12580
    },
    {
      "epoch": 1.7632796075683252,
      "grad_norm": 1.0102722644805908,
      "learning_rate": 0.00012670174599378137,
      "loss": 0.125,
      "step": 12581
    },
    {
      "epoch": 1.7634197617379117,
      "grad_norm": 0.18581822514533997,
      "learning_rate": 0.00012668739535996173,
      "loss": 0.0174,
      "step": 12582
    },
    {
      "epoch": 1.7635599159074982,
      "grad_norm": 0.5525992512702942,
      "learning_rate": 0.00012667304472614206,
      "loss": 0.1618,
      "step": 12583
    },
    {
      "epoch": 1.7637000700770848,
      "grad_norm": 0.6855126023292542,
      "learning_rate": 0.0001266586940923224,
      "loss": 0.0905,
      "step": 12584
    },
    {
      "epoch": 1.7638402242466713,
      "grad_norm": 0.6343996524810791,
      "learning_rate": 0.00012664434345850274,
      "loss": 0.0606,
      "step": 12585
    },
    {
      "epoch": 1.763980378416258,
      "grad_norm": 0.1963779479265213,
      "learning_rate": 0.00012662999282468307,
      "loss": 0.0331,
      "step": 12586
    },
    {
      "epoch": 1.7641205325858444,
      "grad_norm": 0.4166298508644104,
      "learning_rate": 0.00012661564219086343,
      "loss": 0.0606,
      "step": 12587
    },
    {
      "epoch": 1.764260686755431,
      "grad_norm": 0.36177870631217957,
      "learning_rate": 0.00012660129155704376,
      "loss": 0.0648,
      "step": 12588
    },
    {
      "epoch": 1.7644008409250174,
      "grad_norm": 0.3219148516654968,
      "learning_rate": 0.00012658694092322408,
      "loss": 0.0499,
      "step": 12589
    },
    {
      "epoch": 1.764540995094604,
      "grad_norm": 0.5056019425392151,
      "learning_rate": 0.00012657259028940444,
      "loss": 0.0483,
      "step": 12590
    },
    {
      "epoch": 1.7646811492641907,
      "grad_norm": 0.2618730962276459,
      "learning_rate": 0.00012655823965558477,
      "loss": 0.0272,
      "step": 12591
    },
    {
      "epoch": 1.7648213034337772,
      "grad_norm": 0.18048636615276337,
      "learning_rate": 0.00012654388902176512,
      "loss": 0.0182,
      "step": 12592
    },
    {
      "epoch": 1.7649614576033636,
      "grad_norm": 0.1842491179704666,
      "learning_rate": 0.00012652953838794545,
      "loss": 0.0284,
      "step": 12593
    },
    {
      "epoch": 1.76510161177295,
      "grad_norm": 0.23825345933437347,
      "learning_rate": 0.00012651518775412578,
      "loss": 0.0697,
      "step": 12594
    },
    {
      "epoch": 1.7652417659425368,
      "grad_norm": 0.37002116441726685,
      "learning_rate": 0.00012650083712030614,
      "loss": 0.057,
      "step": 12595
    },
    {
      "epoch": 1.7653819201121235,
      "grad_norm": 0.3460192084312439,
      "learning_rate": 0.0001264864864864865,
      "loss": 0.0852,
      "step": 12596
    },
    {
      "epoch": 1.76552207428171,
      "grad_norm": 0.505820631980896,
      "learning_rate": 0.00012647213585266682,
      "loss": 0.0672,
      "step": 12597
    },
    {
      "epoch": 1.7656622284512964,
      "grad_norm": 0.19242654740810394,
      "learning_rate": 0.00012645778521884715,
      "loss": 0.044,
      "step": 12598
    },
    {
      "epoch": 1.7658023826208828,
      "grad_norm": 0.3322543799877167,
      "learning_rate": 0.00012644343458502748,
      "loss": 0.051,
      "step": 12599
    },
    {
      "epoch": 1.7659425367904695,
      "grad_norm": 0.16603419184684753,
      "learning_rate": 0.00012642908395120783,
      "loss": 0.0436,
      "step": 12600
    },
    {
      "epoch": 1.7660826909600562,
      "grad_norm": 0.6716790199279785,
      "learning_rate": 0.0001264147333173882,
      "loss": 0.0463,
      "step": 12601
    },
    {
      "epoch": 1.7662228451296427,
      "grad_norm": 0.20524267852306366,
      "learning_rate": 0.00012640038268356852,
      "loss": 0.0679,
      "step": 12602
    },
    {
      "epoch": 1.7663629992992291,
      "grad_norm": 0.18807309865951538,
      "learning_rate": 0.00012638603204974885,
      "loss": 0.0268,
      "step": 12603
    },
    {
      "epoch": 1.7665031534688156,
      "grad_norm": 0.13584941625595093,
      "learning_rate": 0.0001263716814159292,
      "loss": 0.0225,
      "step": 12604
    },
    {
      "epoch": 1.7666433076384023,
      "grad_norm": 0.22762326896190643,
      "learning_rate": 0.00012635733078210953,
      "loss": 0.0196,
      "step": 12605
    },
    {
      "epoch": 1.766783461807989,
      "grad_norm": 0.37516525387763977,
      "learning_rate": 0.0001263429801482899,
      "loss": 0.0909,
      "step": 12606
    },
    {
      "epoch": 1.7669236159775754,
      "grad_norm": 0.4195827543735504,
      "learning_rate": 0.00012632862951447022,
      "loss": 0.0278,
      "step": 12607
    },
    {
      "epoch": 1.7670637701471619,
      "grad_norm": 0.21018537878990173,
      "learning_rate": 0.00012631427888065054,
      "loss": 0.0535,
      "step": 12608
    },
    {
      "epoch": 1.7672039243167483,
      "grad_norm": 0.40081337094306946,
      "learning_rate": 0.0001262999282468309,
      "loss": 0.0629,
      "step": 12609
    },
    {
      "epoch": 1.767344078486335,
      "grad_norm": 0.2986059784889221,
      "learning_rate": 0.00012628557761301123,
      "loss": 0.0422,
      "step": 12610
    },
    {
      "epoch": 1.7674842326559215,
      "grad_norm": 0.3104020357131958,
      "learning_rate": 0.00012627122697919156,
      "loss": 0.0503,
      "step": 12611
    },
    {
      "epoch": 1.7676243868255082,
      "grad_norm": 0.33469510078430176,
      "learning_rate": 0.0001262568763453719,
      "loss": 0.0558,
      "step": 12612
    },
    {
      "epoch": 1.7677645409950946,
      "grad_norm": 0.11093602329492569,
      "learning_rate": 0.00012624252571155224,
      "loss": 0.0101,
      "step": 12613
    },
    {
      "epoch": 1.767904695164681,
      "grad_norm": 0.29873716831207275,
      "learning_rate": 0.0001262281750777326,
      "loss": 0.0479,
      "step": 12614
    },
    {
      "epoch": 1.7680448493342675,
      "grad_norm": 0.1035333201289177,
      "learning_rate": 0.00012621382444391293,
      "loss": 0.0159,
      "step": 12615
    },
    {
      "epoch": 1.7681850035038542,
      "grad_norm": 0.16268615424633026,
      "learning_rate": 0.00012619947381009325,
      "loss": 0.0242,
      "step": 12616
    },
    {
      "epoch": 1.768325157673441,
      "grad_norm": 0.19221974909305573,
      "learning_rate": 0.0001261851231762736,
      "loss": 0.0137,
      "step": 12617
    },
    {
      "epoch": 1.7684653118430274,
      "grad_norm": 0.15477271378040314,
      "learning_rate": 0.00012617077254245394,
      "loss": 0.0101,
      "step": 12618
    },
    {
      "epoch": 1.7686054660126138,
      "grad_norm": 0.370347797870636,
      "learning_rate": 0.0001261564219086343,
      "loss": 0.0382,
      "step": 12619
    },
    {
      "epoch": 1.7687456201822003,
      "grad_norm": 0.39101099967956543,
      "learning_rate": 0.00012614207127481462,
      "loss": 0.0528,
      "step": 12620
    },
    {
      "epoch": 1.768885774351787,
      "grad_norm": 0.27915164828300476,
      "learning_rate": 0.00012612772064099495,
      "loss": 0.0172,
      "step": 12621
    },
    {
      "epoch": 1.7690259285213736,
      "grad_norm": 0.13251233100891113,
      "learning_rate": 0.0001261133700071753,
      "loss": 0.0067,
      "step": 12622
    },
    {
      "epoch": 1.76916608269096,
      "grad_norm": 0.7182928323745728,
      "learning_rate": 0.00012609901937335566,
      "loss": 0.072,
      "step": 12623
    },
    {
      "epoch": 1.7693062368605466,
      "grad_norm": 0.27497270703315735,
      "learning_rate": 0.000126084668739536,
      "loss": 0.0328,
      "step": 12624
    },
    {
      "epoch": 1.769446391030133,
      "grad_norm": 0.2314102053642273,
      "learning_rate": 0.00012607031810571632,
      "loss": 0.0492,
      "step": 12625
    },
    {
      "epoch": 1.7695865451997197,
      "grad_norm": 0.35429078340530396,
      "learning_rate": 0.00012605596747189665,
      "loss": 0.0135,
      "step": 12626
    },
    {
      "epoch": 1.7697266993693064,
      "grad_norm": 0.3357757031917572,
      "learning_rate": 0.000126041616838077,
      "loss": 0.0564,
      "step": 12627
    },
    {
      "epoch": 1.7698668535388928,
      "grad_norm": 0.29387056827545166,
      "learning_rate": 0.00012602726620425736,
      "loss": 0.0229,
      "step": 12628
    },
    {
      "epoch": 1.7700070077084793,
      "grad_norm": 0.13563747704029083,
      "learning_rate": 0.0001260129155704377,
      "loss": 0.0163,
      "step": 12629
    },
    {
      "epoch": 1.7701471618780658,
      "grad_norm": 0.39580798149108887,
      "learning_rate": 0.00012599856493661802,
      "loss": 0.0618,
      "step": 12630
    },
    {
      "epoch": 1.7702873160476524,
      "grad_norm": 0.3091544210910797,
      "learning_rate": 0.00012598421430279837,
      "loss": 0.0926,
      "step": 12631
    },
    {
      "epoch": 1.7704274702172391,
      "grad_norm": 0.18303082883358002,
      "learning_rate": 0.0001259698636689787,
      "loss": 0.0152,
      "step": 12632
    },
    {
      "epoch": 1.7705676243868256,
      "grad_norm": 0.37664055824279785,
      "learning_rate": 0.00012595551303515906,
      "loss": 0.0202,
      "step": 12633
    },
    {
      "epoch": 1.770707778556412,
      "grad_norm": 3.7317678928375244,
      "learning_rate": 0.00012594116240133938,
      "loss": 0.1951,
      "step": 12634
    },
    {
      "epoch": 1.7708479327259985,
      "grad_norm": 1.39244544506073,
      "learning_rate": 0.0001259268117675197,
      "loss": 0.1723,
      "step": 12635
    },
    {
      "epoch": 1.7709880868955852,
      "grad_norm": 0.5231479406356812,
      "learning_rate": 0.00012591246113370007,
      "loss": 0.0598,
      "step": 12636
    },
    {
      "epoch": 1.7711282410651716,
      "grad_norm": 1.100511908531189,
      "learning_rate": 0.0001258981104998804,
      "loss": 0.1009,
      "step": 12637
    },
    {
      "epoch": 1.7712683952347583,
      "grad_norm": 0.0895213931798935,
      "learning_rate": 0.00012588375986606075,
      "loss": 0.0154,
      "step": 12638
    },
    {
      "epoch": 1.7714085494043448,
      "grad_norm": 0.06310956925153732,
      "learning_rate": 0.00012586940923224108,
      "loss": 0.0054,
      "step": 12639
    },
    {
      "epoch": 1.7715487035739312,
      "grad_norm": 0.350788414478302,
      "learning_rate": 0.0001258550585984214,
      "loss": 0.0839,
      "step": 12640
    },
    {
      "epoch": 1.771688857743518,
      "grad_norm": 0.22423112392425537,
      "learning_rate": 0.00012584070796460177,
      "loss": 0.0286,
      "step": 12641
    },
    {
      "epoch": 1.7718290119131044,
      "grad_norm": 0.0421469509601593,
      "learning_rate": 0.0001258263573307821,
      "loss": 0.0113,
      "step": 12642
    },
    {
      "epoch": 1.771969166082691,
      "grad_norm": 0.08583997935056686,
      "learning_rate": 0.00012581200669696242,
      "loss": 0.0183,
      "step": 12643
    },
    {
      "epoch": 1.7721093202522775,
      "grad_norm": 0.21854151785373688,
      "learning_rate": 0.00012579765606314278,
      "loss": 0.0289,
      "step": 12644
    },
    {
      "epoch": 1.772249474421864,
      "grad_norm": 0.5578779578208923,
      "learning_rate": 0.0001257833054293231,
      "loss": 0.0993,
      "step": 12645
    },
    {
      "epoch": 1.7723896285914504,
      "grad_norm": 0.1693013608455658,
      "learning_rate": 0.00012576895479550346,
      "loss": 0.0239,
      "step": 12646
    },
    {
      "epoch": 1.7725297827610371,
      "grad_norm": 0.3322499096393585,
      "learning_rate": 0.0001257546041616838,
      "loss": 0.0321,
      "step": 12647
    },
    {
      "epoch": 1.7726699369306238,
      "grad_norm": 0.6681264042854309,
      "learning_rate": 0.00012574025352786412,
      "loss": 0.1421,
      "step": 12648
    },
    {
      "epoch": 1.7728100911002103,
      "grad_norm": 0.16810016334056854,
      "learning_rate": 0.00012572590289404448,
      "loss": 0.0326,
      "step": 12649
    },
    {
      "epoch": 1.7729502452697967,
      "grad_norm": 0.14535851776599884,
      "learning_rate": 0.00012571155226022483,
      "loss": 0.0543,
      "step": 12650
    },
    {
      "epoch": 1.7730903994393832,
      "grad_norm": 0.15418435633182526,
      "learning_rate": 0.00012569720162640516,
      "loss": 0.0229,
      "step": 12651
    },
    {
      "epoch": 1.7732305536089699,
      "grad_norm": 0.218855082988739,
      "learning_rate": 0.0001256828509925855,
      "loss": 0.0322,
      "step": 12652
    },
    {
      "epoch": 1.7733707077785565,
      "grad_norm": 0.21407152712345123,
      "learning_rate": 0.00012566850035876582,
      "loss": 0.0532,
      "step": 12653
    },
    {
      "epoch": 1.773510861948143,
      "grad_norm": 0.37991687655448914,
      "learning_rate": 0.00012565414972494617,
      "loss": 0.1043,
      "step": 12654
    },
    {
      "epoch": 1.7736510161177295,
      "grad_norm": 0.09540198743343353,
      "learning_rate": 0.00012563979909112653,
      "loss": 0.0082,
      "step": 12655
    },
    {
      "epoch": 1.773791170287316,
      "grad_norm": 0.2393234223127365,
      "learning_rate": 0.00012562544845730686,
      "loss": 0.0288,
      "step": 12656
    },
    {
      "epoch": 1.7739313244569026,
      "grad_norm": 0.41243407130241394,
      "learning_rate": 0.00012561109782348719,
      "loss": 0.1101,
      "step": 12657
    },
    {
      "epoch": 1.7740714786264893,
      "grad_norm": 0.20797981321811676,
      "learning_rate": 0.00012559674718966754,
      "loss": 0.0546,
      "step": 12658
    },
    {
      "epoch": 1.7742116327960757,
      "grad_norm": 0.14063552021980286,
      "learning_rate": 0.00012558239655584787,
      "loss": 0.0297,
      "step": 12659
    },
    {
      "epoch": 1.7743517869656622,
      "grad_norm": 0.2068670094013214,
      "learning_rate": 0.00012556804592202823,
      "loss": 0.037,
      "step": 12660
    },
    {
      "epoch": 1.7744919411352487,
      "grad_norm": 0.1441810578107834,
      "learning_rate": 0.00012555369528820855,
      "loss": 0.0214,
      "step": 12661
    },
    {
      "epoch": 1.7746320953048353,
      "grad_norm": 0.19155828654766083,
      "learning_rate": 0.00012553934465438888,
      "loss": 0.0416,
      "step": 12662
    },
    {
      "epoch": 1.774772249474422,
      "grad_norm": 0.2164948433637619,
      "learning_rate": 0.00012552499402056924,
      "loss": 0.0271,
      "step": 12663
    },
    {
      "epoch": 1.7749124036440085,
      "grad_norm": 0.4058765769004822,
      "learning_rate": 0.00012551064338674957,
      "loss": 0.0288,
      "step": 12664
    },
    {
      "epoch": 1.775052557813595,
      "grad_norm": 0.22426724433898926,
      "learning_rate": 0.00012549629275292992,
      "loss": 0.0408,
      "step": 12665
    },
    {
      "epoch": 1.7751927119831814,
      "grad_norm": 0.13831987977027893,
      "learning_rate": 0.00012548194211911025,
      "loss": 0.009,
      "step": 12666
    },
    {
      "epoch": 1.775332866152768,
      "grad_norm": 0.559334933757782,
      "learning_rate": 0.00012546759148529058,
      "loss": 0.084,
      "step": 12667
    },
    {
      "epoch": 1.7754730203223545,
      "grad_norm": 0.4013811945915222,
      "learning_rate": 0.00012545324085147094,
      "loss": 0.0505,
      "step": 12668
    },
    {
      "epoch": 1.7756131744919412,
      "grad_norm": 0.40574657917022705,
      "learning_rate": 0.00012543889021765126,
      "loss": 0.0552,
      "step": 12669
    },
    {
      "epoch": 1.7757533286615277,
      "grad_norm": 0.10503213107585907,
      "learning_rate": 0.00012542453958383162,
      "loss": 0.0253,
      "step": 12670
    },
    {
      "epoch": 1.7758934828311141,
      "grad_norm": 0.35956352949142456,
      "learning_rate": 0.00012541018895001195,
      "loss": 0.0913,
      "step": 12671
    },
    {
      "epoch": 1.7760336370007006,
      "grad_norm": 0.14425687491893768,
      "learning_rate": 0.00012539583831619228,
      "loss": 0.0203,
      "step": 12672
    },
    {
      "epoch": 1.7761737911702873,
      "grad_norm": 0.21116703748703003,
      "learning_rate": 0.00012538148768237263,
      "loss": 0.0404,
      "step": 12673
    },
    {
      "epoch": 1.776313945339874,
      "grad_norm": 0.33213019371032715,
      "learning_rate": 0.00012536713704855296,
      "loss": 0.0407,
      "step": 12674
    },
    {
      "epoch": 1.7764540995094604,
      "grad_norm": 0.6757363677024841,
      "learning_rate": 0.0001253527864147333,
      "loss": 0.0416,
      "step": 12675
    },
    {
      "epoch": 1.7765942536790469,
      "grad_norm": 0.5575529336929321,
      "learning_rate": 0.00012533843578091365,
      "loss": 0.0263,
      "step": 12676
    },
    {
      "epoch": 1.7767344078486333,
      "grad_norm": 0.40344178676605225,
      "learning_rate": 0.00012532408514709397,
      "loss": 0.0326,
      "step": 12677
    },
    {
      "epoch": 1.77687456201822,
      "grad_norm": 0.5858974456787109,
      "learning_rate": 0.00012530973451327433,
      "loss": 0.1061,
      "step": 12678
    },
    {
      "epoch": 1.7770147161878067,
      "grad_norm": 0.639039933681488,
      "learning_rate": 0.00012529538387945466,
      "loss": 0.0399,
      "step": 12679
    },
    {
      "epoch": 1.7771548703573932,
      "grad_norm": 0.5654316544532776,
      "learning_rate": 0.000125281033245635,
      "loss": 0.0478,
      "step": 12680
    },
    {
      "epoch": 1.7772950245269796,
      "grad_norm": 0.7055120468139648,
      "learning_rate": 0.00012526668261181534,
      "loss": 0.0757,
      "step": 12681
    },
    {
      "epoch": 1.777435178696566,
      "grad_norm": 0.36441636085510254,
      "learning_rate": 0.0001252523319779957,
      "loss": 0.0681,
      "step": 12682
    },
    {
      "epoch": 1.7775753328661528,
      "grad_norm": 1.6339479684829712,
      "learning_rate": 0.00012523798134417603,
      "loss": 0.1076,
      "step": 12683
    },
    {
      "epoch": 1.7777154870357395,
      "grad_norm": 1.3059263229370117,
      "learning_rate": 0.00012522363071035635,
      "loss": 0.0825,
      "step": 12684
    },
    {
      "epoch": 1.777855641205326,
      "grad_norm": 2.2003870010375977,
      "learning_rate": 0.0001252092800765367,
      "loss": 0.3255,
      "step": 12685
    },
    {
      "epoch": 1.7779957953749124,
      "grad_norm": 0.23578482866287231,
      "learning_rate": 0.00012519492944271704,
      "loss": 0.039,
      "step": 12686
    },
    {
      "epoch": 1.7781359495444988,
      "grad_norm": 0.29941314458847046,
      "learning_rate": 0.0001251805788088974,
      "loss": 0.0408,
      "step": 12687
    },
    {
      "epoch": 1.7782761037140855,
      "grad_norm": 1.0790528059005737,
      "learning_rate": 0.00012516622817507772,
      "loss": 0.0638,
      "step": 12688
    },
    {
      "epoch": 1.7784162578836722,
      "grad_norm": 0.1633813977241516,
      "learning_rate": 0.00012515187754125805,
      "loss": 0.0511,
      "step": 12689
    },
    {
      "epoch": 1.7785564120532587,
      "grad_norm": 0.26883625984191895,
      "learning_rate": 0.0001251375269074384,
      "loss": 0.0601,
      "step": 12690
    },
    {
      "epoch": 1.778696566222845,
      "grad_norm": 0.5229213237762451,
      "learning_rate": 0.00012512317627361874,
      "loss": 0.0897,
      "step": 12691
    },
    {
      "epoch": 1.7788367203924316,
      "grad_norm": 0.1665145605802536,
      "learning_rate": 0.0001251088256397991,
      "loss": 0.0149,
      "step": 12692
    },
    {
      "epoch": 1.7789768745620183,
      "grad_norm": 0.08214559406042099,
      "learning_rate": 0.00012509447500597942,
      "loss": 0.0091,
      "step": 12693
    },
    {
      "epoch": 1.7791170287316047,
      "grad_norm": 0.4949318468570709,
      "learning_rate": 0.00012508012437215975,
      "loss": 0.07,
      "step": 12694
    },
    {
      "epoch": 1.7792571829011914,
      "grad_norm": 0.10292713344097137,
      "learning_rate": 0.0001250657737383401,
      "loss": 0.0095,
      "step": 12695
    },
    {
      "epoch": 1.7793973370707779,
      "grad_norm": 0.18902745842933655,
      "learning_rate": 0.00012505142310452043,
      "loss": 0.0321,
      "step": 12696
    },
    {
      "epoch": 1.7795374912403643,
      "grad_norm": 0.35434672236442566,
      "learning_rate": 0.0001250370724707008,
      "loss": 0.0309,
      "step": 12697
    },
    {
      "epoch": 1.779677645409951,
      "grad_norm": 0.24826212227344513,
      "learning_rate": 0.00012502272183688112,
      "loss": 0.0343,
      "step": 12698
    },
    {
      "epoch": 1.7798177995795375,
      "grad_norm": 0.2808452844619751,
      "learning_rate": 0.00012500837120306145,
      "loss": 0.0457,
      "step": 12699
    },
    {
      "epoch": 1.7799579537491241,
      "grad_norm": 0.35589727759361267,
      "learning_rate": 0.0001249940205692418,
      "loss": 0.0702,
      "step": 12700
    },
    {
      "epoch": 1.7800981079187106,
      "grad_norm": 0.5442264080047607,
      "learning_rate": 0.00012497966993542216,
      "loss": 0.0785,
      "step": 12701
    },
    {
      "epoch": 1.780238262088297,
      "grad_norm": 0.7384634613990784,
      "learning_rate": 0.00012496531930160249,
      "loss": 0.0845,
      "step": 12702
    },
    {
      "epoch": 1.7803784162578835,
      "grad_norm": 0.35767456889152527,
      "learning_rate": 0.00012495096866778281,
      "loss": 0.0625,
      "step": 12703
    },
    {
      "epoch": 1.7805185704274702,
      "grad_norm": 0.6801416873931885,
      "learning_rate": 0.00012493661803396314,
      "loss": 0.0882,
      "step": 12704
    },
    {
      "epoch": 1.7806587245970569,
      "grad_norm": 0.5508906841278076,
      "learning_rate": 0.0001249222674001435,
      "loss": 0.043,
      "step": 12705
    },
    {
      "epoch": 1.7807988787666433,
      "grad_norm": 0.4439547061920166,
      "learning_rate": 0.00012490791676632383,
      "loss": 0.0531,
      "step": 12706
    },
    {
      "epoch": 1.7809390329362298,
      "grad_norm": 0.4452202320098877,
      "learning_rate": 0.00012489356613250416,
      "loss": 0.0475,
      "step": 12707
    },
    {
      "epoch": 1.7810791871058163,
      "grad_norm": 0.1963074654340744,
      "learning_rate": 0.0001248792154986845,
      "loss": 0.0251,
      "step": 12708
    },
    {
      "epoch": 1.781219341275403,
      "grad_norm": 0.09595419466495514,
      "learning_rate": 0.00012486486486486487,
      "loss": 0.0177,
      "step": 12709
    },
    {
      "epoch": 1.7813594954449896,
      "grad_norm": 0.4474094808101654,
      "learning_rate": 0.0001248505142310452,
      "loss": 0.0499,
      "step": 12710
    },
    {
      "epoch": 1.781499649614576,
      "grad_norm": 0.20792241394519806,
      "learning_rate": 0.00012483616359722552,
      "loss": 0.0097,
      "step": 12711
    },
    {
      "epoch": 1.7816398037841625,
      "grad_norm": 0.44448330998420715,
      "learning_rate": 0.00012482181296340585,
      "loss": 0.0267,
      "step": 12712
    },
    {
      "epoch": 1.781779957953749,
      "grad_norm": 0.2350614219903946,
      "learning_rate": 0.0001248074623295862,
      "loss": 0.0746,
      "step": 12713
    },
    {
      "epoch": 1.7819201121233357,
      "grad_norm": 0.3534233272075653,
      "learning_rate": 0.00012479311169576656,
      "loss": 0.0385,
      "step": 12714
    },
    {
      "epoch": 1.7820602662929224,
      "grad_norm": 0.259888231754303,
      "learning_rate": 0.0001247787610619469,
      "loss": 0.0471,
      "step": 12715
    },
    {
      "epoch": 1.7822004204625088,
      "grad_norm": 0.39188113808631897,
      "learning_rate": 0.00012476441042812722,
      "loss": 0.0552,
      "step": 12716
    },
    {
      "epoch": 1.7823405746320953,
      "grad_norm": 0.451839804649353,
      "learning_rate": 0.00012475005979430758,
      "loss": 0.034,
      "step": 12717
    },
    {
      "epoch": 1.7824807288016817,
      "grad_norm": 0.4768160283565521,
      "learning_rate": 0.0001247357091604879,
      "loss": 0.0323,
      "step": 12718
    },
    {
      "epoch": 1.7826208829712684,
      "grad_norm": 0.426025927066803,
      "learning_rate": 0.00012472135852666826,
      "loss": 0.0702,
      "step": 12719
    },
    {
      "epoch": 1.782761037140855,
      "grad_norm": 0.10429257154464722,
      "learning_rate": 0.0001247070078928486,
      "loss": 0.0298,
      "step": 12720
    },
    {
      "epoch": 1.7829011913104416,
      "grad_norm": 0.0650818794965744,
      "learning_rate": 0.00012469265725902892,
      "loss": 0.0103,
      "step": 12721
    },
    {
      "epoch": 1.783041345480028,
      "grad_norm": 0.12811315059661865,
      "learning_rate": 0.00012467830662520927,
      "loss": 0.0182,
      "step": 12722
    },
    {
      "epoch": 1.7831814996496145,
      "grad_norm": 0.76716148853302,
      "learning_rate": 0.0001246639559913896,
      "loss": 0.0923,
      "step": 12723
    },
    {
      "epoch": 1.7833216538192012,
      "grad_norm": 0.16608989238739014,
      "learning_rate": 0.00012464960535756996,
      "loss": 0.0156,
      "step": 12724
    },
    {
      "epoch": 1.7834618079887876,
      "grad_norm": 0.7828342914581299,
      "learning_rate": 0.0001246352547237503,
      "loss": 0.0841,
      "step": 12725
    },
    {
      "epoch": 1.7836019621583743,
      "grad_norm": 0.20947137475013733,
      "learning_rate": 0.00012462090408993062,
      "loss": 0.0497,
      "step": 12726
    },
    {
      "epoch": 1.7837421163279608,
      "grad_norm": 0.11443881690502167,
      "learning_rate": 0.00012460655345611097,
      "loss": 0.0268,
      "step": 12727
    },
    {
      "epoch": 1.7838822704975472,
      "grad_norm": 0.3637041747570038,
      "learning_rate": 0.00012459220282229133,
      "loss": 0.0427,
      "step": 12728
    },
    {
      "epoch": 1.784022424667134,
      "grad_norm": 0.11641380190849304,
      "learning_rate": 0.00012457785218847166,
      "loss": 0.0318,
      "step": 12729
    },
    {
      "epoch": 1.7841625788367204,
      "grad_norm": 0.3642352819442749,
      "learning_rate": 0.00012456350155465198,
      "loss": 0.1057,
      "step": 12730
    },
    {
      "epoch": 1.784302733006307,
      "grad_norm": 0.4932324290275574,
      "learning_rate": 0.0001245491509208323,
      "loss": 0.0619,
      "step": 12731
    },
    {
      "epoch": 1.7844428871758935,
      "grad_norm": 1.1922752857208252,
      "learning_rate": 0.00012453480028701267,
      "loss": 0.0725,
      "step": 12732
    },
    {
      "epoch": 1.78458304134548,
      "grad_norm": 1.504634976387024,
      "learning_rate": 0.00012452044965319302,
      "loss": 0.0547,
      "step": 12733
    },
    {
      "epoch": 1.7847231955150664,
      "grad_norm": 0.5879935622215271,
      "learning_rate": 0.00012450609901937335,
      "loss": 0.0553,
      "step": 12734
    },
    {
      "epoch": 1.784863349684653,
      "grad_norm": 2.0564184188842773,
      "learning_rate": 0.00012449174838555368,
      "loss": 0.2254,
      "step": 12735
    },
    {
      "epoch": 1.7850035038542398,
      "grad_norm": 0.10723566263914108,
      "learning_rate": 0.00012447739775173404,
      "loss": 0.0177,
      "step": 12736
    },
    {
      "epoch": 1.7851436580238262,
      "grad_norm": 0.3034172058105469,
      "learning_rate": 0.00012446304711791436,
      "loss": 0.0379,
      "step": 12737
    },
    {
      "epoch": 1.7852838121934127,
      "grad_norm": 0.5171294808387756,
      "learning_rate": 0.0001244486964840947,
      "loss": 0.0432,
      "step": 12738
    },
    {
      "epoch": 1.7854239663629992,
      "grad_norm": 0.244834303855896,
      "learning_rate": 0.00012443434585027505,
      "loss": 0.0378,
      "step": 12739
    },
    {
      "epoch": 1.7855641205325858,
      "grad_norm": 0.1415349841117859,
      "learning_rate": 0.00012441999521645538,
      "loss": 0.0325,
      "step": 12740
    },
    {
      "epoch": 1.7857042747021725,
      "grad_norm": 0.14750425517559052,
      "learning_rate": 0.00012440564458263573,
      "loss": 0.0069,
      "step": 12741
    },
    {
      "epoch": 1.785844428871759,
      "grad_norm": 0.12571613490581512,
      "learning_rate": 0.00012439129394881606,
      "loss": 0.0087,
      "step": 12742
    },
    {
      "epoch": 1.7859845830413454,
      "grad_norm": 1.6872973442077637,
      "learning_rate": 0.0001243769433149964,
      "loss": 0.0459,
      "step": 12743
    },
    {
      "epoch": 1.786124737210932,
      "grad_norm": 0.7671909928321838,
      "learning_rate": 0.00012436259268117675,
      "loss": 0.0562,
      "step": 12744
    },
    {
      "epoch": 1.7862648913805186,
      "grad_norm": 0.09681243449449539,
      "learning_rate": 0.00012434824204735707,
      "loss": 0.009,
      "step": 12745
    },
    {
      "epoch": 1.7864050455501053,
      "grad_norm": 0.393929123878479,
      "learning_rate": 0.00012433389141353743,
      "loss": 0.0547,
      "step": 12746
    },
    {
      "epoch": 1.7865451997196917,
      "grad_norm": 0.18076857924461365,
      "learning_rate": 0.00012431954077971776,
      "loss": 0.0164,
      "step": 12747
    },
    {
      "epoch": 1.7866853538892782,
      "grad_norm": 0.3891890347003937,
      "learning_rate": 0.0001243051901458981,
      "loss": 0.0754,
      "step": 12748
    },
    {
      "epoch": 1.7868255080588646,
      "grad_norm": 0.8744583129882812,
      "learning_rate": 0.00012429083951207844,
      "loss": 0.0431,
      "step": 12749
    },
    {
      "epoch": 1.7869656622284513,
      "grad_norm": 0.5414014458656311,
      "learning_rate": 0.00012427648887825877,
      "loss": 0.0516,
      "step": 12750
    },
    {
      "epoch": 1.787105816398038,
      "grad_norm": 0.19303114712238312,
      "learning_rate": 0.00012426213824443913,
      "loss": 0.0125,
      "step": 12751
    },
    {
      "epoch": 1.7872459705676245,
      "grad_norm": 0.9245198369026184,
      "learning_rate": 0.00012424778761061946,
      "loss": 0.0468,
      "step": 12752
    },
    {
      "epoch": 1.787386124737211,
      "grad_norm": 0.13063810765743256,
      "learning_rate": 0.00012423343697679978,
      "loss": 0.0081,
      "step": 12753
    },
    {
      "epoch": 1.7875262789067974,
      "grad_norm": 0.3303832709789276,
      "learning_rate": 0.00012421908634298014,
      "loss": 0.0576,
      "step": 12754
    },
    {
      "epoch": 1.787666433076384,
      "grad_norm": 0.27099576592445374,
      "learning_rate": 0.00012420473570916047,
      "loss": 0.0708,
      "step": 12755
    },
    {
      "epoch": 1.7878065872459705,
      "grad_norm": 0.30913522839546204,
      "learning_rate": 0.00012419038507534082,
      "loss": 0.0309,
      "step": 12756
    },
    {
      "epoch": 1.7879467414155572,
      "grad_norm": 0.34400373697280884,
      "learning_rate": 0.00012417603444152115,
      "loss": 0.1017,
      "step": 12757
    },
    {
      "epoch": 1.7880868955851437,
      "grad_norm": 0.3078020513057709,
      "learning_rate": 0.00012416168380770148,
      "loss": 0.0226,
      "step": 12758
    },
    {
      "epoch": 1.7882270497547301,
      "grad_norm": 0.19640223681926727,
      "learning_rate": 0.00012414733317388184,
      "loss": 0.0409,
      "step": 12759
    },
    {
      "epoch": 1.7883672039243166,
      "grad_norm": 0.40916523337364197,
      "learning_rate": 0.0001241329825400622,
      "loss": 0.0618,
      "step": 12760
    },
    {
      "epoch": 1.7885073580939033,
      "grad_norm": 0.3068006932735443,
      "learning_rate": 0.00012411863190624252,
      "loss": 0.0612,
      "step": 12761
    },
    {
      "epoch": 1.78864751226349,
      "grad_norm": 0.0761723741889,
      "learning_rate": 0.00012410428127242285,
      "loss": 0.0072,
      "step": 12762
    },
    {
      "epoch": 1.7887876664330764,
      "grad_norm": 0.14007216691970825,
      "learning_rate": 0.0001240899306386032,
      "loss": 0.0171,
      "step": 12763
    },
    {
      "epoch": 1.7889278206026629,
      "grad_norm": 0.24300192296504974,
      "learning_rate": 0.00012407558000478353,
      "loss": 0.0347,
      "step": 12764
    },
    {
      "epoch": 1.7890679747722493,
      "grad_norm": 0.4248351752758026,
      "learning_rate": 0.0001240612293709639,
      "loss": 0.0481,
      "step": 12765
    },
    {
      "epoch": 1.789208128941836,
      "grad_norm": 0.4369022250175476,
      "learning_rate": 0.00012404687873714422,
      "loss": 0.0442,
      "step": 12766
    },
    {
      "epoch": 1.7893482831114227,
      "grad_norm": 0.09725047647953033,
      "learning_rate": 0.00012403252810332455,
      "loss": 0.0128,
      "step": 12767
    },
    {
      "epoch": 1.7894884372810091,
      "grad_norm": 0.25909674167633057,
      "learning_rate": 0.0001240181774695049,
      "loss": 0.0482,
      "step": 12768
    },
    {
      "epoch": 1.7896285914505956,
      "grad_norm": 0.7365142703056335,
      "learning_rate": 0.00012400382683568523,
      "loss": 0.0662,
      "step": 12769
    },
    {
      "epoch": 1.789768745620182,
      "grad_norm": 0.22100892663002014,
      "learning_rate": 0.00012398947620186556,
      "loss": 0.0367,
      "step": 12770
    },
    {
      "epoch": 1.7899088997897687,
      "grad_norm": 0.17569518089294434,
      "learning_rate": 0.00012397512556804592,
      "loss": 0.0145,
      "step": 12771
    },
    {
      "epoch": 1.7900490539593554,
      "grad_norm": 0.7322267293930054,
      "learning_rate": 0.00012396077493422624,
      "loss": 0.1292,
      "step": 12772
    },
    {
      "epoch": 1.790189208128942,
      "grad_norm": 0.1580660343170166,
      "learning_rate": 0.0001239464243004066,
      "loss": 0.0249,
      "step": 12773
    },
    {
      "epoch": 1.7903293622985283,
      "grad_norm": 0.5480430126190186,
      "learning_rate": 0.00012393207366658693,
      "loss": 0.078,
      "step": 12774
    },
    {
      "epoch": 1.7904695164681148,
      "grad_norm": 0.20228888094425201,
      "learning_rate": 0.00012391772303276726,
      "loss": 0.0221,
      "step": 12775
    },
    {
      "epoch": 1.7906096706377015,
      "grad_norm": 0.22830694913864136,
      "learning_rate": 0.0001239033723989476,
      "loss": 0.0397,
      "step": 12776
    },
    {
      "epoch": 1.7907498248072882,
      "grad_norm": 0.5371941328048706,
      "learning_rate": 0.00012388902176512794,
      "loss": 0.0343,
      "step": 12777
    },
    {
      "epoch": 1.7908899789768746,
      "grad_norm": 0.44322195649147034,
      "learning_rate": 0.0001238746711313083,
      "loss": 0.1148,
      "step": 12778
    },
    {
      "epoch": 1.791030133146461,
      "grad_norm": 0.08946286886930466,
      "learning_rate": 0.00012386032049748863,
      "loss": 0.0045,
      "step": 12779
    },
    {
      "epoch": 1.7911702873160475,
      "grad_norm": 0.24678777158260345,
      "learning_rate": 0.00012384596986366895,
      "loss": 0.0506,
      "step": 12780
    },
    {
      "epoch": 1.7913104414856342,
      "grad_norm": 0.1845226138830185,
      "learning_rate": 0.0001238316192298493,
      "loss": 0.0079,
      "step": 12781
    },
    {
      "epoch": 1.7914505956552207,
      "grad_norm": 0.7750501036643982,
      "learning_rate": 0.00012381726859602964,
      "loss": 0.1014,
      "step": 12782
    },
    {
      "epoch": 1.7915907498248074,
      "grad_norm": 0.8726668357849121,
      "learning_rate": 0.00012380291796221,
      "loss": 0.1029,
      "step": 12783
    },
    {
      "epoch": 1.7917309039943938,
      "grad_norm": 0.4718799591064453,
      "learning_rate": 0.00012378856732839032,
      "loss": 0.0209,
      "step": 12784
    },
    {
      "epoch": 1.7918710581639803,
      "grad_norm": 1.7219072580337524,
      "learning_rate": 0.00012377421669457065,
      "loss": 0.3222,
      "step": 12785
    },
    {
      "epoch": 1.792011212333567,
      "grad_norm": 0.48471567034721375,
      "learning_rate": 0.000123759866060751,
      "loss": 0.1472,
      "step": 12786
    },
    {
      "epoch": 1.7921513665031534,
      "grad_norm": 0.6885779500007629,
      "learning_rate": 0.00012374551542693136,
      "loss": 0.0614,
      "step": 12787
    },
    {
      "epoch": 1.7922915206727401,
      "grad_norm": 0.15752582252025604,
      "learning_rate": 0.0001237311647931117,
      "loss": 0.022,
      "step": 12788
    },
    {
      "epoch": 1.7924316748423266,
      "grad_norm": 0.27776530385017395,
      "learning_rate": 0.00012371681415929202,
      "loss": 0.0356,
      "step": 12789
    },
    {
      "epoch": 1.792571829011913,
      "grad_norm": 0.34364476799964905,
      "learning_rate": 0.00012370246352547235,
      "loss": 0.0325,
      "step": 12790
    },
    {
      "epoch": 1.7927119831814995,
      "grad_norm": 0.13597267866134644,
      "learning_rate": 0.0001236881128916527,
      "loss": 0.0334,
      "step": 12791
    },
    {
      "epoch": 1.7928521373510862,
      "grad_norm": 0.27088266611099243,
      "learning_rate": 0.00012367376225783306,
      "loss": 0.0204,
      "step": 12792
    },
    {
      "epoch": 1.7929922915206729,
      "grad_norm": 0.14142204821109772,
      "learning_rate": 0.0001236594116240134,
      "loss": 0.0406,
      "step": 12793
    },
    {
      "epoch": 1.7931324456902593,
      "grad_norm": 0.355624258518219,
      "learning_rate": 0.00012364506099019372,
      "loss": 0.0506,
      "step": 12794
    },
    {
      "epoch": 1.7932725998598458,
      "grad_norm": 0.2310442328453064,
      "learning_rate": 0.00012363071035637407,
      "loss": 0.0446,
      "step": 12795
    },
    {
      "epoch": 1.7934127540294322,
      "grad_norm": 0.1650094985961914,
      "learning_rate": 0.0001236163597225544,
      "loss": 0.025,
      "step": 12796
    },
    {
      "epoch": 1.793552908199019,
      "grad_norm": 0.17290274798870087,
      "learning_rate": 0.00012360200908873476,
      "loss": 0.0277,
      "step": 12797
    },
    {
      "epoch": 1.7936930623686056,
      "grad_norm": 0.18702958524227142,
      "learning_rate": 0.00012358765845491508,
      "loss": 0.0258,
      "step": 12798
    },
    {
      "epoch": 1.793833216538192,
      "grad_norm": 1.1847370862960815,
      "learning_rate": 0.0001235733078210954,
      "loss": 0.048,
      "step": 12799
    },
    {
      "epoch": 1.7939733707077785,
      "grad_norm": 0.22448472678661346,
      "learning_rate": 0.00012355895718727577,
      "loss": 0.0185,
      "step": 12800
    },
    {
      "epoch": 1.794113524877365,
      "grad_norm": 0.34777554869651794,
      "learning_rate": 0.0001235446065534561,
      "loss": 0.0607,
      "step": 12801
    },
    {
      "epoch": 1.7942536790469517,
      "grad_norm": 0.10799390822649002,
      "learning_rate": 0.00012353025591963643,
      "loss": 0.0132,
      "step": 12802
    },
    {
      "epoch": 1.7943938332165383,
      "grad_norm": 0.17135055363178253,
      "learning_rate": 0.00012351590528581678,
      "loss": 0.0442,
      "step": 12803
    },
    {
      "epoch": 1.7945339873861248,
      "grad_norm": 0.25680282711982727,
      "learning_rate": 0.0001235015546519971,
      "loss": 0.0561,
      "step": 12804
    },
    {
      "epoch": 1.7946741415557113,
      "grad_norm": 0.24472257494926453,
      "learning_rate": 0.00012348720401817747,
      "loss": 0.0595,
      "step": 12805
    },
    {
      "epoch": 1.7948142957252977,
      "grad_norm": 0.3981437683105469,
      "learning_rate": 0.0001234728533843578,
      "loss": 0.033,
      "step": 12806
    },
    {
      "epoch": 1.7949544498948844,
      "grad_norm": 0.33710384368896484,
      "learning_rate": 0.00012345850275053812,
      "loss": 0.0274,
      "step": 12807
    },
    {
      "epoch": 1.795094604064471,
      "grad_norm": 0.5404962301254272,
      "learning_rate": 0.00012344415211671848,
      "loss": 0.0724,
      "step": 12808
    },
    {
      "epoch": 1.7952347582340575,
      "grad_norm": 0.5258068442344666,
      "learning_rate": 0.0001234298014828988,
      "loss": 0.0278,
      "step": 12809
    },
    {
      "epoch": 1.795374912403644,
      "grad_norm": 0.23582759499549866,
      "learning_rate": 0.00012341545084907916,
      "loss": 0.0279,
      "step": 12810
    },
    {
      "epoch": 1.7955150665732305,
      "grad_norm": 0.3797152042388916,
      "learning_rate": 0.0001234011002152595,
      "loss": 0.0589,
      "step": 12811
    },
    {
      "epoch": 1.7956552207428171,
      "grad_norm": 0.18421849608421326,
      "learning_rate": 0.00012338674958143982,
      "loss": 0.0319,
      "step": 12812
    },
    {
      "epoch": 1.7957953749124036,
      "grad_norm": 0.16365164518356323,
      "learning_rate": 0.00012337239894762018,
      "loss": 0.0205,
      "step": 12813
    },
    {
      "epoch": 1.7959355290819903,
      "grad_norm": 0.37158793210983276,
      "learning_rate": 0.00012335804831380053,
      "loss": 0.0319,
      "step": 12814
    },
    {
      "epoch": 1.7960756832515767,
      "grad_norm": 0.46859195828437805,
      "learning_rate": 0.00012334369767998086,
      "loss": 0.0657,
      "step": 12815
    },
    {
      "epoch": 1.7962158374211632,
      "grad_norm": 0.18334749341011047,
      "learning_rate": 0.0001233293470461612,
      "loss": 0.0262,
      "step": 12816
    },
    {
      "epoch": 1.7963559915907499,
      "grad_norm": 0.5150696635246277,
      "learning_rate": 0.00012331499641234152,
      "loss": 0.0284,
      "step": 12817
    },
    {
      "epoch": 1.7964961457603363,
      "grad_norm": 0.4401821494102478,
      "learning_rate": 0.00012330064577852187,
      "loss": 0.0935,
      "step": 12818
    },
    {
      "epoch": 1.796636299929923,
      "grad_norm": 0.4382910132408142,
      "learning_rate": 0.00012328629514470223,
      "loss": 0.0607,
      "step": 12819
    },
    {
      "epoch": 1.7967764540995095,
      "grad_norm": 0.47764065861701965,
      "learning_rate": 0.00012327194451088256,
      "loss": 0.0603,
      "step": 12820
    },
    {
      "epoch": 1.796916608269096,
      "grad_norm": 0.35685819387435913,
      "learning_rate": 0.00012325759387706289,
      "loss": 0.0474,
      "step": 12821
    },
    {
      "epoch": 1.7970567624386824,
      "grad_norm": 0.19071047008037567,
      "learning_rate": 0.00012324324324324324,
      "loss": 0.026,
      "step": 12822
    },
    {
      "epoch": 1.797196916608269,
      "grad_norm": 0.15680713951587677,
      "learning_rate": 0.00012322889260942357,
      "loss": 0.0159,
      "step": 12823
    },
    {
      "epoch": 1.7973370707778558,
      "grad_norm": 0.4171592891216278,
      "learning_rate": 0.00012321454197560393,
      "loss": 0.0324,
      "step": 12824
    },
    {
      "epoch": 1.7974772249474422,
      "grad_norm": 0.3295215964317322,
      "learning_rate": 0.00012320019134178425,
      "loss": 0.0288,
      "step": 12825
    },
    {
      "epoch": 1.7976173791170287,
      "grad_norm": 0.5781238675117493,
      "learning_rate": 0.00012318584070796458,
      "loss": 0.0533,
      "step": 12826
    },
    {
      "epoch": 1.7977575332866151,
      "grad_norm": 0.13872668147087097,
      "learning_rate": 0.00012317149007414494,
      "loss": 0.0139,
      "step": 12827
    },
    {
      "epoch": 1.7978976874562018,
      "grad_norm": 0.7919232249259949,
      "learning_rate": 0.00012315713944032527,
      "loss": 0.0213,
      "step": 12828
    },
    {
      "epoch": 1.7980378416257885,
      "grad_norm": 0.29356151819229126,
      "learning_rate": 0.00012314278880650562,
      "loss": 0.0295,
      "step": 12829
    },
    {
      "epoch": 1.798177995795375,
      "grad_norm": 0.5192530751228333,
      "learning_rate": 0.00012312843817268595,
      "loss": 0.0744,
      "step": 12830
    },
    {
      "epoch": 1.7983181499649614,
      "grad_norm": 0.4007788300514221,
      "learning_rate": 0.00012311408753886628,
      "loss": 0.0138,
      "step": 12831
    },
    {
      "epoch": 1.7984583041345479,
      "grad_norm": 0.6972614526748657,
      "learning_rate": 0.00012309973690504664,
      "loss": 0.1492,
      "step": 12832
    },
    {
      "epoch": 1.7985984583041346,
      "grad_norm": 1.7300293445587158,
      "learning_rate": 0.00012308538627122696,
      "loss": 0.2528,
      "step": 12833
    },
    {
      "epoch": 1.7987386124737212,
      "grad_norm": 1.1460438966751099,
      "learning_rate": 0.0001230710356374073,
      "loss": 0.082,
      "step": 12834
    },
    {
      "epoch": 1.7988787666433077,
      "grad_norm": 1.8754453659057617,
      "learning_rate": 0.00012305668500358765,
      "loss": 0.0555,
      "step": 12835
    },
    {
      "epoch": 1.7990189208128942,
      "grad_norm": 0.2733992636203766,
      "learning_rate": 0.00012304233436976798,
      "loss": 0.0723,
      "step": 12836
    },
    {
      "epoch": 1.7991590749824806,
      "grad_norm": 0.18559584021568298,
      "learning_rate": 0.00012302798373594833,
      "loss": 0.0822,
      "step": 12837
    },
    {
      "epoch": 1.7992992291520673,
      "grad_norm": 0.34309783577919006,
      "learning_rate": 0.00012301363310212866,
      "loss": 0.0942,
      "step": 12838
    },
    {
      "epoch": 1.799439383321654,
      "grad_norm": 0.200308695435524,
      "learning_rate": 0.000122999282468309,
      "loss": 0.0212,
      "step": 12839
    },
    {
      "epoch": 1.7995795374912404,
      "grad_norm": 0.6797464489936829,
      "learning_rate": 0.00012298493183448935,
      "loss": 0.0895,
      "step": 12840
    },
    {
      "epoch": 1.799719691660827,
      "grad_norm": 0.3852284550666809,
      "learning_rate": 0.0001229705812006697,
      "loss": 0.0917,
      "step": 12841
    },
    {
      "epoch": 1.7998598458304134,
      "grad_norm": 0.5513192415237427,
      "learning_rate": 0.00012295623056685003,
      "loss": 0.0811,
      "step": 12842
    },
    {
      "epoch": 1.8,
      "grad_norm": 0.6143665909767151,
      "learning_rate": 0.00012294187993303036,
      "loss": 0.0854,
      "step": 12843
    },
    {
      "epoch": 1.8001401541695865,
      "grad_norm": 0.296474426984787,
      "learning_rate": 0.00012292752929921069,
      "loss": 0.0443,
      "step": 12844
    },
    {
      "epoch": 1.8002803083391732,
      "grad_norm": 0.5858607292175293,
      "learning_rate": 0.00012291317866539104,
      "loss": 0.0267,
      "step": 12845
    },
    {
      "epoch": 1.8004204625087596,
      "grad_norm": 0.50831538438797,
      "learning_rate": 0.0001228988280315714,
      "loss": 0.0776,
      "step": 12846
    },
    {
      "epoch": 1.800560616678346,
      "grad_norm": 0.2200281172990799,
      "learning_rate": 0.00012288447739775173,
      "loss": 0.0608,
      "step": 12847
    },
    {
      "epoch": 1.8007007708479326,
      "grad_norm": 0.4607313275337219,
      "learning_rate": 0.00012287012676393205,
      "loss": 0.0823,
      "step": 12848
    },
    {
      "epoch": 1.8008409250175192,
      "grad_norm": 0.4934716820716858,
      "learning_rate": 0.0001228557761301124,
      "loss": 0.0548,
      "step": 12849
    },
    {
      "epoch": 1.800981079187106,
      "grad_norm": 0.4305703341960907,
      "learning_rate": 0.00012284142549629274,
      "loss": 0.0893,
      "step": 12850
    },
    {
      "epoch": 1.8011212333566924,
      "grad_norm": 0.2147679477930069,
      "learning_rate": 0.0001228270748624731,
      "loss": 0.0237,
      "step": 12851
    },
    {
      "epoch": 1.8012613875262788,
      "grad_norm": 0.19156956672668457,
      "learning_rate": 0.00012281272422865342,
      "loss": 0.046,
      "step": 12852
    },
    {
      "epoch": 1.8014015416958653,
      "grad_norm": 0.2585931122303009,
      "learning_rate": 0.00012279837359483375,
      "loss": 0.049,
      "step": 12853
    },
    {
      "epoch": 1.801541695865452,
      "grad_norm": 0.24206191301345825,
      "learning_rate": 0.0001227840229610141,
      "loss": 0.029,
      "step": 12854
    },
    {
      "epoch": 1.8016818500350387,
      "grad_norm": 0.6593036651611328,
      "learning_rate": 0.00012276967232719444,
      "loss": 0.067,
      "step": 12855
    },
    {
      "epoch": 1.8018220042046251,
      "grad_norm": 0.17387881875038147,
      "learning_rate": 0.0001227553216933748,
      "loss": 0.0155,
      "step": 12856
    },
    {
      "epoch": 1.8019621583742116,
      "grad_norm": 0.20001249015331268,
      "learning_rate": 0.00012274097105955512,
      "loss": 0.0456,
      "step": 12857
    },
    {
      "epoch": 1.802102312543798,
      "grad_norm": 0.1928340494632721,
      "learning_rate": 0.00012272662042573545,
      "loss": 0.0272,
      "step": 12858
    },
    {
      "epoch": 1.8022424667133847,
      "grad_norm": 0.2522693872451782,
      "learning_rate": 0.0001227122697919158,
      "loss": 0.025,
      "step": 12859
    },
    {
      "epoch": 1.8023826208829714,
      "grad_norm": 0.24919652938842773,
      "learning_rate": 0.00012269791915809613,
      "loss": 0.058,
      "step": 12860
    },
    {
      "epoch": 1.8025227750525579,
      "grad_norm": 0.4658038914203644,
      "learning_rate": 0.0001226835685242765,
      "loss": 0.1087,
      "step": 12861
    },
    {
      "epoch": 1.8026629292221443,
      "grad_norm": 0.16299644112586975,
      "learning_rate": 0.00012266921789045682,
      "loss": 0.0226,
      "step": 12862
    },
    {
      "epoch": 1.8028030833917308,
      "grad_norm": 0.19520533084869385,
      "learning_rate": 0.00012265486725663715,
      "loss": 0.0299,
      "step": 12863
    },
    {
      "epoch": 1.8029432375613175,
      "grad_norm": 0.27884581685066223,
      "learning_rate": 0.0001226405166228175,
      "loss": 0.0269,
      "step": 12864
    },
    {
      "epoch": 1.8030833917309042,
      "grad_norm": 0.5366170406341553,
      "learning_rate": 0.00012262616598899783,
      "loss": 0.0769,
      "step": 12865
    },
    {
      "epoch": 1.8032235459004906,
      "grad_norm": 0.8271068334579468,
      "learning_rate": 0.00012261181535517819,
      "loss": 0.1031,
      "step": 12866
    },
    {
      "epoch": 1.803363700070077,
      "grad_norm": 0.609592616558075,
      "learning_rate": 0.00012259746472135851,
      "loss": 0.0556,
      "step": 12867
    },
    {
      "epoch": 1.8035038542396635,
      "grad_norm": 0.3008006811141968,
      "learning_rate": 0.00012258311408753884,
      "loss": 0.0526,
      "step": 12868
    },
    {
      "epoch": 1.8036440084092502,
      "grad_norm": 0.30220529437065125,
      "learning_rate": 0.0001225687634537192,
      "loss": 0.0502,
      "step": 12869
    },
    {
      "epoch": 1.8037841625788367,
      "grad_norm": 0.22283318638801575,
      "learning_rate": 0.00012255441281989953,
      "loss": 0.0398,
      "step": 12870
    },
    {
      "epoch": 1.8039243167484234,
      "grad_norm": 0.21892574429512024,
      "learning_rate": 0.00012254006218607986,
      "loss": 0.0234,
      "step": 12871
    },
    {
      "epoch": 1.8040644709180098,
      "grad_norm": 0.17471952736377716,
      "learning_rate": 0.0001225257115522602,
      "loss": 0.0211,
      "step": 12872
    },
    {
      "epoch": 1.8042046250875963,
      "grad_norm": 0.32636162638664246,
      "learning_rate": 0.00012251136091844057,
      "loss": 0.087,
      "step": 12873
    },
    {
      "epoch": 1.804344779257183,
      "grad_norm": 0.29673856496810913,
      "learning_rate": 0.0001224970102846209,
      "loss": 0.0174,
      "step": 12874
    },
    {
      "epoch": 1.8044849334267694,
      "grad_norm": 0.34801968932151794,
      "learning_rate": 0.00012248265965080122,
      "loss": 0.0722,
      "step": 12875
    },
    {
      "epoch": 1.804625087596356,
      "grad_norm": 0.2269166111946106,
      "learning_rate": 0.00012246830901698158,
      "loss": 0.0395,
      "step": 12876
    },
    {
      "epoch": 1.8047652417659426,
      "grad_norm": 0.4518377184867859,
      "learning_rate": 0.0001224539583831619,
      "loss": 0.043,
      "step": 12877
    },
    {
      "epoch": 1.804905395935529,
      "grad_norm": 0.6025972962379456,
      "learning_rate": 0.00012243960774934226,
      "loss": 0.0827,
      "step": 12878
    },
    {
      "epoch": 1.8050455501051155,
      "grad_norm": 0.27791860699653625,
      "learning_rate": 0.0001224252571155226,
      "loss": 0.0517,
      "step": 12879
    },
    {
      "epoch": 1.8051857042747022,
      "grad_norm": 0.6230483651161194,
      "learning_rate": 0.00012241090648170292,
      "loss": 0.0594,
      "step": 12880
    },
    {
      "epoch": 1.8053258584442888,
      "grad_norm": 0.3419473469257355,
      "learning_rate": 0.00012239655584788328,
      "loss": 0.0554,
      "step": 12881
    },
    {
      "epoch": 1.8054660126138753,
      "grad_norm": 1.1800894737243652,
      "learning_rate": 0.0001223822052140636,
      "loss": 0.0809,
      "step": 12882
    },
    {
      "epoch": 1.8056061667834618,
      "grad_norm": 1.495298147201538,
      "learning_rate": 0.00012236785458024396,
      "loss": 0.155,
      "step": 12883
    },
    {
      "epoch": 1.8057463209530482,
      "grad_norm": 1.0437233448028564,
      "learning_rate": 0.0001223535039464243,
      "loss": 0.1134,
      "step": 12884
    },
    {
      "epoch": 1.805886475122635,
      "grad_norm": 0.05267738923430443,
      "learning_rate": 0.00012233915331260462,
      "loss": 0.0039,
      "step": 12885
    },
    {
      "epoch": 1.8060266292922216,
      "grad_norm": 0.18792273104190826,
      "learning_rate": 0.00012232480267878497,
      "loss": 0.021,
      "step": 12886
    },
    {
      "epoch": 1.806166783461808,
      "grad_norm": 0.30840983986854553,
      "learning_rate": 0.0001223104520449653,
      "loss": 0.0551,
      "step": 12887
    },
    {
      "epoch": 1.8063069376313945,
      "grad_norm": 0.158649742603302,
      "learning_rate": 0.00012229610141114566,
      "loss": 0.0346,
      "step": 12888
    },
    {
      "epoch": 1.806447091800981,
      "grad_norm": 0.2195279598236084,
      "learning_rate": 0.000122281750777326,
      "loss": 0.0431,
      "step": 12889
    },
    {
      "epoch": 1.8065872459705676,
      "grad_norm": 0.18091900646686554,
      "learning_rate": 0.00012226740014350632,
      "loss": 0.0185,
      "step": 12890
    },
    {
      "epoch": 1.8067274001401543,
      "grad_norm": 0.17235256731510162,
      "learning_rate": 0.00012225304950968667,
      "loss": 0.0085,
      "step": 12891
    },
    {
      "epoch": 1.8068675543097408,
      "grad_norm": 0.4248929023742676,
      "learning_rate": 0.00012223869887586703,
      "loss": 0.0924,
      "step": 12892
    },
    {
      "epoch": 1.8070077084793272,
      "grad_norm": 0.1359269618988037,
      "learning_rate": 0.00012222434824204736,
      "loss": 0.0328,
      "step": 12893
    },
    {
      "epoch": 1.8071478626489137,
      "grad_norm": 0.3212231993675232,
      "learning_rate": 0.00012220999760822768,
      "loss": 0.0783,
      "step": 12894
    },
    {
      "epoch": 1.8072880168185004,
      "grad_norm": 0.38217827677726746,
      "learning_rate": 0.000122195646974408,
      "loss": 0.1037,
      "step": 12895
    },
    {
      "epoch": 1.807428170988087,
      "grad_norm": 0.2574213147163391,
      "learning_rate": 0.00012218129634058837,
      "loss": 0.0424,
      "step": 12896
    },
    {
      "epoch": 1.8075683251576735,
      "grad_norm": 1.117431879043579,
      "learning_rate": 0.0001221669457067687,
      "loss": 0.0647,
      "step": 12897
    },
    {
      "epoch": 1.80770847932726,
      "grad_norm": 0.14649498462677002,
      "learning_rate": 0.00012215259507294905,
      "loss": 0.0179,
      "step": 12898
    },
    {
      "epoch": 1.8078486334968464,
      "grad_norm": 0.48996809124946594,
      "learning_rate": 0.00012213824443912938,
      "loss": 0.0621,
      "step": 12899
    },
    {
      "epoch": 1.8079887876664331,
      "grad_norm": 0.19080258905887604,
      "learning_rate": 0.00012212389380530974,
      "loss": 0.0466,
      "step": 12900
    },
    {
      "epoch": 1.8081289418360196,
      "grad_norm": 0.18082423508167267,
      "learning_rate": 0.00012210954317149006,
      "loss": 0.0171,
      "step": 12901
    },
    {
      "epoch": 1.8082690960056063,
      "grad_norm": 0.1817387491464615,
      "learning_rate": 0.0001220951925376704,
      "loss": 0.0219,
      "step": 12902
    },
    {
      "epoch": 1.8084092501751927,
      "grad_norm": 0.31144678592681885,
      "learning_rate": 0.00012208084190385072,
      "loss": 0.0164,
      "step": 12903
    },
    {
      "epoch": 1.8085494043447792,
      "grad_norm": 0.22523410618305206,
      "learning_rate": 0.00012206649127003108,
      "loss": 0.012,
      "step": 12904
    },
    {
      "epoch": 1.8086895585143656,
      "grad_norm": 0.3270754814147949,
      "learning_rate": 0.00012205214063621142,
      "loss": 0.1076,
      "step": 12905
    },
    {
      "epoch": 1.8088297126839523,
      "grad_norm": 0.3382010757923126,
      "learning_rate": 0.00012203779000239176,
      "loss": 0.0256,
      "step": 12906
    },
    {
      "epoch": 1.808969866853539,
      "grad_norm": 1.599377155303955,
      "learning_rate": 0.0001220234393685721,
      "loss": 0.1008,
      "step": 12907
    },
    {
      "epoch": 1.8091100210231255,
      "grad_norm": 0.4510815739631653,
      "learning_rate": 0.00012200908873475245,
      "loss": 0.0194,
      "step": 12908
    },
    {
      "epoch": 1.809250175192712,
      "grad_norm": 0.45227116346359253,
      "learning_rate": 0.00012199473810093277,
      "loss": 0.0672,
      "step": 12909
    },
    {
      "epoch": 1.8093903293622984,
      "grad_norm": 0.13850203156471252,
      "learning_rate": 0.00012198038746711312,
      "loss": 0.0136,
      "step": 12910
    },
    {
      "epoch": 1.809530483531885,
      "grad_norm": 0.12272889167070389,
      "learning_rate": 0.00012196603683329347,
      "loss": 0.0134,
      "step": 12911
    },
    {
      "epoch": 1.8096706377014717,
      "grad_norm": 0.2970481812953949,
      "learning_rate": 0.0001219516861994738,
      "loss": 0.0347,
      "step": 12912
    },
    {
      "epoch": 1.8098107918710582,
      "grad_norm": 0.1268194317817688,
      "learning_rate": 0.00012193733556565414,
      "loss": 0.0148,
      "step": 12913
    },
    {
      "epoch": 1.8099509460406447,
      "grad_norm": 0.3441358208656311,
      "learning_rate": 0.00012192298493183447,
      "loss": 0.0932,
      "step": 12914
    },
    {
      "epoch": 1.8100911002102311,
      "grad_norm": 0.3065379559993744,
      "learning_rate": 0.00012190863429801481,
      "loss": 0.0302,
      "step": 12915
    },
    {
      "epoch": 1.8102312543798178,
      "grad_norm": 0.4870491027832031,
      "learning_rate": 0.00012189428366419517,
      "loss": 0.0886,
      "step": 12916
    },
    {
      "epoch": 1.8103714085494045,
      "grad_norm": 0.2083703726530075,
      "learning_rate": 0.0001218799330303755,
      "loss": 0.0246,
      "step": 12917
    },
    {
      "epoch": 1.810511562718991,
      "grad_norm": 0.2707286775112152,
      "learning_rate": 0.00012186558239655584,
      "loss": 0.0481,
      "step": 12918
    },
    {
      "epoch": 1.8106517168885774,
      "grad_norm": 0.5525835156440735,
      "learning_rate": 0.00012185123176273618,
      "loss": 0.0246,
      "step": 12919
    },
    {
      "epoch": 1.8107918710581639,
      "grad_norm": 0.19326196610927582,
      "learning_rate": 0.00012183688112891651,
      "loss": 0.0569,
      "step": 12920
    },
    {
      "epoch": 1.8109320252277505,
      "grad_norm": 0.4657191336154938,
      "learning_rate": 0.00012182253049509685,
      "loss": 0.0782,
      "step": 12921
    },
    {
      "epoch": 1.8110721793973372,
      "grad_norm": 0.07895074784755707,
      "learning_rate": 0.0001218081798612772,
      "loss": 0.0203,
      "step": 12922
    },
    {
      "epoch": 1.8112123335669237,
      "grad_norm": 0.4483071267604828,
      "learning_rate": 0.00012179382922745754,
      "loss": 0.0315,
      "step": 12923
    },
    {
      "epoch": 1.8113524877365101,
      "grad_norm": 0.4019063711166382,
      "learning_rate": 0.00012177947859363788,
      "loss": 0.0706,
      "step": 12924
    },
    {
      "epoch": 1.8114926419060966,
      "grad_norm": 0.314506471157074,
      "learning_rate": 0.00012176512795981821,
      "loss": 0.0625,
      "step": 12925
    },
    {
      "epoch": 1.8116327960756833,
      "grad_norm": 0.6163290143013,
      "learning_rate": 0.00012175077732599855,
      "loss": 0.1359,
      "step": 12926
    },
    {
      "epoch": 1.8117729502452697,
      "grad_norm": 1.1162049770355225,
      "learning_rate": 0.0001217364266921789,
      "loss": 0.1204,
      "step": 12927
    },
    {
      "epoch": 1.8119131044148564,
      "grad_norm": 0.13889525830745697,
      "learning_rate": 0.00012172207605835923,
      "loss": 0.0369,
      "step": 12928
    },
    {
      "epoch": 1.8120532585844429,
      "grad_norm": 0.5765511989593506,
      "learning_rate": 0.00012170772542453958,
      "loss": 0.0563,
      "step": 12929
    },
    {
      "epoch": 1.8121934127540293,
      "grad_norm": 0.7851291298866272,
      "learning_rate": 0.0001216933747907199,
      "loss": 0.0659,
      "step": 12930
    },
    {
      "epoch": 1.812333566923616,
      "grad_norm": 0.20594029128551483,
      "learning_rate": 0.00012167902415690025,
      "loss": 0.0333,
      "step": 12931
    },
    {
      "epoch": 1.8124737210932025,
      "grad_norm": 0.5577102303504944,
      "learning_rate": 0.0001216646735230806,
      "loss": 0.0596,
      "step": 12932
    },
    {
      "epoch": 1.8126138752627892,
      "grad_norm": 0.18293729424476624,
      "learning_rate": 0.00012165032288926093,
      "loss": 0.0154,
      "step": 12933
    },
    {
      "epoch": 1.8127540294323756,
      "grad_norm": 4.890850067138672,
      "learning_rate": 0.00012163597225544127,
      "loss": 0.145,
      "step": 12934
    },
    {
      "epoch": 1.812894183601962,
      "grad_norm": 0.7400285005569458,
      "learning_rate": 0.00012162162162162162,
      "loss": 0.0453,
      "step": 12935
    },
    {
      "epoch": 1.8130343377715485,
      "grad_norm": 0.2621132731437683,
      "learning_rate": 0.00012160727098780194,
      "loss": 0.016,
      "step": 12936
    },
    {
      "epoch": 1.8131744919411352,
      "grad_norm": 0.39345619082450867,
      "learning_rate": 0.00012159292035398229,
      "loss": 0.0355,
      "step": 12937
    },
    {
      "epoch": 1.813314646110722,
      "grad_norm": 0.14441049098968506,
      "learning_rate": 0.00012157856972016263,
      "loss": 0.0184,
      "step": 12938
    },
    {
      "epoch": 1.8134548002803084,
      "grad_norm": 0.15110982954502106,
      "learning_rate": 0.00012156421908634297,
      "loss": 0.0097,
      "step": 12939
    },
    {
      "epoch": 1.8135949544498948,
      "grad_norm": 0.33224159479141235,
      "learning_rate": 0.00012154986845252331,
      "loss": 0.0579,
      "step": 12940
    },
    {
      "epoch": 1.8137351086194813,
      "grad_norm": 0.2408864051103592,
      "learning_rate": 0.00012153551781870364,
      "loss": 0.0449,
      "step": 12941
    },
    {
      "epoch": 1.813875262789068,
      "grad_norm": 0.4061248302459717,
      "learning_rate": 0.00012152116718488398,
      "loss": 0.0769,
      "step": 12942
    },
    {
      "epoch": 1.8140154169586546,
      "grad_norm": 0.270393967628479,
      "learning_rate": 0.00012150681655106434,
      "loss": 0.0347,
      "step": 12943
    },
    {
      "epoch": 1.814155571128241,
      "grad_norm": 0.20464079082012177,
      "learning_rate": 0.00012149246591724467,
      "loss": 0.0385,
      "step": 12944
    },
    {
      "epoch": 1.8142957252978276,
      "grad_norm": 0.39568743109703064,
      "learning_rate": 0.00012147811528342501,
      "loss": 0.066,
      "step": 12945
    },
    {
      "epoch": 1.814435879467414,
      "grad_norm": 0.2845861613750458,
      "learning_rate": 0.00012146376464960535,
      "loss": 0.0206,
      "step": 12946
    },
    {
      "epoch": 1.8145760336370007,
      "grad_norm": 0.39516395330429077,
      "learning_rate": 0.00012144941401578568,
      "loss": 0.0989,
      "step": 12947
    },
    {
      "epoch": 1.8147161878065874,
      "grad_norm": 0.31977009773254395,
      "learning_rate": 0.00012143506338196604,
      "loss": 0.0327,
      "step": 12948
    },
    {
      "epoch": 1.8148563419761738,
      "grad_norm": 0.20780211687088013,
      "learning_rate": 0.00012142071274814636,
      "loss": 0.0263,
      "step": 12949
    },
    {
      "epoch": 1.8149964961457603,
      "grad_norm": 1.5113365650177002,
      "learning_rate": 0.0001214063621143267,
      "loss": 0.0165,
      "step": 12950
    },
    {
      "epoch": 1.8151366503153468,
      "grad_norm": 0.15647585690021515,
      "learning_rate": 0.00012139201148050705,
      "loss": 0.0255,
      "step": 12951
    },
    {
      "epoch": 1.8152768044849334,
      "grad_norm": 0.1031777411699295,
      "learning_rate": 0.00012137766084668738,
      "loss": 0.0304,
      "step": 12952
    },
    {
      "epoch": 1.8154169586545201,
      "grad_norm": 0.5230974555015564,
      "learning_rate": 0.00012136331021286772,
      "loss": 0.1189,
      "step": 12953
    },
    {
      "epoch": 1.8155571128241066,
      "grad_norm": 0.34056246280670166,
      "learning_rate": 0.00012134895957904807,
      "loss": 0.0957,
      "step": 12954
    },
    {
      "epoch": 1.815697266993693,
      "grad_norm": 0.2997777462005615,
      "learning_rate": 0.0001213346089452284,
      "loss": 0.044,
      "step": 12955
    },
    {
      "epoch": 1.8158374211632795,
      "grad_norm": 0.15355022251605988,
      "learning_rate": 0.00012132025831140875,
      "loss": 0.0153,
      "step": 12956
    },
    {
      "epoch": 1.8159775753328662,
      "grad_norm": 0.3450574576854706,
      "learning_rate": 0.00012130590767758907,
      "loss": 0.0336,
      "step": 12957
    },
    {
      "epoch": 1.8161177295024526,
      "grad_norm": 0.10439668595790863,
      "learning_rate": 0.00012129155704376942,
      "loss": 0.0146,
      "step": 12958
    },
    {
      "epoch": 1.8162578836720393,
      "grad_norm": 0.2745637893676758,
      "learning_rate": 0.00012127720640994977,
      "loss": 0.0187,
      "step": 12959
    },
    {
      "epoch": 1.8163980378416258,
      "grad_norm": 0.21283739805221558,
      "learning_rate": 0.0001212628557761301,
      "loss": 0.0284,
      "step": 12960
    },
    {
      "epoch": 1.8165381920112122,
      "grad_norm": 0.05814182385802269,
      "learning_rate": 0.00012124850514231044,
      "loss": 0.0125,
      "step": 12961
    },
    {
      "epoch": 1.816678346180799,
      "grad_norm": 0.10886926203966141,
      "learning_rate": 0.00012123415450849078,
      "loss": 0.0119,
      "step": 12962
    },
    {
      "epoch": 1.8168185003503854,
      "grad_norm": 0.06237927824258804,
      "learning_rate": 0.00012121980387467111,
      "loss": 0.0051,
      "step": 12963
    },
    {
      "epoch": 1.816958654519972,
      "grad_norm": 0.4345255196094513,
      "learning_rate": 0.00012120545324085147,
      "loss": 0.1363,
      "step": 12964
    },
    {
      "epoch": 1.8170988086895585,
      "grad_norm": 0.2464209944009781,
      "learning_rate": 0.0001211911026070318,
      "loss": 0.0291,
      "step": 12965
    },
    {
      "epoch": 1.817238962859145,
      "grad_norm": 0.05634305626153946,
      "learning_rate": 0.00012117675197321214,
      "loss": 0.0044,
      "step": 12966
    },
    {
      "epoch": 1.8173791170287315,
      "grad_norm": 0.41657713055610657,
      "learning_rate": 0.00012116240133939248,
      "loss": 0.0844,
      "step": 12967
    },
    {
      "epoch": 1.8175192711983181,
      "grad_norm": 0.20880872011184692,
      "learning_rate": 0.00012114805070557281,
      "loss": 0.0271,
      "step": 12968
    },
    {
      "epoch": 1.8176594253679048,
      "grad_norm": 0.20800060033798218,
      "learning_rate": 0.00012113370007175317,
      "loss": 0.0341,
      "step": 12969
    },
    {
      "epoch": 1.8177995795374913,
      "grad_norm": 0.6321690082550049,
      "learning_rate": 0.00012111934943793351,
      "loss": 0.1202,
      "step": 12970
    },
    {
      "epoch": 1.8179397337070777,
      "grad_norm": 0.3648313581943512,
      "learning_rate": 0.00012110499880411384,
      "loss": 0.0777,
      "step": 12971
    },
    {
      "epoch": 1.8180798878766642,
      "grad_norm": 0.1709165722131729,
      "learning_rate": 0.00012109064817029418,
      "loss": 0.0421,
      "step": 12972
    },
    {
      "epoch": 1.8182200420462509,
      "grad_norm": 0.4314466118812561,
      "learning_rate": 0.00012107629753647451,
      "loss": 0.0814,
      "step": 12973
    },
    {
      "epoch": 1.8183601962158376,
      "grad_norm": 0.3605118989944458,
      "learning_rate": 0.00012106194690265485,
      "loss": 0.038,
      "step": 12974
    },
    {
      "epoch": 1.818500350385424,
      "grad_norm": 0.19191895425319672,
      "learning_rate": 0.0001210475962688352,
      "loss": 0.0215,
      "step": 12975
    },
    {
      "epoch": 1.8186405045550105,
      "grad_norm": 0.3105887770652771,
      "learning_rate": 0.00012103324563501553,
      "loss": 0.056,
      "step": 12976
    },
    {
      "epoch": 1.818780658724597,
      "grad_norm": 0.7323061227798462,
      "learning_rate": 0.00012101889500119588,
      "loss": 0.0867,
      "step": 12977
    },
    {
      "epoch": 1.8189208128941836,
      "grad_norm": 0.38156676292419434,
      "learning_rate": 0.00012100454436737622,
      "loss": 0.0299,
      "step": 12978
    },
    {
      "epoch": 1.8190609670637703,
      "grad_norm": 1.1498442888259888,
      "learning_rate": 0.00012099019373355655,
      "loss": 0.1193,
      "step": 12979
    },
    {
      "epoch": 1.8192011212333568,
      "grad_norm": 0.37885749340057373,
      "learning_rate": 0.0001209758430997369,
      "loss": 0.0684,
      "step": 12980
    },
    {
      "epoch": 1.8193412754029432,
      "grad_norm": 1.0525306463241577,
      "learning_rate": 0.00012096149246591723,
      "loss": 0.0725,
      "step": 12981
    },
    {
      "epoch": 1.8194814295725297,
      "grad_norm": 0.6396930813789368,
      "learning_rate": 0.00012094714183209757,
      "loss": 0.231,
      "step": 12982
    },
    {
      "epoch": 1.8196215837421164,
      "grad_norm": 0.31932976841926575,
      "learning_rate": 0.00012093279119827791,
      "loss": 0.0281,
      "step": 12983
    },
    {
      "epoch": 1.819761737911703,
      "grad_norm": 2.0052361488342285,
      "learning_rate": 0.00012091844056445824,
      "loss": 0.1293,
      "step": 12984
    },
    {
      "epoch": 1.8199018920812895,
      "grad_norm": 0.32648447155952454,
      "learning_rate": 0.0001209040899306386,
      "loss": 0.0603,
      "step": 12985
    },
    {
      "epoch": 1.820042046250876,
      "grad_norm": 0.2711527943611145,
      "learning_rate": 0.00012088973929681894,
      "loss": 0.0393,
      "step": 12986
    },
    {
      "epoch": 1.8201822004204624,
      "grad_norm": 0.2042703926563263,
      "learning_rate": 0.00012087538866299927,
      "loss": 0.0398,
      "step": 12987
    },
    {
      "epoch": 1.820322354590049,
      "grad_norm": 0.26983964443206787,
      "learning_rate": 0.00012086103802917961,
      "loss": 0.0357,
      "step": 12988
    },
    {
      "epoch": 1.8204625087596356,
      "grad_norm": 0.7581669688224792,
      "learning_rate": 0.00012084668739535995,
      "loss": 0.0962,
      "step": 12989
    },
    {
      "epoch": 1.8206026629292222,
      "grad_norm": 0.5755071640014648,
      "learning_rate": 0.00012083233676154028,
      "loss": 0.0493,
      "step": 12990
    },
    {
      "epoch": 1.8207428170988087,
      "grad_norm": 0.15472157299518585,
      "learning_rate": 0.00012081798612772064,
      "loss": 0.0256,
      "step": 12991
    },
    {
      "epoch": 1.8208829712683952,
      "grad_norm": 0.14568139612674713,
      "learning_rate": 0.00012080363549390097,
      "loss": 0.0365,
      "step": 12992
    },
    {
      "epoch": 1.8210231254379816,
      "grad_norm": 0.43495064973831177,
      "learning_rate": 0.00012078928486008131,
      "loss": 0.045,
      "step": 12993
    },
    {
      "epoch": 1.8211632796075683,
      "grad_norm": 0.3852024972438812,
      "learning_rate": 0.00012077493422626165,
      "loss": 0.0351,
      "step": 12994
    },
    {
      "epoch": 1.821303433777155,
      "grad_norm": 0.45781028270721436,
      "learning_rate": 0.00012076058359244198,
      "loss": 0.0785,
      "step": 12995
    },
    {
      "epoch": 1.8214435879467414,
      "grad_norm": 0.12383861839771271,
      "learning_rate": 0.00012074623295862234,
      "loss": 0.0258,
      "step": 12996
    },
    {
      "epoch": 1.821583742116328,
      "grad_norm": 0.10406908392906189,
      "learning_rate": 0.00012073188232480268,
      "loss": 0.0116,
      "step": 12997
    },
    {
      "epoch": 1.8217238962859144,
      "grad_norm": 0.1576005518436432,
      "learning_rate": 0.000120717531690983,
      "loss": 0.0116,
      "step": 12998
    },
    {
      "epoch": 1.821864050455501,
      "grad_norm": 0.17741477489471436,
      "learning_rate": 0.00012070318105716335,
      "loss": 0.019,
      "step": 12999
    },
    {
      "epoch": 1.8220042046250877,
      "grad_norm": 0.15079960227012634,
      "learning_rate": 0.00012068883042334368,
      "loss": 0.0087,
      "step": 13000
    },
    {
      "epoch": 1.8221443587946742,
      "grad_norm": 0.16749387979507446,
      "learning_rate": 0.00012067447978952403,
      "loss": 0.0267,
      "step": 13001
    },
    {
      "epoch": 1.8222845129642606,
      "grad_norm": 0.24852505326271057,
      "learning_rate": 0.00012066012915570437,
      "loss": 0.0528,
      "step": 13002
    },
    {
      "epoch": 1.822424667133847,
      "grad_norm": 0.4245162010192871,
      "learning_rate": 0.0001206457785218847,
      "loss": 0.0867,
      "step": 13003
    },
    {
      "epoch": 1.8225648213034338,
      "grad_norm": 0.1511314958333969,
      "learning_rate": 0.00012063142788806504,
      "loss": 0.0185,
      "step": 13004
    },
    {
      "epoch": 1.8227049754730205,
      "grad_norm": 0.14931067824363708,
      "learning_rate": 0.00012061707725424539,
      "loss": 0.026,
      "step": 13005
    },
    {
      "epoch": 1.822845129642607,
      "grad_norm": 0.2274581342935562,
      "learning_rate": 0.00012060272662042572,
      "loss": 0.0451,
      "step": 13006
    },
    {
      "epoch": 1.8229852838121934,
      "grad_norm": 0.24408139288425446,
      "learning_rate": 0.00012058837598660607,
      "loss": 0.0454,
      "step": 13007
    },
    {
      "epoch": 1.8231254379817798,
      "grad_norm": 0.12239329516887665,
      "learning_rate": 0.0001205740253527864,
      "loss": 0.0152,
      "step": 13008
    },
    {
      "epoch": 1.8232655921513665,
      "grad_norm": 0.18465597927570343,
      "learning_rate": 0.00012055967471896674,
      "loss": 0.0391,
      "step": 13009
    },
    {
      "epoch": 1.8234057463209532,
      "grad_norm": 0.27876588702201843,
      "learning_rate": 0.00012054532408514708,
      "loss": 0.0457,
      "step": 13010
    },
    {
      "epoch": 1.8235459004905397,
      "grad_norm": 0.24993014335632324,
      "learning_rate": 0.00012053097345132741,
      "loss": 0.054,
      "step": 13011
    },
    {
      "epoch": 1.8236860546601261,
      "grad_norm": 0.4324030578136444,
      "learning_rate": 0.00012051662281750777,
      "loss": 0.0357,
      "step": 13012
    },
    {
      "epoch": 1.8238262088297126,
      "grad_norm": 0.36240458488464355,
      "learning_rate": 0.00012050227218368811,
      "loss": 0.0658,
      "step": 13013
    },
    {
      "epoch": 1.8239663629992993,
      "grad_norm": 0.1839447319507599,
      "learning_rate": 0.00012048792154986844,
      "loss": 0.0501,
      "step": 13014
    },
    {
      "epoch": 1.8241065171688857,
      "grad_norm": 0.14985217154026031,
      "learning_rate": 0.00012047357091604878,
      "loss": 0.0355,
      "step": 13015
    },
    {
      "epoch": 1.8242466713384724,
      "grad_norm": 0.10741608589887619,
      "learning_rate": 0.00012045922028222911,
      "loss": 0.0101,
      "step": 13016
    },
    {
      "epoch": 1.8243868255080589,
      "grad_norm": 0.36963754892349243,
      "learning_rate": 0.00012044486964840947,
      "loss": 0.0444,
      "step": 13017
    },
    {
      "epoch": 1.8245269796776453,
      "grad_norm": 0.14281608164310455,
      "learning_rate": 0.00012043051901458981,
      "loss": 0.0195,
      "step": 13018
    },
    {
      "epoch": 1.824667133847232,
      "grad_norm": 0.49174901843070984,
      "learning_rate": 0.00012041616838077014,
      "loss": 0.0208,
      "step": 13019
    },
    {
      "epoch": 1.8248072880168185,
      "grad_norm": 0.3527078926563263,
      "learning_rate": 0.00012040181774695048,
      "loss": 0.0724,
      "step": 13020
    },
    {
      "epoch": 1.8249474421864051,
      "grad_norm": 0.11536964774131775,
      "learning_rate": 0.00012038746711313082,
      "loss": 0.0134,
      "step": 13021
    },
    {
      "epoch": 1.8250875963559916,
      "grad_norm": 0.19288687407970428,
      "learning_rate": 0.00012037311647931115,
      "loss": 0.0367,
      "step": 13022
    },
    {
      "epoch": 1.825227750525578,
      "grad_norm": 0.5221089720726013,
      "learning_rate": 0.0001203587658454915,
      "loss": 0.0236,
      "step": 13023
    },
    {
      "epoch": 1.8253679046951645,
      "grad_norm": 0.18207323551177979,
      "learning_rate": 0.00012034441521167185,
      "loss": 0.014,
      "step": 13024
    },
    {
      "epoch": 1.8255080588647512,
      "grad_norm": 0.2753743529319763,
      "learning_rate": 0.00012033006457785218,
      "loss": 0.0546,
      "step": 13025
    },
    {
      "epoch": 1.8256482130343379,
      "grad_norm": 0.528946578502655,
      "learning_rate": 0.00012031571394403252,
      "loss": 0.0622,
      "step": 13026
    },
    {
      "epoch": 1.8257883672039243,
      "grad_norm": 0.6245681643486023,
      "learning_rate": 0.00012030136331021285,
      "loss": 0.0868,
      "step": 13027
    },
    {
      "epoch": 1.8259285213735108,
      "grad_norm": 0.24446775019168854,
      "learning_rate": 0.0001202870126763932,
      "loss": 0.021,
      "step": 13028
    },
    {
      "epoch": 1.8260686755430973,
      "grad_norm": 0.13140496611595154,
      "learning_rate": 0.00012027266204257354,
      "loss": 0.0085,
      "step": 13029
    },
    {
      "epoch": 1.826208829712684,
      "grad_norm": 0.3186090588569641,
      "learning_rate": 0.00012025831140875387,
      "loss": 0.0363,
      "step": 13030
    },
    {
      "epoch": 1.8263489838822706,
      "grad_norm": 0.5285914540290833,
      "learning_rate": 0.00012024396077493421,
      "loss": 0.0428,
      "step": 13031
    },
    {
      "epoch": 1.826489138051857,
      "grad_norm": 0.7337438464164734,
      "learning_rate": 0.00012022961014111456,
      "loss": 0.1334,
      "step": 13032
    },
    {
      "epoch": 1.8266292922214435,
      "grad_norm": 1.9644092321395874,
      "learning_rate": 0.0001202152595072949,
      "loss": 0.3768,
      "step": 13033
    },
    {
      "epoch": 1.82676944639103,
      "grad_norm": 0.9033927917480469,
      "learning_rate": 0.00012020090887347524,
      "loss": 0.0594,
      "step": 13034
    },
    {
      "epoch": 1.8269096005606167,
      "grad_norm": 0.8508265018463135,
      "learning_rate": 0.00012018655823965557,
      "loss": 0.0897,
      "step": 13035
    },
    {
      "epoch": 1.8270497547302034,
      "grad_norm": 0.3438679277896881,
      "learning_rate": 0.00012017220760583591,
      "loss": 0.0452,
      "step": 13036
    },
    {
      "epoch": 1.8271899088997898,
      "grad_norm": 0.1351635456085205,
      "learning_rate": 0.00012015785697201625,
      "loss": 0.0303,
      "step": 13037
    },
    {
      "epoch": 1.8273300630693763,
      "grad_norm": 0.16365383565425873,
      "learning_rate": 0.00012014350633819658,
      "loss": 0.0331,
      "step": 13038
    },
    {
      "epoch": 1.8274702172389627,
      "grad_norm": 0.1686399281024933,
      "learning_rate": 0.00012012915570437694,
      "loss": 0.0231,
      "step": 13039
    },
    {
      "epoch": 1.8276103714085494,
      "grad_norm": 0.3161714971065521,
      "learning_rate": 0.00012011480507055728,
      "loss": 0.0602,
      "step": 13040
    },
    {
      "epoch": 1.827750525578136,
      "grad_norm": 0.23480254411697388,
      "learning_rate": 0.00012010045443673761,
      "loss": 0.0287,
      "step": 13041
    },
    {
      "epoch": 1.8278906797477226,
      "grad_norm": 0.3092454969882965,
      "learning_rate": 0.00012008610380291795,
      "loss": 0.0774,
      "step": 13042
    },
    {
      "epoch": 1.828030833917309,
      "grad_norm": 0.37827104330062866,
      "learning_rate": 0.00012007175316909828,
      "loss": 0.0558,
      "step": 13043
    },
    {
      "epoch": 1.8281709880868955,
      "grad_norm": 0.20225325226783752,
      "learning_rate": 0.00012005740253527863,
      "loss": 0.0205,
      "step": 13044
    },
    {
      "epoch": 1.8283111422564822,
      "grad_norm": 0.20479802787303925,
      "learning_rate": 0.00012004305190145898,
      "loss": 0.0885,
      "step": 13045
    },
    {
      "epoch": 1.8284512964260686,
      "grad_norm": 0.47692474722862244,
      "learning_rate": 0.0001200287012676393,
      "loss": 0.0658,
      "step": 13046
    },
    {
      "epoch": 1.8285914505956553,
      "grad_norm": 0.17206741869449615,
      "learning_rate": 0.00012001435063381965,
      "loss": 0.0229,
      "step": 13047
    },
    {
      "epoch": 1.8287316047652418,
      "grad_norm": 0.2441057413816452,
      "learning_rate": 0.00011999999999999999,
      "loss": 0.0624,
      "step": 13048
    },
    {
      "epoch": 1.8288717589348282,
      "grad_norm": 0.2826390564441681,
      "learning_rate": 0.00011998564936618033,
      "loss": 0.0657,
      "step": 13049
    },
    {
      "epoch": 1.829011913104415,
      "grad_norm": 0.4242059588432312,
      "learning_rate": 0.00011997129873236067,
      "loss": 0.076,
      "step": 13050
    },
    {
      "epoch": 1.8291520672740014,
      "grad_norm": 0.08126556873321533,
      "learning_rate": 0.000119956948098541,
      "loss": 0.0125,
      "step": 13051
    },
    {
      "epoch": 1.829292221443588,
      "grad_norm": 0.38126716017723083,
      "learning_rate": 0.00011994259746472134,
      "loss": 0.0705,
      "step": 13052
    },
    {
      "epoch": 1.8294323756131745,
      "grad_norm": 0.2914828658103943,
      "learning_rate": 0.00011992824683090169,
      "loss": 0.0447,
      "step": 13053
    },
    {
      "epoch": 1.829572529782761,
      "grad_norm": 0.2624664902687073,
      "learning_rate": 0.00011991389619708202,
      "loss": 0.0215,
      "step": 13054
    },
    {
      "epoch": 1.8297126839523474,
      "grad_norm": 0.1292945146560669,
      "learning_rate": 0.00011989954556326237,
      "loss": 0.0202,
      "step": 13055
    },
    {
      "epoch": 1.829852838121934,
      "grad_norm": 0.31137487292289734,
      "learning_rate": 0.00011988519492944271,
      "loss": 0.0487,
      "step": 13056
    },
    {
      "epoch": 1.8299929922915208,
      "grad_norm": 0.3116886019706726,
      "learning_rate": 0.00011987084429562304,
      "loss": 0.0722,
      "step": 13057
    },
    {
      "epoch": 1.8301331464611073,
      "grad_norm": 0.3327091336250305,
      "learning_rate": 0.00011985649366180338,
      "loss": 0.0785,
      "step": 13058
    },
    {
      "epoch": 1.8302733006306937,
      "grad_norm": 0.1800590604543686,
      "learning_rate": 0.00011984214302798374,
      "loss": 0.0343,
      "step": 13059
    },
    {
      "epoch": 1.8304134548002802,
      "grad_norm": 0.2878572642803192,
      "learning_rate": 0.00011982779239416407,
      "loss": 0.0333,
      "step": 13060
    },
    {
      "epoch": 1.8305536089698669,
      "grad_norm": 0.1222718209028244,
      "learning_rate": 0.00011981344176034441,
      "loss": 0.0126,
      "step": 13061
    },
    {
      "epoch": 1.8306937631394535,
      "grad_norm": 0.18908029794692993,
      "learning_rate": 0.00011979909112652474,
      "loss": 0.0199,
      "step": 13062
    },
    {
      "epoch": 1.83083391730904,
      "grad_norm": 0.41595008969306946,
      "learning_rate": 0.00011978474049270508,
      "loss": 0.0354,
      "step": 13063
    },
    {
      "epoch": 1.8309740714786265,
      "grad_norm": 0.09385739266872406,
      "learning_rate": 0.00011977038985888542,
      "loss": 0.011,
      "step": 13064
    },
    {
      "epoch": 1.831114225648213,
      "grad_norm": 0.1520908623933792,
      "learning_rate": 0.00011975603922506576,
      "loss": 0.0247,
      "step": 13065
    },
    {
      "epoch": 1.8312543798177996,
      "grad_norm": 0.34486567974090576,
      "learning_rate": 0.00011974168859124611,
      "loss": 0.0363,
      "step": 13066
    },
    {
      "epoch": 1.8313945339873863,
      "grad_norm": 0.08466723561286926,
      "learning_rate": 0.00011972733795742645,
      "loss": 0.0061,
      "step": 13067
    },
    {
      "epoch": 1.8315346881569727,
      "grad_norm": 0.2776062488555908,
      "learning_rate": 0.00011971298732360678,
      "loss": 0.0531,
      "step": 13068
    },
    {
      "epoch": 1.8316748423265592,
      "grad_norm": 0.3212437927722931,
      "learning_rate": 0.00011969863668978712,
      "loss": 0.0496,
      "step": 13069
    },
    {
      "epoch": 1.8318149964961457,
      "grad_norm": 0.835310161113739,
      "learning_rate": 0.00011968428605596745,
      "loss": 0.0871,
      "step": 13070
    },
    {
      "epoch": 1.8319551506657323,
      "grad_norm": 0.6809616088867188,
      "learning_rate": 0.0001196699354221478,
      "loss": 0.0445,
      "step": 13071
    },
    {
      "epoch": 1.832095304835319,
      "grad_norm": 0.3452107012271881,
      "learning_rate": 0.00011965558478832815,
      "loss": 0.0174,
      "step": 13072
    },
    {
      "epoch": 1.8322354590049055,
      "grad_norm": 0.32924070954322815,
      "learning_rate": 0.00011964123415450847,
      "loss": 0.0418,
      "step": 13073
    },
    {
      "epoch": 1.832375613174492,
      "grad_norm": 0.4984101951122284,
      "learning_rate": 0.00011962688352068882,
      "loss": 0.1248,
      "step": 13074
    },
    {
      "epoch": 1.8325157673440784,
      "grad_norm": 0.23525528609752655,
      "learning_rate": 0.00011961253288686917,
      "loss": 0.0481,
      "step": 13075
    },
    {
      "epoch": 1.832655921513665,
      "grad_norm": 0.05106350779533386,
      "learning_rate": 0.0001195981822530495,
      "loss": 0.0056,
      "step": 13076
    },
    {
      "epoch": 1.8327960756832515,
      "grad_norm": 0.10208006948232651,
      "learning_rate": 0.00011958383161922984,
      "loss": 0.0093,
      "step": 13077
    },
    {
      "epoch": 1.8329362298528382,
      "grad_norm": 0.23808462917804718,
      "learning_rate": 0.00011956948098541017,
      "loss": 0.0092,
      "step": 13078
    },
    {
      "epoch": 1.8330763840224247,
      "grad_norm": 0.2991667091846466,
      "learning_rate": 0.00011955513035159051,
      "loss": 0.0297,
      "step": 13079
    },
    {
      "epoch": 1.8332165381920111,
      "grad_norm": 0.323784738779068,
      "learning_rate": 0.00011954077971777087,
      "loss": 0.0275,
      "step": 13080
    },
    {
      "epoch": 1.8333566923615976,
      "grad_norm": 0.25618141889572144,
      "learning_rate": 0.0001195264290839512,
      "loss": 0.0147,
      "step": 13081
    },
    {
      "epoch": 1.8334968465311843,
      "grad_norm": 0.8678697943687439,
      "learning_rate": 0.00011951207845013154,
      "loss": 0.0665,
      "step": 13082
    },
    {
      "epoch": 1.833637000700771,
      "grad_norm": 2.700031042098999,
      "learning_rate": 0.00011949772781631188,
      "loss": 0.313,
      "step": 13083
    },
    {
      "epoch": 1.8337771548703574,
      "grad_norm": 3.0952396392822266,
      "learning_rate": 0.00011948337718249221,
      "loss": 0.2181,
      "step": 13084
    },
    {
      "epoch": 1.8339173090399439,
      "grad_norm": 0.32221120595932007,
      "learning_rate": 0.00011946902654867255,
      "loss": 0.0144,
      "step": 13085
    },
    {
      "epoch": 1.8340574632095303,
      "grad_norm": 0.14368171989917755,
      "learning_rate": 0.00011945467591485288,
      "loss": 0.0122,
      "step": 13086
    },
    {
      "epoch": 1.834197617379117,
      "grad_norm": 0.4254034161567688,
      "learning_rate": 0.00011944032528103324,
      "loss": 0.0475,
      "step": 13087
    },
    {
      "epoch": 1.8343377715487037,
      "grad_norm": 0.37998393177986145,
      "learning_rate": 0.00011942597464721358,
      "loss": 0.0426,
      "step": 13088
    },
    {
      "epoch": 1.8344779257182902,
      "grad_norm": 0.3881828486919403,
      "learning_rate": 0.00011941162401339391,
      "loss": 0.0797,
      "step": 13089
    },
    {
      "epoch": 1.8346180798878766,
      "grad_norm": 0.12122989445924759,
      "learning_rate": 0.00011939727337957425,
      "loss": 0.0085,
      "step": 13090
    },
    {
      "epoch": 1.834758234057463,
      "grad_norm": 0.2777736485004425,
      "learning_rate": 0.0001193829227457546,
      "loss": 0.0341,
      "step": 13091
    },
    {
      "epoch": 1.8348983882270498,
      "grad_norm": 0.31072843074798584,
      "learning_rate": 0.00011936857211193493,
      "loss": 0.026,
      "step": 13092
    },
    {
      "epoch": 1.8350385423966364,
      "grad_norm": 0.24548600614070892,
      "learning_rate": 0.00011935422147811528,
      "loss": 0.0229,
      "step": 13093
    },
    {
      "epoch": 1.835178696566223,
      "grad_norm": 0.15506403148174286,
      "learning_rate": 0.00011933987084429562,
      "loss": 0.0331,
      "step": 13094
    },
    {
      "epoch": 1.8353188507358094,
      "grad_norm": 0.4133717715740204,
      "learning_rate": 0.00011932552021047595,
      "loss": 0.0764,
      "step": 13095
    },
    {
      "epoch": 1.8354590049053958,
      "grad_norm": 0.10784870386123657,
      "learning_rate": 0.0001193111695766563,
      "loss": 0.0293,
      "step": 13096
    },
    {
      "epoch": 1.8355991590749825,
      "grad_norm": 0.32498976588249207,
      "learning_rate": 0.00011929681894283663,
      "loss": 0.0433,
      "step": 13097
    },
    {
      "epoch": 1.8357393132445692,
      "grad_norm": 0.4950914978981018,
      "learning_rate": 0.00011928246830901697,
      "loss": 0.055,
      "step": 13098
    },
    {
      "epoch": 1.8358794674141556,
      "grad_norm": 0.34596705436706543,
      "learning_rate": 0.00011926811767519732,
      "loss": 0.0179,
      "step": 13099
    },
    {
      "epoch": 1.836019621583742,
      "grad_norm": 0.22688059508800507,
      "learning_rate": 0.00011925376704137764,
      "loss": 0.0346,
      "step": 13100
    },
    {
      "epoch": 1.8361597757533286,
      "grad_norm": 0.295215368270874,
      "learning_rate": 0.00011923941640755799,
      "loss": 0.0324,
      "step": 13101
    },
    {
      "epoch": 1.8362999299229152,
      "grad_norm": 0.17926205694675446,
      "learning_rate": 0.00011922506577373834,
      "loss": 0.0205,
      "step": 13102
    },
    {
      "epoch": 1.8364400840925017,
      "grad_norm": 0.3611327111721039,
      "learning_rate": 0.00011921071513991867,
      "loss": 0.0884,
      "step": 13103
    },
    {
      "epoch": 1.8365802382620884,
      "grad_norm": 0.47480544447898865,
      "learning_rate": 0.00011919636450609901,
      "loss": 0.2169,
      "step": 13104
    },
    {
      "epoch": 1.8367203924316748,
      "grad_norm": 0.3176404535770416,
      "learning_rate": 0.00011918201387227934,
      "loss": 0.03,
      "step": 13105
    },
    {
      "epoch": 1.8368605466012613,
      "grad_norm": 0.23722144961357117,
      "learning_rate": 0.00011916766323845968,
      "loss": 0.0496,
      "step": 13106
    },
    {
      "epoch": 1.837000700770848,
      "grad_norm": 0.395290732383728,
      "learning_rate": 0.00011915331260464004,
      "loss": 0.1526,
      "step": 13107
    },
    {
      "epoch": 1.8371408549404344,
      "grad_norm": 0.38038206100463867,
      "learning_rate": 0.00011913896197082037,
      "loss": 0.0216,
      "step": 13108
    },
    {
      "epoch": 1.8372810091100211,
      "grad_norm": 0.3218480944633484,
      "learning_rate": 0.00011912461133700071,
      "loss": 0.0707,
      "step": 13109
    },
    {
      "epoch": 1.8374211632796076,
      "grad_norm": 0.12077045440673828,
      "learning_rate": 0.00011911026070318105,
      "loss": 0.0135,
      "step": 13110
    },
    {
      "epoch": 1.837561317449194,
      "grad_norm": 0.24675005674362183,
      "learning_rate": 0.00011909591006936138,
      "loss": 0.0237,
      "step": 13111
    },
    {
      "epoch": 1.8377014716187805,
      "grad_norm": 0.4440574645996094,
      "learning_rate": 0.00011908155943554174,
      "loss": 0.0926,
      "step": 13112
    },
    {
      "epoch": 1.8378416257883672,
      "grad_norm": 0.2965434193611145,
      "learning_rate": 0.00011906720880172206,
      "loss": 0.0222,
      "step": 13113
    },
    {
      "epoch": 1.8379817799579539,
      "grad_norm": 0.34264495968818665,
      "learning_rate": 0.0001190528581679024,
      "loss": 0.0372,
      "step": 13114
    },
    {
      "epoch": 1.8381219341275403,
      "grad_norm": 0.45088109374046326,
      "learning_rate": 0.00011903850753408275,
      "loss": 0.0431,
      "step": 13115
    },
    {
      "epoch": 1.8382620882971268,
      "grad_norm": 0.1931804120540619,
      "learning_rate": 0.00011902415690026308,
      "loss": 0.0196,
      "step": 13116
    },
    {
      "epoch": 1.8384022424667132,
      "grad_norm": 0.2493153214454651,
      "learning_rate": 0.00011900980626644342,
      "loss": 0.0377,
      "step": 13117
    },
    {
      "epoch": 1.8385423966363,
      "grad_norm": 0.22799772024154663,
      "learning_rate": 0.00011899545563262377,
      "loss": 0.0368,
      "step": 13118
    },
    {
      "epoch": 1.8386825508058866,
      "grad_norm": 0.20331008732318878,
      "learning_rate": 0.0001189811049988041,
      "loss": 0.0401,
      "step": 13119
    },
    {
      "epoch": 1.838822704975473,
      "grad_norm": 0.32094836235046387,
      "learning_rate": 0.00011896675436498445,
      "loss": 0.0812,
      "step": 13120
    },
    {
      "epoch": 1.8389628591450595,
      "grad_norm": 0.4043622612953186,
      "learning_rate": 0.00011895240373116477,
      "loss": 0.0286,
      "step": 13121
    },
    {
      "epoch": 1.839103013314646,
      "grad_norm": 0.18666526675224304,
      "learning_rate": 0.00011893805309734512,
      "loss": 0.0235,
      "step": 13122
    },
    {
      "epoch": 1.8392431674842327,
      "grad_norm": 0.2039700299501419,
      "learning_rate": 0.00011892370246352547,
      "loss": 0.0478,
      "step": 13123
    },
    {
      "epoch": 1.8393833216538193,
      "grad_norm": 0.44583046436309814,
      "learning_rate": 0.0001189093518297058,
      "loss": 0.0219,
      "step": 13124
    },
    {
      "epoch": 1.8395234758234058,
      "grad_norm": 0.2930704951286316,
      "learning_rate": 0.00011889500119588614,
      "loss": 0.0748,
      "step": 13125
    },
    {
      "epoch": 1.8396636299929923,
      "grad_norm": 0.4853978455066681,
      "learning_rate": 0.00011888065056206648,
      "loss": 0.075,
      "step": 13126
    },
    {
      "epoch": 1.8398037841625787,
      "grad_norm": 0.26997554302215576,
      "learning_rate": 0.00011886629992824681,
      "loss": 0.0337,
      "step": 13127
    },
    {
      "epoch": 1.8399439383321654,
      "grad_norm": 0.20511342585086823,
      "learning_rate": 0.00011885194929442717,
      "loss": 0.02,
      "step": 13128
    },
    {
      "epoch": 1.840084092501752,
      "grad_norm": 0.39991262555122375,
      "learning_rate": 0.0001188375986606075,
      "loss": 0.073,
      "step": 13129
    },
    {
      "epoch": 1.8402242466713385,
      "grad_norm": 0.4542100131511688,
      "learning_rate": 0.00011882324802678784,
      "loss": 0.0787,
      "step": 13130
    },
    {
      "epoch": 1.840364400840925,
      "grad_norm": 0.8485069870948792,
      "learning_rate": 0.00011880889739296818,
      "loss": 0.0352,
      "step": 13131
    },
    {
      "epoch": 1.8405045550105115,
      "grad_norm": 2.933495283126831,
      "learning_rate": 0.00011879454675914851,
      "loss": 0.1002,
      "step": 13132
    },
    {
      "epoch": 1.8406447091800981,
      "grad_norm": 0.8393083214759827,
      "learning_rate": 0.00011878019612532885,
      "loss": 0.2263,
      "step": 13133
    },
    {
      "epoch": 1.8407848633496846,
      "grad_norm": 2.1371407508850098,
      "learning_rate": 0.00011876584549150921,
      "loss": 0.2124,
      "step": 13134
    },
    {
      "epoch": 1.8409250175192713,
      "grad_norm": 1.579941987991333,
      "learning_rate": 0.00011875149485768954,
      "loss": 0.2545,
      "step": 13135
    },
    {
      "epoch": 1.8410651716888577,
      "grad_norm": 0.1906174123287201,
      "learning_rate": 0.00011873714422386988,
      "loss": 0.0195,
      "step": 13136
    },
    {
      "epoch": 1.8412053258584442,
      "grad_norm": 0.24619857966899872,
      "learning_rate": 0.00011872279359005022,
      "loss": 0.08,
      "step": 13137
    },
    {
      "epoch": 1.8413454800280307,
      "grad_norm": 0.24982307851314545,
      "learning_rate": 0.00011870844295623055,
      "loss": 0.0417,
      "step": 13138
    },
    {
      "epoch": 1.8414856341976173,
      "grad_norm": 0.27078914642333984,
      "learning_rate": 0.0001186940923224109,
      "loss": 0.0725,
      "step": 13139
    },
    {
      "epoch": 1.841625788367204,
      "grad_norm": 0.08704311400651932,
      "learning_rate": 0.00011867974168859123,
      "loss": 0.0108,
      "step": 13140
    },
    {
      "epoch": 1.8417659425367905,
      "grad_norm": 0.3021286427974701,
      "learning_rate": 0.00011866539105477158,
      "loss": 0.0433,
      "step": 13141
    },
    {
      "epoch": 1.841906096706377,
      "grad_norm": 0.12301019579172134,
      "learning_rate": 0.00011865104042095192,
      "loss": 0.0335,
      "step": 13142
    },
    {
      "epoch": 1.8420462508759634,
      "grad_norm": 0.13123148679733276,
      "learning_rate": 0.00011863668978713225,
      "loss": 0.017,
      "step": 13143
    },
    {
      "epoch": 1.84218640504555,
      "grad_norm": 0.633851945400238,
      "learning_rate": 0.0001186223391533126,
      "loss": 0.0919,
      "step": 13144
    },
    {
      "epoch": 1.8423265592151368,
      "grad_norm": 0.09904909133911133,
      "learning_rate": 0.00011860798851949294,
      "loss": 0.0171,
      "step": 13145
    },
    {
      "epoch": 1.8424667133847232,
      "grad_norm": 0.3920518755912781,
      "learning_rate": 0.00011859363788567327,
      "loss": 0.0883,
      "step": 13146
    },
    {
      "epoch": 1.8426068675543097,
      "grad_norm": 0.3847196102142334,
      "learning_rate": 0.00011857928725185361,
      "loss": 0.0804,
      "step": 13147
    },
    {
      "epoch": 1.8427470217238962,
      "grad_norm": 0.2734045088291168,
      "learning_rate": 0.00011856493661803394,
      "loss": 0.0622,
      "step": 13148
    },
    {
      "epoch": 1.8428871758934828,
      "grad_norm": 0.15843774378299713,
      "learning_rate": 0.00011855058598421429,
      "loss": 0.0267,
      "step": 13149
    },
    {
      "epoch": 1.8430273300630695,
      "grad_norm": 0.18920204043388367,
      "learning_rate": 0.00011853623535039464,
      "loss": 0.0146,
      "step": 13150
    },
    {
      "epoch": 1.843167484232656,
      "grad_norm": 0.32722994685173035,
      "learning_rate": 0.00011852188471657497,
      "loss": 0.0932,
      "step": 13151
    },
    {
      "epoch": 1.8433076384022424,
      "grad_norm": 0.3440050482749939,
      "learning_rate": 0.00011850753408275531,
      "loss": 0.0602,
      "step": 13152
    },
    {
      "epoch": 1.843447792571829,
      "grad_norm": 0.25374898314476013,
      "learning_rate": 0.00011849318344893565,
      "loss": 0.0258,
      "step": 13153
    },
    {
      "epoch": 1.8435879467414156,
      "grad_norm": 0.1849697381258011,
      "learning_rate": 0.00011847883281511598,
      "loss": 0.0389,
      "step": 13154
    },
    {
      "epoch": 1.8437281009110023,
      "grad_norm": 0.1742199957370758,
      "learning_rate": 0.00011846448218129634,
      "loss": 0.0489,
      "step": 13155
    },
    {
      "epoch": 1.8438682550805887,
      "grad_norm": 0.28863275051116943,
      "learning_rate": 0.00011845013154747667,
      "loss": 0.0616,
      "step": 13156
    },
    {
      "epoch": 1.8440084092501752,
      "grad_norm": 0.22597481310367584,
      "learning_rate": 0.00011843578091365701,
      "loss": 0.0513,
      "step": 13157
    },
    {
      "epoch": 1.8441485634197616,
      "grad_norm": 0.3605732023715973,
      "learning_rate": 0.00011842143027983735,
      "loss": 0.028,
      "step": 13158
    },
    {
      "epoch": 1.8442887175893483,
      "grad_norm": 0.14195258915424347,
      "learning_rate": 0.00011840707964601768,
      "loss": 0.0091,
      "step": 13159
    },
    {
      "epoch": 1.8444288717589348,
      "grad_norm": 0.43894365429878235,
      "learning_rate": 0.00011839272901219804,
      "loss": 0.0561,
      "step": 13160
    },
    {
      "epoch": 1.8445690259285215,
      "grad_norm": 0.1146656721830368,
      "learning_rate": 0.00011837837837837838,
      "loss": 0.0167,
      "step": 13161
    },
    {
      "epoch": 1.844709180098108,
      "grad_norm": 0.2076784074306488,
      "learning_rate": 0.0001183640277445587,
      "loss": 0.0139,
      "step": 13162
    },
    {
      "epoch": 1.8448493342676944,
      "grad_norm": 0.9799286723136902,
      "learning_rate": 0.00011834967711073905,
      "loss": 0.0545,
      "step": 13163
    },
    {
      "epoch": 1.844989488437281,
      "grad_norm": 0.31785622239112854,
      "learning_rate": 0.00011833532647691938,
      "loss": 0.0349,
      "step": 13164
    },
    {
      "epoch": 1.8451296426068675,
      "grad_norm": 0.5915439128875732,
      "learning_rate": 0.00011832097584309972,
      "loss": 0.05,
      "step": 13165
    },
    {
      "epoch": 1.8452697967764542,
      "grad_norm": 0.4077807664871216,
      "learning_rate": 0.00011830662520928007,
      "loss": 0.0764,
      "step": 13166
    },
    {
      "epoch": 1.8454099509460407,
      "grad_norm": 0.388322651386261,
      "learning_rate": 0.0001182922745754604,
      "loss": 0.0525,
      "step": 13167
    },
    {
      "epoch": 1.8455501051156271,
      "grad_norm": 0.21979929506778717,
      "learning_rate": 0.00011827792394164074,
      "loss": 0.0775,
      "step": 13168
    },
    {
      "epoch": 1.8456902592852136,
      "grad_norm": 0.6163580417633057,
      "learning_rate": 0.00011826357330782109,
      "loss": 0.0745,
      "step": 13169
    },
    {
      "epoch": 1.8458304134548003,
      "grad_norm": 0.20840896666049957,
      "learning_rate": 0.00011824922267400142,
      "loss": 0.0627,
      "step": 13170
    },
    {
      "epoch": 1.845970567624387,
      "grad_norm": 0.3919309079647064,
      "learning_rate": 0.00011823487204018177,
      "loss": 0.0567,
      "step": 13171
    },
    {
      "epoch": 1.8461107217939734,
      "grad_norm": 0.28132322430610657,
      "learning_rate": 0.00011822052140636211,
      "loss": 0.0637,
      "step": 13172
    },
    {
      "epoch": 1.8462508759635599,
      "grad_norm": 0.17737574875354767,
      "learning_rate": 0.00011820617077254244,
      "loss": 0.0229,
      "step": 13173
    },
    {
      "epoch": 1.8463910301331463,
      "grad_norm": 0.41007745265960693,
      "learning_rate": 0.00011819182013872278,
      "loss": 0.0304,
      "step": 13174
    },
    {
      "epoch": 1.846531184302733,
      "grad_norm": 0.08945126086473465,
      "learning_rate": 0.00011817746950490311,
      "loss": 0.0093,
      "step": 13175
    },
    {
      "epoch": 1.8466713384723197,
      "grad_norm": 0.1429409682750702,
      "learning_rate": 0.00011816311887108347,
      "loss": 0.0187,
      "step": 13176
    },
    {
      "epoch": 1.8468114926419061,
      "grad_norm": 0.06563121825456619,
      "learning_rate": 0.00011814876823726381,
      "loss": 0.0043,
      "step": 13177
    },
    {
      "epoch": 1.8469516468114926,
      "grad_norm": 0.10587241500616074,
      "learning_rate": 0.00011813441760344414,
      "loss": 0.0162,
      "step": 13178
    },
    {
      "epoch": 1.847091800981079,
      "grad_norm": 0.6637721657752991,
      "learning_rate": 0.00011812006696962448,
      "loss": 0.1071,
      "step": 13179
    },
    {
      "epoch": 1.8472319551506657,
      "grad_norm": 0.9505621194839478,
      "learning_rate": 0.00011810571633580482,
      "loss": 0.0999,
      "step": 13180
    },
    {
      "epoch": 1.8473721093202524,
      "grad_norm": 0.7461830377578735,
      "learning_rate": 0.00011809136570198515,
      "loss": 0.1124,
      "step": 13181
    },
    {
      "epoch": 1.8475122634898389,
      "grad_norm": 0.9445149302482605,
      "learning_rate": 0.00011807701506816551,
      "loss": 0.0807,
      "step": 13182
    },
    {
      "epoch": 1.8476524176594253,
      "grad_norm": 0.1189340353012085,
      "learning_rate": 0.00011806266443434584,
      "loss": 0.0063,
      "step": 13183
    },
    {
      "epoch": 1.8477925718290118,
      "grad_norm": 0.2214917689561844,
      "learning_rate": 0.00011804831380052618,
      "loss": 0.0426,
      "step": 13184
    },
    {
      "epoch": 1.8479327259985985,
      "grad_norm": 4.704756259918213,
      "learning_rate": 0.00011803396316670652,
      "loss": 0.1249,
      "step": 13185
    },
    {
      "epoch": 1.8480728801681852,
      "grad_norm": 0.34630975127220154,
      "learning_rate": 0.00011801961253288685,
      "loss": 0.0417,
      "step": 13186
    },
    {
      "epoch": 1.8482130343377716,
      "grad_norm": 0.6147617697715759,
      "learning_rate": 0.0001180052618990672,
      "loss": 0.0617,
      "step": 13187
    },
    {
      "epoch": 1.848353188507358,
      "grad_norm": 0.16629427671432495,
      "learning_rate": 0.00011799091126524755,
      "loss": 0.0322,
      "step": 13188
    },
    {
      "epoch": 1.8484933426769445,
      "grad_norm": 0.1443949043750763,
      "learning_rate": 0.00011797656063142788,
      "loss": 0.0194,
      "step": 13189
    },
    {
      "epoch": 1.8486334968465312,
      "grad_norm": 0.3427322804927826,
      "learning_rate": 0.00011796220999760822,
      "loss": 0.0437,
      "step": 13190
    },
    {
      "epoch": 1.8487736510161177,
      "grad_norm": 0.2636445164680481,
      "learning_rate": 0.00011794785936378855,
      "loss": 0.0466,
      "step": 13191
    },
    {
      "epoch": 1.8489138051857044,
      "grad_norm": 0.4148591458797455,
      "learning_rate": 0.0001179335087299689,
      "loss": 0.0606,
      "step": 13192
    },
    {
      "epoch": 1.8490539593552908,
      "grad_norm": 0.29563140869140625,
      "learning_rate": 0.00011791915809614924,
      "loss": 0.0468,
      "step": 13193
    },
    {
      "epoch": 1.8491941135248773,
      "grad_norm": 0.18425562977790833,
      "learning_rate": 0.00011790480746232957,
      "loss": 0.0187,
      "step": 13194
    },
    {
      "epoch": 1.849334267694464,
      "grad_norm": 0.08078820258378983,
      "learning_rate": 0.00011789045682850991,
      "loss": 0.0094,
      "step": 13195
    },
    {
      "epoch": 1.8494744218640504,
      "grad_norm": 0.17210640013217926,
      "learning_rate": 0.00011787610619469026,
      "loss": 0.0098,
      "step": 13196
    },
    {
      "epoch": 1.849614576033637,
      "grad_norm": 0.5614300966262817,
      "learning_rate": 0.00011786175556087058,
      "loss": 0.0292,
      "step": 13197
    },
    {
      "epoch": 1.8497547302032236,
      "grad_norm": 0.2318727821111679,
      "learning_rate": 0.00011784740492705094,
      "loss": 0.0152,
      "step": 13198
    },
    {
      "epoch": 1.84989488437281,
      "grad_norm": 0.34381574392318726,
      "learning_rate": 0.00011783305429323127,
      "loss": 0.0361,
      "step": 13199
    },
    {
      "epoch": 1.8500350385423965,
      "grad_norm": 0.6980080008506775,
      "learning_rate": 0.00011781870365941161,
      "loss": 0.0579,
      "step": 13200
    },
    {
      "epoch": 1.8501751927119832,
      "grad_norm": 0.6064243912696838,
      "learning_rate": 0.00011780435302559195,
      "loss": 0.0627,
      "step": 13201
    },
    {
      "epoch": 1.8503153468815698,
      "grad_norm": 0.30239546298980713,
      "learning_rate": 0.00011779000239177228,
      "loss": 0.0353,
      "step": 13202
    },
    {
      "epoch": 1.8504555010511563,
      "grad_norm": 0.35314688086509705,
      "learning_rate": 0.00011777565175795264,
      "loss": 0.0942,
      "step": 13203
    },
    {
      "epoch": 1.8505956552207428,
      "grad_norm": 0.3479951024055481,
      "learning_rate": 0.00011776130112413298,
      "loss": 0.0804,
      "step": 13204
    },
    {
      "epoch": 1.8507358093903292,
      "grad_norm": 0.2140364795923233,
      "learning_rate": 0.00011774695049031331,
      "loss": 0.0383,
      "step": 13205
    },
    {
      "epoch": 1.850875963559916,
      "grad_norm": 0.284807413816452,
      "learning_rate": 0.00011773259985649365,
      "loss": 0.0318,
      "step": 13206
    },
    {
      "epoch": 1.8510161177295026,
      "grad_norm": 0.22400060296058655,
      "learning_rate": 0.000117718249222674,
      "loss": 0.0334,
      "step": 13207
    },
    {
      "epoch": 1.851156271899089,
      "grad_norm": 0.2021048367023468,
      "learning_rate": 0.00011770389858885433,
      "loss": 0.0387,
      "step": 13208
    },
    {
      "epoch": 1.8512964260686755,
      "grad_norm": 0.22780652344226837,
      "learning_rate": 0.00011768954795503468,
      "loss": 0.0232,
      "step": 13209
    },
    {
      "epoch": 1.851436580238262,
      "grad_norm": 0.34028196334838867,
      "learning_rate": 0.000117675197321215,
      "loss": 0.0531,
      "step": 13210
    },
    {
      "epoch": 1.8515767344078486,
      "grad_norm": 0.3480367362499237,
      "learning_rate": 0.00011766084668739535,
      "loss": 0.0864,
      "step": 13211
    },
    {
      "epoch": 1.8517168885774353,
      "grad_norm": 0.18290720880031586,
      "learning_rate": 0.00011764649605357569,
      "loss": 0.0512,
      "step": 13212
    },
    {
      "epoch": 1.8518570427470218,
      "grad_norm": 0.389321506023407,
      "learning_rate": 0.00011763214541975602,
      "loss": 0.0498,
      "step": 13213
    },
    {
      "epoch": 1.8519971969166082,
      "grad_norm": 0.2584618926048279,
      "learning_rate": 0.00011761779478593637,
      "loss": 0.049,
      "step": 13214
    },
    {
      "epoch": 1.8521373510861947,
      "grad_norm": 0.6915590763092041,
      "learning_rate": 0.00011760344415211672,
      "loss": 0.1029,
      "step": 13215
    },
    {
      "epoch": 1.8522775052557814,
      "grad_norm": 0.10391825437545776,
      "learning_rate": 0.00011758909351829704,
      "loss": 0.0139,
      "step": 13216
    },
    {
      "epoch": 1.852417659425368,
      "grad_norm": 0.17216764390468597,
      "learning_rate": 0.00011757474288447739,
      "loss": 0.012,
      "step": 13217
    },
    {
      "epoch": 1.8525578135949545,
      "grad_norm": 0.9722307920455933,
      "learning_rate": 0.00011756039225065771,
      "loss": 0.0632,
      "step": 13218
    },
    {
      "epoch": 1.852697967764541,
      "grad_norm": 0.8123739361763,
      "learning_rate": 0.00011754604161683807,
      "loss": 0.027,
      "step": 13219
    },
    {
      "epoch": 1.8528381219341274,
      "grad_norm": 0.21997202932834625,
      "learning_rate": 0.00011753169098301841,
      "loss": 0.0396,
      "step": 13220
    },
    {
      "epoch": 1.8529782761037141,
      "grad_norm": 0.5752045512199402,
      "learning_rate": 0.00011751734034919874,
      "loss": 0.0561,
      "step": 13221
    },
    {
      "epoch": 1.8531184302733006,
      "grad_norm": 0.2269778996706009,
      "learning_rate": 0.00011750298971537908,
      "loss": 0.076,
      "step": 13222
    },
    {
      "epoch": 1.8532585844428873,
      "grad_norm": 0.45925530791282654,
      "learning_rate": 0.00011748863908155944,
      "loss": 0.0347,
      "step": 13223
    },
    {
      "epoch": 1.8533987386124737,
      "grad_norm": 0.4658682942390442,
      "learning_rate": 0.00011747428844773977,
      "loss": 0.0206,
      "step": 13224
    },
    {
      "epoch": 1.8535388927820602,
      "grad_norm": 0.20311835408210754,
      "learning_rate": 0.00011745993781392011,
      "loss": 0.0418,
      "step": 13225
    },
    {
      "epoch": 1.8536790469516466,
      "grad_norm": 0.12614500522613525,
      "learning_rate": 0.00011744558718010044,
      "loss": 0.0072,
      "step": 13226
    },
    {
      "epoch": 1.8538192011212333,
      "grad_norm": 0.2045486569404602,
      "learning_rate": 0.00011743123654628078,
      "loss": 0.0137,
      "step": 13227
    },
    {
      "epoch": 1.85395935529082,
      "grad_norm": 0.6752736568450928,
      "learning_rate": 0.00011741688591246112,
      "loss": 0.0345,
      "step": 13228
    },
    {
      "epoch": 1.8540995094604065,
      "grad_norm": 0.6215781569480896,
      "learning_rate": 0.00011740253527864145,
      "loss": 0.0481,
      "step": 13229
    },
    {
      "epoch": 1.854239663629993,
      "grad_norm": 0.716833233833313,
      "learning_rate": 0.0001173881846448218,
      "loss": 0.2071,
      "step": 13230
    },
    {
      "epoch": 1.8543798177995794,
      "grad_norm": 1.8723562955856323,
      "learning_rate": 0.00011737383401100215,
      "loss": 0.1326,
      "step": 13231
    },
    {
      "epoch": 1.854519971969166,
      "grad_norm": 0.43276792764663696,
      "learning_rate": 0.00011735948337718248,
      "loss": 0.0206,
      "step": 13232
    },
    {
      "epoch": 1.8546601261387528,
      "grad_norm": 0.4123082756996155,
      "learning_rate": 0.00011734513274336282,
      "loss": 0.053,
      "step": 13233
    },
    {
      "epoch": 1.8548002803083392,
      "grad_norm": 0.42152661085128784,
      "learning_rate": 0.00011733078210954315,
      "loss": 0.023,
      "step": 13234
    },
    {
      "epoch": 1.8549404344779257,
      "grad_norm": 1.80049467086792,
      "learning_rate": 0.0001173164314757235,
      "loss": 0.1437,
      "step": 13235
    },
    {
      "epoch": 1.8550805886475121,
      "grad_norm": 0.161837637424469,
      "learning_rate": 0.00011730208084190385,
      "loss": 0.0256,
      "step": 13236
    },
    {
      "epoch": 1.8552207428170988,
      "grad_norm": 0.23803015053272247,
      "learning_rate": 0.00011728773020808417,
      "loss": 0.0207,
      "step": 13237
    },
    {
      "epoch": 1.8553608969866855,
      "grad_norm": 0.2239723652601242,
      "learning_rate": 0.00011727337957426452,
      "loss": 0.0276,
      "step": 13238
    },
    {
      "epoch": 1.855501051156272,
      "grad_norm": 0.376230388879776,
      "learning_rate": 0.00011725902894044487,
      "loss": 0.0379,
      "step": 13239
    },
    {
      "epoch": 1.8556412053258584,
      "grad_norm": 0.31867870688438416,
      "learning_rate": 0.0001172446783066252,
      "loss": 0.0396,
      "step": 13240
    },
    {
      "epoch": 1.8557813594954449,
      "grad_norm": 0.36151254177093506,
      "learning_rate": 0.00011723032767280554,
      "loss": 0.0346,
      "step": 13241
    },
    {
      "epoch": 1.8559215136650316,
      "grad_norm": 0.36161887645721436,
      "learning_rate": 0.00011721597703898589,
      "loss": 0.0416,
      "step": 13242
    },
    {
      "epoch": 1.8560616678346182,
      "grad_norm": 0.3505237102508545,
      "learning_rate": 0.00011720162640516621,
      "loss": 0.0314,
      "step": 13243
    },
    {
      "epoch": 1.8562018220042047,
      "grad_norm": 0.22928068041801453,
      "learning_rate": 0.00011718727577134656,
      "loss": 0.0257,
      "step": 13244
    },
    {
      "epoch": 1.8563419761737912,
      "grad_norm": 0.17889830470085144,
      "learning_rate": 0.00011717292513752688,
      "loss": 0.0284,
      "step": 13245
    },
    {
      "epoch": 1.8564821303433776,
      "grad_norm": 0.09752125293016434,
      "learning_rate": 0.00011715857450370724,
      "loss": 0.014,
      "step": 13246
    },
    {
      "epoch": 1.8566222845129643,
      "grad_norm": 0.10232482850551605,
      "learning_rate": 0.00011714422386988758,
      "loss": 0.0078,
      "step": 13247
    },
    {
      "epoch": 1.8567624386825508,
      "grad_norm": 0.43635377287864685,
      "learning_rate": 0.00011712987323606791,
      "loss": 0.0474,
      "step": 13248
    },
    {
      "epoch": 1.8569025928521374,
      "grad_norm": 1.17537522315979,
      "learning_rate": 0.00011711552260224825,
      "loss": 0.0882,
      "step": 13249
    },
    {
      "epoch": 1.857042747021724,
      "grad_norm": 0.15354828536510468,
      "learning_rate": 0.00011710117196842861,
      "loss": 0.0452,
      "step": 13250
    },
    {
      "epoch": 1.8571829011913104,
      "grad_norm": 0.4120674729347229,
      "learning_rate": 0.00011708682133460894,
      "loss": 0.0814,
      "step": 13251
    },
    {
      "epoch": 1.857323055360897,
      "grad_norm": 0.37464696168899536,
      "learning_rate": 0.00011707247070078928,
      "loss": 0.081,
      "step": 13252
    },
    {
      "epoch": 1.8574632095304835,
      "grad_norm": 0.4091629385948181,
      "learning_rate": 0.00011705812006696961,
      "loss": 0.0498,
      "step": 13253
    },
    {
      "epoch": 1.8576033637000702,
      "grad_norm": 0.23688896000385284,
      "learning_rate": 0.00011704376943314995,
      "loss": 0.0269,
      "step": 13254
    },
    {
      "epoch": 1.8577435178696566,
      "grad_norm": 0.4084051549434662,
      "learning_rate": 0.0001170294187993303,
      "loss": 0.0582,
      "step": 13255
    },
    {
      "epoch": 1.857883672039243,
      "grad_norm": 0.2864155173301697,
      "learning_rate": 0.00011701506816551063,
      "loss": 0.0415,
      "step": 13256
    },
    {
      "epoch": 1.8580238262088296,
      "grad_norm": 0.0915154367685318,
      "learning_rate": 0.00011700071753169098,
      "loss": 0.0286,
      "step": 13257
    },
    {
      "epoch": 1.8581639803784162,
      "grad_norm": 0.8650240898132324,
      "learning_rate": 0.00011698636689787132,
      "loss": 0.11,
      "step": 13258
    },
    {
      "epoch": 1.858304134548003,
      "grad_norm": 0.3484264016151428,
      "learning_rate": 0.00011697201626405165,
      "loss": 0.0299,
      "step": 13259
    },
    {
      "epoch": 1.8584442887175894,
      "grad_norm": 0.7601109147071838,
      "learning_rate": 0.00011695766563023199,
      "loss": 0.1608,
      "step": 13260
    },
    {
      "epoch": 1.8585844428871758,
      "grad_norm": 0.41416794061660767,
      "learning_rate": 0.00011694331499641232,
      "loss": 0.0376,
      "step": 13261
    },
    {
      "epoch": 1.8587245970567623,
      "grad_norm": 0.29231420159339905,
      "learning_rate": 0.00011692896436259267,
      "loss": 0.0314,
      "step": 13262
    },
    {
      "epoch": 1.858864751226349,
      "grad_norm": 0.281376451253891,
      "learning_rate": 0.00011691461372877302,
      "loss": 0.0776,
      "step": 13263
    },
    {
      "epoch": 1.8590049053959357,
      "grad_norm": 0.4767308831214905,
      "learning_rate": 0.00011690026309495334,
      "loss": 0.0943,
      "step": 13264
    },
    {
      "epoch": 1.8591450595655221,
      "grad_norm": 0.2560347020626068,
      "learning_rate": 0.00011688591246113369,
      "loss": 0.0667,
      "step": 13265
    },
    {
      "epoch": 1.8592852137351086,
      "grad_norm": 0.12228003889322281,
      "learning_rate": 0.00011687156182731404,
      "loss": 0.0162,
      "step": 13266
    },
    {
      "epoch": 1.859425367904695,
      "grad_norm": 0.48055019974708557,
      "learning_rate": 0.00011685721119349437,
      "loss": 0.0344,
      "step": 13267
    },
    {
      "epoch": 1.8595655220742817,
      "grad_norm": 0.0845029354095459,
      "learning_rate": 0.00011684286055967471,
      "loss": 0.0156,
      "step": 13268
    },
    {
      "epoch": 1.8597056762438684,
      "grad_norm": 0.1355154663324356,
      "learning_rate": 0.00011682850992585504,
      "loss": 0.0125,
      "step": 13269
    },
    {
      "epoch": 1.8598458304134549,
      "grad_norm": 0.5049238204956055,
      "learning_rate": 0.00011681415929203538,
      "loss": 0.0577,
      "step": 13270
    },
    {
      "epoch": 1.8599859845830413,
      "grad_norm": 0.7007641196250916,
      "learning_rate": 0.00011679980865821574,
      "loss": 0.1806,
      "step": 13271
    },
    {
      "epoch": 1.8601261387526278,
      "grad_norm": 0.46290716528892517,
      "learning_rate": 0.00011678545802439607,
      "loss": 0.0437,
      "step": 13272
    },
    {
      "epoch": 1.8602662929222145,
      "grad_norm": 0.249790221452713,
      "learning_rate": 0.00011677110739057641,
      "loss": 0.0219,
      "step": 13273
    },
    {
      "epoch": 1.8604064470918011,
      "grad_norm": 0.5716654658317566,
      "learning_rate": 0.00011675675675675675,
      "loss": 0.0768,
      "step": 13274
    },
    {
      "epoch": 1.8605466012613876,
      "grad_norm": 0.6225055456161499,
      "learning_rate": 0.00011674240612293708,
      "loss": 0.1216,
      "step": 13275
    },
    {
      "epoch": 1.860686755430974,
      "grad_norm": 0.5695887804031372,
      "learning_rate": 0.00011672805548911742,
      "loss": 0.1039,
      "step": 13276
    },
    {
      "epoch": 1.8608269096005605,
      "grad_norm": 0.3829694986343384,
      "learning_rate": 0.00011671370485529775,
      "loss": 0.0681,
      "step": 13277
    },
    {
      "epoch": 1.8609670637701472,
      "grad_norm": 1.835770606994629,
      "learning_rate": 0.0001166993542214781,
      "loss": 0.0846,
      "step": 13278
    },
    {
      "epoch": 1.8611072179397337,
      "grad_norm": 0.23133142292499542,
      "learning_rate": 0.00011668500358765845,
      "loss": 0.0208,
      "step": 13279
    },
    {
      "epoch": 1.8612473721093203,
      "grad_norm": 0.751875102519989,
      "learning_rate": 0.00011667065295383878,
      "loss": 0.073,
      "step": 13280
    },
    {
      "epoch": 1.8613875262789068,
      "grad_norm": 0.299029678106308,
      "learning_rate": 0.00011665630232001912,
      "loss": 0.0347,
      "step": 13281
    },
    {
      "epoch": 1.8615276804484933,
      "grad_norm": 0.642019510269165,
      "learning_rate": 0.00011664195168619947,
      "loss": 0.1066,
      "step": 13282
    },
    {
      "epoch": 1.86166783461808,
      "grad_norm": 0.9072075486183167,
      "learning_rate": 0.0001166276010523798,
      "loss": 0.03,
      "step": 13283
    },
    {
      "epoch": 1.8618079887876664,
      "grad_norm": 2.1316585540771484,
      "learning_rate": 0.00011661325041856015,
      "loss": 0.2839,
      "step": 13284
    },
    {
      "epoch": 1.861948142957253,
      "grad_norm": 0.6896312832832336,
      "learning_rate": 0.00011659889978474049,
      "loss": 0.2553,
      "step": 13285
    },
    {
      "epoch": 1.8620882971268395,
      "grad_norm": 0.23766830563545227,
      "learning_rate": 0.00011658454915092082,
      "loss": 0.0155,
      "step": 13286
    },
    {
      "epoch": 1.862228451296426,
      "grad_norm": 0.37232038378715515,
      "learning_rate": 0.00011657019851710117,
      "loss": 0.0761,
      "step": 13287
    },
    {
      "epoch": 1.8623686054660125,
      "grad_norm": 0.16449019312858582,
      "learning_rate": 0.0001165558478832815,
      "loss": 0.0179,
      "step": 13288
    },
    {
      "epoch": 1.8625087596355991,
      "grad_norm": 0.21596352756023407,
      "learning_rate": 0.00011654149724946184,
      "loss": 0.0231,
      "step": 13289
    },
    {
      "epoch": 1.8626489138051858,
      "grad_norm": 0.28764578700065613,
      "learning_rate": 0.00011652714661564218,
      "loss": 0.0565,
      "step": 13290
    },
    {
      "epoch": 1.8627890679747723,
      "grad_norm": 0.10288781672716141,
      "learning_rate": 0.00011651279598182251,
      "loss": 0.0227,
      "step": 13291
    },
    {
      "epoch": 1.8629292221443587,
      "grad_norm": 0.4076229929924011,
      "learning_rate": 0.00011649844534800286,
      "loss": 0.0459,
      "step": 13292
    },
    {
      "epoch": 1.8630693763139452,
      "grad_norm": 0.24998854100704193,
      "learning_rate": 0.00011648409471418321,
      "loss": 0.0396,
      "step": 13293
    },
    {
      "epoch": 1.8632095304835319,
      "grad_norm": 0.24248023331165314,
      "learning_rate": 0.00011646974408036354,
      "loss": 0.0404,
      "step": 13294
    },
    {
      "epoch": 1.8633496846531186,
      "grad_norm": 0.09918013215065002,
      "learning_rate": 0.00011645539344654388,
      "loss": 0.0191,
      "step": 13295
    },
    {
      "epoch": 1.863489838822705,
      "grad_norm": 0.2902102470397949,
      "learning_rate": 0.00011644104281272421,
      "loss": 0.0554,
      "step": 13296
    },
    {
      "epoch": 1.8636299929922915,
      "grad_norm": 0.26920413970947266,
      "learning_rate": 0.00011642669217890455,
      "loss": 0.0315,
      "step": 13297
    },
    {
      "epoch": 1.863770147161878,
      "grad_norm": 0.2771225869655609,
      "learning_rate": 0.00011641234154508491,
      "loss": 0.0781,
      "step": 13298
    },
    {
      "epoch": 1.8639103013314646,
      "grad_norm": 1.0889365673065186,
      "learning_rate": 0.00011639799091126524,
      "loss": 0.0228,
      "step": 13299
    },
    {
      "epoch": 1.8640504555010513,
      "grad_norm": 0.3517109453678131,
      "learning_rate": 0.00011638364027744558,
      "loss": 0.1046,
      "step": 13300
    },
    {
      "epoch": 1.8641906096706378,
      "grad_norm": 0.11087671667337418,
      "learning_rate": 0.00011636928964362592,
      "loss": 0.0099,
      "step": 13301
    },
    {
      "epoch": 1.8643307638402242,
      "grad_norm": 0.5309032201766968,
      "learning_rate": 0.00011635493900980625,
      "loss": 0.0606,
      "step": 13302
    },
    {
      "epoch": 1.8644709180098107,
      "grad_norm": 0.14224471151828766,
      "learning_rate": 0.0001163405883759866,
      "loss": 0.0225,
      "step": 13303
    },
    {
      "epoch": 1.8646110721793974,
      "grad_norm": 0.20649072527885437,
      "learning_rate": 0.00011632623774216693,
      "loss": 0.0683,
      "step": 13304
    },
    {
      "epoch": 1.864751226348984,
      "grad_norm": 0.2530348300933838,
      "learning_rate": 0.00011631188710834728,
      "loss": 0.0499,
      "step": 13305
    },
    {
      "epoch": 1.8648913805185705,
      "grad_norm": 0.4077099561691284,
      "learning_rate": 0.00011629753647452762,
      "loss": 0.0448,
      "step": 13306
    },
    {
      "epoch": 1.865031534688157,
      "grad_norm": 0.34534546732902527,
      "learning_rate": 0.00011628318584070795,
      "loss": 0.0574,
      "step": 13307
    },
    {
      "epoch": 1.8651716888577434,
      "grad_norm": 0.24342651665210724,
      "learning_rate": 0.00011626883520688829,
      "loss": 0.0349,
      "step": 13308
    },
    {
      "epoch": 1.86531184302733,
      "grad_norm": 0.16251760721206665,
      "learning_rate": 0.00011625448457306864,
      "loss": 0.0151,
      "step": 13309
    },
    {
      "epoch": 1.8654519971969166,
      "grad_norm": 0.2958175241947174,
      "learning_rate": 0.00011624013393924897,
      "loss": 0.0408,
      "step": 13310
    },
    {
      "epoch": 1.8655921513665032,
      "grad_norm": 0.17052926123142242,
      "learning_rate": 0.00011622578330542931,
      "loss": 0.0445,
      "step": 13311
    },
    {
      "epoch": 1.8657323055360897,
      "grad_norm": 0.3912261426448822,
      "learning_rate": 0.00011621143267160964,
      "loss": 0.0982,
      "step": 13312
    },
    {
      "epoch": 1.8658724597056762,
      "grad_norm": 0.4915083050727844,
      "learning_rate": 0.00011619708203778999,
      "loss": 0.0347,
      "step": 13313
    },
    {
      "epoch": 1.8660126138752626,
      "grad_norm": 0.2878972887992859,
      "learning_rate": 0.00011618273140397034,
      "loss": 0.0283,
      "step": 13314
    },
    {
      "epoch": 1.8661527680448493,
      "grad_norm": 0.37967249751091003,
      "learning_rate": 0.00011616838077015067,
      "loss": 0.0611,
      "step": 13315
    },
    {
      "epoch": 1.866292922214436,
      "grad_norm": 0.07954419404268265,
      "learning_rate": 0.00011615403013633101,
      "loss": 0.017,
      "step": 13316
    },
    {
      "epoch": 1.8664330763840224,
      "grad_norm": 0.6028661727905273,
      "learning_rate": 0.00011613967950251135,
      "loss": 0.1259,
      "step": 13317
    },
    {
      "epoch": 1.866573230553609,
      "grad_norm": 0.14035560190677643,
      "learning_rate": 0.00011612532886869168,
      "loss": 0.0152,
      "step": 13318
    },
    {
      "epoch": 1.8667133847231954,
      "grad_norm": 0.4251324236392975,
      "learning_rate": 0.00011611097823487204,
      "loss": 0.0507,
      "step": 13319
    },
    {
      "epoch": 1.866853538892782,
      "grad_norm": 0.3765099346637726,
      "learning_rate": 0.00011609662760105238,
      "loss": 0.0396,
      "step": 13320
    },
    {
      "epoch": 1.8669936930623687,
      "grad_norm": 0.12300407886505127,
      "learning_rate": 0.00011608227696723271,
      "loss": 0.0098,
      "step": 13321
    },
    {
      "epoch": 1.8671338472319552,
      "grad_norm": 0.1625182330608368,
      "learning_rate": 0.00011606792633341305,
      "loss": 0.0064,
      "step": 13322
    },
    {
      "epoch": 1.8672740014015416,
      "grad_norm": 0.18274950981140137,
      "learning_rate": 0.00011605357569959338,
      "loss": 0.0375,
      "step": 13323
    },
    {
      "epoch": 1.867414155571128,
      "grad_norm": 0.15846355259418488,
      "learning_rate": 0.00011603922506577372,
      "loss": 0.0099,
      "step": 13324
    },
    {
      "epoch": 1.8675543097407148,
      "grad_norm": 0.6078698039054871,
      "learning_rate": 0.00011602487443195408,
      "loss": 0.0851,
      "step": 13325
    },
    {
      "epoch": 1.8676944639103015,
      "grad_norm": 0.2566559314727783,
      "learning_rate": 0.0001160105237981344,
      "loss": 0.0256,
      "step": 13326
    },
    {
      "epoch": 1.867834618079888,
      "grad_norm": 0.27305838465690613,
      "learning_rate": 0.00011599617316431475,
      "loss": 0.1789,
      "step": 13327
    },
    {
      "epoch": 1.8679747722494744,
      "grad_norm": 0.5287879109382629,
      "learning_rate": 0.00011598182253049509,
      "loss": 0.0471,
      "step": 13328
    },
    {
      "epoch": 1.8681149264190609,
      "grad_norm": 0.15730054676532745,
      "learning_rate": 0.00011596747189667542,
      "loss": 0.0125,
      "step": 13329
    },
    {
      "epoch": 1.8682550805886475,
      "grad_norm": 0.17922146618366241,
      "learning_rate": 0.00011595312126285577,
      "loss": 0.0462,
      "step": 13330
    },
    {
      "epoch": 1.8683952347582342,
      "grad_norm": 0.20706181228160858,
      "learning_rate": 0.0001159387706290361,
      "loss": 0.0158,
      "step": 13331
    },
    {
      "epoch": 1.8685353889278207,
      "grad_norm": 0.19337230920791626,
      "learning_rate": 0.00011592441999521644,
      "loss": 0.0379,
      "step": 13332
    },
    {
      "epoch": 1.8686755430974071,
      "grad_norm": 0.5239232182502747,
      "learning_rate": 0.00011591006936139679,
      "loss": 0.0408,
      "step": 13333
    },
    {
      "epoch": 1.8688156972669936,
      "grad_norm": 0.5003645420074463,
      "learning_rate": 0.00011589571872757712,
      "loss": 0.1373,
      "step": 13334
    },
    {
      "epoch": 1.8689558514365803,
      "grad_norm": 5.551768779754639,
      "learning_rate": 0.00011588136809375747,
      "loss": 0.443,
      "step": 13335
    },
    {
      "epoch": 1.8690960056061667,
      "grad_norm": 0.20352433621883392,
      "learning_rate": 0.00011586701745993781,
      "loss": 0.0428,
      "step": 13336
    },
    {
      "epoch": 1.8692361597757534,
      "grad_norm": 0.720059871673584,
      "learning_rate": 0.00011585266682611814,
      "loss": 0.0797,
      "step": 13337
    },
    {
      "epoch": 1.8693763139453399,
      "grad_norm": 0.2068159431219101,
      "learning_rate": 0.00011583831619229848,
      "loss": 0.0173,
      "step": 13338
    },
    {
      "epoch": 1.8695164681149263,
      "grad_norm": 0.1925693154335022,
      "learning_rate": 0.00011582396555847881,
      "loss": 0.0415,
      "step": 13339
    },
    {
      "epoch": 1.869656622284513,
      "grad_norm": 0.1934211552143097,
      "learning_rate": 0.00011580961492465915,
      "loss": 0.0188,
      "step": 13340
    },
    {
      "epoch": 1.8697967764540995,
      "grad_norm": 0.8721495866775513,
      "learning_rate": 0.00011579526429083951,
      "loss": 0.0739,
      "step": 13341
    },
    {
      "epoch": 1.8699369306236862,
      "grad_norm": 0.24747636914253235,
      "learning_rate": 0.00011578091365701984,
      "loss": 0.0303,
      "step": 13342
    },
    {
      "epoch": 1.8700770847932726,
      "grad_norm": 0.08373694866895676,
      "learning_rate": 0.00011576656302320018,
      "loss": 0.0149,
      "step": 13343
    },
    {
      "epoch": 1.870217238962859,
      "grad_norm": 0.39172282814979553,
      "learning_rate": 0.00011575221238938052,
      "loss": 0.0743,
      "step": 13344
    },
    {
      "epoch": 1.8703573931324455,
      "grad_norm": 0.2763004004955292,
      "learning_rate": 0.00011573786175556085,
      "loss": 0.0218,
      "step": 13345
    },
    {
      "epoch": 1.8704975473020322,
      "grad_norm": 0.5017855167388916,
      "learning_rate": 0.00011572351112174121,
      "loss": 0.0292,
      "step": 13346
    },
    {
      "epoch": 1.870637701471619,
      "grad_norm": 0.6171298027038574,
      "learning_rate": 0.00011570916048792154,
      "loss": 0.0437,
      "step": 13347
    },
    {
      "epoch": 1.8707778556412054,
      "grad_norm": 0.38493019342422485,
      "learning_rate": 0.00011569480985410188,
      "loss": 0.0366,
      "step": 13348
    },
    {
      "epoch": 1.8709180098107918,
      "grad_norm": 0.2922995686531067,
      "learning_rate": 0.00011568045922028222,
      "loss": 0.0563,
      "step": 13349
    },
    {
      "epoch": 1.8710581639803783,
      "grad_norm": 0.09337473660707474,
      "learning_rate": 0.00011566610858646255,
      "loss": 0.0097,
      "step": 13350
    },
    {
      "epoch": 1.871198318149965,
      "grad_norm": 1.8487067222595215,
      "learning_rate": 0.0001156517579526429,
      "loss": 0.0553,
      "step": 13351
    },
    {
      "epoch": 1.8713384723195516,
      "grad_norm": 0.9333266615867615,
      "learning_rate": 0.00011563740731882325,
      "loss": 0.0809,
      "step": 13352
    },
    {
      "epoch": 1.871478626489138,
      "grad_norm": 0.37149691581726074,
      "learning_rate": 0.00011562305668500357,
      "loss": 0.0361,
      "step": 13353
    },
    {
      "epoch": 1.8716187806587246,
      "grad_norm": 0.4017590284347534,
      "learning_rate": 0.00011560870605118392,
      "loss": 0.0731,
      "step": 13354
    },
    {
      "epoch": 1.871758934828311,
      "grad_norm": 0.211772620677948,
      "learning_rate": 0.00011559435541736426,
      "loss": 0.0626,
      "step": 13355
    },
    {
      "epoch": 1.8718990889978977,
      "grad_norm": 0.11263607442378998,
      "learning_rate": 0.00011558000478354459,
      "loss": 0.0204,
      "step": 13356
    },
    {
      "epoch": 1.8720392431674844,
      "grad_norm": 0.16496983170509338,
      "learning_rate": 0.00011556565414972494,
      "loss": 0.0404,
      "step": 13357
    },
    {
      "epoch": 1.8721793973370708,
      "grad_norm": 0.08869792520999908,
      "learning_rate": 0.00011555130351590527,
      "loss": 0.0078,
      "step": 13358
    },
    {
      "epoch": 1.8723195515066573,
      "grad_norm": 0.22677567601203918,
      "learning_rate": 0.00011553695288208561,
      "loss": 0.047,
      "step": 13359
    },
    {
      "epoch": 1.8724597056762438,
      "grad_norm": 0.23768877983093262,
      "learning_rate": 0.00011552260224826596,
      "loss": 0.0224,
      "step": 13360
    },
    {
      "epoch": 1.8725998598458304,
      "grad_norm": 0.1602317988872528,
      "learning_rate": 0.00011550825161444628,
      "loss": 0.0225,
      "step": 13361
    },
    {
      "epoch": 1.8727400140154171,
      "grad_norm": 0.21547986567020416,
      "learning_rate": 0.00011549390098062664,
      "loss": 0.0432,
      "step": 13362
    },
    {
      "epoch": 1.8728801681850036,
      "grad_norm": 0.3699752688407898,
      "learning_rate": 0.00011547955034680698,
      "loss": 0.0313,
      "step": 13363
    },
    {
      "epoch": 1.87302032235459,
      "grad_norm": 0.35498467087745667,
      "learning_rate": 0.00011546519971298731,
      "loss": 0.0423,
      "step": 13364
    },
    {
      "epoch": 1.8731604765241765,
      "grad_norm": 0.4760517179965973,
      "learning_rate": 0.00011545084907916765,
      "loss": 0.0793,
      "step": 13365
    },
    {
      "epoch": 1.8733006306937632,
      "grad_norm": 0.12012570351362228,
      "learning_rate": 0.00011543649844534798,
      "loss": 0.019,
      "step": 13366
    },
    {
      "epoch": 1.8734407848633496,
      "grad_norm": 0.5168591141700745,
      "learning_rate": 0.00011542214781152834,
      "loss": 0.0674,
      "step": 13367
    },
    {
      "epoch": 1.8735809390329363,
      "grad_norm": 0.22395719587802887,
      "learning_rate": 0.00011540779717770868,
      "loss": 0.0678,
      "step": 13368
    },
    {
      "epoch": 1.8737210932025228,
      "grad_norm": 0.19130194187164307,
      "learning_rate": 0.00011539344654388901,
      "loss": 0.0528,
      "step": 13369
    },
    {
      "epoch": 1.8738612473721092,
      "grad_norm": 0.5229623913764954,
      "learning_rate": 0.00011537909591006935,
      "loss": 0.0415,
      "step": 13370
    },
    {
      "epoch": 1.8740014015416957,
      "grad_norm": 0.265043169260025,
      "learning_rate": 0.00011536474527624969,
      "loss": 0.032,
      "step": 13371
    },
    {
      "epoch": 1.8741415557112824,
      "grad_norm": 0.2419590950012207,
      "learning_rate": 0.00011535039464243002,
      "loss": 0.0369,
      "step": 13372
    },
    {
      "epoch": 1.874281709880869,
      "grad_norm": 0.14468783140182495,
      "learning_rate": 0.00011533604400861038,
      "loss": 0.0214,
      "step": 13373
    },
    {
      "epoch": 1.8744218640504555,
      "grad_norm": 0.33427003026008606,
      "learning_rate": 0.0001153216933747907,
      "loss": 0.0239,
      "step": 13374
    },
    {
      "epoch": 1.874562018220042,
      "grad_norm": 0.4960840344429016,
      "learning_rate": 0.00011530734274097105,
      "loss": 0.0913,
      "step": 13375
    },
    {
      "epoch": 1.8747021723896284,
      "grad_norm": 0.475669801235199,
      "learning_rate": 0.00011529299210715139,
      "loss": 0.0624,
      "step": 13376
    },
    {
      "epoch": 1.8748423265592151,
      "grad_norm": 0.21983139216899872,
      "learning_rate": 0.00011527864147333172,
      "loss": 0.0396,
      "step": 13377
    },
    {
      "epoch": 1.8749824807288018,
      "grad_norm": 0.4598732888698578,
      "learning_rate": 0.00011526429083951207,
      "loss": 0.0301,
      "step": 13378
    },
    {
      "epoch": 1.8751226348983883,
      "grad_norm": 0.46474209427833557,
      "learning_rate": 0.00011524994020569242,
      "loss": 0.0325,
      "step": 13379
    },
    {
      "epoch": 1.8752627890679747,
      "grad_norm": 0.04625300318002701,
      "learning_rate": 0.00011523558957187274,
      "loss": 0.002,
      "step": 13380
    },
    {
      "epoch": 1.8754029432375612,
      "grad_norm": 0.9333197474479675,
      "learning_rate": 0.00011522123893805309,
      "loss": 0.2254,
      "step": 13381
    },
    {
      "epoch": 1.8755430974071479,
      "grad_norm": 1.1528195142745972,
      "learning_rate": 0.00011520688830423341,
      "loss": 0.1393,
      "step": 13382
    },
    {
      "epoch": 1.8756832515767345,
      "grad_norm": 1.5066992044448853,
      "learning_rate": 0.00011519253767041377,
      "loss": 0.0915,
      "step": 13383
    },
    {
      "epoch": 1.875823405746321,
      "grad_norm": 0.21653881669044495,
      "learning_rate": 0.00011517818703659411,
      "loss": 0.0076,
      "step": 13384
    },
    {
      "epoch": 1.8759635599159075,
      "grad_norm": 0.12956270575523376,
      "learning_rate": 0.00011516383640277444,
      "loss": 0.0055,
      "step": 13385
    },
    {
      "epoch": 1.876103714085494,
      "grad_norm": 0.2553445100784302,
      "learning_rate": 0.00011514948576895478,
      "loss": 0.0573,
      "step": 13386
    },
    {
      "epoch": 1.8762438682550806,
      "grad_norm": 0.32053250074386597,
      "learning_rate": 0.00011513513513513513,
      "loss": 0.013,
      "step": 13387
    },
    {
      "epoch": 1.8763840224246673,
      "grad_norm": 0.4977538287639618,
      "learning_rate": 0.00011512078450131545,
      "loss": 0.0439,
      "step": 13388
    },
    {
      "epoch": 1.8765241765942537,
      "grad_norm": 0.17351041734218597,
      "learning_rate": 0.00011510643386749581,
      "loss": 0.0277,
      "step": 13389
    },
    {
      "epoch": 1.8766643307638402,
      "grad_norm": 0.23659555613994598,
      "learning_rate": 0.00011509208323367615,
      "loss": 0.0624,
      "step": 13390
    },
    {
      "epoch": 1.8768044849334267,
      "grad_norm": 0.3162183463573456,
      "learning_rate": 0.00011507773259985648,
      "loss": 0.0555,
      "step": 13391
    },
    {
      "epoch": 1.8769446391030133,
      "grad_norm": 0.1986086070537567,
      "learning_rate": 0.00011506338196603682,
      "loss": 0.0381,
      "step": 13392
    },
    {
      "epoch": 1.8770847932725998,
      "grad_norm": 0.24596472084522247,
      "learning_rate": 0.00011504903133221715,
      "loss": 0.034,
      "step": 13393
    },
    {
      "epoch": 1.8772249474421865,
      "grad_norm": 0.2564314901828766,
      "learning_rate": 0.0001150346806983975,
      "loss": 0.0227,
      "step": 13394
    },
    {
      "epoch": 1.877365101611773,
      "grad_norm": 0.08028414845466614,
      "learning_rate": 0.00011502033006457785,
      "loss": 0.0091,
      "step": 13395
    },
    {
      "epoch": 1.8775052557813594,
      "grad_norm": 0.17415869235992432,
      "learning_rate": 0.00011500597943075818,
      "loss": 0.037,
      "step": 13396
    },
    {
      "epoch": 1.877645409950946,
      "grad_norm": 0.3385166823863983,
      "learning_rate": 0.00011499162879693852,
      "loss": 0.0992,
      "step": 13397
    },
    {
      "epoch": 1.8777855641205325,
      "grad_norm": 0.19357110559940338,
      "learning_rate": 0.00011497727816311888,
      "loss": 0.0274,
      "step": 13398
    },
    {
      "epoch": 1.8779257182901192,
      "grad_norm": 0.2873781621456146,
      "learning_rate": 0.0001149629275292992,
      "loss": 0.0111,
      "step": 13399
    },
    {
      "epoch": 1.8780658724597057,
      "grad_norm": 0.3342207968235016,
      "learning_rate": 0.00011494857689547955,
      "loss": 0.0432,
      "step": 13400
    },
    {
      "epoch": 1.8782060266292921,
      "grad_norm": 0.32632869482040405,
      "learning_rate": 0.00011493422626165987,
      "loss": 0.0339,
      "step": 13401
    },
    {
      "epoch": 1.8783461807988786,
      "grad_norm": 0.15519414842128754,
      "learning_rate": 0.00011491987562784022,
      "loss": 0.0437,
      "step": 13402
    },
    {
      "epoch": 1.8784863349684653,
      "grad_norm": 0.297125905752182,
      "learning_rate": 0.00011490552499402056,
      "loss": 0.0413,
      "step": 13403
    },
    {
      "epoch": 1.878626489138052,
      "grad_norm": 0.18203961849212646,
      "learning_rate": 0.00011489117436020089,
      "loss": 0.0426,
      "step": 13404
    },
    {
      "epoch": 1.8787666433076384,
      "grad_norm": 0.3117711842060089,
      "learning_rate": 0.00011487682372638124,
      "loss": 0.0311,
      "step": 13405
    },
    {
      "epoch": 1.8789067974772249,
      "grad_norm": 0.3444058299064636,
      "learning_rate": 0.00011486247309256158,
      "loss": 0.0363,
      "step": 13406
    },
    {
      "epoch": 1.8790469516468113,
      "grad_norm": 0.30491572618484497,
      "learning_rate": 0.00011484812245874191,
      "loss": 0.0298,
      "step": 13407
    },
    {
      "epoch": 1.879187105816398,
      "grad_norm": 0.29263925552368164,
      "learning_rate": 0.00011483377182492226,
      "loss": 0.0356,
      "step": 13408
    },
    {
      "epoch": 1.8793272599859847,
      "grad_norm": 0.32042330503463745,
      "learning_rate": 0.00011481942119110258,
      "loss": 0.0281,
      "step": 13409
    },
    {
      "epoch": 1.8794674141555712,
      "grad_norm": 0.05516209825873375,
      "learning_rate": 0.00011480507055728294,
      "loss": 0.0084,
      "step": 13410
    },
    {
      "epoch": 1.8796075683251576,
      "grad_norm": 0.2328115999698639,
      "learning_rate": 0.00011479071992346328,
      "loss": 0.0492,
      "step": 13411
    },
    {
      "epoch": 1.879747722494744,
      "grad_norm": 0.5882803797721863,
      "learning_rate": 0.00011477636928964361,
      "loss": 0.0607,
      "step": 13412
    },
    {
      "epoch": 1.8798878766643308,
      "grad_norm": 0.25868362188339233,
      "learning_rate": 0.00011476201865582395,
      "loss": 0.0269,
      "step": 13413
    },
    {
      "epoch": 1.8800280308339175,
      "grad_norm": 0.28339165449142456,
      "learning_rate": 0.00011474766802200431,
      "loss": 0.0343,
      "step": 13414
    },
    {
      "epoch": 1.880168185003504,
      "grad_norm": 0.27832159399986267,
      "learning_rate": 0.00011473331738818464,
      "loss": 0.0145,
      "step": 13415
    },
    {
      "epoch": 1.8803083391730904,
      "grad_norm": 0.4135292172431946,
      "learning_rate": 0.00011471896675436498,
      "loss": 0.1034,
      "step": 13416
    },
    {
      "epoch": 1.8804484933426768,
      "grad_norm": 0.3380087614059448,
      "learning_rate": 0.00011470461612054531,
      "loss": 0.0611,
      "step": 13417
    },
    {
      "epoch": 1.8805886475122635,
      "grad_norm": 0.5146254301071167,
      "learning_rate": 0.00011469026548672565,
      "loss": 0.0226,
      "step": 13418
    },
    {
      "epoch": 1.8807288016818502,
      "grad_norm": 0.45673611760139465,
      "learning_rate": 0.00011467591485290599,
      "loss": 0.0177,
      "step": 13419
    },
    {
      "epoch": 1.8808689558514367,
      "grad_norm": 0.455418199300766,
      "learning_rate": 0.00011466156421908633,
      "loss": 0.0729,
      "step": 13420
    },
    {
      "epoch": 1.8810091100210231,
      "grad_norm": 0.12285628914833069,
      "learning_rate": 0.00011464721358526668,
      "loss": 0.0156,
      "step": 13421
    },
    {
      "epoch": 1.8811492641906096,
      "grad_norm": 0.39038726687431335,
      "learning_rate": 0.00011463286295144702,
      "loss": 0.0572,
      "step": 13422
    },
    {
      "epoch": 1.8812894183601963,
      "grad_norm": 0.29715901613235474,
      "learning_rate": 0.00011461851231762735,
      "loss": 0.0723,
      "step": 13423
    },
    {
      "epoch": 1.8814295725297827,
      "grad_norm": 0.26968151330947876,
      "learning_rate": 0.00011460416168380769,
      "loss": 0.031,
      "step": 13424
    },
    {
      "epoch": 1.8815697266993694,
      "grad_norm": 0.1891375631093979,
      "learning_rate": 0.00011458981104998802,
      "loss": 0.0432,
      "step": 13425
    },
    {
      "epoch": 1.8817098808689559,
      "grad_norm": 0.17729321122169495,
      "learning_rate": 0.00011457546041616837,
      "loss": 0.0242,
      "step": 13426
    },
    {
      "epoch": 1.8818500350385423,
      "grad_norm": 0.04632085934281349,
      "learning_rate": 0.00011456110978234872,
      "loss": 0.004,
      "step": 13427
    },
    {
      "epoch": 1.881990189208129,
      "grad_norm": 0.24560733139514923,
      "learning_rate": 0.00011454675914852904,
      "loss": 0.0197,
      "step": 13428
    },
    {
      "epoch": 1.8821303433777155,
      "grad_norm": 1.5941994190216064,
      "learning_rate": 0.00011453240851470939,
      "loss": 0.0926,
      "step": 13429
    },
    {
      "epoch": 1.8822704975473021,
      "grad_norm": 1.6707119941711426,
      "learning_rate": 0.00011451805788088974,
      "loss": 0.0491,
      "step": 13430
    },
    {
      "epoch": 1.8824106517168886,
      "grad_norm": 0.49443477392196655,
      "learning_rate": 0.00011450370724707007,
      "loss": 0.0455,
      "step": 13431
    },
    {
      "epoch": 1.882550805886475,
      "grad_norm": 0.35893669724464417,
      "learning_rate": 0.00011448935661325041,
      "loss": 0.0324,
      "step": 13432
    },
    {
      "epoch": 1.8826909600560615,
      "grad_norm": 0.35242122411727905,
      "learning_rate": 0.00011447500597943075,
      "loss": 0.0444,
      "step": 13433
    },
    {
      "epoch": 1.8828311142256482,
      "grad_norm": 2.934565544128418,
      "learning_rate": 0.00011446065534561108,
      "loss": 0.1014,
      "step": 13434
    },
    {
      "epoch": 1.8829712683952349,
      "grad_norm": 4.257355690002441,
      "learning_rate": 0.00011444630471179142,
      "loss": 0.2638,
      "step": 13435
    },
    {
      "epoch": 1.8831114225648213,
      "grad_norm": 0.23855766654014587,
      "learning_rate": 0.00011443195407797177,
      "loss": 0.0191,
      "step": 13436
    },
    {
      "epoch": 1.8832515767344078,
      "grad_norm": 0.349558025598526,
      "learning_rate": 0.00011441760344415211,
      "loss": 0.089,
      "step": 13437
    },
    {
      "epoch": 1.8833917309039943,
      "grad_norm": 0.28910204768180847,
      "learning_rate": 0.00011440325281033245,
      "loss": 0.1031,
      "step": 13438
    },
    {
      "epoch": 1.883531885073581,
      "grad_norm": 0.7620351910591125,
      "learning_rate": 0.00011438890217651278,
      "loss": 0.0279,
      "step": 13439
    },
    {
      "epoch": 1.8836720392431676,
      "grad_norm": 0.2556101381778717,
      "learning_rate": 0.00011437455154269312,
      "loss": 0.0393,
      "step": 13440
    },
    {
      "epoch": 1.883812193412754,
      "grad_norm": 0.17038565874099731,
      "learning_rate": 0.00011436020090887348,
      "loss": 0.0565,
      "step": 13441
    },
    {
      "epoch": 1.8839523475823405,
      "grad_norm": 0.06055064499378204,
      "learning_rate": 0.0001143458502750538,
      "loss": 0.0137,
      "step": 13442
    },
    {
      "epoch": 1.884092501751927,
      "grad_norm": 0.44050848484039307,
      "learning_rate": 0.00011433149964123415,
      "loss": 0.0429,
      "step": 13443
    },
    {
      "epoch": 1.8842326559215137,
      "grad_norm": 0.5867749452590942,
      "learning_rate": 0.00011431714900741448,
      "loss": 0.1403,
      "step": 13444
    },
    {
      "epoch": 1.8843728100911004,
      "grad_norm": 0.1588454395532608,
      "learning_rate": 0.00011430279837359482,
      "loss": 0.0172,
      "step": 13445
    },
    {
      "epoch": 1.8845129642606868,
      "grad_norm": 0.3927476704120636,
      "learning_rate": 0.00011428844773977517,
      "loss": 0.0399,
      "step": 13446
    },
    {
      "epoch": 1.8846531184302733,
      "grad_norm": 0.26721999049186707,
      "learning_rate": 0.0001142740971059555,
      "loss": 0.0679,
      "step": 13447
    },
    {
      "epoch": 1.8847932725998597,
      "grad_norm": 0.24815945327281952,
      "learning_rate": 0.00011425974647213585,
      "loss": 0.047,
      "step": 13448
    },
    {
      "epoch": 1.8849334267694464,
      "grad_norm": 0.3501009941101074,
      "learning_rate": 0.00011424539583831619,
      "loss": 0.1487,
      "step": 13449
    },
    {
      "epoch": 1.885073580939033,
      "grad_norm": 0.3207671046257019,
      "learning_rate": 0.00011423104520449652,
      "loss": 0.0566,
      "step": 13450
    },
    {
      "epoch": 1.8852137351086196,
      "grad_norm": 0.29027116298675537,
      "learning_rate": 0.00011421669457067686,
      "loss": 0.0892,
      "step": 13451
    },
    {
      "epoch": 1.885353889278206,
      "grad_norm": 0.12084417790174484,
      "learning_rate": 0.0001142023439368572,
      "loss": 0.0141,
      "step": 13452
    },
    {
      "epoch": 1.8854940434477925,
      "grad_norm": 0.555729329586029,
      "learning_rate": 0.00011418799330303754,
      "loss": 0.0427,
      "step": 13453
    },
    {
      "epoch": 1.8856341976173792,
      "grad_norm": 0.1798364371061325,
      "learning_rate": 0.00011417364266921788,
      "loss": 0.0707,
      "step": 13454
    },
    {
      "epoch": 1.8857743517869656,
      "grad_norm": 0.18271052837371826,
      "learning_rate": 0.00011415929203539821,
      "loss": 0.0276,
      "step": 13455
    },
    {
      "epoch": 1.8859145059565523,
      "grad_norm": 0.33206313848495483,
      "learning_rate": 0.00011414494140157856,
      "loss": 0.0507,
      "step": 13456
    },
    {
      "epoch": 1.8860546601261388,
      "grad_norm": 0.2393684685230255,
      "learning_rate": 0.00011413059076775891,
      "loss": 0.0369,
      "step": 13457
    },
    {
      "epoch": 1.8861948142957252,
      "grad_norm": 0.15933044254779816,
      "learning_rate": 0.00011411624013393924,
      "loss": 0.0447,
      "step": 13458
    },
    {
      "epoch": 1.8863349684653117,
      "grad_norm": 0.11679864674806595,
      "learning_rate": 0.00011410188950011958,
      "loss": 0.0239,
      "step": 13459
    },
    {
      "epoch": 1.8864751226348984,
      "grad_norm": 0.2846401035785675,
      "learning_rate": 0.00011408753886629991,
      "loss": 0.0446,
      "step": 13460
    },
    {
      "epoch": 1.886615276804485,
      "grad_norm": 0.33770138025283813,
      "learning_rate": 0.00011407318823248025,
      "loss": 0.0441,
      "step": 13461
    },
    {
      "epoch": 1.8867554309740715,
      "grad_norm": 0.13469266891479492,
      "learning_rate": 0.00011405883759866061,
      "loss": 0.0227,
      "step": 13462
    },
    {
      "epoch": 1.886895585143658,
      "grad_norm": 0.22757545113563538,
      "learning_rate": 0.00011404448696484094,
      "loss": 0.0615,
      "step": 13463
    },
    {
      "epoch": 1.8870357393132444,
      "grad_norm": 0.26594793796539307,
      "learning_rate": 0.00011403013633102128,
      "loss": 0.0311,
      "step": 13464
    },
    {
      "epoch": 1.887175893482831,
      "grad_norm": 0.1401015967130661,
      "learning_rate": 0.00011401578569720162,
      "loss": 0.0369,
      "step": 13465
    },
    {
      "epoch": 1.8873160476524178,
      "grad_norm": 0.40369248390197754,
      "learning_rate": 0.00011400143506338195,
      "loss": 0.0435,
      "step": 13466
    },
    {
      "epoch": 1.8874562018220042,
      "grad_norm": 0.315899133682251,
      "learning_rate": 0.00011398708442956229,
      "loss": 0.0347,
      "step": 13467
    },
    {
      "epoch": 1.8875963559915907,
      "grad_norm": 0.21384172141551971,
      "learning_rate": 0.00011397273379574265,
      "loss": 0.0241,
      "step": 13468
    },
    {
      "epoch": 1.8877365101611772,
      "grad_norm": 0.13499832153320312,
      "learning_rate": 0.00011395838316192298,
      "loss": 0.0485,
      "step": 13469
    },
    {
      "epoch": 1.8878766643307638,
      "grad_norm": 0.504688024520874,
      "learning_rate": 0.00011394403252810332,
      "loss": 0.0651,
      "step": 13470
    },
    {
      "epoch": 1.8880168185003505,
      "grad_norm": 0.2866809070110321,
      "learning_rate": 0.00011392968189428365,
      "loss": 0.056,
      "step": 13471
    },
    {
      "epoch": 1.888156972669937,
      "grad_norm": 0.3693518340587616,
      "learning_rate": 0.00011391533126046399,
      "loss": 0.0274,
      "step": 13472
    },
    {
      "epoch": 1.8882971268395234,
      "grad_norm": 0.3757310211658478,
      "learning_rate": 0.00011390098062664434,
      "loss": 0.042,
      "step": 13473
    },
    {
      "epoch": 1.88843728100911,
      "grad_norm": 0.4082602560520172,
      "learning_rate": 0.00011388662999282467,
      "loss": 0.0278,
      "step": 13474
    },
    {
      "epoch": 1.8885774351786966,
      "grad_norm": 0.1054476946592331,
      "learning_rate": 0.00011387227935900501,
      "loss": 0.0087,
      "step": 13475
    },
    {
      "epoch": 1.8887175893482833,
      "grad_norm": 0.22146852314472198,
      "learning_rate": 0.00011385792872518536,
      "loss": 0.0226,
      "step": 13476
    },
    {
      "epoch": 1.8888577435178697,
      "grad_norm": 0.11555270850658417,
      "learning_rate": 0.00011384357809136569,
      "loss": 0.0043,
      "step": 13477
    },
    {
      "epoch": 1.8889978976874562,
      "grad_norm": 0.27244818210601807,
      "learning_rate": 0.00011382922745754604,
      "loss": 0.0699,
      "step": 13478
    },
    {
      "epoch": 1.8891380518570426,
      "grad_norm": 0.2782879173755646,
      "learning_rate": 0.00011381487682372637,
      "loss": 0.026,
      "step": 13479
    },
    {
      "epoch": 1.8892782060266293,
      "grad_norm": 0.21919305622577667,
      "learning_rate": 0.00011380052618990671,
      "loss": 0.0188,
      "step": 13480
    },
    {
      "epoch": 1.8894183601962158,
      "grad_norm": 0.9740397930145264,
      "learning_rate": 0.00011378617555608705,
      "loss": 0.1612,
      "step": 13481
    },
    {
      "epoch": 1.8895585143658025,
      "grad_norm": 1.249995470046997,
      "learning_rate": 0.00011377182492226738,
      "loss": 0.0655,
      "step": 13482
    },
    {
      "epoch": 1.889698668535389,
      "grad_norm": 0.8391216993331909,
      "learning_rate": 0.00011375747428844772,
      "loss": 0.0429,
      "step": 13483
    },
    {
      "epoch": 1.8898388227049754,
      "grad_norm": 0.7672373652458191,
      "learning_rate": 0.00011374312365462808,
      "loss": 0.1755,
      "step": 13484
    },
    {
      "epoch": 1.889978976874562,
      "grad_norm": 0.7418503761291504,
      "learning_rate": 0.00011372877302080841,
      "loss": 0.1489,
      "step": 13485
    },
    {
      "epoch": 1.8901191310441485,
      "grad_norm": 0.393179327249527,
      "learning_rate": 0.00011371442238698875,
      "loss": 0.0773,
      "step": 13486
    },
    {
      "epoch": 1.8902592852137352,
      "grad_norm": 0.17001232504844666,
      "learning_rate": 0.00011370007175316908,
      "loss": 0.019,
      "step": 13487
    },
    {
      "epoch": 1.8903994393833217,
      "grad_norm": 0.1701052486896515,
      "learning_rate": 0.00011368572111934942,
      "loss": 0.0182,
      "step": 13488
    },
    {
      "epoch": 1.8905395935529081,
      "grad_norm": 0.4775990843772888,
      "learning_rate": 0.00011367137048552978,
      "loss": 0.0445,
      "step": 13489
    },
    {
      "epoch": 1.8906797477224946,
      "grad_norm": 0.21463489532470703,
      "learning_rate": 0.0001136570198517101,
      "loss": 0.029,
      "step": 13490
    },
    {
      "epoch": 1.8908199018920813,
      "grad_norm": 0.31704986095428467,
      "learning_rate": 0.00011364266921789045,
      "loss": 0.0432,
      "step": 13491
    },
    {
      "epoch": 1.890960056061668,
      "grad_norm": 0.21317845582962036,
      "learning_rate": 0.00011362831858407079,
      "loss": 0.0328,
      "step": 13492
    },
    {
      "epoch": 1.8911002102312544,
      "grad_norm": 0.6330044269561768,
      "learning_rate": 0.00011361396795025112,
      "loss": 0.0722,
      "step": 13493
    },
    {
      "epoch": 1.8912403644008409,
      "grad_norm": 0.41525593400001526,
      "learning_rate": 0.00011359961731643147,
      "loss": 0.1009,
      "step": 13494
    },
    {
      "epoch": 1.8913805185704273,
      "grad_norm": 0.3246195614337921,
      "learning_rate": 0.0001135852666826118,
      "loss": 0.0593,
      "step": 13495
    },
    {
      "epoch": 1.891520672740014,
      "grad_norm": 0.4455593526363373,
      "learning_rate": 0.00011357091604879214,
      "loss": 0.052,
      "step": 13496
    },
    {
      "epoch": 1.8916608269096007,
      "grad_norm": 0.4324365556240082,
      "learning_rate": 0.00011355656541497249,
      "loss": 0.0558,
      "step": 13497
    },
    {
      "epoch": 1.8918009810791871,
      "grad_norm": 0.2009086161851883,
      "learning_rate": 0.00011354221478115282,
      "loss": 0.0356,
      "step": 13498
    },
    {
      "epoch": 1.8919411352487736,
      "grad_norm": 0.29308587312698364,
      "learning_rate": 0.00011352786414733316,
      "loss": 0.0215,
      "step": 13499
    },
    {
      "epoch": 1.89208128941836,
      "grad_norm": 0.23277276754379272,
      "learning_rate": 0.00011351351351351351,
      "loss": 0.0932,
      "step": 13500
    },
    {
      "epoch": 1.8922214435879467,
      "grad_norm": 0.40751221776008606,
      "learning_rate": 0.00011349916287969384,
      "loss": 0.1203,
      "step": 13501
    },
    {
      "epoch": 1.8923615977575334,
      "grad_norm": 0.4604926109313965,
      "learning_rate": 0.00011348481224587418,
      "loss": 0.0525,
      "step": 13502
    },
    {
      "epoch": 1.89250175192712,
      "grad_norm": 0.29643258452415466,
      "learning_rate": 0.00011347046161205453,
      "loss": 0.019,
      "step": 13503
    },
    {
      "epoch": 1.8926419060967064,
      "grad_norm": 0.1702430248260498,
      "learning_rate": 0.00011345611097823485,
      "loss": 0.0286,
      "step": 13504
    },
    {
      "epoch": 1.8927820602662928,
      "grad_norm": 0.21273918449878693,
      "learning_rate": 0.00011344176034441521,
      "loss": 0.0557,
      "step": 13505
    },
    {
      "epoch": 1.8929222144358795,
      "grad_norm": 0.3799642026424408,
      "learning_rate": 0.00011342740971059554,
      "loss": 0.0455,
      "step": 13506
    },
    {
      "epoch": 1.8930623686054662,
      "grad_norm": 0.18924139440059662,
      "learning_rate": 0.00011341305907677588,
      "loss": 0.0492,
      "step": 13507
    },
    {
      "epoch": 1.8932025227750526,
      "grad_norm": 0.2973273992538452,
      "learning_rate": 0.00011339870844295622,
      "loss": 0.046,
      "step": 13508
    },
    {
      "epoch": 1.893342676944639,
      "grad_norm": 0.22490397095680237,
      "learning_rate": 0.00011338435780913655,
      "loss": 0.0166,
      "step": 13509
    },
    {
      "epoch": 1.8934828311142256,
      "grad_norm": 0.13594576716423035,
      "learning_rate": 0.00011337000717531691,
      "loss": 0.0196,
      "step": 13510
    },
    {
      "epoch": 1.8936229852838122,
      "grad_norm": 0.2492450475692749,
      "learning_rate": 0.00011335565654149725,
      "loss": 0.0521,
      "step": 13511
    },
    {
      "epoch": 1.8937631394533987,
      "grad_norm": 0.17327338457107544,
      "learning_rate": 0.00011334130590767758,
      "loss": 0.023,
      "step": 13512
    },
    {
      "epoch": 1.8939032936229854,
      "grad_norm": 0.49359628558158875,
      "learning_rate": 0.00011332695527385792,
      "loss": 0.0612,
      "step": 13513
    },
    {
      "epoch": 1.8940434477925718,
      "grad_norm": 0.2705613672733307,
      "learning_rate": 0.00011331260464003825,
      "loss": 0.0758,
      "step": 13514
    },
    {
      "epoch": 1.8941836019621583,
      "grad_norm": 0.4242211878299713,
      "learning_rate": 0.00011329825400621859,
      "loss": 0.018,
      "step": 13515
    },
    {
      "epoch": 1.894323756131745,
      "grad_norm": 0.4368789792060852,
      "learning_rate": 0.00011328390337239895,
      "loss": 0.066,
      "step": 13516
    },
    {
      "epoch": 1.8944639103013314,
      "grad_norm": 0.16248489916324615,
      "learning_rate": 0.00011326955273857927,
      "loss": 0.0131,
      "step": 13517
    },
    {
      "epoch": 1.8946040644709181,
      "grad_norm": 0.04456055909395218,
      "learning_rate": 0.00011325520210475962,
      "loss": 0.0053,
      "step": 13518
    },
    {
      "epoch": 1.8947442186405046,
      "grad_norm": 0.33838513493537903,
      "learning_rate": 0.00011324085147093996,
      "loss": 0.0891,
      "step": 13519
    },
    {
      "epoch": 1.894884372810091,
      "grad_norm": 0.1119290366768837,
      "learning_rate": 0.00011322650083712029,
      "loss": 0.0262,
      "step": 13520
    },
    {
      "epoch": 1.8950245269796775,
      "grad_norm": 0.5778630375862122,
      "learning_rate": 0.00011321215020330064,
      "loss": 0.058,
      "step": 13521
    },
    {
      "epoch": 1.8951646811492642,
      "grad_norm": 0.309861958026886,
      "learning_rate": 0.00011319779956948097,
      "loss": 0.036,
      "step": 13522
    },
    {
      "epoch": 1.8953048353188509,
      "grad_norm": 0.19732381403446198,
      "learning_rate": 0.00011318344893566131,
      "loss": 0.0314,
      "step": 13523
    },
    {
      "epoch": 1.8954449894884373,
      "grad_norm": 0.2120002657175064,
      "learning_rate": 0.00011316909830184166,
      "loss": 0.0237,
      "step": 13524
    },
    {
      "epoch": 1.8955851436580238,
      "grad_norm": 0.11605087667703629,
      "learning_rate": 0.00011315474766802198,
      "loss": 0.0248,
      "step": 13525
    },
    {
      "epoch": 1.8957252978276102,
      "grad_norm": 0.24174977838993073,
      "learning_rate": 0.00011314039703420234,
      "loss": 0.0679,
      "step": 13526
    },
    {
      "epoch": 1.895865451997197,
      "grad_norm": 0.26389169692993164,
      "learning_rate": 0.00011312604640038268,
      "loss": 0.0431,
      "step": 13527
    },
    {
      "epoch": 1.8960056061667836,
      "grad_norm": 0.37816429138183594,
      "learning_rate": 0.00011311169576656301,
      "loss": 0.0553,
      "step": 13528
    },
    {
      "epoch": 1.89614576033637,
      "grad_norm": 0.23505020141601562,
      "learning_rate": 0.00011309734513274335,
      "loss": 0.0955,
      "step": 13529
    },
    {
      "epoch": 1.8962859145059565,
      "grad_norm": 0.5466634035110474,
      "learning_rate": 0.00011308299449892368,
      "loss": 0.1658,
      "step": 13530
    },
    {
      "epoch": 1.896426068675543,
      "grad_norm": 0.4026920199394226,
      "learning_rate": 0.00011306864386510404,
      "loss": 0.0173,
      "step": 13531
    },
    {
      "epoch": 1.8965662228451297,
      "grad_norm": 0.3126298785209656,
      "learning_rate": 0.00011305429323128438,
      "loss": 0.0472,
      "step": 13532
    },
    {
      "epoch": 1.8967063770147163,
      "grad_norm": 1.9589697122573853,
      "learning_rate": 0.00011303994259746471,
      "loss": 0.0255,
      "step": 13533
    },
    {
      "epoch": 1.8968465311843028,
      "grad_norm": 0.16841769218444824,
      "learning_rate": 0.00011302559196364505,
      "loss": 0.0101,
      "step": 13534
    },
    {
      "epoch": 1.8969866853538893,
      "grad_norm": 1.3485615253448486,
      "learning_rate": 0.00011301124132982539,
      "loss": 0.1446,
      "step": 13535
    },
    {
      "epoch": 1.8971268395234757,
      "grad_norm": 0.12767793238162994,
      "learning_rate": 0.00011299689069600572,
      "loss": 0.0145,
      "step": 13536
    },
    {
      "epoch": 1.8972669936930624,
      "grad_norm": 0.33895644545555115,
      "learning_rate": 0.00011298254006218608,
      "loss": 0.0339,
      "step": 13537
    },
    {
      "epoch": 1.897407147862649,
      "grad_norm": 0.26316457986831665,
      "learning_rate": 0.0001129681894283664,
      "loss": 0.0388,
      "step": 13538
    },
    {
      "epoch": 1.8975473020322355,
      "grad_norm": 0.3911347985267639,
      "learning_rate": 0.00011295383879454675,
      "loss": 0.0338,
      "step": 13539
    },
    {
      "epoch": 1.897687456201822,
      "grad_norm": 0.1672990620136261,
      "learning_rate": 0.00011293948816072709,
      "loss": 0.0158,
      "step": 13540
    },
    {
      "epoch": 1.8978276103714085,
      "grad_norm": 0.9989466667175293,
      "learning_rate": 0.00011292513752690742,
      "loss": 0.0723,
      "step": 13541
    },
    {
      "epoch": 1.8979677645409951,
      "grad_norm": 0.2344716340303421,
      "learning_rate": 0.00011291078689308777,
      "loss": 0.0172,
      "step": 13542
    },
    {
      "epoch": 1.8981079187105816,
      "grad_norm": 0.5946636199951172,
      "learning_rate": 0.00011289643625926812,
      "loss": 0.1419,
      "step": 13543
    },
    {
      "epoch": 1.8982480728801683,
      "grad_norm": 0.21360154449939728,
      "learning_rate": 0.00011288208562544844,
      "loss": 0.0428,
      "step": 13544
    },
    {
      "epoch": 1.8983882270497547,
      "grad_norm": 0.2801905572414398,
      "learning_rate": 0.00011286773499162879,
      "loss": 0.0217,
      "step": 13545
    },
    {
      "epoch": 1.8985283812193412,
      "grad_norm": 0.3123037815093994,
      "learning_rate": 0.00011285338435780913,
      "loss": 0.0437,
      "step": 13546
    },
    {
      "epoch": 1.8986685353889277,
      "grad_norm": 0.2466706484556198,
      "learning_rate": 0.00011283903372398947,
      "loss": 0.0712,
      "step": 13547
    },
    {
      "epoch": 1.8988086895585143,
      "grad_norm": 0.25357887148857117,
      "learning_rate": 0.00011282468309016981,
      "loss": 0.0434,
      "step": 13548
    },
    {
      "epoch": 1.898948843728101,
      "grad_norm": 0.4434691071510315,
      "learning_rate": 0.00011281033245635014,
      "loss": 0.0526,
      "step": 13549
    },
    {
      "epoch": 1.8990889978976875,
      "grad_norm": 0.22153626382350922,
      "learning_rate": 0.00011279598182253048,
      "loss": 0.0434,
      "step": 13550
    },
    {
      "epoch": 1.899229152067274,
      "grad_norm": 0.24759724736213684,
      "learning_rate": 0.00011278163118871083,
      "loss": 0.043,
      "step": 13551
    },
    {
      "epoch": 1.8993693062368604,
      "grad_norm": 0.21771380305290222,
      "learning_rate": 0.00011276728055489115,
      "loss": 0.0605,
      "step": 13552
    },
    {
      "epoch": 1.899509460406447,
      "grad_norm": 0.11064166575670242,
      "learning_rate": 0.00011275292992107151,
      "loss": 0.011,
      "step": 13553
    },
    {
      "epoch": 1.8996496145760338,
      "grad_norm": 0.48809659481048584,
      "learning_rate": 0.00011273857928725185,
      "loss": 0.0526,
      "step": 13554
    },
    {
      "epoch": 1.8997897687456202,
      "grad_norm": 0.8010419607162476,
      "learning_rate": 0.00011272422865343218,
      "loss": 0.0903,
      "step": 13555
    },
    {
      "epoch": 1.8999299229152067,
      "grad_norm": 0.1474427729845047,
      "learning_rate": 0.00011270987801961252,
      "loss": 0.0155,
      "step": 13556
    },
    {
      "epoch": 1.9000700770847931,
      "grad_norm": 0.25691649317741394,
      "learning_rate": 0.00011269552738579285,
      "loss": 0.0327,
      "step": 13557
    },
    {
      "epoch": 1.9002102312543798,
      "grad_norm": 0.2793200612068176,
      "learning_rate": 0.0001126811767519732,
      "loss": 0.0297,
      "step": 13558
    },
    {
      "epoch": 1.9003503854239665,
      "grad_norm": 0.28630396723747253,
      "learning_rate": 0.00011266682611815355,
      "loss": 0.043,
      "step": 13559
    },
    {
      "epoch": 1.900490539593553,
      "grad_norm": 0.4067237675189972,
      "learning_rate": 0.00011265247548433388,
      "loss": 0.0815,
      "step": 13560
    },
    {
      "epoch": 1.9006306937631394,
      "grad_norm": 0.0867854431271553,
      "learning_rate": 0.00011263812485051422,
      "loss": 0.0218,
      "step": 13561
    },
    {
      "epoch": 1.9007708479327259,
      "grad_norm": 0.40777266025543213,
      "learning_rate": 0.00011262377421669456,
      "loss": 0.0615,
      "step": 13562
    },
    {
      "epoch": 1.9009110021023126,
      "grad_norm": 0.3923787474632263,
      "learning_rate": 0.0001126094235828749,
      "loss": 0.0323,
      "step": 13563
    },
    {
      "epoch": 1.9010511562718992,
      "grad_norm": 0.2128615528345108,
      "learning_rate": 0.00011259507294905525,
      "loss": 0.0179,
      "step": 13564
    },
    {
      "epoch": 1.9011913104414857,
      "grad_norm": 0.2645399570465088,
      "learning_rate": 0.00011258072231523557,
      "loss": 0.0585,
      "step": 13565
    },
    {
      "epoch": 1.9013314646110722,
      "grad_norm": 0.20695510506629944,
      "learning_rate": 0.00011256637168141592,
      "loss": 0.0175,
      "step": 13566
    },
    {
      "epoch": 1.9014716187806586,
      "grad_norm": 0.22354115545749664,
      "learning_rate": 0.00011255202104759626,
      "loss": 0.0359,
      "step": 13567
    },
    {
      "epoch": 1.9016117729502453,
      "grad_norm": 0.2357221394777298,
      "learning_rate": 0.00011253767041377659,
      "loss": 0.0274,
      "step": 13568
    },
    {
      "epoch": 1.9017519271198318,
      "grad_norm": 0.2637893855571747,
      "learning_rate": 0.00011252331977995694,
      "loss": 0.0269,
      "step": 13569
    },
    {
      "epoch": 1.9018920812894184,
      "grad_norm": 0.19776253402233124,
      "learning_rate": 0.00011250896914613728,
      "loss": 0.0238,
      "step": 13570
    },
    {
      "epoch": 1.902032235459005,
      "grad_norm": 0.18680135905742645,
      "learning_rate": 0.00011249461851231761,
      "loss": 0.0204,
      "step": 13571
    },
    {
      "epoch": 1.9021723896285914,
      "grad_norm": 0.3026014566421509,
      "learning_rate": 0.00011248026787849796,
      "loss": 0.0148,
      "step": 13572
    },
    {
      "epoch": 1.902312543798178,
      "grad_norm": 0.32537463307380676,
      "learning_rate": 0.00011246591724467828,
      "loss": 0.0351,
      "step": 13573
    },
    {
      "epoch": 1.9024526979677645,
      "grad_norm": 0.9825728535652161,
      "learning_rate": 0.00011245156661085864,
      "loss": 0.1462,
      "step": 13574
    },
    {
      "epoch": 1.9025928521373512,
      "grad_norm": 0.9745528697967529,
      "learning_rate": 0.00011243721597703898,
      "loss": 0.0425,
      "step": 13575
    },
    {
      "epoch": 1.9027330063069376,
      "grad_norm": 0.5819140076637268,
      "learning_rate": 0.00011242286534321931,
      "loss": 0.0507,
      "step": 13576
    },
    {
      "epoch": 1.902873160476524,
      "grad_norm": 0.3542453944683075,
      "learning_rate": 0.00011240851470939965,
      "loss": 0.0719,
      "step": 13577
    },
    {
      "epoch": 1.9030133146461106,
      "grad_norm": 0.3804064393043518,
      "learning_rate": 0.00011239416407558,
      "loss": 0.0441,
      "step": 13578
    },
    {
      "epoch": 1.9031534688156972,
      "grad_norm": 0.6226317882537842,
      "learning_rate": 0.00011237981344176034,
      "loss": 0.1228,
      "step": 13579
    },
    {
      "epoch": 1.903293622985284,
      "grad_norm": 0.2324400395154953,
      "learning_rate": 0.00011236546280794068,
      "loss": 0.0431,
      "step": 13580
    },
    {
      "epoch": 1.9034337771548704,
      "grad_norm": 0.9894178509712219,
      "learning_rate": 0.00011235111217412102,
      "loss": 0.0332,
      "step": 13581
    },
    {
      "epoch": 1.9035739313244568,
      "grad_norm": 1.1990857124328613,
      "learning_rate": 0.00011233676154030135,
      "loss": 0.2815,
      "step": 13582
    },
    {
      "epoch": 1.9037140854940433,
      "grad_norm": 0.1801864057779312,
      "learning_rate": 0.00011232241090648169,
      "loss": 0.01,
      "step": 13583
    },
    {
      "epoch": 1.90385423966363,
      "grad_norm": 0.24877943098545074,
      "learning_rate": 0.00011230806027266202,
      "loss": 0.0165,
      "step": 13584
    },
    {
      "epoch": 1.9039943938332167,
      "grad_norm": 0.37404167652130127,
      "learning_rate": 0.00011229370963884238,
      "loss": 0.0305,
      "step": 13585
    },
    {
      "epoch": 1.9041345480028031,
      "grad_norm": 0.41422197222709656,
      "learning_rate": 0.00011227935900502272,
      "loss": 0.0691,
      "step": 13586
    },
    {
      "epoch": 1.9042747021723896,
      "grad_norm": 0.20202675461769104,
      "learning_rate": 0.00011226500837120305,
      "loss": 0.0298,
      "step": 13587
    },
    {
      "epoch": 1.904414856341976,
      "grad_norm": 0.17427481710910797,
      "learning_rate": 0.00011225065773738339,
      "loss": 0.0199,
      "step": 13588
    },
    {
      "epoch": 1.9045550105115627,
      "grad_norm": 0.36209309101104736,
      "learning_rate": 0.00011223630710356374,
      "loss": 0.0555,
      "step": 13589
    },
    {
      "epoch": 1.9046951646811494,
      "grad_norm": 0.1634976714849472,
      "learning_rate": 0.00011222195646974407,
      "loss": 0.0311,
      "step": 13590
    },
    {
      "epoch": 1.9048353188507359,
      "grad_norm": 0.358043372631073,
      "learning_rate": 0.00011220760583592441,
      "loss": 0.0421,
      "step": 13591
    },
    {
      "epoch": 1.9049754730203223,
      "grad_norm": 0.5043150782585144,
      "learning_rate": 0.00011219325520210474,
      "loss": 0.0298,
      "step": 13592
    },
    {
      "epoch": 1.9051156271899088,
      "grad_norm": 0.1375071406364441,
      "learning_rate": 0.00011217890456828509,
      "loss": 0.023,
      "step": 13593
    },
    {
      "epoch": 1.9052557813594955,
      "grad_norm": 0.17262771725654602,
      "learning_rate": 0.00011216455393446543,
      "loss": 0.0472,
      "step": 13594
    },
    {
      "epoch": 1.9053959355290822,
      "grad_norm": 0.6882639527320862,
      "learning_rate": 0.00011215020330064577,
      "loss": 0.0569,
      "step": 13595
    },
    {
      "epoch": 1.9055360896986686,
      "grad_norm": 0.23287679255008698,
      "learning_rate": 0.00011213585266682611,
      "loss": 0.04,
      "step": 13596
    },
    {
      "epoch": 1.905676243868255,
      "grad_norm": 0.19731824100017548,
      "learning_rate": 0.00011212150203300645,
      "loss": 0.0376,
      "step": 13597
    },
    {
      "epoch": 1.9058163980378415,
      "grad_norm": 0.5704525113105774,
      "learning_rate": 0.00011210715139918678,
      "loss": 0.0499,
      "step": 13598
    },
    {
      "epoch": 1.9059565522074282,
      "grad_norm": 0.23459932208061218,
      "learning_rate": 0.00011209280076536712,
      "loss": 0.0594,
      "step": 13599
    },
    {
      "epoch": 1.9060967063770147,
      "grad_norm": 0.08259747177362442,
      "learning_rate": 0.00011207845013154745,
      "loss": 0.0108,
      "step": 13600
    },
    {
      "epoch": 1.9062368605466014,
      "grad_norm": 0.2248906046152115,
      "learning_rate": 0.00011206409949772781,
      "loss": 0.0234,
      "step": 13601
    },
    {
      "epoch": 1.9063770147161878,
      "grad_norm": 0.2571329176425934,
      "learning_rate": 0.00011204974886390815,
      "loss": 0.0389,
      "step": 13602
    },
    {
      "epoch": 1.9065171688857743,
      "grad_norm": 0.33607983589172363,
      "learning_rate": 0.00011203539823008848,
      "loss": 0.0716,
      "step": 13603
    },
    {
      "epoch": 1.9066573230553607,
      "grad_norm": 0.06593748182058334,
      "learning_rate": 0.00011202104759626882,
      "loss": 0.0383,
      "step": 13604
    },
    {
      "epoch": 1.9067974772249474,
      "grad_norm": 0.21620528399944305,
      "learning_rate": 0.00011200669696244918,
      "loss": 0.021,
      "step": 13605
    },
    {
      "epoch": 1.906937631394534,
      "grad_norm": 0.4865710735321045,
      "learning_rate": 0.0001119923463286295,
      "loss": 0.1057,
      "step": 13606
    },
    {
      "epoch": 1.9070777855641206,
      "grad_norm": 0.44378310441970825,
      "learning_rate": 0.00011197799569480985,
      "loss": 0.0803,
      "step": 13607
    },
    {
      "epoch": 1.907217939733707,
      "grad_norm": 0.18585270643234253,
      "learning_rate": 0.00011196364506099018,
      "loss": 0.0467,
      "step": 13608
    },
    {
      "epoch": 1.9073580939032935,
      "grad_norm": 0.2271266132593155,
      "learning_rate": 0.00011194929442717052,
      "loss": 0.0272,
      "step": 13609
    },
    {
      "epoch": 1.9074982480728802,
      "grad_norm": 0.14281751215457916,
      "learning_rate": 0.00011193494379335086,
      "loss": 0.0116,
      "step": 13610
    },
    {
      "epoch": 1.9076384022424668,
      "grad_norm": 0.2914571166038513,
      "learning_rate": 0.0001119205931595312,
      "loss": 0.0499,
      "step": 13611
    },
    {
      "epoch": 1.9077785564120533,
      "grad_norm": 0.21913349628448486,
      "learning_rate": 0.00011190624252571155,
      "loss": 0.0742,
      "step": 13612
    },
    {
      "epoch": 1.9079187105816398,
      "grad_norm": 0.264310359954834,
      "learning_rate": 0.00011189189189189189,
      "loss": 0.0588,
      "step": 13613
    },
    {
      "epoch": 1.9080588647512262,
      "grad_norm": 0.3651709258556366,
      "learning_rate": 0.00011187754125807222,
      "loss": 0.0341,
      "step": 13614
    },
    {
      "epoch": 1.908199018920813,
      "grad_norm": 0.32444310188293457,
      "learning_rate": 0.00011186319062425256,
      "loss": 0.0593,
      "step": 13615
    },
    {
      "epoch": 1.9083391730903996,
      "grad_norm": 0.09272079169750214,
      "learning_rate": 0.00011184883999043291,
      "loss": 0.0112,
      "step": 13616
    },
    {
      "epoch": 1.908479327259986,
      "grad_norm": 0.21247078478336334,
      "learning_rate": 0.00011183448935661324,
      "loss": 0.0378,
      "step": 13617
    },
    {
      "epoch": 1.9086194814295725,
      "grad_norm": 0.27204954624176025,
      "learning_rate": 0.00011182013872279358,
      "loss": 0.0559,
      "step": 13618
    },
    {
      "epoch": 1.908759635599159,
      "grad_norm": 0.5331076383590698,
      "learning_rate": 0.00011180578808897391,
      "loss": 0.0873,
      "step": 13619
    },
    {
      "epoch": 1.9088997897687456,
      "grad_norm": 0.20795750617980957,
      "learning_rate": 0.00011179143745515425,
      "loss": 0.0291,
      "step": 13620
    },
    {
      "epoch": 1.9090399439383323,
      "grad_norm": 0.3403012156486511,
      "learning_rate": 0.00011177708682133461,
      "loss": 0.041,
      "step": 13621
    },
    {
      "epoch": 1.9091800981079188,
      "grad_norm": 0.3852470815181732,
      "learning_rate": 0.00011176273618751494,
      "loss": 0.0225,
      "step": 13622
    },
    {
      "epoch": 1.9093202522775052,
      "grad_norm": 0.46531111001968384,
      "learning_rate": 0.00011174838555369528,
      "loss": 0.0907,
      "step": 13623
    },
    {
      "epoch": 1.9094604064470917,
      "grad_norm": 0.28256070613861084,
      "learning_rate": 0.00011173403491987562,
      "loss": 0.0856,
      "step": 13624
    },
    {
      "epoch": 1.9096005606166784,
      "grad_norm": 0.2915250062942505,
      "learning_rate": 0.00011171968428605595,
      "loss": 0.0394,
      "step": 13625
    },
    {
      "epoch": 1.9097407147862648,
      "grad_norm": 0.3901909589767456,
      "learning_rate": 0.0001117053336522363,
      "loss": 0.0321,
      "step": 13626
    },
    {
      "epoch": 1.9098808689558515,
      "grad_norm": 0.1341322511434555,
      "learning_rate": 0.00011169098301841664,
      "loss": 0.0363,
      "step": 13627
    },
    {
      "epoch": 1.910021023125438,
      "grad_norm": 0.6977105736732483,
      "learning_rate": 0.00011167663238459698,
      "loss": 0.0528,
      "step": 13628
    },
    {
      "epoch": 1.9101611772950244,
      "grad_norm": 0.4452248215675354,
      "learning_rate": 0.00011166228175077732,
      "loss": 0.0687,
      "step": 13629
    },
    {
      "epoch": 1.9103013314646111,
      "grad_norm": 1.505567193031311,
      "learning_rate": 0.00011164793111695765,
      "loss": 0.0767,
      "step": 13630
    },
    {
      "epoch": 1.9104414856341976,
      "grad_norm": 1.6543866395950317,
      "learning_rate": 0.00011163358048313799,
      "loss": 0.0266,
      "step": 13631
    },
    {
      "epoch": 1.9105816398037843,
      "grad_norm": 0.6628902554512024,
      "learning_rate": 0.00011161922984931835,
      "loss": 0.0967,
      "step": 13632
    },
    {
      "epoch": 1.9107217939733707,
      "grad_norm": 1.0607616901397705,
      "learning_rate": 0.00011160487921549868,
      "loss": 0.0571,
      "step": 13633
    },
    {
      "epoch": 1.9108619481429572,
      "grad_norm": 0.14026914536952972,
      "learning_rate": 0.00011159052858167902,
      "loss": 0.0088,
      "step": 13634
    },
    {
      "epoch": 1.9110021023125436,
      "grad_norm": 0.04002803936600685,
      "learning_rate": 0.00011157617794785935,
      "loss": 0.003,
      "step": 13635
    },
    {
      "epoch": 1.9111422564821303,
      "grad_norm": 0.2342625856399536,
      "learning_rate": 0.00011156182731403969,
      "loss": 0.0309,
      "step": 13636
    },
    {
      "epoch": 1.911282410651717,
      "grad_norm": 0.11075586080551147,
      "learning_rate": 0.00011154747668022004,
      "loss": 0.0248,
      "step": 13637
    },
    {
      "epoch": 1.9114225648213035,
      "grad_norm": 0.44621115922927856,
      "learning_rate": 0.00011153312604640037,
      "loss": 0.0383,
      "step": 13638
    },
    {
      "epoch": 1.91156271899089,
      "grad_norm": 0.1257069706916809,
      "learning_rate": 0.00011151877541258071,
      "loss": 0.0061,
      "step": 13639
    },
    {
      "epoch": 1.9117028731604764,
      "grad_norm": 0.4938739240169525,
      "learning_rate": 0.00011150442477876106,
      "loss": 0.031,
      "step": 13640
    },
    {
      "epoch": 1.911843027330063,
      "grad_norm": 0.21167363226413727,
      "learning_rate": 0.00011149007414494139,
      "loss": 0.0149,
      "step": 13641
    },
    {
      "epoch": 1.9119831814996497,
      "grad_norm": 0.22203145921230316,
      "learning_rate": 0.00011147572351112174,
      "loss": 0.0533,
      "step": 13642
    },
    {
      "epoch": 1.9121233356692362,
      "grad_norm": 0.24504150450229645,
      "learning_rate": 0.00011146137287730207,
      "loss": 0.0374,
      "step": 13643
    },
    {
      "epoch": 1.9122634898388227,
      "grad_norm": 0.1304401010274887,
      "learning_rate": 0.00011144702224348241,
      "loss": 0.0117,
      "step": 13644
    },
    {
      "epoch": 1.9124036440084091,
      "grad_norm": 0.34737083315849304,
      "learning_rate": 0.00011143267160966275,
      "loss": 0.0373,
      "step": 13645
    },
    {
      "epoch": 1.9125437981779958,
      "grad_norm": 0.2828051745891571,
      "learning_rate": 0.00011141832097584308,
      "loss": 0.0247,
      "step": 13646
    },
    {
      "epoch": 1.9126839523475825,
      "grad_norm": 0.3984811305999756,
      "learning_rate": 0.00011140397034202342,
      "loss": 0.0302,
      "step": 13647
    },
    {
      "epoch": 1.912824106517169,
      "grad_norm": 0.2794347405433655,
      "learning_rate": 0.00011138961970820378,
      "loss": 0.0946,
      "step": 13648
    },
    {
      "epoch": 1.9129642606867554,
      "grad_norm": 0.2458021193742752,
      "learning_rate": 0.00011137526907438411,
      "loss": 0.0353,
      "step": 13649
    },
    {
      "epoch": 1.9131044148563419,
      "grad_norm": 0.24065767228603363,
      "learning_rate": 0.00011136091844056445,
      "loss": 0.0148,
      "step": 13650
    },
    {
      "epoch": 1.9132445690259285,
      "grad_norm": 0.20720089972019196,
      "learning_rate": 0.00011134656780674479,
      "loss": 0.0524,
      "step": 13651
    },
    {
      "epoch": 1.9133847231955152,
      "grad_norm": 0.36210453510284424,
      "learning_rate": 0.00011133221717292512,
      "loss": 0.0519,
      "step": 13652
    },
    {
      "epoch": 1.9135248773651017,
      "grad_norm": 0.23472703993320465,
      "learning_rate": 0.00011131786653910548,
      "loss": 0.0205,
      "step": 13653
    },
    {
      "epoch": 1.9136650315346881,
      "grad_norm": 0.17534111440181732,
      "learning_rate": 0.0001113035159052858,
      "loss": 0.0204,
      "step": 13654
    },
    {
      "epoch": 1.9138051857042746,
      "grad_norm": 0.1530742198228836,
      "learning_rate": 0.00011128916527146615,
      "loss": 0.0175,
      "step": 13655
    },
    {
      "epoch": 1.9139453398738613,
      "grad_norm": 0.24701647460460663,
      "learning_rate": 0.00011127481463764649,
      "loss": 0.016,
      "step": 13656
    },
    {
      "epoch": 1.9140854940434477,
      "grad_norm": 0.2545458972454071,
      "learning_rate": 0.00011126046400382682,
      "loss": 0.0806,
      "step": 13657
    },
    {
      "epoch": 1.9142256482130344,
      "grad_norm": 0.4718727767467499,
      "learning_rate": 0.00011124611337000717,
      "loss": 0.0489,
      "step": 13658
    },
    {
      "epoch": 1.9143658023826209,
      "grad_norm": 0.7278096675872803,
      "learning_rate": 0.00011123176273618752,
      "loss": 0.0858,
      "step": 13659
    },
    {
      "epoch": 1.9145059565522073,
      "grad_norm": 0.08965340256690979,
      "learning_rate": 0.00011121741210236784,
      "loss": 0.0098,
      "step": 13660
    },
    {
      "epoch": 1.914646110721794,
      "grad_norm": 0.211187481880188,
      "learning_rate": 0.00011120306146854819,
      "loss": 0.014,
      "step": 13661
    },
    {
      "epoch": 1.9147862648913805,
      "grad_norm": 0.520647406578064,
      "learning_rate": 0.00011118871083472852,
      "loss": 0.1383,
      "step": 13662
    },
    {
      "epoch": 1.9149264190609672,
      "grad_norm": 0.7984415292739868,
      "learning_rate": 0.00011117436020090886,
      "loss": 0.0993,
      "step": 13663
    },
    {
      "epoch": 1.9150665732305536,
      "grad_norm": 0.07336806505918503,
      "learning_rate": 0.00011116000956708921,
      "loss": 0.0068,
      "step": 13664
    },
    {
      "epoch": 1.91520672740014,
      "grad_norm": 0.25393521785736084,
      "learning_rate": 0.00011114565893326954,
      "loss": 0.0699,
      "step": 13665
    },
    {
      "epoch": 1.9153468815697265,
      "grad_norm": 0.10439901798963547,
      "learning_rate": 0.00011113130829944988,
      "loss": 0.0105,
      "step": 13666
    },
    {
      "epoch": 1.9154870357393132,
      "grad_norm": 0.32841578125953674,
      "learning_rate": 0.00011111695766563023,
      "loss": 0.0249,
      "step": 13667
    },
    {
      "epoch": 1.9156271899089,
      "grad_norm": 0.6405636668205261,
      "learning_rate": 0.00011110260703181055,
      "loss": 0.0366,
      "step": 13668
    },
    {
      "epoch": 1.9157673440784864,
      "grad_norm": 0.6857269406318665,
      "learning_rate": 0.00011108825639799091,
      "loss": 0.0974,
      "step": 13669
    },
    {
      "epoch": 1.9159074982480728,
      "grad_norm": 0.28608301281929016,
      "learning_rate": 0.00011107390576417124,
      "loss": 0.0925,
      "step": 13670
    },
    {
      "epoch": 1.9160476524176593,
      "grad_norm": 0.862321138381958,
      "learning_rate": 0.00011105955513035158,
      "loss": 0.0852,
      "step": 13671
    },
    {
      "epoch": 1.916187806587246,
      "grad_norm": 0.3136264979839325,
      "learning_rate": 0.00011104520449653192,
      "loss": 0.0695,
      "step": 13672
    },
    {
      "epoch": 1.9163279607568326,
      "grad_norm": 0.5510244965553284,
      "learning_rate": 0.00011103085386271225,
      "loss": 0.0729,
      "step": 13673
    },
    {
      "epoch": 1.916468114926419,
      "grad_norm": 0.38377177715301514,
      "learning_rate": 0.00011101650322889261,
      "loss": 0.0411,
      "step": 13674
    },
    {
      "epoch": 1.9166082690960056,
      "grad_norm": 0.4233558773994446,
      "learning_rate": 0.00011100215259507295,
      "loss": 0.0535,
      "step": 13675
    },
    {
      "epoch": 1.916748423265592,
      "grad_norm": 0.2139747440814972,
      "learning_rate": 0.00011098780196125328,
      "loss": 0.051,
      "step": 13676
    },
    {
      "epoch": 1.9168885774351787,
      "grad_norm": 0.2551210820674896,
      "learning_rate": 0.00011097345132743362,
      "loss": 0.0227,
      "step": 13677
    },
    {
      "epoch": 1.9170287316047654,
      "grad_norm": 0.45006346702575684,
      "learning_rate": 0.00011095910069361395,
      "loss": 0.0527,
      "step": 13678
    },
    {
      "epoch": 1.9171688857743518,
      "grad_norm": 0.6150220036506653,
      "learning_rate": 0.00011094475005979429,
      "loss": 0.0294,
      "step": 13679
    },
    {
      "epoch": 1.9173090399439383,
      "grad_norm": 0.6374920606613159,
      "learning_rate": 0.00011093039942597465,
      "loss": 0.0715,
      "step": 13680
    },
    {
      "epoch": 1.9174491941135248,
      "grad_norm": 0.2077314257621765,
      "learning_rate": 0.00011091604879215497,
      "loss": 0.0164,
      "step": 13681
    },
    {
      "epoch": 1.9175893482831115,
      "grad_norm": 0.20097874104976654,
      "learning_rate": 0.00011090169815833532,
      "loss": 0.0449,
      "step": 13682
    },
    {
      "epoch": 1.9177295024526981,
      "grad_norm": 0.6967077851295471,
      "learning_rate": 0.00011088734752451566,
      "loss": 0.0787,
      "step": 13683
    },
    {
      "epoch": 1.9178696566222846,
      "grad_norm": 0.2360992133617401,
      "learning_rate": 0.00011087299689069599,
      "loss": 0.0237,
      "step": 13684
    },
    {
      "epoch": 1.918009810791871,
      "grad_norm": 0.39261049032211304,
      "learning_rate": 0.00011085864625687634,
      "loss": 0.0359,
      "step": 13685
    },
    {
      "epoch": 1.9181499649614575,
      "grad_norm": 0.08414167165756226,
      "learning_rate": 0.00011084429562305667,
      "loss": 0.0109,
      "step": 13686
    },
    {
      "epoch": 1.9182901191310442,
      "grad_norm": 0.20710989832878113,
      "learning_rate": 0.00011082994498923701,
      "loss": 0.0197,
      "step": 13687
    },
    {
      "epoch": 1.9184302733006307,
      "grad_norm": 0.38900843262672424,
      "learning_rate": 0.00011081559435541736,
      "loss": 0.0294,
      "step": 13688
    },
    {
      "epoch": 1.9185704274702173,
      "grad_norm": 0.38893353939056396,
      "learning_rate": 0.00011080124372159768,
      "loss": 0.0262,
      "step": 13689
    },
    {
      "epoch": 1.9187105816398038,
      "grad_norm": 0.09258971363306046,
      "learning_rate": 0.00011078689308777804,
      "loss": 0.019,
      "step": 13690
    },
    {
      "epoch": 1.9188507358093903,
      "grad_norm": 0.20317251980304718,
      "learning_rate": 0.00011077254245395838,
      "loss": 0.026,
      "step": 13691
    },
    {
      "epoch": 1.9189908899789767,
      "grad_norm": 0.36524906754493713,
      "learning_rate": 0.00011075819182013871,
      "loss": 0.0991,
      "step": 13692
    },
    {
      "epoch": 1.9191310441485634,
      "grad_norm": 0.08668703585863113,
      "learning_rate": 0.00011074384118631905,
      "loss": 0.0144,
      "step": 13693
    },
    {
      "epoch": 1.91927119831815,
      "grad_norm": 0.15800528228282928,
      "learning_rate": 0.0001107294905524994,
      "loss": 0.0204,
      "step": 13694
    },
    {
      "epoch": 1.9194113524877365,
      "grad_norm": 0.5117565393447876,
      "learning_rate": 0.00011071513991867972,
      "loss": 0.1666,
      "step": 13695
    },
    {
      "epoch": 1.919551506657323,
      "grad_norm": 0.3248744606971741,
      "learning_rate": 0.00011070078928486008,
      "loss": 0.0638,
      "step": 13696
    },
    {
      "epoch": 1.9196916608269095,
      "grad_norm": 0.3702656030654907,
      "learning_rate": 0.00011068643865104041,
      "loss": 0.1141,
      "step": 13697
    },
    {
      "epoch": 1.9198318149964961,
      "grad_norm": 0.2507188320159912,
      "learning_rate": 0.00011067208801722075,
      "loss": 0.0379,
      "step": 13698
    },
    {
      "epoch": 1.9199719691660828,
      "grad_norm": 0.2232222855091095,
      "learning_rate": 0.00011065773738340109,
      "loss": 0.0753,
      "step": 13699
    },
    {
      "epoch": 1.9201121233356693,
      "grad_norm": 0.553311824798584,
      "learning_rate": 0.00011064338674958142,
      "loss": 0.0721,
      "step": 13700
    },
    {
      "epoch": 1.9202522775052557,
      "grad_norm": 0.37850871682167053,
      "learning_rate": 0.00011062903611576178,
      "loss": 0.0804,
      "step": 13701
    },
    {
      "epoch": 1.9203924316748422,
      "grad_norm": 0.16052445769309998,
      "learning_rate": 0.00011061468548194212,
      "loss": 0.0137,
      "step": 13702
    },
    {
      "epoch": 1.9205325858444289,
      "grad_norm": 0.21905580163002014,
      "learning_rate": 0.00011060033484812245,
      "loss": 0.0399,
      "step": 13703
    },
    {
      "epoch": 1.9206727400140156,
      "grad_norm": 0.29196351766586304,
      "learning_rate": 0.00011058598421430279,
      "loss": 0.0295,
      "step": 13704
    },
    {
      "epoch": 1.920812894183602,
      "grad_norm": 0.2661282420158386,
      "learning_rate": 0.00011057163358048312,
      "loss": 0.0186,
      "step": 13705
    },
    {
      "epoch": 1.9209530483531885,
      "grad_norm": 0.43304118514060974,
      "learning_rate": 0.00011055728294666347,
      "loss": 0.0624,
      "step": 13706
    },
    {
      "epoch": 1.921093202522775,
      "grad_norm": 0.146284818649292,
      "learning_rate": 0.00011054293231284382,
      "loss": 0.0194,
      "step": 13707
    },
    {
      "epoch": 1.9212333566923616,
      "grad_norm": 0.3591146767139435,
      "learning_rate": 0.00011052858167902414,
      "loss": 0.0316,
      "step": 13708
    },
    {
      "epoch": 1.9213735108619483,
      "grad_norm": 0.4111204445362091,
      "learning_rate": 0.00011051423104520449,
      "loss": 0.064,
      "step": 13709
    },
    {
      "epoch": 1.9215136650315348,
      "grad_norm": 0.21412111818790436,
      "learning_rate": 0.00011049988041138483,
      "loss": 0.0172,
      "step": 13710
    },
    {
      "epoch": 1.9216538192011212,
      "grad_norm": 0.12354470789432526,
      "learning_rate": 0.00011048552977756516,
      "loss": 0.0276,
      "step": 13711
    },
    {
      "epoch": 1.9217939733707077,
      "grad_norm": 0.137582466006279,
      "learning_rate": 0.00011047117914374551,
      "loss": 0.0318,
      "step": 13712
    },
    {
      "epoch": 1.9219341275402944,
      "grad_norm": 0.5220590233802795,
      "learning_rate": 0.00011045682850992584,
      "loss": 0.1214,
      "step": 13713
    },
    {
      "epoch": 1.9220742817098808,
      "grad_norm": 0.6386471390724182,
      "learning_rate": 0.00011044247787610618,
      "loss": 0.0373,
      "step": 13714
    },
    {
      "epoch": 1.9222144358794675,
      "grad_norm": 0.04840155318379402,
      "learning_rate": 0.00011042812724228653,
      "loss": 0.0047,
      "step": 13715
    },
    {
      "epoch": 1.922354590049054,
      "grad_norm": 0.27740180492401123,
      "learning_rate": 0.00011041377660846685,
      "loss": 0.04,
      "step": 13716
    },
    {
      "epoch": 1.9224947442186404,
      "grad_norm": 0.5502803921699524,
      "learning_rate": 0.00011039942597464721,
      "loss": 0.0292,
      "step": 13717
    },
    {
      "epoch": 1.922634898388227,
      "grad_norm": 0.27440428733825684,
      "learning_rate": 0.00011038507534082755,
      "loss": 0.0786,
      "step": 13718
    },
    {
      "epoch": 1.9227750525578136,
      "grad_norm": 1.0464221239089966,
      "learning_rate": 0.00011037072470700788,
      "loss": 0.0929,
      "step": 13719
    },
    {
      "epoch": 1.9229152067274002,
      "grad_norm": 0.11623378098011017,
      "learning_rate": 0.00011035637407318822,
      "loss": 0.0163,
      "step": 13720
    },
    {
      "epoch": 1.9230553608969867,
      "grad_norm": 0.15276794135570526,
      "learning_rate": 0.00011034202343936855,
      "loss": 0.0137,
      "step": 13721
    },
    {
      "epoch": 1.9231955150665732,
      "grad_norm": 0.5811693072319031,
      "learning_rate": 0.0001103276728055489,
      "loss": 0.0824,
      "step": 13722
    },
    {
      "epoch": 1.9233356692361596,
      "grad_norm": 0.416728675365448,
      "learning_rate": 0.00011031332217172925,
      "loss": 0.0526,
      "step": 13723
    },
    {
      "epoch": 1.9234758234057463,
      "grad_norm": 0.33003339171409607,
      "learning_rate": 0.00011029897153790958,
      "loss": 0.0795,
      "step": 13724
    },
    {
      "epoch": 1.923615977575333,
      "grad_norm": 0.4260284900665283,
      "learning_rate": 0.00011028462090408992,
      "loss": 0.0505,
      "step": 13725
    },
    {
      "epoch": 1.9237561317449194,
      "grad_norm": 0.491764098405838,
      "learning_rate": 0.00011027027027027026,
      "loss": 0.071,
      "step": 13726
    },
    {
      "epoch": 1.923896285914506,
      "grad_norm": 0.6291949152946472,
      "learning_rate": 0.00011025591963645059,
      "loss": 0.0306,
      "step": 13727
    },
    {
      "epoch": 1.9240364400840924,
      "grad_norm": 0.26882901787757874,
      "learning_rate": 0.00011024156900263095,
      "loss": 0.0195,
      "step": 13728
    },
    {
      "epoch": 1.924176594253679,
      "grad_norm": 0.4803031086921692,
      "learning_rate": 0.00011022721836881129,
      "loss": 0.0484,
      "step": 13729
    },
    {
      "epoch": 1.9243167484232657,
      "grad_norm": 0.6390761733055115,
      "learning_rate": 0.00011021286773499162,
      "loss": 0.0549,
      "step": 13730
    },
    {
      "epoch": 1.9244569025928522,
      "grad_norm": 0.4836127460002899,
      "learning_rate": 0.00011019851710117196,
      "loss": 0.0135,
      "step": 13731
    },
    {
      "epoch": 1.9245970567624386,
      "grad_norm": 0.30497288703918457,
      "learning_rate": 0.00011018416646735229,
      "loss": 0.0222,
      "step": 13732
    },
    {
      "epoch": 1.924737210932025,
      "grad_norm": 0.22047258913516998,
      "learning_rate": 0.00011016981583353264,
      "loss": 0.0325,
      "step": 13733
    },
    {
      "epoch": 1.9248773651016118,
      "grad_norm": 0.43384021520614624,
      "learning_rate": 0.00011015546519971298,
      "loss": 0.021,
      "step": 13734
    },
    {
      "epoch": 1.9250175192711985,
      "grad_norm": 0.6336798667907715,
      "learning_rate": 0.00011014111456589331,
      "loss": 0.0415,
      "step": 13735
    },
    {
      "epoch": 1.925157673440785,
      "grad_norm": 0.06143474578857422,
      "learning_rate": 0.00011012676393207366,
      "loss": 0.0135,
      "step": 13736
    },
    {
      "epoch": 1.9252978276103714,
      "grad_norm": 0.33726197481155396,
      "learning_rate": 0.000110112413298254,
      "loss": 0.0394,
      "step": 13737
    },
    {
      "epoch": 1.9254379817799578,
      "grad_norm": 0.22072440385818481,
      "learning_rate": 0.00011009806266443434,
      "loss": 0.0475,
      "step": 13738
    },
    {
      "epoch": 1.9255781359495445,
      "grad_norm": 0.8299367427825928,
      "learning_rate": 0.00011008371203061468,
      "loss": 0.0529,
      "step": 13739
    },
    {
      "epoch": 1.9257182901191312,
      "grad_norm": 0.1965607851743698,
      "learning_rate": 0.00011006936139679501,
      "loss": 0.0118,
      "step": 13740
    },
    {
      "epoch": 1.9258584442887177,
      "grad_norm": 0.24947310984134674,
      "learning_rate": 0.00011005501076297535,
      "loss": 0.0304,
      "step": 13741
    },
    {
      "epoch": 1.9259985984583041,
      "grad_norm": 0.3248518705368042,
      "learning_rate": 0.0001100406601291557,
      "loss": 0.074,
      "step": 13742
    },
    {
      "epoch": 1.9261387526278906,
      "grad_norm": 0.4450487494468689,
      "learning_rate": 0.00011002630949533602,
      "loss": 0.0623,
      "step": 13743
    },
    {
      "epoch": 1.9262789067974773,
      "grad_norm": 0.5626828074455261,
      "learning_rate": 0.00011001195886151638,
      "loss": 0.0394,
      "step": 13744
    },
    {
      "epoch": 1.9264190609670637,
      "grad_norm": 0.3531736135482788,
      "learning_rate": 0.00010999760822769672,
      "loss": 0.064,
      "step": 13745
    },
    {
      "epoch": 1.9265592151366504,
      "grad_norm": 0.08108167350292206,
      "learning_rate": 0.00010998325759387705,
      "loss": 0.0097,
      "step": 13746
    },
    {
      "epoch": 1.9266993693062369,
      "grad_norm": 0.11380761116743088,
      "learning_rate": 0.00010996890696005739,
      "loss": 0.0176,
      "step": 13747
    },
    {
      "epoch": 1.9268395234758233,
      "grad_norm": 0.15786680579185486,
      "learning_rate": 0.00010995455632623772,
      "loss": 0.0164,
      "step": 13748
    },
    {
      "epoch": 1.92697967764541,
      "grad_norm": 0.22282624244689941,
      "learning_rate": 0.00010994020569241808,
      "loss": 0.0802,
      "step": 13749
    },
    {
      "epoch": 1.9271198318149965,
      "grad_norm": 0.34158846735954285,
      "learning_rate": 0.00010992585505859842,
      "loss": 0.0749,
      "step": 13750
    },
    {
      "epoch": 1.9272599859845831,
      "grad_norm": 0.23161712288856506,
      "learning_rate": 0.00010991150442477875,
      "loss": 0.0402,
      "step": 13751
    },
    {
      "epoch": 1.9274001401541696,
      "grad_norm": 0.19699764251708984,
      "learning_rate": 0.00010989715379095909,
      "loss": 0.0222,
      "step": 13752
    },
    {
      "epoch": 1.927540294323756,
      "grad_norm": 0.3704795241355896,
      "learning_rate": 0.00010988280315713943,
      "loss": 0.013,
      "step": 13753
    },
    {
      "epoch": 1.9276804484933425,
      "grad_norm": 0.38436684012413025,
      "learning_rate": 0.00010986845252331977,
      "loss": 0.013,
      "step": 13754
    },
    {
      "epoch": 1.9278206026629292,
      "grad_norm": 0.22594177722930908,
      "learning_rate": 0.00010985410188950011,
      "loss": 0.031,
      "step": 13755
    },
    {
      "epoch": 1.9279607568325159,
      "grad_norm": 0.09729012101888657,
      "learning_rate": 0.00010983975125568044,
      "loss": 0.0152,
      "step": 13756
    },
    {
      "epoch": 1.9281009110021023,
      "grad_norm": 0.3148127496242523,
      "learning_rate": 0.00010982540062186079,
      "loss": 0.0314,
      "step": 13757
    },
    {
      "epoch": 1.9282410651716888,
      "grad_norm": 0.26471808552742004,
      "learning_rate": 0.00010981104998804113,
      "loss": 0.0418,
      "step": 13758
    },
    {
      "epoch": 1.9283812193412753,
      "grad_norm": 0.09002430737018585,
      "learning_rate": 0.00010979669935422146,
      "loss": 0.0071,
      "step": 13759
    },
    {
      "epoch": 1.928521373510862,
      "grad_norm": 0.2948862910270691,
      "learning_rate": 0.00010978234872040181,
      "loss": 0.0539,
      "step": 13760
    },
    {
      "epoch": 1.9286615276804486,
      "grad_norm": 0.1825835108757019,
      "learning_rate": 0.00010976799808658215,
      "loss": 0.0193,
      "step": 13761
    },
    {
      "epoch": 1.928801681850035,
      "grad_norm": 0.1326032280921936,
      "learning_rate": 0.00010975364745276248,
      "loss": 0.0272,
      "step": 13762
    },
    {
      "epoch": 1.9289418360196215,
      "grad_norm": 0.7449473142623901,
      "learning_rate": 0.00010973929681894282,
      "loss": 0.1853,
      "step": 13763
    },
    {
      "epoch": 1.929081990189208,
      "grad_norm": 0.21338985860347748,
      "learning_rate": 0.00010972494618512318,
      "loss": 0.056,
      "step": 13764
    },
    {
      "epoch": 1.9292221443587947,
      "grad_norm": 0.6736727356910706,
      "learning_rate": 0.00010971059555130351,
      "loss": 0.0164,
      "step": 13765
    },
    {
      "epoch": 1.9293622985283814,
      "grad_norm": 0.36937427520751953,
      "learning_rate": 0.00010969624491748385,
      "loss": 0.024,
      "step": 13766
    },
    {
      "epoch": 1.9295024526979678,
      "grad_norm": 0.14926302433013916,
      "learning_rate": 0.00010968189428366418,
      "loss": 0.0137,
      "step": 13767
    },
    {
      "epoch": 1.9296426068675543,
      "grad_norm": 0.3490966856479645,
      "learning_rate": 0.00010966754364984452,
      "loss": 0.0288,
      "step": 13768
    },
    {
      "epoch": 1.9297827610371407,
      "grad_norm": 0.4471503794193268,
      "learning_rate": 0.00010965319301602488,
      "loss": 0.0268,
      "step": 13769
    },
    {
      "epoch": 1.9299229152067274,
      "grad_norm": 0.22939524054527283,
      "learning_rate": 0.0001096388423822052,
      "loss": 0.0694,
      "step": 13770
    },
    {
      "epoch": 1.9300630693763141,
      "grad_norm": 0.5468003749847412,
      "learning_rate": 0.00010962449174838555,
      "loss": 0.0599,
      "step": 13771
    },
    {
      "epoch": 1.9302032235459006,
      "grad_norm": 0.2246396392583847,
      "learning_rate": 0.00010961014111456589,
      "loss": 0.0338,
      "step": 13772
    },
    {
      "epoch": 1.930343377715487,
      "grad_norm": 0.3756606876850128,
      "learning_rate": 0.00010959579048074622,
      "loss": 0.0501,
      "step": 13773
    },
    {
      "epoch": 1.9304835318850735,
      "grad_norm": 0.2372695356607437,
      "learning_rate": 0.00010958143984692656,
      "loss": 0.0467,
      "step": 13774
    },
    {
      "epoch": 1.9306236860546602,
      "grad_norm": 0.20574170351028442,
      "learning_rate": 0.00010956708921310689,
      "loss": 0.0221,
      "step": 13775
    },
    {
      "epoch": 1.9307638402242466,
      "grad_norm": 0.45478641986846924,
      "learning_rate": 0.00010955273857928725,
      "loss": 0.0319,
      "step": 13776
    },
    {
      "epoch": 1.9309039943938333,
      "grad_norm": 0.39161354303359985,
      "learning_rate": 0.00010953838794546759,
      "loss": 0.0298,
      "step": 13777
    },
    {
      "epoch": 1.9310441485634198,
      "grad_norm": 0.3739454746246338,
      "learning_rate": 0.00010952403731164792,
      "loss": 0.0221,
      "step": 13778
    },
    {
      "epoch": 1.9311843027330062,
      "grad_norm": 0.315547376871109,
      "learning_rate": 0.00010950968667782826,
      "loss": 0.0253,
      "step": 13779
    },
    {
      "epoch": 1.9313244569025927,
      "grad_norm": 0.4347879886627197,
      "learning_rate": 0.00010949533604400861,
      "loss": 0.0678,
      "step": 13780
    },
    {
      "epoch": 1.9314646110721794,
      "grad_norm": 0.319716215133667,
      "learning_rate": 0.00010948098541018894,
      "loss": 0.0601,
      "step": 13781
    },
    {
      "epoch": 1.931604765241766,
      "grad_norm": 0.46149200201034546,
      "learning_rate": 0.00010946663477636928,
      "loss": 0.0187,
      "step": 13782
    },
    {
      "epoch": 1.9317449194113525,
      "grad_norm": 0.6473413109779358,
      "learning_rate": 0.00010945228414254961,
      "loss": 0.0762,
      "step": 13783
    },
    {
      "epoch": 1.931885073580939,
      "grad_norm": 0.6601626873016357,
      "learning_rate": 0.00010943793350872995,
      "loss": 0.0351,
      "step": 13784
    },
    {
      "epoch": 1.9320252277505254,
      "grad_norm": 0.7963165640830994,
      "learning_rate": 0.00010942358287491031,
      "loss": 0.0157,
      "step": 13785
    },
    {
      "epoch": 1.9321653819201121,
      "grad_norm": 0.4127991795539856,
      "learning_rate": 0.00010940923224109064,
      "loss": 0.0388,
      "step": 13786
    },
    {
      "epoch": 1.9323055360896988,
      "grad_norm": 0.6259059309959412,
      "learning_rate": 0.00010939488160727098,
      "loss": 0.0548,
      "step": 13787
    },
    {
      "epoch": 1.9324456902592853,
      "grad_norm": 0.31176838278770447,
      "learning_rate": 0.00010938053097345132,
      "loss": 0.0493,
      "step": 13788
    },
    {
      "epoch": 1.9325858444288717,
      "grad_norm": 0.4198831617832184,
      "learning_rate": 0.00010936618033963165,
      "loss": 0.0446,
      "step": 13789
    },
    {
      "epoch": 1.9327259985984582,
      "grad_norm": 0.2061156928539276,
      "learning_rate": 0.000109351829705812,
      "loss": 0.0413,
      "step": 13790
    },
    {
      "epoch": 1.9328661527680449,
      "grad_norm": 0.3667908012866974,
      "learning_rate": 0.00010933747907199232,
      "loss": 0.0732,
      "step": 13791
    },
    {
      "epoch": 1.9330063069376315,
      "grad_norm": 0.15379749238491058,
      "learning_rate": 0.00010932312843817268,
      "loss": 0.0379,
      "step": 13792
    },
    {
      "epoch": 1.933146461107218,
      "grad_norm": 0.4355453550815582,
      "learning_rate": 0.00010930877780435302,
      "loss": 0.0285,
      "step": 13793
    },
    {
      "epoch": 1.9332866152768045,
      "grad_norm": 0.6018511652946472,
      "learning_rate": 0.00010929442717053335,
      "loss": 0.1155,
      "step": 13794
    },
    {
      "epoch": 1.933426769446391,
      "grad_norm": 0.35737666487693787,
      "learning_rate": 0.00010928007653671369,
      "loss": 0.0854,
      "step": 13795
    },
    {
      "epoch": 1.9335669236159776,
      "grad_norm": 0.27383071184158325,
      "learning_rate": 0.00010926572590289405,
      "loss": 0.0535,
      "step": 13796
    },
    {
      "epoch": 1.9337070777855643,
      "grad_norm": 0.46503695845603943,
      "learning_rate": 0.00010925137526907438,
      "loss": 0.1117,
      "step": 13797
    },
    {
      "epoch": 1.9338472319551507,
      "grad_norm": 0.2172660380601883,
      "learning_rate": 0.00010923702463525472,
      "loss": 0.0359,
      "step": 13798
    },
    {
      "epoch": 1.9339873861247372,
      "grad_norm": 0.4931662082672119,
      "learning_rate": 0.00010922267400143506,
      "loss": 0.048,
      "step": 13799
    },
    {
      "epoch": 1.9341275402943237,
      "grad_norm": 0.505435585975647,
      "learning_rate": 0.00010920832336761539,
      "loss": 0.059,
      "step": 13800
    },
    {
      "epoch": 1.9342676944639103,
      "grad_norm": 0.11063232272863388,
      "learning_rate": 0.00010919397273379574,
      "loss": 0.0099,
      "step": 13801
    },
    {
      "epoch": 1.9344078486334968,
      "grad_norm": 0.2406204640865326,
      "learning_rate": 0.00010917962209997607,
      "loss": 0.089,
      "step": 13802
    },
    {
      "epoch": 1.9345480028030835,
      "grad_norm": 0.25254330039024353,
      "learning_rate": 0.00010916527146615641,
      "loss": 0.0476,
      "step": 13803
    },
    {
      "epoch": 1.93468815697267,
      "grad_norm": 0.11822552978992462,
      "learning_rate": 0.00010915092083233676,
      "loss": 0.0471,
      "step": 13804
    },
    {
      "epoch": 1.9348283111422564,
      "grad_norm": 0.2360810488462448,
      "learning_rate": 0.00010913657019851708,
      "loss": 0.0337,
      "step": 13805
    },
    {
      "epoch": 1.934968465311843,
      "grad_norm": 0.11549448221921921,
      "learning_rate": 0.00010912221956469743,
      "loss": 0.014,
      "step": 13806
    },
    {
      "epoch": 1.9351086194814295,
      "grad_norm": 0.2151952087879181,
      "learning_rate": 0.00010910786893087778,
      "loss": 0.0172,
      "step": 13807
    },
    {
      "epoch": 1.9352487736510162,
      "grad_norm": 0.1471911370754242,
      "learning_rate": 0.00010909351829705811,
      "loss": 0.0178,
      "step": 13808
    },
    {
      "epoch": 1.9353889278206027,
      "grad_norm": 0.22184059023857117,
      "learning_rate": 0.00010907916766323845,
      "loss": 0.0379,
      "step": 13809
    },
    {
      "epoch": 1.9355290819901891,
      "grad_norm": 0.21752691268920898,
      "learning_rate": 0.00010906481702941878,
      "loss": 0.0628,
      "step": 13810
    },
    {
      "epoch": 1.9356692361597756,
      "grad_norm": 0.17189821600914001,
      "learning_rate": 0.00010905046639559912,
      "loss": 0.0191,
      "step": 13811
    },
    {
      "epoch": 1.9358093903293623,
      "grad_norm": 0.5695016980171204,
      "learning_rate": 0.00010903611576177948,
      "loss": 0.071,
      "step": 13812
    },
    {
      "epoch": 1.935949544498949,
      "grad_norm": 0.15487004816532135,
      "learning_rate": 0.00010902176512795981,
      "loss": 0.0103,
      "step": 13813
    },
    {
      "epoch": 1.9360896986685354,
      "grad_norm": 0.08888904750347137,
      "learning_rate": 0.00010900741449414015,
      "loss": 0.0265,
      "step": 13814
    },
    {
      "epoch": 1.9362298528381219,
      "grad_norm": 0.2852155566215515,
      "learning_rate": 0.00010899306386032049,
      "loss": 0.0424,
      "step": 13815
    },
    {
      "epoch": 1.9363700070077083,
      "grad_norm": 0.355835497379303,
      "learning_rate": 0.00010897871322650082,
      "loss": 0.0324,
      "step": 13816
    },
    {
      "epoch": 1.936510161177295,
      "grad_norm": 0.23602645099163055,
      "learning_rate": 0.00010896436259268118,
      "loss": 0.0827,
      "step": 13817
    },
    {
      "epoch": 1.9366503153468817,
      "grad_norm": 0.08828553557395935,
      "learning_rate": 0.0001089500119588615,
      "loss": 0.0128,
      "step": 13818
    },
    {
      "epoch": 1.9367904695164682,
      "grad_norm": 0.48711979389190674,
      "learning_rate": 0.00010893566132504185,
      "loss": 0.1301,
      "step": 13819
    },
    {
      "epoch": 1.9369306236860546,
      "grad_norm": 0.33350110054016113,
      "learning_rate": 0.00010892131069122219,
      "loss": 0.0712,
      "step": 13820
    },
    {
      "epoch": 1.937070777855641,
      "grad_norm": 0.17934256792068481,
      "learning_rate": 0.00010890696005740252,
      "loss": 0.0381,
      "step": 13821
    },
    {
      "epoch": 1.9372109320252278,
      "grad_norm": 0.48867979645729065,
      "learning_rate": 0.00010889260942358286,
      "loss": 0.0469,
      "step": 13822
    },
    {
      "epoch": 1.9373510861948144,
      "grad_norm": 1.0411688089370728,
      "learning_rate": 0.00010887825878976322,
      "loss": 0.1012,
      "step": 13823
    },
    {
      "epoch": 1.937491240364401,
      "grad_norm": 0.2419024258852005,
      "learning_rate": 0.00010886390815594354,
      "loss": 0.0169,
      "step": 13824
    },
    {
      "epoch": 1.9376313945339874,
      "grad_norm": 0.6210082173347473,
      "learning_rate": 0.00010884955752212389,
      "loss": 0.1389,
      "step": 13825
    },
    {
      "epoch": 1.9377715487035738,
      "grad_norm": 0.2601433992385864,
      "learning_rate": 0.00010883520688830422,
      "loss": 0.0431,
      "step": 13826
    },
    {
      "epoch": 1.9379117028731605,
      "grad_norm": 0.19944724440574646,
      "learning_rate": 0.00010882085625448456,
      "loss": 0.0165,
      "step": 13827
    },
    {
      "epoch": 1.9380518570427472,
      "grad_norm": 0.3502379357814789,
      "learning_rate": 0.00010880650562066491,
      "loss": 0.0274,
      "step": 13828
    },
    {
      "epoch": 1.9381920112123336,
      "grad_norm": 0.1394815593957901,
      "learning_rate": 0.00010879215498684524,
      "loss": 0.0237,
      "step": 13829
    },
    {
      "epoch": 1.93833216538192,
      "grad_norm": 0.6292027831077576,
      "learning_rate": 0.00010877780435302558,
      "loss": 0.1062,
      "step": 13830
    },
    {
      "epoch": 1.9384723195515066,
      "grad_norm": 0.5791199207305908,
      "learning_rate": 0.00010876345371920593,
      "loss": 0.1374,
      "step": 13831
    },
    {
      "epoch": 1.9386124737210932,
      "grad_norm": 0.2543894648551941,
      "learning_rate": 0.00010874910308538625,
      "loss": 0.0341,
      "step": 13832
    },
    {
      "epoch": 1.9387526278906797,
      "grad_norm": 0.43626680970191956,
      "learning_rate": 0.00010873475245156661,
      "loss": 0.069,
      "step": 13833
    },
    {
      "epoch": 1.9388927820602664,
      "grad_norm": 0.8026769161224365,
      "learning_rate": 0.00010872040181774694,
      "loss": 0.1655,
      "step": 13834
    },
    {
      "epoch": 1.9390329362298528,
      "grad_norm": 1.4881922006607056,
      "learning_rate": 0.00010870605118392728,
      "loss": 0.3615,
      "step": 13835
    },
    {
      "epoch": 1.9391730903994393,
      "grad_norm": 0.2582043707370758,
      "learning_rate": 0.00010869170055010762,
      "loss": 0.0547,
      "step": 13836
    },
    {
      "epoch": 1.9393132445690258,
      "grad_norm": 0.1846296638250351,
      "learning_rate": 0.00010867734991628795,
      "loss": 0.0165,
      "step": 13837
    },
    {
      "epoch": 1.9394533987386124,
      "grad_norm": 0.3504028916358948,
      "learning_rate": 0.0001086629992824683,
      "loss": 0.0551,
      "step": 13838
    },
    {
      "epoch": 1.9395935529081991,
      "grad_norm": 0.3042311668395996,
      "learning_rate": 0.00010864864864864865,
      "loss": 0.0574,
      "step": 13839
    },
    {
      "epoch": 1.9397337070777856,
      "grad_norm": 0.15060807764530182,
      "learning_rate": 0.00010863429801482898,
      "loss": 0.0237,
      "step": 13840
    },
    {
      "epoch": 1.939873861247372,
      "grad_norm": 0.3377339243888855,
      "learning_rate": 0.00010861994738100932,
      "loss": 0.0243,
      "step": 13841
    },
    {
      "epoch": 1.9400140154169585,
      "grad_norm": 0.47248032689094543,
      "learning_rate": 0.00010860559674718966,
      "loss": 0.0276,
      "step": 13842
    },
    {
      "epoch": 1.9401541695865452,
      "grad_norm": 0.2184755951166153,
      "learning_rate": 0.00010859124611336999,
      "loss": 0.0159,
      "step": 13843
    },
    {
      "epoch": 1.9402943237561319,
      "grad_norm": 0.1103387251496315,
      "learning_rate": 0.00010857689547955035,
      "loss": 0.0278,
      "step": 13844
    },
    {
      "epoch": 1.9404344779257183,
      "grad_norm": 0.3250039517879486,
      "learning_rate": 0.00010856254484573067,
      "loss": 0.0296,
      "step": 13845
    },
    {
      "epoch": 1.9405746320953048,
      "grad_norm": 0.30050671100616455,
      "learning_rate": 0.00010854819421191102,
      "loss": 0.0476,
      "step": 13846
    },
    {
      "epoch": 1.9407147862648912,
      "grad_norm": 0.3676130771636963,
      "learning_rate": 0.00010853384357809136,
      "loss": 0.0519,
      "step": 13847
    },
    {
      "epoch": 1.940854940434478,
      "grad_norm": 0.5416865944862366,
      "learning_rate": 0.00010851949294427169,
      "loss": 0.0487,
      "step": 13848
    },
    {
      "epoch": 1.9409950946040646,
      "grad_norm": 0.3596961200237274,
      "learning_rate": 0.00010850514231045204,
      "loss": 0.0525,
      "step": 13849
    },
    {
      "epoch": 1.941135248773651,
      "grad_norm": 0.1657838225364685,
      "learning_rate": 0.00010849079167663239,
      "loss": 0.0116,
      "step": 13850
    },
    {
      "epoch": 1.9412754029432375,
      "grad_norm": 0.11259095370769501,
      "learning_rate": 0.00010847644104281271,
      "loss": 0.0181,
      "step": 13851
    },
    {
      "epoch": 1.941415557112824,
      "grad_norm": 0.11800874024629593,
      "learning_rate": 0.00010846209040899306,
      "loss": 0.011,
      "step": 13852
    },
    {
      "epoch": 1.9415557112824107,
      "grad_norm": 0.9125251173973083,
      "learning_rate": 0.00010844773977517338,
      "loss": 0.0403,
      "step": 13853
    },
    {
      "epoch": 1.9416958654519973,
      "grad_norm": 0.6194085478782654,
      "learning_rate": 0.00010843338914135373,
      "loss": 0.0887,
      "step": 13854
    },
    {
      "epoch": 1.9418360196215838,
      "grad_norm": 0.20562759041786194,
      "learning_rate": 0.00010841903850753408,
      "loss": 0.0334,
      "step": 13855
    },
    {
      "epoch": 1.9419761737911703,
      "grad_norm": 0.381644070148468,
      "learning_rate": 0.00010840468787371441,
      "loss": 0.1484,
      "step": 13856
    },
    {
      "epoch": 1.9421163279607567,
      "grad_norm": 0.188826784491539,
      "learning_rate": 0.00010839033723989475,
      "loss": 0.0472,
      "step": 13857
    },
    {
      "epoch": 1.9422564821303434,
      "grad_norm": 0.16431394219398499,
      "learning_rate": 0.0001083759866060751,
      "loss": 0.0337,
      "step": 13858
    },
    {
      "epoch": 1.9423966362999299,
      "grad_norm": 0.9686247110366821,
      "learning_rate": 0.00010836163597225542,
      "loss": 0.0764,
      "step": 13859
    },
    {
      "epoch": 1.9425367904695165,
      "grad_norm": 0.3906697928905487,
      "learning_rate": 0.00010834728533843578,
      "loss": 0.0448,
      "step": 13860
    },
    {
      "epoch": 1.942676944639103,
      "grad_norm": 0.5008538961410522,
      "learning_rate": 0.00010833293470461611,
      "loss": 0.0672,
      "step": 13861
    },
    {
      "epoch": 1.9428170988086895,
      "grad_norm": 0.3694528639316559,
      "learning_rate": 0.00010831858407079645,
      "loss": 0.054,
      "step": 13862
    },
    {
      "epoch": 1.9429572529782762,
      "grad_norm": 0.29367485642433167,
      "learning_rate": 0.00010830423343697679,
      "loss": 0.0182,
      "step": 13863
    },
    {
      "epoch": 1.9430974071478626,
      "grad_norm": 0.18175604939460754,
      "learning_rate": 0.00010828988280315712,
      "loss": 0.0274,
      "step": 13864
    },
    {
      "epoch": 1.9432375613174493,
      "grad_norm": 0.16520561277866364,
      "learning_rate": 0.00010827553216933748,
      "loss": 0.0208,
      "step": 13865
    },
    {
      "epoch": 1.9433777154870358,
      "grad_norm": 0.5899354219436646,
      "learning_rate": 0.00010826118153551782,
      "loss": 0.0953,
      "step": 13866
    },
    {
      "epoch": 1.9435178696566222,
      "grad_norm": 0.194908007979393,
      "learning_rate": 0.00010824683090169815,
      "loss": 0.0356,
      "step": 13867
    },
    {
      "epoch": 1.9436580238262087,
      "grad_norm": 0.08533184975385666,
      "learning_rate": 0.00010823248026787849,
      "loss": 0.0317,
      "step": 13868
    },
    {
      "epoch": 1.9437981779957954,
      "grad_norm": 0.22282400727272034,
      "learning_rate": 0.00010821812963405882,
      "loss": 0.0406,
      "step": 13869
    },
    {
      "epoch": 1.943938332165382,
      "grad_norm": 0.179804265499115,
      "learning_rate": 0.00010820377900023916,
      "loss": 0.0153,
      "step": 13870
    },
    {
      "epoch": 1.9440784863349685,
      "grad_norm": 0.1297025829553604,
      "learning_rate": 0.00010818942836641952,
      "loss": 0.0187,
      "step": 13871
    },
    {
      "epoch": 1.944218640504555,
      "grad_norm": 0.11439014971256256,
      "learning_rate": 0.00010817507773259984,
      "loss": 0.0148,
      "step": 13872
    },
    {
      "epoch": 1.9443587946741414,
      "grad_norm": 0.7564060688018799,
      "learning_rate": 0.00010816072709878019,
      "loss": 0.067,
      "step": 13873
    },
    {
      "epoch": 1.944498948843728,
      "grad_norm": 0.17752495408058167,
      "learning_rate": 0.00010814637646496053,
      "loss": 0.0152,
      "step": 13874
    },
    {
      "epoch": 1.9446391030133148,
      "grad_norm": 0.18348746001720428,
      "learning_rate": 0.00010813202583114086,
      "loss": 0.0124,
      "step": 13875
    },
    {
      "epoch": 1.9447792571829012,
      "grad_norm": 0.37344425916671753,
      "learning_rate": 0.00010811767519732121,
      "loss": 0.0463,
      "step": 13876
    },
    {
      "epoch": 1.9449194113524877,
      "grad_norm": 0.2197548747062683,
      "learning_rate": 0.00010810332456350155,
      "loss": 0.0185,
      "step": 13877
    },
    {
      "epoch": 1.9450595655220742,
      "grad_norm": 0.1182716116309166,
      "learning_rate": 0.00010808897392968188,
      "loss": 0.0099,
      "step": 13878
    },
    {
      "epoch": 1.9451997196916608,
      "grad_norm": 0.8102246522903442,
      "learning_rate": 0.00010807462329586223,
      "loss": 0.0724,
      "step": 13879
    },
    {
      "epoch": 1.9453398738612475,
      "grad_norm": 0.48709583282470703,
      "learning_rate": 0.00010806027266204255,
      "loss": 0.0202,
      "step": 13880
    },
    {
      "epoch": 1.945480028030834,
      "grad_norm": 0.5163368582725525,
      "learning_rate": 0.00010804592202822291,
      "loss": 0.0427,
      "step": 13881
    },
    {
      "epoch": 1.9456201822004204,
      "grad_norm": 1.0289500951766968,
      "learning_rate": 0.00010803157139440325,
      "loss": 0.0403,
      "step": 13882
    },
    {
      "epoch": 1.945760336370007,
      "grad_norm": 1.6088106632232666,
      "learning_rate": 0.00010801722076058358,
      "loss": 0.2601,
      "step": 13883
    },
    {
      "epoch": 1.9459004905395936,
      "grad_norm": 0.749770998954773,
      "learning_rate": 0.00010800287012676392,
      "loss": 0.0889,
      "step": 13884
    },
    {
      "epoch": 1.9460406447091803,
      "grad_norm": 1.7017935514450073,
      "learning_rate": 0.00010798851949294426,
      "loss": 0.2107,
      "step": 13885
    },
    {
      "epoch": 1.9461807988787667,
      "grad_norm": 0.5391717553138733,
      "learning_rate": 0.00010797416885912459,
      "loss": 0.0717,
      "step": 13886
    },
    {
      "epoch": 1.9463209530483532,
      "grad_norm": 0.21586287021636963,
      "learning_rate": 0.00010795981822530495,
      "loss": 0.0241,
      "step": 13887
    },
    {
      "epoch": 1.9464611072179396,
      "grad_norm": 0.23221690952777863,
      "learning_rate": 0.00010794546759148528,
      "loss": 0.0254,
      "step": 13888
    },
    {
      "epoch": 1.9466012613875263,
      "grad_norm": 0.6500512957572937,
      "learning_rate": 0.00010793111695766562,
      "loss": 0.0439,
      "step": 13889
    },
    {
      "epoch": 1.9467414155571128,
      "grad_norm": 0.25360721349716187,
      "learning_rate": 0.00010791676632384596,
      "loss": 0.0482,
      "step": 13890
    },
    {
      "epoch": 1.9468815697266995,
      "grad_norm": 0.46393564343452454,
      "learning_rate": 0.00010790241569002629,
      "loss": 0.0297,
      "step": 13891
    },
    {
      "epoch": 1.947021723896286,
      "grad_norm": 0.22118176519870758,
      "learning_rate": 0.00010788806505620665,
      "loss": 0.018,
      "step": 13892
    },
    {
      "epoch": 1.9471618780658724,
      "grad_norm": 0.6282756924629211,
      "learning_rate": 0.00010787371442238699,
      "loss": 0.0809,
      "step": 13893
    },
    {
      "epoch": 1.947302032235459,
      "grad_norm": 0.3335075080394745,
      "learning_rate": 0.00010785936378856732,
      "loss": 0.0357,
      "step": 13894
    },
    {
      "epoch": 1.9474421864050455,
      "grad_norm": 0.09476706385612488,
      "learning_rate": 0.00010784501315474766,
      "loss": 0.0141,
      "step": 13895
    },
    {
      "epoch": 1.9475823405746322,
      "grad_norm": 0.24879112839698792,
      "learning_rate": 0.00010783066252092799,
      "loss": 0.0437,
      "step": 13896
    },
    {
      "epoch": 1.9477224947442187,
      "grad_norm": 0.2767227590084076,
      "learning_rate": 0.00010781631188710834,
      "loss": 0.0426,
      "step": 13897
    },
    {
      "epoch": 1.9478626489138051,
      "grad_norm": 0.3350938856601715,
      "learning_rate": 0.00010780196125328868,
      "loss": 0.0444,
      "step": 13898
    },
    {
      "epoch": 1.9480028030833916,
      "grad_norm": 0.2501748204231262,
      "learning_rate": 0.00010778761061946901,
      "loss": 0.0387,
      "step": 13899
    },
    {
      "epoch": 1.9481429572529783,
      "grad_norm": 0.1859581470489502,
      "learning_rate": 0.00010777325998564936,
      "loss": 0.0202,
      "step": 13900
    },
    {
      "epoch": 1.948283111422565,
      "grad_norm": 0.5712987780570984,
      "learning_rate": 0.0001077589093518297,
      "loss": 0.049,
      "step": 13901
    },
    {
      "epoch": 1.9484232655921514,
      "grad_norm": 0.3544265925884247,
      "learning_rate": 0.00010774455871801003,
      "loss": 0.0345,
      "step": 13902
    },
    {
      "epoch": 1.9485634197617379,
      "grad_norm": 0.21755433082580566,
      "learning_rate": 0.00010773020808419038,
      "loss": 0.0378,
      "step": 13903
    },
    {
      "epoch": 1.9487035739313243,
      "grad_norm": 0.29317066073417664,
      "learning_rate": 0.00010771585745037071,
      "loss": 0.0343,
      "step": 13904
    },
    {
      "epoch": 1.948843728100911,
      "grad_norm": 0.36446142196655273,
      "learning_rate": 0.00010770150681655105,
      "loss": 0.0468,
      "step": 13905
    },
    {
      "epoch": 1.9489838822704977,
      "grad_norm": 0.29610416293144226,
      "learning_rate": 0.0001076871561827314,
      "loss": 0.0319,
      "step": 13906
    },
    {
      "epoch": 1.9491240364400841,
      "grad_norm": 0.4245496988296509,
      "learning_rate": 0.00010767280554891172,
      "loss": 0.0499,
      "step": 13907
    },
    {
      "epoch": 1.9492641906096706,
      "grad_norm": 0.6212261319160461,
      "learning_rate": 0.00010765845491509208,
      "loss": 0.063,
      "step": 13908
    },
    {
      "epoch": 1.949404344779257,
      "grad_norm": 0.19561579823493958,
      "learning_rate": 0.00010764410428127242,
      "loss": 0.0469,
      "step": 13909
    },
    {
      "epoch": 1.9495444989488437,
      "grad_norm": 0.24210990965366364,
      "learning_rate": 0.00010762975364745275,
      "loss": 0.0574,
      "step": 13910
    },
    {
      "epoch": 1.9496846531184304,
      "grad_norm": 0.4476587474346161,
      "learning_rate": 0.00010761540301363309,
      "loss": 0.026,
      "step": 13911
    },
    {
      "epoch": 1.9498248072880169,
      "grad_norm": 0.2300807535648346,
      "learning_rate": 0.00010760105237981345,
      "loss": 0.0641,
      "step": 13912
    },
    {
      "epoch": 1.9499649614576033,
      "grad_norm": 0.18532782793045044,
      "learning_rate": 0.00010758670174599378,
      "loss": 0.049,
      "step": 13913
    },
    {
      "epoch": 1.9501051156271898,
      "grad_norm": 0.17468349635601044,
      "learning_rate": 0.00010757235111217412,
      "loss": 0.0349,
      "step": 13914
    },
    {
      "epoch": 1.9502452697967765,
      "grad_norm": 0.22240892052650452,
      "learning_rate": 0.00010755800047835445,
      "loss": 0.0599,
      "step": 13915
    },
    {
      "epoch": 1.9503854239663632,
      "grad_norm": 0.2816687822341919,
      "learning_rate": 0.00010754364984453479,
      "loss": 0.0578,
      "step": 13916
    },
    {
      "epoch": 1.9505255781359496,
      "grad_norm": 0.3398972749710083,
      "learning_rate": 0.00010752929921071513,
      "loss": 0.1,
      "step": 13917
    },
    {
      "epoch": 1.950665732305536,
      "grad_norm": 0.25512760877609253,
      "learning_rate": 0.00010751494857689546,
      "loss": 0.016,
      "step": 13918
    },
    {
      "epoch": 1.9508058864751225,
      "grad_norm": 0.18065519630908966,
      "learning_rate": 0.00010750059794307581,
      "loss": 0.0249,
      "step": 13919
    },
    {
      "epoch": 1.9509460406447092,
      "grad_norm": 0.36691734194755554,
      "learning_rate": 0.00010748624730925616,
      "loss": 0.0273,
      "step": 13920
    },
    {
      "epoch": 1.9510861948142957,
      "grad_norm": 0.061061400920152664,
      "learning_rate": 0.00010747189667543649,
      "loss": 0.0049,
      "step": 13921
    },
    {
      "epoch": 1.9512263489838824,
      "grad_norm": 0.31019407510757446,
      "learning_rate": 0.00010745754604161683,
      "loss": 0.0385,
      "step": 13922
    },
    {
      "epoch": 1.9513665031534688,
      "grad_norm": 0.2908509075641632,
      "learning_rate": 0.00010744319540779716,
      "loss": 0.0921,
      "step": 13923
    },
    {
      "epoch": 1.9515066573230553,
      "grad_norm": 0.30684125423431396,
      "learning_rate": 0.00010742884477397751,
      "loss": 0.0318,
      "step": 13924
    },
    {
      "epoch": 1.9516468114926417,
      "grad_norm": 0.15897582471370697,
      "learning_rate": 0.00010741449414015785,
      "loss": 0.0238,
      "step": 13925
    },
    {
      "epoch": 1.9517869656622284,
      "grad_norm": 0.2699630558490753,
      "learning_rate": 0.00010740014350633818,
      "loss": 0.069,
      "step": 13926
    },
    {
      "epoch": 1.951927119831815,
      "grad_norm": 2.184220790863037,
      "learning_rate": 0.00010738579287251852,
      "loss": 0.0643,
      "step": 13927
    },
    {
      "epoch": 1.9520672740014016,
      "grad_norm": 0.27847978472709656,
      "learning_rate": 0.00010737144223869888,
      "loss": 0.0424,
      "step": 13928
    },
    {
      "epoch": 1.952207428170988,
      "grad_norm": 0.5898178219795227,
      "learning_rate": 0.00010735709160487921,
      "loss": 0.0426,
      "step": 13929
    },
    {
      "epoch": 1.9523475823405745,
      "grad_norm": 0.9590038657188416,
      "learning_rate": 0.00010734274097105955,
      "loss": 0.1187,
      "step": 13930
    },
    {
      "epoch": 1.9524877365101612,
      "grad_norm": 0.3945847153663635,
      "learning_rate": 0.00010732839033723988,
      "loss": 0.0351,
      "step": 13931
    },
    {
      "epoch": 1.9526278906797478,
      "grad_norm": 1.1869257688522339,
      "learning_rate": 0.00010731403970342022,
      "loss": 0.179,
      "step": 13932
    },
    {
      "epoch": 1.9527680448493343,
      "grad_norm": 2.1791369915008545,
      "learning_rate": 0.00010729968906960056,
      "loss": 0.1583,
      "step": 13933
    },
    {
      "epoch": 1.9529081990189208,
      "grad_norm": 0.1371636986732483,
      "learning_rate": 0.00010728533843578089,
      "loss": 0.0067,
      "step": 13934
    },
    {
      "epoch": 1.9530483531885072,
      "grad_norm": 2.6841821670532227,
      "learning_rate": 0.00010727098780196125,
      "loss": 0.3381,
      "step": 13935
    },
    {
      "epoch": 1.953188507358094,
      "grad_norm": 0.11485984176397324,
      "learning_rate": 0.00010725663716814159,
      "loss": 0.0203,
      "step": 13936
    },
    {
      "epoch": 1.9533286615276806,
      "grad_norm": 0.7508218288421631,
      "learning_rate": 0.00010724228653432192,
      "loss": 0.0697,
      "step": 13937
    },
    {
      "epoch": 1.953468815697267,
      "grad_norm": 0.211881622672081,
      "learning_rate": 0.00010722793590050226,
      "loss": 0.0559,
      "step": 13938
    },
    {
      "epoch": 1.9536089698668535,
      "grad_norm": 0.12432222813367844,
      "learning_rate": 0.00010721358526668259,
      "loss": 0.0316,
      "step": 13939
    },
    {
      "epoch": 1.95374912403644,
      "grad_norm": 0.2746027112007141,
      "learning_rate": 0.00010719923463286294,
      "loss": 0.0131,
      "step": 13940
    },
    {
      "epoch": 1.9538892782060266,
      "grad_norm": 0.3552861511707306,
      "learning_rate": 0.00010718488399904329,
      "loss": 0.0408,
      "step": 13941
    },
    {
      "epoch": 1.9540294323756133,
      "grad_norm": 0.4911704957485199,
      "learning_rate": 0.00010717053336522362,
      "loss": 0.0735,
      "step": 13942
    },
    {
      "epoch": 1.9541695865451998,
      "grad_norm": 0.5915058851242065,
      "learning_rate": 0.00010715618273140396,
      "loss": 0.0907,
      "step": 13943
    },
    {
      "epoch": 1.9543097407147862,
      "grad_norm": 0.08755096048116684,
      "learning_rate": 0.00010714183209758431,
      "loss": 0.0105,
      "step": 13944
    },
    {
      "epoch": 1.9544498948843727,
      "grad_norm": 0.4453171491622925,
      "learning_rate": 0.00010712748146376464,
      "loss": 0.0501,
      "step": 13945
    },
    {
      "epoch": 1.9545900490539594,
      "grad_norm": 0.4702605903148651,
      "learning_rate": 0.00010711313082994498,
      "loss": 0.0249,
      "step": 13946
    },
    {
      "epoch": 1.9547302032235458,
      "grad_norm": 0.2845425605773926,
      "learning_rate": 0.00010709878019612533,
      "loss": 0.0333,
      "step": 13947
    },
    {
      "epoch": 1.9548703573931325,
      "grad_norm": 0.09130755811929703,
      "learning_rate": 0.00010708442956230565,
      "loss": 0.0128,
      "step": 13948
    },
    {
      "epoch": 1.955010511562719,
      "grad_norm": 0.12655296921730042,
      "learning_rate": 0.000107070078928486,
      "loss": 0.0131,
      "step": 13949
    },
    {
      "epoch": 1.9551506657323054,
      "grad_norm": 0.4547109603881836,
      "learning_rate": 0.00010705572829466633,
      "loss": 0.0392,
      "step": 13950
    },
    {
      "epoch": 1.9552908199018921,
      "grad_norm": 0.1656562089920044,
      "learning_rate": 0.00010704137766084668,
      "loss": 0.0682,
      "step": 13951
    },
    {
      "epoch": 1.9554309740714786,
      "grad_norm": 0.3922971487045288,
      "learning_rate": 0.00010702702702702702,
      "loss": 0.0484,
      "step": 13952
    },
    {
      "epoch": 1.9555711282410653,
      "grad_norm": 0.23750469088554382,
      "learning_rate": 0.00010701267639320735,
      "loss": 0.0517,
      "step": 13953
    },
    {
      "epoch": 1.9557112824106517,
      "grad_norm": 0.25085195899009705,
      "learning_rate": 0.0001069983257593877,
      "loss": 0.049,
      "step": 13954
    },
    {
      "epoch": 1.9558514365802382,
      "grad_norm": 0.49034756422042847,
      "learning_rate": 0.00010698397512556805,
      "loss": 0.0161,
      "step": 13955
    },
    {
      "epoch": 1.9559915907498246,
      "grad_norm": 0.15123486518859863,
      "learning_rate": 0.00010696962449174838,
      "loss": 0.0427,
      "step": 13956
    },
    {
      "epoch": 1.9561317449194113,
      "grad_norm": 0.43760210275650024,
      "learning_rate": 0.00010695527385792872,
      "loss": 0.0825,
      "step": 13957
    },
    {
      "epoch": 1.956271899088998,
      "grad_norm": 0.11341866850852966,
      "learning_rate": 0.00010694092322410905,
      "loss": 0.0199,
      "step": 13958
    },
    {
      "epoch": 1.9564120532585845,
      "grad_norm": 0.48485666513442993,
      "learning_rate": 0.00010692657259028939,
      "loss": 0.0445,
      "step": 13959
    },
    {
      "epoch": 1.956552207428171,
      "grad_norm": 0.19133444130420685,
      "learning_rate": 0.00010691222195646975,
      "loss": 0.0452,
      "step": 13960
    },
    {
      "epoch": 1.9566923615977574,
      "grad_norm": 0.18332277238368988,
      "learning_rate": 0.00010689787132265008,
      "loss": 0.0379,
      "step": 13961
    },
    {
      "epoch": 1.956832515767344,
      "grad_norm": 0.5678544640541077,
      "learning_rate": 0.00010688352068883042,
      "loss": 0.12,
      "step": 13962
    },
    {
      "epoch": 1.9569726699369308,
      "grad_norm": 0.17859381437301636,
      "learning_rate": 0.00010686917005501076,
      "loss": 0.0176,
      "step": 13963
    },
    {
      "epoch": 1.9571128241065172,
      "grad_norm": 0.12010553479194641,
      "learning_rate": 0.00010685481942119109,
      "loss": 0.0151,
      "step": 13964
    },
    {
      "epoch": 1.9572529782761037,
      "grad_norm": 0.22215235233306885,
      "learning_rate": 0.00010684046878737143,
      "loss": 0.0355,
      "step": 13965
    },
    {
      "epoch": 1.9573931324456901,
      "grad_norm": 0.42693641781806946,
      "learning_rate": 0.00010682611815355176,
      "loss": 0.0793,
      "step": 13966
    },
    {
      "epoch": 1.9575332866152768,
      "grad_norm": 0.3330225646495819,
      "learning_rate": 0.00010681176751973211,
      "loss": 0.0533,
      "step": 13967
    },
    {
      "epoch": 1.9576734407848635,
      "grad_norm": 0.27477332949638367,
      "learning_rate": 0.00010679741688591246,
      "loss": 0.0213,
      "step": 13968
    },
    {
      "epoch": 1.95781359495445,
      "grad_norm": 0.22648563981056213,
      "learning_rate": 0.00010678306625209278,
      "loss": 0.042,
      "step": 13969
    },
    {
      "epoch": 1.9579537491240364,
      "grad_norm": 0.23525995016098022,
      "learning_rate": 0.00010676871561827313,
      "loss": 0.0189,
      "step": 13970
    },
    {
      "epoch": 1.9580939032936229,
      "grad_norm": 0.22855354845523834,
      "learning_rate": 0.00010675436498445348,
      "loss": 0.0298,
      "step": 13971
    },
    {
      "epoch": 1.9582340574632096,
      "grad_norm": 0.7233769297599792,
      "learning_rate": 0.00010674001435063381,
      "loss": 0.1841,
      "step": 13972
    },
    {
      "epoch": 1.9583742116327962,
      "grad_norm": 0.12312700599431992,
      "learning_rate": 0.00010672566371681415,
      "loss": 0.0065,
      "step": 13973
    },
    {
      "epoch": 1.9585143658023827,
      "grad_norm": 0.7623341679573059,
      "learning_rate": 0.00010671131308299448,
      "loss": 0.0476,
      "step": 13974
    },
    {
      "epoch": 1.9586545199719692,
      "grad_norm": 0.484224796295166,
      "learning_rate": 0.00010669696244917482,
      "loss": 0.0998,
      "step": 13975
    },
    {
      "epoch": 1.9587946741415556,
      "grad_norm": 0.8296681642532349,
      "learning_rate": 0.00010668261181535518,
      "loss": 0.0459,
      "step": 13976
    },
    {
      "epoch": 1.9589348283111423,
      "grad_norm": 0.12233518064022064,
      "learning_rate": 0.00010666826118153551,
      "loss": 0.0157,
      "step": 13977
    },
    {
      "epoch": 1.9590749824807288,
      "grad_norm": 0.029155075550079346,
      "learning_rate": 0.00010665391054771585,
      "loss": 0.0019,
      "step": 13978
    },
    {
      "epoch": 1.9592151366503154,
      "grad_norm": 0.20978274941444397,
      "learning_rate": 0.00010663955991389619,
      "loss": 0.0165,
      "step": 13979
    },
    {
      "epoch": 1.959355290819902,
      "grad_norm": 0.5570623874664307,
      "learning_rate": 0.00010662520928007652,
      "loss": 0.188,
      "step": 13980
    },
    {
      "epoch": 1.9594954449894884,
      "grad_norm": 0.4756470322608948,
      "learning_rate": 0.00010661085864625686,
      "loss": 0.042,
      "step": 13981
    },
    {
      "epoch": 1.959635599159075,
      "grad_norm": 0.17194361984729767,
      "learning_rate": 0.00010659650801243719,
      "loss": 0.0291,
      "step": 13982
    },
    {
      "epoch": 1.9597757533286615,
      "grad_norm": 1.010820746421814,
      "learning_rate": 0.00010658215737861755,
      "loss": 0.1147,
      "step": 13983
    },
    {
      "epoch": 1.9599159074982482,
      "grad_norm": 0.3724798262119293,
      "learning_rate": 0.00010656780674479789,
      "loss": 0.041,
      "step": 13984
    },
    {
      "epoch": 1.9600560616678346,
      "grad_norm": 0.9576537013053894,
      "learning_rate": 0.00010655345611097822,
      "loss": 0.0944,
      "step": 13985
    },
    {
      "epoch": 1.960196215837421,
      "grad_norm": 0.1325252801179886,
      "learning_rate": 0.00010653910547715856,
      "loss": 0.0217,
      "step": 13986
    },
    {
      "epoch": 1.9603363700070076,
      "grad_norm": 0.2720743417739868,
      "learning_rate": 0.00010652475484333892,
      "loss": 0.0436,
      "step": 13987
    },
    {
      "epoch": 1.9604765241765942,
      "grad_norm": 0.6176046133041382,
      "learning_rate": 0.00010651040420951924,
      "loss": 0.0949,
      "step": 13988
    },
    {
      "epoch": 1.960616678346181,
      "grad_norm": 0.311335027217865,
      "learning_rate": 0.00010649605357569959,
      "loss": 0.0793,
      "step": 13989
    },
    {
      "epoch": 1.9607568325157674,
      "grad_norm": 0.3555140495300293,
      "learning_rate": 0.00010648170294187993,
      "loss": 0.0415,
      "step": 13990
    },
    {
      "epoch": 1.9608969866853538,
      "grad_norm": 0.4476960301399231,
      "learning_rate": 0.00010646735230806026,
      "loss": 0.0423,
      "step": 13991
    },
    {
      "epoch": 1.9610371408549403,
      "grad_norm": 0.49478137493133545,
      "learning_rate": 0.00010645300167424061,
      "loss": 0.0595,
      "step": 13992
    },
    {
      "epoch": 1.961177295024527,
      "grad_norm": 0.43799328804016113,
      "learning_rate": 0.00010643865104042094,
      "loss": 0.1071,
      "step": 13993
    },
    {
      "epoch": 1.9613174491941137,
      "grad_norm": 0.4966183304786682,
      "learning_rate": 0.00010642430040660128,
      "loss": 0.1129,
      "step": 13994
    },
    {
      "epoch": 1.9614576033637001,
      "grad_norm": 0.33611536026000977,
      "learning_rate": 0.00010640994977278163,
      "loss": 0.0442,
      "step": 13995
    },
    {
      "epoch": 1.9615977575332866,
      "grad_norm": 0.27683529257774353,
      "learning_rate": 0.00010639559913896195,
      "loss": 0.0546,
      "step": 13996
    },
    {
      "epoch": 1.961737911702873,
      "grad_norm": 0.15661093592643738,
      "learning_rate": 0.0001063812485051423,
      "loss": 0.0289,
      "step": 13997
    },
    {
      "epoch": 1.9618780658724597,
      "grad_norm": 0.16219112277030945,
      "learning_rate": 0.00010636689787132265,
      "loss": 0.039,
      "step": 13998
    },
    {
      "epoch": 1.9620182200420464,
      "grad_norm": 0.22393590211868286,
      "learning_rate": 0.00010635254723750298,
      "loss": 0.0355,
      "step": 13999
    },
    {
      "epoch": 1.9621583742116329,
      "grad_norm": 0.4806915819644928,
      "learning_rate": 0.00010633819660368332,
      "loss": 0.0757,
      "step": 14000
    },
    {
      "epoch": 1.9622985283812193,
      "grad_norm": 0.2986256182193756,
      "learning_rate": 0.00010632384596986365,
      "loss": 0.0324,
      "step": 14001
    },
    {
      "epoch": 1.9624386825508058,
      "grad_norm": 0.20005184412002563,
      "learning_rate": 0.000106309495336044,
      "loss": 0.0834,
      "step": 14002
    },
    {
      "epoch": 1.9625788367203925,
      "grad_norm": 0.19226433336734772,
      "learning_rate": 0.00010629514470222435,
      "loss": 0.0302,
      "step": 14003
    },
    {
      "epoch": 1.9627189908899791,
      "grad_norm": 0.15790244936943054,
      "learning_rate": 0.00010628079406840468,
      "loss": 0.0353,
      "step": 14004
    },
    {
      "epoch": 1.9628591450595656,
      "grad_norm": 0.25119757652282715,
      "learning_rate": 0.00010626644343458502,
      "loss": 0.0287,
      "step": 14005
    },
    {
      "epoch": 1.962999299229152,
      "grad_norm": 0.20292729139328003,
      "learning_rate": 0.00010625209280076536,
      "loss": 0.0289,
      "step": 14006
    },
    {
      "epoch": 1.9631394533987385,
      "grad_norm": 0.19640420377254486,
      "learning_rate": 0.00010623774216694569,
      "loss": 0.0327,
      "step": 14007
    },
    {
      "epoch": 1.9632796075683252,
      "grad_norm": 0.20290902256965637,
      "learning_rate": 0.00010622339153312605,
      "loss": 0.0412,
      "step": 14008
    },
    {
      "epoch": 1.9634197617379117,
      "grad_norm": 0.3083365857601166,
      "learning_rate": 0.00010620904089930637,
      "loss": 0.0434,
      "step": 14009
    },
    {
      "epoch": 1.9635599159074983,
      "grad_norm": 0.38759079575538635,
      "learning_rate": 0.00010619469026548672,
      "loss": 0.0354,
      "step": 14010
    },
    {
      "epoch": 1.9637000700770848,
      "grad_norm": 0.20216983556747437,
      "learning_rate": 0.00010618033963166706,
      "loss": 0.0542,
      "step": 14011
    },
    {
      "epoch": 1.9638402242466713,
      "grad_norm": 0.20141395926475525,
      "learning_rate": 0.00010616598899784739,
      "loss": 0.0393,
      "step": 14012
    },
    {
      "epoch": 1.9639803784162577,
      "grad_norm": 0.20983704924583435,
      "learning_rate": 0.00010615163836402773,
      "loss": 0.0142,
      "step": 14013
    },
    {
      "epoch": 1.9641205325858444,
      "grad_norm": 0.26330599188804626,
      "learning_rate": 0.00010613728773020809,
      "loss": 0.0542,
      "step": 14014
    },
    {
      "epoch": 1.964260686755431,
      "grad_norm": 0.1540800929069519,
      "learning_rate": 0.00010612293709638841,
      "loss": 0.0411,
      "step": 14015
    },
    {
      "epoch": 1.9644008409250175,
      "grad_norm": 0.3687482178211212,
      "learning_rate": 0.00010610858646256876,
      "loss": 0.0716,
      "step": 14016
    },
    {
      "epoch": 1.964540995094604,
      "grad_norm": 0.24748332798480988,
      "learning_rate": 0.00010609423582874908,
      "loss": 0.0309,
      "step": 14017
    },
    {
      "epoch": 1.9646811492641905,
      "grad_norm": 0.3238883316516876,
      "learning_rate": 0.00010607988519492943,
      "loss": 0.088,
      "step": 14018
    },
    {
      "epoch": 1.9648213034337771,
      "grad_norm": 0.3410910665988922,
      "learning_rate": 0.00010606553456110978,
      "loss": 0.0896,
      "step": 14019
    },
    {
      "epoch": 1.9649614576033638,
      "grad_norm": 0.1531667560338974,
      "learning_rate": 0.00010605118392729011,
      "loss": 0.023,
      "step": 14020
    },
    {
      "epoch": 1.9651016117729503,
      "grad_norm": 0.24498090147972107,
      "learning_rate": 0.00010603683329347045,
      "loss": 0.0586,
      "step": 14021
    },
    {
      "epoch": 1.9652417659425367,
      "grad_norm": 0.11039438843727112,
      "learning_rate": 0.0001060224826596508,
      "loss": 0.0131,
      "step": 14022
    },
    {
      "epoch": 1.9653819201121232,
      "grad_norm": 0.4878736734390259,
      "learning_rate": 0.00010600813202583112,
      "loss": 0.0508,
      "step": 14023
    },
    {
      "epoch": 1.9655220742817099,
      "grad_norm": 0.270585298538208,
      "learning_rate": 0.00010599378139201148,
      "loss": 0.0605,
      "step": 14024
    },
    {
      "epoch": 1.9656622284512966,
      "grad_norm": 0.4651467800140381,
      "learning_rate": 0.00010597943075819182,
      "loss": 0.0542,
      "step": 14025
    },
    {
      "epoch": 1.965802382620883,
      "grad_norm": 0.4888472259044647,
      "learning_rate": 0.00010596508012437215,
      "loss": 0.0455,
      "step": 14026
    },
    {
      "epoch": 1.9659425367904695,
      "grad_norm": 0.23348116874694824,
      "learning_rate": 0.00010595072949055249,
      "loss": 0.0358,
      "step": 14027
    },
    {
      "epoch": 1.966082690960056,
      "grad_norm": 0.505035400390625,
      "learning_rate": 0.00010593637885673282,
      "loss": 0.0683,
      "step": 14028
    },
    {
      "epoch": 1.9662228451296426,
      "grad_norm": 0.5941540598869324,
      "learning_rate": 0.00010592202822291316,
      "loss": 0.0607,
      "step": 14029
    },
    {
      "epoch": 1.9663629992992293,
      "grad_norm": 1.0807098150253296,
      "learning_rate": 0.00010590767758909352,
      "loss": 0.055,
      "step": 14030
    },
    {
      "epoch": 1.9665031534688158,
      "grad_norm": 0.4120032787322998,
      "learning_rate": 0.00010589332695527385,
      "loss": 0.0313,
      "step": 14031
    },
    {
      "epoch": 1.9666433076384022,
      "grad_norm": 0.519198477268219,
      "learning_rate": 0.00010587897632145419,
      "loss": 0.0519,
      "step": 14032
    },
    {
      "epoch": 1.9667834618079887,
      "grad_norm": 0.6556109189987183,
      "learning_rate": 0.00010586462568763453,
      "loss": 0.0692,
      "step": 14033
    },
    {
      "epoch": 1.9669236159775754,
      "grad_norm": 1.7057074308395386,
      "learning_rate": 0.00010585027505381486,
      "loss": 0.2809,
      "step": 14034
    },
    {
      "epoch": 1.9670637701471618,
      "grad_norm": 1.1197335720062256,
      "learning_rate": 0.00010583592441999522,
      "loss": 0.0688,
      "step": 14035
    },
    {
      "epoch": 1.9672039243167485,
      "grad_norm": 0.16459771990776062,
      "learning_rate": 0.00010582157378617554,
      "loss": 0.0272,
      "step": 14036
    },
    {
      "epoch": 1.967344078486335,
      "grad_norm": 0.20394645631313324,
      "learning_rate": 0.00010580722315235589,
      "loss": 0.0428,
      "step": 14037
    },
    {
      "epoch": 1.9674842326559214,
      "grad_norm": 0.31463733315467834,
      "learning_rate": 0.00010579287251853623,
      "loss": 0.1215,
      "step": 14038
    },
    {
      "epoch": 1.967624386825508,
      "grad_norm": 0.3070388436317444,
      "learning_rate": 0.00010577852188471656,
      "loss": 0.1286,
      "step": 14039
    },
    {
      "epoch": 1.9677645409950946,
      "grad_norm": 0.34140148758888245,
      "learning_rate": 0.00010576417125089691,
      "loss": 0.0406,
      "step": 14040
    },
    {
      "epoch": 1.9679046951646813,
      "grad_norm": 0.16881799697875977,
      "learning_rate": 0.00010574982061707725,
      "loss": 0.0208,
      "step": 14041
    },
    {
      "epoch": 1.9680448493342677,
      "grad_norm": 0.1559797078371048,
      "learning_rate": 0.00010573546998325758,
      "loss": 0.0175,
      "step": 14042
    },
    {
      "epoch": 1.9681850035038542,
      "grad_norm": 0.567139208316803,
      "learning_rate": 0.00010572111934943793,
      "loss": 0.0233,
      "step": 14043
    },
    {
      "epoch": 1.9683251576734406,
      "grad_norm": 0.08762841671705246,
      "learning_rate": 0.00010570676871561825,
      "loss": 0.014,
      "step": 14044
    },
    {
      "epoch": 1.9684653118430273,
      "grad_norm": 0.1431482434272766,
      "learning_rate": 0.0001056924180817986,
      "loss": 0.0149,
      "step": 14045
    },
    {
      "epoch": 1.968605466012614,
      "grad_norm": 0.20340314507484436,
      "learning_rate": 0.00010567806744797895,
      "loss": 0.0328,
      "step": 14046
    },
    {
      "epoch": 1.9687456201822005,
      "grad_norm": 0.504148542881012,
      "learning_rate": 0.00010566371681415928,
      "loss": 0.0497,
      "step": 14047
    },
    {
      "epoch": 1.968885774351787,
      "grad_norm": 0.1360362470149994,
      "learning_rate": 0.00010564936618033962,
      "loss": 0.0257,
      "step": 14048
    },
    {
      "epoch": 1.9690259285213734,
      "grad_norm": 0.5313668251037598,
      "learning_rate": 0.00010563501554651996,
      "loss": 0.0756,
      "step": 14049
    },
    {
      "epoch": 1.96916608269096,
      "grad_norm": 0.2683827877044678,
      "learning_rate": 0.00010562066491270029,
      "loss": 0.0511,
      "step": 14050
    },
    {
      "epoch": 1.9693062368605467,
      "grad_norm": 0.32181647419929504,
      "learning_rate": 0.00010560631427888065,
      "loss": 0.0396,
      "step": 14051
    },
    {
      "epoch": 1.9694463910301332,
      "grad_norm": 0.2120240032672882,
      "learning_rate": 0.00010559196364506098,
      "loss": 0.0154,
      "step": 14052
    },
    {
      "epoch": 1.9695865451997197,
      "grad_norm": 0.4073394238948822,
      "learning_rate": 0.00010557761301124132,
      "loss": 0.0753,
      "step": 14053
    },
    {
      "epoch": 1.969726699369306,
      "grad_norm": 0.16407668590545654,
      "learning_rate": 0.00010556326237742166,
      "loss": 0.0392,
      "step": 14054
    },
    {
      "epoch": 1.9698668535388928,
      "grad_norm": 0.4192562401294708,
      "learning_rate": 0.00010554891174360199,
      "loss": 0.0356,
      "step": 14055
    },
    {
      "epoch": 1.9700070077084795,
      "grad_norm": 0.16742397844791412,
      "learning_rate": 0.00010553456110978235,
      "loss": 0.0675,
      "step": 14056
    },
    {
      "epoch": 1.970147161878066,
      "grad_norm": 0.14298264682292938,
      "learning_rate": 0.00010552021047596269,
      "loss": 0.0501,
      "step": 14057
    },
    {
      "epoch": 1.9702873160476524,
      "grad_norm": 0.6647476553916931,
      "learning_rate": 0.00010550585984214302,
      "loss": 0.0715,
      "step": 14058
    },
    {
      "epoch": 1.9704274702172389,
      "grad_norm": 0.2449726015329361,
      "learning_rate": 0.00010549150920832336,
      "loss": 0.0394,
      "step": 14059
    },
    {
      "epoch": 1.9705676243868255,
      "grad_norm": 0.3318365514278412,
      "learning_rate": 0.0001054771585745037,
      "loss": 0.0409,
      "step": 14060
    },
    {
      "epoch": 1.9707077785564122,
      "grad_norm": 0.09152483940124512,
      "learning_rate": 0.00010546280794068403,
      "loss": 0.0149,
      "step": 14061
    },
    {
      "epoch": 1.9708479327259987,
      "grad_norm": 0.29178494215011597,
      "learning_rate": 0.00010544845730686438,
      "loss": 0.0379,
      "step": 14062
    },
    {
      "epoch": 1.9709880868955851,
      "grad_norm": 0.4171029031276703,
      "learning_rate": 0.00010543410667304471,
      "loss": 0.1133,
      "step": 14063
    },
    {
      "epoch": 1.9711282410651716,
      "grad_norm": 0.18766044080257416,
      "learning_rate": 0.00010541975603922506,
      "loss": 0.0386,
      "step": 14064
    },
    {
      "epoch": 1.9712683952347583,
      "grad_norm": 0.1554962396621704,
      "learning_rate": 0.0001054054054054054,
      "loss": 0.036,
      "step": 14065
    },
    {
      "epoch": 1.9714085494043447,
      "grad_norm": 0.24288548529148102,
      "learning_rate": 0.00010539105477158573,
      "loss": 0.0656,
      "step": 14066
    },
    {
      "epoch": 1.9715487035739314,
      "grad_norm": 0.9484646320343018,
      "learning_rate": 0.00010537670413776608,
      "loss": 0.1434,
      "step": 14067
    },
    {
      "epoch": 1.9716888577435179,
      "grad_norm": 0.26609301567077637,
      "learning_rate": 0.00010536235350394642,
      "loss": 0.0392,
      "step": 14068
    },
    {
      "epoch": 1.9718290119131043,
      "grad_norm": 0.29580777883529663,
      "learning_rate": 0.00010534800287012675,
      "loss": 0.0468,
      "step": 14069
    },
    {
      "epoch": 1.9719691660826908,
      "grad_norm": 0.23241661489009857,
      "learning_rate": 0.0001053336522363071,
      "loss": 0.0368,
      "step": 14070
    },
    {
      "epoch": 1.9721093202522775,
      "grad_norm": 0.4995517432689667,
      "learning_rate": 0.00010531930160248742,
      "loss": 0.081,
      "step": 14071
    },
    {
      "epoch": 1.9722494744218642,
      "grad_norm": 0.37326884269714355,
      "learning_rate": 0.00010530495096866778,
      "loss": 0.0372,
      "step": 14072
    },
    {
      "epoch": 1.9723896285914506,
      "grad_norm": 0.3576323390007019,
      "learning_rate": 0.00010529060033484812,
      "loss": 0.0441,
      "step": 14073
    },
    {
      "epoch": 1.972529782761037,
      "grad_norm": 0.6952236890792847,
      "learning_rate": 0.00010527624970102845,
      "loss": 0.033,
      "step": 14074
    },
    {
      "epoch": 1.9726699369306235,
      "grad_norm": 0.23206666111946106,
      "learning_rate": 0.00010526189906720879,
      "loss": 0.0346,
      "step": 14075
    },
    {
      "epoch": 1.9728100911002102,
      "grad_norm": 0.2733270823955536,
      "learning_rate": 0.00010524754843338913,
      "loss": 0.0595,
      "step": 14076
    },
    {
      "epoch": 1.972950245269797,
      "grad_norm": 0.2883847653865814,
      "learning_rate": 0.00010523319779956946,
      "loss": 0.0419,
      "step": 14077
    },
    {
      "epoch": 1.9730903994393834,
      "grad_norm": 0.4231919050216675,
      "learning_rate": 0.00010521884716574982,
      "loss": 0.0554,
      "step": 14078
    },
    {
      "epoch": 1.9732305536089698,
      "grad_norm": 0.30283451080322266,
      "learning_rate": 0.00010520449653193015,
      "loss": 0.0238,
      "step": 14079
    },
    {
      "epoch": 1.9733707077785563,
      "grad_norm": 0.07608125358819962,
      "learning_rate": 0.00010519014589811049,
      "loss": 0.0061,
      "step": 14080
    },
    {
      "epoch": 1.973510861948143,
      "grad_norm": 2.808887481689453,
      "learning_rate": 0.00010517579526429083,
      "loss": 0.0529,
      "step": 14081
    },
    {
      "epoch": 1.9736510161177296,
      "grad_norm": 0.5500365495681763,
      "learning_rate": 0.00010516144463047116,
      "loss": 0.0316,
      "step": 14082
    },
    {
      "epoch": 1.973791170287316,
      "grad_norm": 0.5552535653114319,
      "learning_rate": 0.00010514709399665151,
      "loss": 0.1317,
      "step": 14083
    },
    {
      "epoch": 1.9739313244569026,
      "grad_norm": 5.861975193023682,
      "learning_rate": 0.00010513274336283186,
      "loss": 0.3374,
      "step": 14084
    },
    {
      "epoch": 1.974071478626489,
      "grad_norm": 0.9901712536811829,
      "learning_rate": 0.00010511839272901219,
      "loss": 0.1485,
      "step": 14085
    },
    {
      "epoch": 1.9742116327960757,
      "grad_norm": 0.118989959359169,
      "learning_rate": 0.00010510404209519253,
      "loss": 0.0317,
      "step": 14086
    },
    {
      "epoch": 1.9743517869656624,
      "grad_norm": 0.25722867250442505,
      "learning_rate": 0.00010508969146137286,
      "loss": 0.0414,
      "step": 14087
    },
    {
      "epoch": 1.9744919411352488,
      "grad_norm": 0.19089168310165405,
      "learning_rate": 0.00010507534082755321,
      "loss": 0.0287,
      "step": 14088
    },
    {
      "epoch": 1.9746320953048353,
      "grad_norm": 0.3456478714942932,
      "learning_rate": 0.00010506099019373355,
      "loss": 0.0849,
      "step": 14089
    },
    {
      "epoch": 1.9747722494744218,
      "grad_norm": 0.331366628408432,
      "learning_rate": 0.00010504663955991388,
      "loss": 0.0549,
      "step": 14090
    },
    {
      "epoch": 1.9749124036440084,
      "grad_norm": 0.3205987513065338,
      "learning_rate": 0.00010503228892609422,
      "loss": 0.0355,
      "step": 14091
    },
    {
      "epoch": 1.975052557813595,
      "grad_norm": 0.6546750068664551,
      "learning_rate": 0.00010501793829227457,
      "loss": 0.0468,
      "step": 14092
    },
    {
      "epoch": 1.9751927119831816,
      "grad_norm": 0.1530638486146927,
      "learning_rate": 0.0001050035876584549,
      "loss": 0.0262,
      "step": 14093
    },
    {
      "epoch": 1.975332866152768,
      "grad_norm": 0.22343894839286804,
      "learning_rate": 0.00010498923702463525,
      "loss": 0.0401,
      "step": 14094
    },
    {
      "epoch": 1.9754730203223545,
      "grad_norm": 0.3264727294445038,
      "learning_rate": 0.00010497488639081558,
      "loss": 0.0321,
      "step": 14095
    },
    {
      "epoch": 1.9756131744919412,
      "grad_norm": 0.29887378215789795,
      "learning_rate": 0.00010496053575699592,
      "loss": 0.084,
      "step": 14096
    },
    {
      "epoch": 1.9757533286615276,
      "grad_norm": 0.19343771040439606,
      "learning_rate": 0.00010494618512317626,
      "loss": 0.0257,
      "step": 14097
    },
    {
      "epoch": 1.9758934828311143,
      "grad_norm": 0.07619180530309677,
      "learning_rate": 0.00010493183448935659,
      "loss": 0.008,
      "step": 14098
    },
    {
      "epoch": 1.9760336370007008,
      "grad_norm": 0.4226943552494049,
      "learning_rate": 0.00010491748385553695,
      "loss": 0.0454,
      "step": 14099
    },
    {
      "epoch": 1.9761737911702872,
      "grad_norm": 0.5277911424636841,
      "learning_rate": 0.00010490313322171729,
      "loss": 0.0566,
      "step": 14100
    },
    {
      "epoch": 1.9763139453398737,
      "grad_norm": 0.6614856123924255,
      "learning_rate": 0.00010488878258789762,
      "loss": 0.0116,
      "step": 14101
    },
    {
      "epoch": 1.9764540995094604,
      "grad_norm": 0.33292704820632935,
      "learning_rate": 0.00010487443195407796,
      "loss": 0.0419,
      "step": 14102
    },
    {
      "epoch": 1.976594253679047,
      "grad_norm": 0.7209381461143494,
      "learning_rate": 0.00010486008132025832,
      "loss": 0.0298,
      "step": 14103
    },
    {
      "epoch": 1.9767344078486335,
      "grad_norm": 0.17458872497081757,
      "learning_rate": 0.00010484573068643864,
      "loss": 0.0318,
      "step": 14104
    },
    {
      "epoch": 1.97687456201822,
      "grad_norm": 0.35672834515571594,
      "learning_rate": 0.00010483138005261899,
      "loss": 0.0539,
      "step": 14105
    },
    {
      "epoch": 1.9770147161878064,
      "grad_norm": 0.5295037627220154,
      "learning_rate": 0.00010481702941879932,
      "loss": 0.0458,
      "step": 14106
    },
    {
      "epoch": 1.9771548703573931,
      "grad_norm": 0.05617177113890648,
      "learning_rate": 0.00010480267878497966,
      "loss": 0.013,
      "step": 14107
    },
    {
      "epoch": 1.9772950245269798,
      "grad_norm": 0.28012824058532715,
      "learning_rate": 0.00010478832815116,
      "loss": 0.039,
      "step": 14108
    },
    {
      "epoch": 1.9774351786965663,
      "grad_norm": 0.23234710097312927,
      "learning_rate": 0.00010477397751734034,
      "loss": 0.041,
      "step": 14109
    },
    {
      "epoch": 1.9775753328661527,
      "grad_norm": 0.5331157445907593,
      "learning_rate": 0.00010475962688352068,
      "loss": 0.0196,
      "step": 14110
    },
    {
      "epoch": 1.9777154870357392,
      "grad_norm": 0.40266895294189453,
      "learning_rate": 0.00010474527624970103,
      "loss": 0.0819,
      "step": 14111
    },
    {
      "epoch": 1.9778556412053259,
      "grad_norm": 0.3187224864959717,
      "learning_rate": 0.00010473092561588135,
      "loss": 0.0358,
      "step": 14112
    },
    {
      "epoch": 1.9779957953749125,
      "grad_norm": 0.6463882327079773,
      "learning_rate": 0.0001047165749820617,
      "loss": 0.0945,
      "step": 14113
    },
    {
      "epoch": 1.978135949544499,
      "grad_norm": 0.22417816519737244,
      "learning_rate": 0.00010470222434824203,
      "loss": 0.0319,
      "step": 14114
    },
    {
      "epoch": 1.9782761037140855,
      "grad_norm": 0.7379664182662964,
      "learning_rate": 0.00010468787371442238,
      "loss": 0.0565,
      "step": 14115
    },
    {
      "epoch": 1.978416257883672,
      "grad_norm": 0.17094933986663818,
      "learning_rate": 0.00010467352308060272,
      "loss": 0.0134,
      "step": 14116
    },
    {
      "epoch": 1.9785564120532586,
      "grad_norm": 0.0786542072892189,
      "learning_rate": 0.00010465917244678305,
      "loss": 0.0116,
      "step": 14117
    },
    {
      "epoch": 1.9786965662228453,
      "grad_norm": 0.41802674531936646,
      "learning_rate": 0.0001046448218129634,
      "loss": 0.0422,
      "step": 14118
    },
    {
      "epoch": 1.9788367203924317,
      "grad_norm": 0.29712462425231934,
      "learning_rate": 0.00010463047117914375,
      "loss": 0.0981,
      "step": 14119
    },
    {
      "epoch": 1.9789768745620182,
      "grad_norm": 0.1856394112110138,
      "learning_rate": 0.00010461612054532408,
      "loss": 0.0453,
      "step": 14120
    },
    {
      "epoch": 1.9791170287316047,
      "grad_norm": 0.23553521931171417,
      "learning_rate": 0.00010460176991150442,
      "loss": 0.0364,
      "step": 14121
    },
    {
      "epoch": 1.9792571829011913,
      "grad_norm": 0.6078153848648071,
      "learning_rate": 0.00010458741927768475,
      "loss": 0.0384,
      "step": 14122
    },
    {
      "epoch": 1.9793973370707778,
      "grad_norm": 0.5511221885681152,
      "learning_rate": 0.00010457306864386509,
      "loss": 0.04,
      "step": 14123
    },
    {
      "epoch": 1.9795374912403645,
      "grad_norm": 0.9764310121536255,
      "learning_rate": 0.00010455871801004543,
      "loss": 0.081,
      "step": 14124
    },
    {
      "epoch": 1.979677645409951,
      "grad_norm": 0.9053546786308289,
      "learning_rate": 0.00010454436737622577,
      "loss": 0.09,
      "step": 14125
    },
    {
      "epoch": 1.9798177995795374,
      "grad_norm": 0.22252118587493896,
      "learning_rate": 0.00010453001674240612,
      "loss": 0.0271,
      "step": 14126
    },
    {
      "epoch": 1.979957953749124,
      "grad_norm": 0.1598093956708908,
      "learning_rate": 0.00010451566610858646,
      "loss": 0.0133,
      "step": 14127
    },
    {
      "epoch": 1.9800981079187105,
      "grad_norm": 0.4796890914440155,
      "learning_rate": 0.00010450131547476679,
      "loss": 0.0758,
      "step": 14128
    },
    {
      "epoch": 1.9802382620882972,
      "grad_norm": 0.3614867329597473,
      "learning_rate": 0.00010448696484094713,
      "loss": 0.0299,
      "step": 14129
    },
    {
      "epoch": 1.9803784162578837,
      "grad_norm": 0.13194267451763153,
      "learning_rate": 0.00010447261420712746,
      "loss": 0.0171,
      "step": 14130
    },
    {
      "epoch": 1.9805185704274701,
      "grad_norm": 0.598003089427948,
      "learning_rate": 0.00010445826357330781,
      "loss": 0.1502,
      "step": 14131
    },
    {
      "epoch": 1.9806587245970566,
      "grad_norm": 0.7155991196632385,
      "learning_rate": 0.00010444391293948816,
      "loss": 0.0321,
      "step": 14132
    },
    {
      "epoch": 1.9807988787666433,
      "grad_norm": 0.490131676197052,
      "learning_rate": 0.00010442956230566848,
      "loss": 0.0418,
      "step": 14133
    },
    {
      "epoch": 1.98093903293623,
      "grad_norm": 0.47156602144241333,
      "learning_rate": 0.00010441521167184883,
      "loss": 0.0303,
      "step": 14134
    },
    {
      "epoch": 1.9810791871058164,
      "grad_norm": 1.3970472812652588,
      "learning_rate": 0.00010440086103802918,
      "loss": 0.0927,
      "step": 14135
    },
    {
      "epoch": 1.981219341275403,
      "grad_norm": 0.4822499454021454,
      "learning_rate": 0.00010438651040420951,
      "loss": 0.0566,
      "step": 14136
    },
    {
      "epoch": 1.9813594954449893,
      "grad_norm": 0.20475226640701294,
      "learning_rate": 0.00010437215977038985,
      "loss": 0.0389,
      "step": 14137
    },
    {
      "epoch": 1.981499649614576,
      "grad_norm": 0.551353931427002,
      "learning_rate": 0.0001043578091365702,
      "loss": 0.0355,
      "step": 14138
    },
    {
      "epoch": 1.9816398037841627,
      "grad_norm": 0.24566631019115448,
      "learning_rate": 0.00010434345850275052,
      "loss": 0.0694,
      "step": 14139
    },
    {
      "epoch": 1.9817799579537492,
      "grad_norm": 0.49406033754348755,
      "learning_rate": 0.00010432910786893087,
      "loss": 0.0594,
      "step": 14140
    },
    {
      "epoch": 1.9819201121233356,
      "grad_norm": 0.20493844151496887,
      "learning_rate": 0.00010431475723511121,
      "loss": 0.0938,
      "step": 14141
    },
    {
      "epoch": 1.982060266292922,
      "grad_norm": 0.4261718690395355,
      "learning_rate": 0.00010430040660129155,
      "loss": 0.0726,
      "step": 14142
    },
    {
      "epoch": 1.9822004204625088,
      "grad_norm": 0.22277750074863434,
      "learning_rate": 0.00010428605596747189,
      "loss": 0.0169,
      "step": 14143
    },
    {
      "epoch": 1.9823405746320955,
      "grad_norm": 0.21987660229206085,
      "learning_rate": 0.00010427170533365222,
      "loss": 0.0213,
      "step": 14144
    },
    {
      "epoch": 1.982480728801682,
      "grad_norm": 0.46280214190483093,
      "learning_rate": 0.00010425735469983256,
      "loss": 0.0525,
      "step": 14145
    },
    {
      "epoch": 1.9826208829712684,
      "grad_norm": 0.3055768609046936,
      "learning_rate": 0.00010424300406601292,
      "loss": 0.0706,
      "step": 14146
    },
    {
      "epoch": 1.9827610371408548,
      "grad_norm": 0.2504782974720001,
      "learning_rate": 0.00010422865343219325,
      "loss": 0.0256,
      "step": 14147
    },
    {
      "epoch": 1.9829011913104415,
      "grad_norm": 0.09164685755968094,
      "learning_rate": 0.00010421430279837359,
      "loss": 0.0091,
      "step": 14148
    },
    {
      "epoch": 1.9830413454800282,
      "grad_norm": 0.643365740776062,
      "learning_rate": 0.00010419995216455392,
      "loss": 0.0428,
      "step": 14149
    },
    {
      "epoch": 1.9831814996496147,
      "grad_norm": 0.2546672523021698,
      "learning_rate": 0.00010418560153073426,
      "loss": 0.0175,
      "step": 14150
    },
    {
      "epoch": 1.9833216538192011,
      "grad_norm": 0.18915030360221863,
      "learning_rate": 0.00010417125089691462,
      "loss": 0.012,
      "step": 14151
    },
    {
      "epoch": 1.9834618079887876,
      "grad_norm": 0.24554553627967834,
      "learning_rate": 0.00010415690026309494,
      "loss": 0.0397,
      "step": 14152
    },
    {
      "epoch": 1.9836019621583743,
      "grad_norm": 0.9851786494255066,
      "learning_rate": 0.00010414254962927529,
      "loss": 0.1146,
      "step": 14153
    },
    {
      "epoch": 1.9837421163279607,
      "grad_norm": 0.1032838299870491,
      "learning_rate": 0.00010412819899545563,
      "loss": 0.0091,
      "step": 14154
    },
    {
      "epoch": 1.9838822704975474,
      "grad_norm": 0.24289239943027496,
      "learning_rate": 0.00010411384836163596,
      "loss": 0.0494,
      "step": 14155
    },
    {
      "epoch": 1.9840224246671339,
      "grad_norm": 0.27520713210105896,
      "learning_rate": 0.0001040994977278163,
      "loss": 0.0189,
      "step": 14156
    },
    {
      "epoch": 1.9841625788367203,
      "grad_norm": 0.1735120564699173,
      "learning_rate": 0.00010408514709399664,
      "loss": 0.0124,
      "step": 14157
    },
    {
      "epoch": 1.9843027330063068,
      "grad_norm": 0.273334264755249,
      "learning_rate": 0.00010407079646017698,
      "loss": 0.0555,
      "step": 14158
    },
    {
      "epoch": 1.9844428871758935,
      "grad_norm": 0.18390794098377228,
      "learning_rate": 0.00010405644582635733,
      "loss": 0.0344,
      "step": 14159
    },
    {
      "epoch": 1.9845830413454801,
      "grad_norm": 0.36073118448257446,
      "learning_rate": 0.00010404209519253765,
      "loss": 0.0507,
      "step": 14160
    },
    {
      "epoch": 1.9847231955150666,
      "grad_norm": 0.21442502737045288,
      "learning_rate": 0.000104027744558718,
      "loss": 0.0286,
      "step": 14161
    },
    {
      "epoch": 1.984863349684653,
      "grad_norm": 0.26301631331443787,
      "learning_rate": 0.00010401339392489835,
      "loss": 0.0713,
      "step": 14162
    },
    {
      "epoch": 1.9850035038542395,
      "grad_norm": 0.4175613820552826,
      "learning_rate": 0.00010399904329107868,
      "loss": 0.0496,
      "step": 14163
    },
    {
      "epoch": 1.9851436580238262,
      "grad_norm": 0.24470297992229462,
      "learning_rate": 0.00010398469265725902,
      "loss": 0.0578,
      "step": 14164
    },
    {
      "epoch": 1.9852838121934129,
      "grad_norm": 0.32175448536872864,
      "learning_rate": 0.00010397034202343935,
      "loss": 0.0808,
      "step": 14165
    },
    {
      "epoch": 1.9854239663629993,
      "grad_norm": 0.9071288108825684,
      "learning_rate": 0.00010395599138961969,
      "loss": 0.0697,
      "step": 14166
    },
    {
      "epoch": 1.9855641205325858,
      "grad_norm": 0.35176801681518555,
      "learning_rate": 0.00010394164075580005,
      "loss": 0.0163,
      "step": 14167
    },
    {
      "epoch": 1.9857042747021723,
      "grad_norm": 0.15793417394161224,
      "learning_rate": 0.00010392729012198038,
      "loss": 0.0281,
      "step": 14168
    },
    {
      "epoch": 1.985844428871759,
      "grad_norm": 0.3313787281513214,
      "learning_rate": 0.00010391293948816072,
      "loss": 0.0496,
      "step": 14169
    },
    {
      "epoch": 1.9859845830413456,
      "grad_norm": 0.5487057566642761,
      "learning_rate": 0.00010389858885434106,
      "loss": 0.0862,
      "step": 14170
    },
    {
      "epoch": 1.986124737210932,
      "grad_norm": 0.14403605461120605,
      "learning_rate": 0.00010388423822052139,
      "loss": 0.0097,
      "step": 14171
    },
    {
      "epoch": 1.9862648913805185,
      "grad_norm": 0.2174629420042038,
      "learning_rate": 0.00010386988758670173,
      "loss": 0.0222,
      "step": 14172
    },
    {
      "epoch": 1.986405045550105,
      "grad_norm": 0.8246282339096069,
      "learning_rate": 0.00010385553695288209,
      "loss": 0.0449,
      "step": 14173
    },
    {
      "epoch": 1.9865451997196917,
      "grad_norm": 0.5739886164665222,
      "learning_rate": 0.00010384118631906242,
      "loss": 0.022,
      "step": 14174
    },
    {
      "epoch": 1.9866853538892784,
      "grad_norm": 1.486023187637329,
      "learning_rate": 0.00010382683568524276,
      "loss": 0.063,
      "step": 14175
    },
    {
      "epoch": 1.9868255080588648,
      "grad_norm": 0.362218976020813,
      "learning_rate": 0.00010381248505142309,
      "loss": 0.069,
      "step": 14176
    },
    {
      "epoch": 1.9869656622284513,
      "grad_norm": 0.4330422282218933,
      "learning_rate": 0.00010379813441760343,
      "loss": 0.0433,
      "step": 14177
    },
    {
      "epoch": 1.9871058163980377,
      "grad_norm": 0.4412398636341095,
      "learning_rate": 0.00010378378378378378,
      "loss": 0.018,
      "step": 14178
    },
    {
      "epoch": 1.9872459705676244,
      "grad_norm": 0.1058543473482132,
      "learning_rate": 0.00010376943314996411,
      "loss": 0.0073,
      "step": 14179
    },
    {
      "epoch": 1.9873861247372109,
      "grad_norm": 0.1804308146238327,
      "learning_rate": 0.00010375508251614446,
      "loss": 0.0422,
      "step": 14180
    },
    {
      "epoch": 1.9875262789067976,
      "grad_norm": 0.2622183561325073,
      "learning_rate": 0.0001037407318823248,
      "loss": 0.032,
      "step": 14181
    },
    {
      "epoch": 1.987666433076384,
      "grad_norm": 0.3941948413848877,
      "learning_rate": 0.00010372638124850513,
      "loss": 0.0447,
      "step": 14182
    },
    {
      "epoch": 1.9878065872459705,
      "grad_norm": 1.0045377016067505,
      "learning_rate": 0.00010371203061468548,
      "loss": 0.0499,
      "step": 14183
    },
    {
      "epoch": 1.9879467414155572,
      "grad_norm": 0.19761860370635986,
      "learning_rate": 0.00010369767998086581,
      "loss": 0.0295,
      "step": 14184
    },
    {
      "epoch": 1.9880868955851436,
      "grad_norm": 2.321589469909668,
      "learning_rate": 0.00010368332934704615,
      "loss": 0.1747,
      "step": 14185
    },
    {
      "epoch": 1.9882270497547303,
      "grad_norm": 0.28479689359664917,
      "learning_rate": 0.0001036689787132265,
      "loss": 0.0238,
      "step": 14186
    },
    {
      "epoch": 1.9883672039243168,
      "grad_norm": 0.24905124306678772,
      "learning_rate": 0.00010365462807940682,
      "loss": 0.0152,
      "step": 14187
    },
    {
      "epoch": 1.9885073580939032,
      "grad_norm": 0.16113144159317017,
      "learning_rate": 0.00010364027744558717,
      "loss": 0.0241,
      "step": 14188
    },
    {
      "epoch": 1.9886475122634897,
      "grad_norm": 0.2999199628829956,
      "learning_rate": 0.00010362592681176752,
      "loss": 0.0387,
      "step": 14189
    },
    {
      "epoch": 1.9887876664330764,
      "grad_norm": 0.301959365606308,
      "learning_rate": 0.00010361157617794785,
      "loss": 0.0583,
      "step": 14190
    },
    {
      "epoch": 1.988927820602663,
      "grad_norm": 0.21860761940479279,
      "learning_rate": 0.00010359722554412819,
      "loss": 0.0207,
      "step": 14191
    },
    {
      "epoch": 1.9890679747722495,
      "grad_norm": 0.260749489068985,
      "learning_rate": 0.00010358287491030852,
      "loss": 0.0166,
      "step": 14192
    },
    {
      "epoch": 1.989208128941836,
      "grad_norm": 0.3328326940536499,
      "learning_rate": 0.00010356852427648886,
      "loss": 0.0972,
      "step": 14193
    },
    {
      "epoch": 1.9893482831114224,
      "grad_norm": 0.2738496661186218,
      "learning_rate": 0.00010355417364266922,
      "loss": 0.0228,
      "step": 14194
    },
    {
      "epoch": 1.989488437281009,
      "grad_norm": 0.20044095814228058,
      "learning_rate": 0.00010353982300884955,
      "loss": 0.0264,
      "step": 14195
    },
    {
      "epoch": 1.9896285914505958,
      "grad_norm": 0.322775661945343,
      "learning_rate": 0.00010352547237502989,
      "loss": 0.0439,
      "step": 14196
    },
    {
      "epoch": 1.9897687456201822,
      "grad_norm": 0.24739785492420197,
      "learning_rate": 0.00010351112174121023,
      "loss": 0.0369,
      "step": 14197
    },
    {
      "epoch": 1.9899088997897687,
      "grad_norm": 0.11036693304777145,
      "learning_rate": 0.00010349677110739056,
      "loss": 0.0128,
      "step": 14198
    },
    {
      "epoch": 1.9900490539593552,
      "grad_norm": 0.4505460858345032,
      "learning_rate": 0.00010348242047357092,
      "loss": 0.0628,
      "step": 14199
    },
    {
      "epoch": 1.9901892081289418,
      "grad_norm": 0.28146296739578247,
      "learning_rate": 0.00010346806983975124,
      "loss": 0.0522,
      "step": 14200
    },
    {
      "epoch": 1.9903293622985285,
      "grad_norm": 0.23998992145061493,
      "learning_rate": 0.00010345371920593159,
      "loss": 0.0427,
      "step": 14201
    },
    {
      "epoch": 1.990469516468115,
      "grad_norm": 0.19448859989643097,
      "learning_rate": 0.00010343936857211193,
      "loss": 0.0192,
      "step": 14202
    },
    {
      "epoch": 1.9906096706377014,
      "grad_norm": 0.22926148772239685,
      "learning_rate": 0.00010342501793829226,
      "loss": 0.0447,
      "step": 14203
    },
    {
      "epoch": 1.990749824807288,
      "grad_norm": 0.2581430673599243,
      "learning_rate": 0.0001034106673044726,
      "loss": 0.0284,
      "step": 14204
    },
    {
      "epoch": 1.9908899789768746,
      "grad_norm": 0.19479742646217346,
      "learning_rate": 0.00010339631667065295,
      "loss": 0.0281,
      "step": 14205
    },
    {
      "epoch": 1.9910301331464613,
      "grad_norm": 0.3242775499820709,
      "learning_rate": 0.00010338196603683328,
      "loss": 0.0247,
      "step": 14206
    },
    {
      "epoch": 1.9911702873160477,
      "grad_norm": 0.33048728108406067,
      "learning_rate": 0.00010336761540301362,
      "loss": 0.1099,
      "step": 14207
    },
    {
      "epoch": 1.9913104414856342,
      "grad_norm": 0.08701975643634796,
      "learning_rate": 0.00010335326476919397,
      "loss": 0.0075,
      "step": 14208
    },
    {
      "epoch": 1.9914505956552206,
      "grad_norm": 0.42110908031463623,
      "learning_rate": 0.0001033389141353743,
      "loss": 0.0247,
      "step": 14209
    },
    {
      "epoch": 1.9915907498248073,
      "grad_norm": 0.2061358094215393,
      "learning_rate": 0.00010332456350155465,
      "loss": 0.02,
      "step": 14210
    },
    {
      "epoch": 1.9917309039943938,
      "grad_norm": 0.7582136988639832,
      "learning_rate": 0.00010331021286773498,
      "loss": 0.0274,
      "step": 14211
    },
    {
      "epoch": 1.9918710581639805,
      "grad_norm": 0.18745380640029907,
      "learning_rate": 0.00010329586223391532,
      "loss": 0.0276,
      "step": 14212
    },
    {
      "epoch": 1.992011212333567,
      "grad_norm": 0.26254788041114807,
      "learning_rate": 0.00010328151160009566,
      "loss": 0.0372,
      "step": 14213
    },
    {
      "epoch": 1.9921513665031534,
      "grad_norm": 0.4268210828304291,
      "learning_rate": 0.00010326716096627599,
      "loss": 0.026,
      "step": 14214
    },
    {
      "epoch": 1.99229152067274,
      "grad_norm": 0.8060597777366638,
      "learning_rate": 0.00010325281033245635,
      "loss": 0.0357,
      "step": 14215
    },
    {
      "epoch": 1.9924316748423265,
      "grad_norm": 0.411082923412323,
      "learning_rate": 0.00010323845969863669,
      "loss": 0.0911,
      "step": 14216
    },
    {
      "epoch": 1.9925718290119132,
      "grad_norm": 0.17483334243297577,
      "learning_rate": 0.00010322410906481702,
      "loss": 0.0162,
      "step": 14217
    },
    {
      "epoch": 1.9927119831814997,
      "grad_norm": 0.998288631439209,
      "learning_rate": 0.00010320975843099736,
      "loss": 0.0714,
      "step": 14218
    },
    {
      "epoch": 1.9928521373510861,
      "grad_norm": 0.4254349172115326,
      "learning_rate": 0.00010319540779717769,
      "loss": 0.0398,
      "step": 14219
    },
    {
      "epoch": 1.9929922915206726,
      "grad_norm": 0.6960335969924927,
      "learning_rate": 0.00010318105716335805,
      "loss": 0.1023,
      "step": 14220
    },
    {
      "epoch": 1.9931324456902593,
      "grad_norm": 0.2798539102077484,
      "learning_rate": 0.00010316670652953839,
      "loss": 0.0415,
      "step": 14221
    },
    {
      "epoch": 1.993272599859846,
      "grad_norm": 0.09360259026288986,
      "learning_rate": 0.00010315235589571872,
      "loss": 0.0237,
      "step": 14222
    },
    {
      "epoch": 1.9934127540294324,
      "grad_norm": 0.2600257098674774,
      "learning_rate": 0.00010313800526189906,
      "loss": 0.0222,
      "step": 14223
    },
    {
      "epoch": 1.9935529081990189,
      "grad_norm": 0.5302936434745789,
      "learning_rate": 0.0001031236546280794,
      "loss": 0.0556,
      "step": 14224
    },
    {
      "epoch": 1.9936930623686053,
      "grad_norm": 0.3347436487674713,
      "learning_rate": 0.00010310930399425973,
      "loss": 0.0258,
      "step": 14225
    },
    {
      "epoch": 1.993833216538192,
      "grad_norm": 0.24498754739761353,
      "learning_rate": 0.00010309495336044008,
      "loss": 0.0334,
      "step": 14226
    },
    {
      "epoch": 1.9939733707077787,
      "grad_norm": 0.16731178760528564,
      "learning_rate": 0.00010308060272662041,
      "loss": 0.0148,
      "step": 14227
    },
    {
      "epoch": 1.9941135248773652,
      "grad_norm": 0.6915204524993896,
      "learning_rate": 0.00010306625209280076,
      "loss": 0.0776,
      "step": 14228
    },
    {
      "epoch": 1.9942536790469516,
      "grad_norm": 0.09304513782262802,
      "learning_rate": 0.0001030519014589811,
      "loss": 0.0082,
      "step": 14229
    },
    {
      "epoch": 1.994393833216538,
      "grad_norm": 0.056376006454229355,
      "learning_rate": 0.00010303755082516143,
      "loss": 0.004,
      "step": 14230
    },
    {
      "epoch": 1.9945339873861248,
      "grad_norm": 0.6417027711868286,
      "learning_rate": 0.00010302320019134178,
      "loss": 0.0605,
      "step": 14231
    },
    {
      "epoch": 1.9946741415557114,
      "grad_norm": 0.5420274138450623,
      "learning_rate": 0.00010300884955752212,
      "loss": 0.0969,
      "step": 14232
    },
    {
      "epoch": 1.994814295725298,
      "grad_norm": 0.8372558951377869,
      "learning_rate": 0.00010299449892370245,
      "loss": 0.0789,
      "step": 14233
    },
    {
      "epoch": 1.9949544498948844,
      "grad_norm": 1.303813099861145,
      "learning_rate": 0.0001029801482898828,
      "loss": 0.0694,
      "step": 14234
    },
    {
      "epoch": 1.9950946040644708,
      "grad_norm": 0.2677852213382721,
      "learning_rate": 0.00010296579765606312,
      "loss": 0.0538,
      "step": 14235
    },
    {
      "epoch": 1.9952347582340575,
      "grad_norm": 0.2184418886899948,
      "learning_rate": 0.00010295144702224348,
      "loss": 0.0224,
      "step": 14236
    },
    {
      "epoch": 1.9953749124036442,
      "grad_norm": 0.4506930410861969,
      "learning_rate": 0.00010293709638842382,
      "loss": 0.0767,
      "step": 14237
    },
    {
      "epoch": 1.9955150665732306,
      "grad_norm": 0.31031256914138794,
      "learning_rate": 0.00010292274575460415,
      "loss": 0.0573,
      "step": 14238
    },
    {
      "epoch": 1.995655220742817,
      "grad_norm": 0.16498108208179474,
      "learning_rate": 0.00010290839512078449,
      "loss": 0.0117,
      "step": 14239
    },
    {
      "epoch": 1.9957953749124036,
      "grad_norm": 0.5877910256385803,
      "learning_rate": 0.00010289404448696483,
      "loss": 0.0762,
      "step": 14240
    },
    {
      "epoch": 1.9959355290819902,
      "grad_norm": 0.19379130005836487,
      "learning_rate": 0.00010287969385314516,
      "loss": 0.0345,
      "step": 14241
    },
    {
      "epoch": 1.9960756832515767,
      "grad_norm": 0.13378974795341492,
      "learning_rate": 0.00010286534321932552,
      "loss": 0.0227,
      "step": 14242
    },
    {
      "epoch": 1.9962158374211634,
      "grad_norm": 0.35973796248435974,
      "learning_rate": 0.00010285099258550585,
      "loss": 0.0734,
      "step": 14243
    },
    {
      "epoch": 1.9963559915907498,
      "grad_norm": 0.13812102377414703,
      "learning_rate": 0.00010283664195168619,
      "loss": 0.0127,
      "step": 14244
    },
    {
      "epoch": 1.9964961457603363,
      "grad_norm": 0.3664568364620209,
      "learning_rate": 0.00010282229131786653,
      "loss": 0.0692,
      "step": 14245
    },
    {
      "epoch": 1.9966362999299228,
      "grad_norm": 0.26672065258026123,
      "learning_rate": 0.00010280794068404686,
      "loss": 0.0718,
      "step": 14246
    },
    {
      "epoch": 1.9967764540995094,
      "grad_norm": 0.375955730676651,
      "learning_rate": 0.00010279359005022721,
      "loss": 0.0571,
      "step": 14247
    },
    {
      "epoch": 1.9969166082690961,
      "grad_norm": 0.8177602291107178,
      "learning_rate": 0.00010277923941640756,
      "loss": 0.0375,
      "step": 14248
    },
    {
      "epoch": 1.9970567624386826,
      "grad_norm": 0.31063029170036316,
      "learning_rate": 0.00010276488878258789,
      "loss": 0.1102,
      "step": 14249
    },
    {
      "epoch": 1.997196916608269,
      "grad_norm": 0.6209132075309753,
      "learning_rate": 0.00010275053814876823,
      "loss": 0.1039,
      "step": 14250
    },
    {
      "epoch": 1.9973370707778555,
      "grad_norm": 0.22595059871673584,
      "learning_rate": 0.00010273618751494857,
      "loss": 0.0312,
      "step": 14251
    },
    {
      "epoch": 1.9974772249474422,
      "grad_norm": 0.05967971310019493,
      "learning_rate": 0.00010272183688112891,
      "loss": 0.0067,
      "step": 14252
    },
    {
      "epoch": 1.9976173791170289,
      "grad_norm": 0.2264532446861267,
      "learning_rate": 0.00010270748624730925,
      "loss": 0.0409,
      "step": 14253
    },
    {
      "epoch": 1.9977575332866153,
      "grad_norm": 0.4629777669906616,
      "learning_rate": 0.00010269313561348958,
      "loss": 0.0735,
      "step": 14254
    },
    {
      "epoch": 1.9978976874562018,
      "grad_norm": 0.27865928411483765,
      "learning_rate": 0.00010267878497966992,
      "loss": 0.0429,
      "step": 14255
    },
    {
      "epoch": 1.9980378416257882,
      "grad_norm": 0.21995767951011658,
      "learning_rate": 0.00010266443434585027,
      "loss": 0.0286,
      "step": 14256
    },
    {
      "epoch": 1.998177995795375,
      "grad_norm": 0.2889644503593445,
      "learning_rate": 0.0001026500837120306,
      "loss": 0.0552,
      "step": 14257
    },
    {
      "epoch": 1.9983181499649616,
      "grad_norm": 0.2833872437477112,
      "learning_rate": 0.00010263573307821095,
      "loss": 0.0241,
      "step": 14258
    },
    {
      "epoch": 1.998458304134548,
      "grad_norm": 0.1358339488506317,
      "learning_rate": 0.00010262138244439129,
      "loss": 0.0151,
      "step": 14259
    },
    {
      "epoch": 1.9985984583041345,
      "grad_norm": 0.7646512389183044,
      "learning_rate": 0.00010260703181057162,
      "loss": 0.0738,
      "step": 14260
    },
    {
      "epoch": 1.998738612473721,
      "grad_norm": 0.3783266544342041,
      "learning_rate": 0.00010259268117675196,
      "loss": 0.0595,
      "step": 14261
    },
    {
      "epoch": 1.9988787666433077,
      "grad_norm": 0.6213785409927368,
      "learning_rate": 0.00010257833054293229,
      "loss": 0.1944,
      "step": 14262
    },
    {
      "epoch": 1.9990189208128943,
      "grad_norm": 0.3035310208797455,
      "learning_rate": 0.00010256397990911265,
      "loss": 0.013,
      "step": 14263
    },
    {
      "epoch": 1.9991590749824808,
      "grad_norm": 0.22235266864299774,
      "learning_rate": 0.00010254962927529299,
      "loss": 0.0399,
      "step": 14264
    },
    {
      "epoch": 1.9992992291520673,
      "grad_norm": 0.2720561921596527,
      "learning_rate": 0.00010253527864147332,
      "loss": 0.0261,
      "step": 14265
    },
    {
      "epoch": 1.9994393833216537,
      "grad_norm": 0.2022603452205658,
      "learning_rate": 0.00010252092800765366,
      "loss": 0.019,
      "step": 14266
    },
    {
      "epoch": 1.9995795374912404,
      "grad_norm": 0.4893212616443634,
      "learning_rate": 0.000102506577373834,
      "loss": 0.0503,
      "step": 14267
    },
    {
      "epoch": 1.9997196916608269,
      "grad_norm": 0.283733606338501,
      "learning_rate": 0.00010249222674001434,
      "loss": 0.0101,
      "step": 14268
    },
    {
      "epoch": 1.9998598458304135,
      "grad_norm": 3.0472166538238525,
      "learning_rate": 0.00010247787610619469,
      "loss": 0.2659,
      "step": 14269
    },
    {
      "epoch": 2.0,
      "grad_norm": 3.6867873668670654,
      "learning_rate": 0.00010246352547237502,
      "loss": 0.3818,
      "step": 14270
    },
    {
      "epoch": 2.0001401541695865,
      "grad_norm": 0.23136372864246368,
      "learning_rate": 0.00010244917483855536,
      "loss": 0.0377,
      "step": 14271
    },
    {
      "epoch": 2.000280308339173,
      "grad_norm": 0.3771960735321045,
      "learning_rate": 0.0001024348242047357,
      "loss": 0.0314,
      "step": 14272
    },
    {
      "epoch": 2.00042046250876,
      "grad_norm": 0.18012498319149017,
      "learning_rate": 0.00010242047357091603,
      "loss": 0.0263,
      "step": 14273
    },
    {
      "epoch": 2.0005606166783463,
      "grad_norm": 0.48979452252388,
      "learning_rate": 0.00010240612293709638,
      "loss": 0.0296,
      "step": 14274
    },
    {
      "epoch": 2.0007007708479327,
      "grad_norm": 0.5342593789100647,
      "learning_rate": 0.00010239177230327673,
      "loss": 0.0347,
      "step": 14275
    },
    {
      "epoch": 2.000840925017519,
      "grad_norm": 0.3313027024269104,
      "learning_rate": 0.00010237742166945705,
      "loss": 0.0557,
      "step": 14276
    },
    {
      "epoch": 2.0009810791871057,
      "grad_norm": 0.4729999899864197,
      "learning_rate": 0.0001023630710356374,
      "loss": 0.0338,
      "step": 14277
    },
    {
      "epoch": 2.0011212333566926,
      "grad_norm": 0.18270646035671234,
      "learning_rate": 0.00010234872040181773,
      "loss": 0.0404,
      "step": 14278
    },
    {
      "epoch": 2.001261387526279,
      "grad_norm": 0.2502637803554535,
      "learning_rate": 0.00010233436976799808,
      "loss": 0.0435,
      "step": 14279
    },
    {
      "epoch": 2.0014015416958655,
      "grad_norm": 0.23573294281959534,
      "learning_rate": 0.00010232001913417842,
      "loss": 0.0401,
      "step": 14280
    },
    {
      "epoch": 2.001541695865452,
      "grad_norm": 0.21577972173690796,
      "learning_rate": 0.00010230566850035875,
      "loss": 0.0194,
      "step": 14281
    },
    {
      "epoch": 2.0016818500350384,
      "grad_norm": 0.3586162328720093,
      "learning_rate": 0.0001022913178665391,
      "loss": 0.0583,
      "step": 14282
    },
    {
      "epoch": 2.0018220042046253,
      "grad_norm": 0.35905852913856506,
      "learning_rate": 0.00010227696723271944,
      "loss": 0.0536,
      "step": 14283
    },
    {
      "epoch": 2.0019621583742118,
      "grad_norm": 0.1588849574327469,
      "learning_rate": 0.00010226261659889978,
      "loss": 0.014,
      "step": 14284
    },
    {
      "epoch": 2.0021023125437982,
      "grad_norm": 0.13074587285518646,
      "learning_rate": 0.00010224826596508012,
      "loss": 0.0092,
      "step": 14285
    },
    {
      "epoch": 2.0022424667133847,
      "grad_norm": 0.2219579666852951,
      "learning_rate": 0.00010223391533126046,
      "loss": 0.0183,
      "step": 14286
    },
    {
      "epoch": 2.002382620882971,
      "grad_norm": 0.29415473341941833,
      "learning_rate": 0.00010221956469744079,
      "loss": 0.0787,
      "step": 14287
    },
    {
      "epoch": 2.0025227750525576,
      "grad_norm": 0.4591987133026123,
      "learning_rate": 0.00010220521406362113,
      "loss": 0.0667,
      "step": 14288
    },
    {
      "epoch": 2.0026629292221445,
      "grad_norm": 0.4064667522907257,
      "learning_rate": 0.00010219086342980146,
      "loss": 0.0423,
      "step": 14289
    },
    {
      "epoch": 2.002803083391731,
      "grad_norm": 0.22674909234046936,
      "learning_rate": 0.00010217651279598182,
      "loss": 0.0452,
      "step": 14290
    },
    {
      "epoch": 2.0029432375613174,
      "grad_norm": 0.10505981743335724,
      "learning_rate": 0.00010216216216216216,
      "loss": 0.0111,
      "step": 14291
    },
    {
      "epoch": 2.003083391730904,
      "grad_norm": 0.3039432168006897,
      "learning_rate": 0.00010214781152834249,
      "loss": 0.098,
      "step": 14292
    },
    {
      "epoch": 2.0032235459004903,
      "grad_norm": 0.11148194968700409,
      "learning_rate": 0.00010213346089452283,
      "loss": 0.0048,
      "step": 14293
    },
    {
      "epoch": 2.0033637000700772,
      "grad_norm": 0.11342445015907288,
      "learning_rate": 0.00010211911026070319,
      "loss": 0.0086,
      "step": 14294
    },
    {
      "epoch": 2.0035038542396637,
      "grad_norm": 0.19621862471103668,
      "learning_rate": 0.00010210475962688351,
      "loss": 0.0227,
      "step": 14295
    },
    {
      "epoch": 2.00364400840925,
      "grad_norm": 0.20716015994548798,
      "learning_rate": 0.00010209040899306386,
      "loss": 0.0496,
      "step": 14296
    },
    {
      "epoch": 2.0037841625788366,
      "grad_norm": 0.5497312545776367,
      "learning_rate": 0.00010207605835924418,
      "loss": 0.0781,
      "step": 14297
    },
    {
      "epoch": 2.003924316748423,
      "grad_norm": 0.22204463183879852,
      "learning_rate": 0.00010206170772542453,
      "loss": 0.0081,
      "step": 14298
    },
    {
      "epoch": 2.00406447091801,
      "grad_norm": 0.1989804357290268,
      "learning_rate": 0.00010204735709160487,
      "loss": 0.0221,
      "step": 14299
    },
    {
      "epoch": 2.0042046250875964,
      "grad_norm": 0.26693519949913025,
      "learning_rate": 0.00010203300645778521,
      "loss": 0.0461,
      "step": 14300
    },
    {
      "epoch": 2.004344779257183,
      "grad_norm": 0.11593149602413177,
      "learning_rate": 0.00010201865582396555,
      "loss": 0.0402,
      "step": 14301
    },
    {
      "epoch": 2.0044849334267694,
      "grad_norm": 0.255950927734375,
      "learning_rate": 0.0001020043051901459,
      "loss": 0.0129,
      "step": 14302
    },
    {
      "epoch": 2.004625087596356,
      "grad_norm": 0.1897752434015274,
      "learning_rate": 0.00010198995455632622,
      "loss": 0.015,
      "step": 14303
    },
    {
      "epoch": 2.0047652417659427,
      "grad_norm": 0.8016980886459351,
      "learning_rate": 0.00010197560392250657,
      "loss": 0.1002,
      "step": 14304
    },
    {
      "epoch": 2.004905395935529,
      "grad_norm": 0.0957736074924469,
      "learning_rate": 0.0001019612532886869,
      "loss": 0.0129,
      "step": 14305
    },
    {
      "epoch": 2.0050455501051156,
      "grad_norm": 0.1342475414276123,
      "learning_rate": 0.00010194690265486725,
      "loss": 0.0053,
      "step": 14306
    },
    {
      "epoch": 2.005185704274702,
      "grad_norm": 0.10899920761585236,
      "learning_rate": 0.00010193255202104759,
      "loss": 0.0058,
      "step": 14307
    },
    {
      "epoch": 2.0053258584442886,
      "grad_norm": 0.03692729398608208,
      "learning_rate": 0.00010191820138722792,
      "loss": 0.0022,
      "step": 14308
    },
    {
      "epoch": 2.0054660126138755,
      "grad_norm": 0.08919638395309448,
      "learning_rate": 0.00010190385075340826,
      "loss": 0.0059,
      "step": 14309
    },
    {
      "epoch": 2.005606166783462,
      "grad_norm": 0.23241807520389557,
      "learning_rate": 0.00010188950011958862,
      "loss": 0.0122,
      "step": 14310
    },
    {
      "epoch": 2.0057463209530484,
      "grad_norm": 0.06809527426958084,
      "learning_rate": 0.00010187514948576895,
      "loss": 0.0053,
      "step": 14311
    },
    {
      "epoch": 2.005886475122635,
      "grad_norm": 0.23260357975959778,
      "learning_rate": 0.00010186079885194929,
      "loss": 0.0394,
      "step": 14312
    },
    {
      "epoch": 2.0060266292922213,
      "grad_norm": 0.45001575350761414,
      "learning_rate": 0.00010184644821812962,
      "loss": 0.0242,
      "step": 14313
    },
    {
      "epoch": 2.0061667834618078,
      "grad_norm": 0.46052512526512146,
      "learning_rate": 0.00010183209758430996,
      "loss": 0.0139,
      "step": 14314
    },
    {
      "epoch": 2.0063069376313947,
      "grad_norm": 0.45768025517463684,
      "learning_rate": 0.0001018177469504903,
      "loss": 0.0123,
      "step": 14315
    },
    {
      "epoch": 2.006447091800981,
      "grad_norm": 0.3107963800430298,
      "learning_rate": 0.00010180339631667064,
      "loss": 0.059,
      "step": 14316
    },
    {
      "epoch": 2.0065872459705676,
      "grad_norm": 0.7314419150352478,
      "learning_rate": 0.00010178904568285099,
      "loss": 0.0566,
      "step": 14317
    },
    {
      "epoch": 2.006727400140154,
      "grad_norm": 0.29502925276756287,
      "learning_rate": 0.00010177469504903133,
      "loss": 0.0108,
      "step": 14318
    },
    {
      "epoch": 2.0068675543097405,
      "grad_norm": 0.2972310781478882,
      "learning_rate": 0.00010176034441521166,
      "loss": 0.0128,
      "step": 14319
    },
    {
      "epoch": 2.0070077084793274,
      "grad_norm": 0.7865574955940247,
      "learning_rate": 0.000101745993781392,
      "loss": 0.0275,
      "step": 14320
    },
    {
      "epoch": 2.007147862648914,
      "grad_norm": 0.20951853692531586,
      "learning_rate": 0.00010173164314757235,
      "loss": 0.0219,
      "step": 14321
    },
    {
      "epoch": 2.0072880168185003,
      "grad_norm": 0.3417162299156189,
      "learning_rate": 0.00010171729251375268,
      "loss": 0.1068,
      "step": 14322
    },
    {
      "epoch": 2.007428170988087,
      "grad_norm": 0.18037255108356476,
      "learning_rate": 0.00010170294187993303,
      "loss": 0.073,
      "step": 14323
    },
    {
      "epoch": 2.0075683251576733,
      "grad_norm": 0.5002044439315796,
      "learning_rate": 0.00010168859124611335,
      "loss": 0.0334,
      "step": 14324
    },
    {
      "epoch": 2.00770847932726,
      "grad_norm": 0.24219584465026855,
      "learning_rate": 0.0001016742406122937,
      "loss": 0.0197,
      "step": 14325
    },
    {
      "epoch": 2.0078486334968466,
      "grad_norm": 0.028247453272342682,
      "learning_rate": 0.00010165988997847405,
      "loss": 0.0021,
      "step": 14326
    },
    {
      "epoch": 2.007988787666433,
      "grad_norm": 1.0345940589904785,
      "learning_rate": 0.00010164553934465438,
      "loss": 0.0626,
      "step": 14327
    },
    {
      "epoch": 2.0081289418360195,
      "grad_norm": 0.2338320016860962,
      "learning_rate": 0.00010163118871083472,
      "loss": 0.0308,
      "step": 14328
    },
    {
      "epoch": 2.008269096005606,
      "grad_norm": 0.5210989117622375,
      "learning_rate": 0.00010161683807701506,
      "loss": 0.0639,
      "step": 14329
    },
    {
      "epoch": 2.008409250175193,
      "grad_norm": 0.3688915967941284,
      "learning_rate": 0.00010160248744319539,
      "loss": 0.0263,
      "step": 14330
    },
    {
      "epoch": 2.0085494043447794,
      "grad_norm": 0.08547557145357132,
      "learning_rate": 0.00010158813680937575,
      "loss": 0.0035,
      "step": 14331
    },
    {
      "epoch": 2.008689558514366,
      "grad_norm": 0.12328723818063736,
      "learning_rate": 0.00010157378617555608,
      "loss": 0.0229,
      "step": 14332
    },
    {
      "epoch": 2.0088297126839523,
      "grad_norm": 0.3097442090511322,
      "learning_rate": 0.00010155943554173642,
      "loss": 0.0425,
      "step": 14333
    },
    {
      "epoch": 2.0089698668535387,
      "grad_norm": 0.04403432458639145,
      "learning_rate": 0.00010154508490791676,
      "loss": 0.0048,
      "step": 14334
    },
    {
      "epoch": 2.0091100210231256,
      "grad_norm": 0.21485382318496704,
      "learning_rate": 0.00010153073427409709,
      "loss": 0.0421,
      "step": 14335
    },
    {
      "epoch": 2.009250175192712,
      "grad_norm": 0.15131321549415588,
      "learning_rate": 0.00010151638364027743,
      "loss": 0.0168,
      "step": 14336
    },
    {
      "epoch": 2.0093903293622986,
      "grad_norm": 0.1158590093255043,
      "learning_rate": 0.00010150203300645779,
      "loss": 0.0193,
      "step": 14337
    },
    {
      "epoch": 2.009530483531885,
      "grad_norm": 0.126034677028656,
      "learning_rate": 0.00010148768237263812,
      "loss": 0.0239,
      "step": 14338
    },
    {
      "epoch": 2.0096706377014715,
      "grad_norm": 1.0763146877288818,
      "learning_rate": 0.00010147333173881846,
      "loss": 0.0428,
      "step": 14339
    },
    {
      "epoch": 2.0098107918710584,
      "grad_norm": 0.24444782733917236,
      "learning_rate": 0.00010145898110499879,
      "loss": 0.0231,
      "step": 14340
    },
    {
      "epoch": 2.009950946040645,
      "grad_norm": 0.3235735297203064,
      "learning_rate": 0.00010144463047117913,
      "loss": 0.0782,
      "step": 14341
    },
    {
      "epoch": 2.0100911002102313,
      "grad_norm": 0.3248410224914551,
      "learning_rate": 0.00010143027983735948,
      "loss": 0.0172,
      "step": 14342
    },
    {
      "epoch": 2.0102312543798178,
      "grad_norm": 0.3081238269805908,
      "learning_rate": 0.00010141592920353981,
      "loss": 0.0245,
      "step": 14343
    },
    {
      "epoch": 2.010371408549404,
      "grad_norm": 0.16979824006557465,
      "learning_rate": 0.00010140157856972016,
      "loss": 0.0427,
      "step": 14344
    },
    {
      "epoch": 2.0105115627189907,
      "grad_norm": 0.06817808747291565,
      "learning_rate": 0.0001013872279359005,
      "loss": 0.0035,
      "step": 14345
    },
    {
      "epoch": 2.0106517168885776,
      "grad_norm": 0.21799680590629578,
      "learning_rate": 0.00010137287730208083,
      "loss": 0.0292,
      "step": 14346
    },
    {
      "epoch": 2.010791871058164,
      "grad_norm": 0.17592798173427582,
      "learning_rate": 0.00010135852666826118,
      "loss": 0.0151,
      "step": 14347
    },
    {
      "epoch": 2.0109320252277505,
      "grad_norm": 0.12155366688966751,
      "learning_rate": 0.00010134417603444151,
      "loss": 0.0215,
      "step": 14348
    },
    {
      "epoch": 2.011072179397337,
      "grad_norm": 0.46024125814437866,
      "learning_rate": 0.00010132982540062185,
      "loss": 0.0284,
      "step": 14349
    },
    {
      "epoch": 2.0112123335669234,
      "grad_norm": 0.015230483375489712,
      "learning_rate": 0.0001013154747668022,
      "loss": 0.0008,
      "step": 14350
    },
    {
      "epoch": 2.0113524877365103,
      "grad_norm": 0.03840211033821106,
      "learning_rate": 0.00010130112413298252,
      "loss": 0.0016,
      "step": 14351
    },
    {
      "epoch": 2.011492641906097,
      "grad_norm": 0.2641845643520355,
      "learning_rate": 0.00010128677349916287,
      "loss": 0.0515,
      "step": 14352
    },
    {
      "epoch": 2.0116327960756832,
      "grad_norm": 0.31310683488845825,
      "learning_rate": 0.00010127242286534322,
      "loss": 0.0128,
      "step": 14353
    },
    {
      "epoch": 2.0117729502452697,
      "grad_norm": 0.39936432242393494,
      "learning_rate": 0.00010125807223152355,
      "loss": 0.0242,
      "step": 14354
    },
    {
      "epoch": 2.011913104414856,
      "grad_norm": 0.09522376954555511,
      "learning_rate": 0.00010124372159770389,
      "loss": 0.0044,
      "step": 14355
    },
    {
      "epoch": 2.012053258584443,
      "grad_norm": 0.2599260210990906,
      "learning_rate": 0.00010122937096388423,
      "loss": 0.0094,
      "step": 14356
    },
    {
      "epoch": 2.0121934127540295,
      "grad_norm": 0.1651524305343628,
      "learning_rate": 0.00010121502033006456,
      "loss": 0.0195,
      "step": 14357
    },
    {
      "epoch": 2.012333566923616,
      "grad_norm": 0.6188377737998962,
      "learning_rate": 0.00010120066969624492,
      "loss": 0.118,
      "step": 14358
    },
    {
      "epoch": 2.0124737210932024,
      "grad_norm": 0.11369770765304565,
      "learning_rate": 0.00010118631906242525,
      "loss": 0.0056,
      "step": 14359
    },
    {
      "epoch": 2.012613875262789,
      "grad_norm": 0.2961896061897278,
      "learning_rate": 0.00010117196842860559,
      "loss": 0.0234,
      "step": 14360
    },
    {
      "epoch": 2.012754029432376,
      "grad_norm": 0.3469317555427551,
      "learning_rate": 0.00010115761779478593,
      "loss": 0.0916,
      "step": 14361
    },
    {
      "epoch": 2.0128941836019623,
      "grad_norm": 0.37067559361457825,
      "learning_rate": 0.00010114326716096626,
      "loss": 0.0402,
      "step": 14362
    },
    {
      "epoch": 2.0130343377715487,
      "grad_norm": 0.4539516866207123,
      "learning_rate": 0.00010112891652714662,
      "loss": 0.0261,
      "step": 14363
    },
    {
      "epoch": 2.013174491941135,
      "grad_norm": 0.32531091570854187,
      "learning_rate": 0.00010111456589332696,
      "loss": 0.0446,
      "step": 14364
    },
    {
      "epoch": 2.0133146461107216,
      "grad_norm": 0.262358158826828,
      "learning_rate": 0.00010110021525950729,
      "loss": 0.0312,
      "step": 14365
    },
    {
      "epoch": 2.0134548002803085,
      "grad_norm": 0.3419927954673767,
      "learning_rate": 0.00010108586462568763,
      "loss": 0.0404,
      "step": 14366
    },
    {
      "epoch": 2.013594954449895,
      "grad_norm": 0.9540210962295532,
      "learning_rate": 0.00010107151399186796,
      "loss": 0.0316,
      "step": 14367
    },
    {
      "epoch": 2.0137351086194815,
      "grad_norm": 0.7896899580955505,
      "learning_rate": 0.0001010571633580483,
      "loss": 0.1242,
      "step": 14368
    },
    {
      "epoch": 2.013875262789068,
      "grad_norm": 0.6743713021278381,
      "learning_rate": 0.00010104281272422865,
      "loss": 0.0419,
      "step": 14369
    },
    {
      "epoch": 2.0140154169586544,
      "grad_norm": 2.9574239253997803,
      "learning_rate": 0.00010102846209040898,
      "loss": 0.0469,
      "step": 14370
    },
    {
      "epoch": 2.0141555711282413,
      "grad_norm": 0.11079802364110947,
      "learning_rate": 0.00010101411145658932,
      "loss": 0.0077,
      "step": 14371
    },
    {
      "epoch": 2.0142957252978277,
      "grad_norm": 0.7975444197654724,
      "learning_rate": 0.00010099976082276967,
      "loss": 0.0245,
      "step": 14372
    },
    {
      "epoch": 2.014435879467414,
      "grad_norm": 0.30389925837516785,
      "learning_rate": 0.00010098541018895,
      "loss": 0.0119,
      "step": 14373
    },
    {
      "epoch": 2.0145760336370007,
      "grad_norm": 0.09679289162158966,
      "learning_rate": 0.00010097105955513035,
      "loss": 0.0197,
      "step": 14374
    },
    {
      "epoch": 2.014716187806587,
      "grad_norm": 0.4335888922214508,
      "learning_rate": 0.00010095670892131068,
      "loss": 0.0384,
      "step": 14375
    },
    {
      "epoch": 2.0148563419761736,
      "grad_norm": 0.38530582189559937,
      "learning_rate": 0.00010094235828749102,
      "loss": 0.0594,
      "step": 14376
    },
    {
      "epoch": 2.0149964961457605,
      "grad_norm": 0.13853639364242554,
      "learning_rate": 0.00010092800765367136,
      "loss": 0.0287,
      "step": 14377
    },
    {
      "epoch": 2.015136650315347,
      "grad_norm": 0.23372463881969452,
      "learning_rate": 0.00010091365701985169,
      "loss": 0.0307,
      "step": 14378
    },
    {
      "epoch": 2.0152768044849334,
      "grad_norm": 0.4094722270965576,
      "learning_rate": 0.00010089930638603205,
      "loss": 0.0145,
      "step": 14379
    },
    {
      "epoch": 2.01541695865452,
      "grad_norm": 0.3784569203853607,
      "learning_rate": 0.00010088495575221239,
      "loss": 0.0108,
      "step": 14380
    },
    {
      "epoch": 2.0155571128241063,
      "grad_norm": 0.1826721429824829,
      "learning_rate": 0.00010087060511839272,
      "loss": 0.026,
      "step": 14381
    },
    {
      "epoch": 2.0156972669936932,
      "grad_norm": 0.3056948781013489,
      "learning_rate": 0.00010085625448457306,
      "loss": 0.0313,
      "step": 14382
    },
    {
      "epoch": 2.0158374211632797,
      "grad_norm": 0.3778788149356842,
      "learning_rate": 0.00010084190385075339,
      "loss": 0.0329,
      "step": 14383
    },
    {
      "epoch": 2.015977575332866,
      "grad_norm": 0.15764489769935608,
      "learning_rate": 0.00010082755321693373,
      "loss": 0.0331,
      "step": 14384
    },
    {
      "epoch": 2.0161177295024526,
      "grad_norm": 0.2387782484292984,
      "learning_rate": 0.00010081320258311409,
      "loss": 0.0142,
      "step": 14385
    },
    {
      "epoch": 2.016257883672039,
      "grad_norm": 0.12807060778141022,
      "learning_rate": 0.00010079885194929442,
      "loss": 0.0148,
      "step": 14386
    },
    {
      "epoch": 2.016398037841626,
      "grad_norm": 0.12051426619291306,
      "learning_rate": 0.00010078450131547476,
      "loss": 0.0179,
      "step": 14387
    },
    {
      "epoch": 2.0165381920112124,
      "grad_norm": 0.20849700272083282,
      "learning_rate": 0.0001007701506816551,
      "loss": 0.0101,
      "step": 14388
    },
    {
      "epoch": 2.016678346180799,
      "grad_norm": 0.20255093276500702,
      "learning_rate": 0.00010075580004783543,
      "loss": 0.0095,
      "step": 14389
    },
    {
      "epoch": 2.0168185003503853,
      "grad_norm": 0.1897495687007904,
      "learning_rate": 0.00010074144941401578,
      "loss": 0.0295,
      "step": 14390
    },
    {
      "epoch": 2.016958654519972,
      "grad_norm": 0.3409312963485718,
      "learning_rate": 0.00010072709878019611,
      "loss": 0.0255,
      "step": 14391
    },
    {
      "epoch": 2.0170988086895587,
      "grad_norm": 0.29313597083091736,
      "learning_rate": 0.00010071274814637645,
      "loss": 0.0359,
      "step": 14392
    },
    {
      "epoch": 2.017238962859145,
      "grad_norm": 0.2857000529766083,
      "learning_rate": 0.0001006983975125568,
      "loss": 0.0463,
      "step": 14393
    },
    {
      "epoch": 2.0173791170287316,
      "grad_norm": 0.31748104095458984,
      "learning_rate": 0.00010068404687873713,
      "loss": 0.0403,
      "step": 14394
    },
    {
      "epoch": 2.017519271198318,
      "grad_norm": 0.36438286304473877,
      "learning_rate": 0.00010066969624491748,
      "loss": 0.0432,
      "step": 14395
    },
    {
      "epoch": 2.0176594253679045,
      "grad_norm": 0.12800566852092743,
      "learning_rate": 0.00010065534561109782,
      "loss": 0.0337,
      "step": 14396
    },
    {
      "epoch": 2.0177995795374915,
      "grad_norm": 0.2623644173145294,
      "learning_rate": 0.00010064099497727815,
      "loss": 0.0314,
      "step": 14397
    },
    {
      "epoch": 2.017939733707078,
      "grad_norm": 0.5782424211502075,
      "learning_rate": 0.0001006266443434585,
      "loss": 0.0472,
      "step": 14398
    },
    {
      "epoch": 2.0180798878766644,
      "grad_norm": 0.14538604021072388,
      "learning_rate": 0.00010061229370963884,
      "loss": 0.0162,
      "step": 14399
    },
    {
      "epoch": 2.018220042046251,
      "grad_norm": 0.4025847613811493,
      "learning_rate": 0.00010059794307581916,
      "loss": 0.0559,
      "step": 14400
    },
    {
      "epoch": 2.0183601962158373,
      "grad_norm": 0.31601834297180176,
      "learning_rate": 0.00010058359244199952,
      "loss": 0.018,
      "step": 14401
    },
    {
      "epoch": 2.0185003503854237,
      "grad_norm": 0.06269515305757523,
      "learning_rate": 0.00010056924180817985,
      "loss": 0.0067,
      "step": 14402
    },
    {
      "epoch": 2.0186405045550107,
      "grad_norm": 0.18927812576293945,
      "learning_rate": 0.00010055489117436019,
      "loss": 0.0229,
      "step": 14403
    },
    {
      "epoch": 2.018780658724597,
      "grad_norm": 0.5609269142150879,
      "learning_rate": 0.00010054054054054053,
      "loss": 0.0519,
      "step": 14404
    },
    {
      "epoch": 2.0189208128941836,
      "grad_norm": 0.18753154575824738,
      "learning_rate": 0.00010052618990672086,
      "loss": 0.0349,
      "step": 14405
    },
    {
      "epoch": 2.01906096706377,
      "grad_norm": 0.3656015396118164,
      "learning_rate": 0.00010051183927290122,
      "loss": 0.0285,
      "step": 14406
    },
    {
      "epoch": 2.0192011212333565,
      "grad_norm": 0.7377094626426697,
      "learning_rate": 0.00010049748863908156,
      "loss": 0.0314,
      "step": 14407
    },
    {
      "epoch": 2.0193412754029434,
      "grad_norm": 0.29533520340919495,
      "learning_rate": 0.00010048313800526189,
      "loss": 0.0856,
      "step": 14408
    },
    {
      "epoch": 2.01948142957253,
      "grad_norm": 0.12153347581624985,
      "learning_rate": 0.00010046878737144223,
      "loss": 0.0308,
      "step": 14409
    },
    {
      "epoch": 2.0196215837421163,
      "grad_norm": 0.2440887838602066,
      "learning_rate": 0.00010045443673762256,
      "loss": 0.0115,
      "step": 14410
    },
    {
      "epoch": 2.0197617379117028,
      "grad_norm": 0.09669745713472366,
      "learning_rate": 0.00010044008610380291,
      "loss": 0.0383,
      "step": 14411
    },
    {
      "epoch": 2.0199018920812892,
      "grad_norm": 0.14960606396198273,
      "learning_rate": 0.00010042573546998326,
      "loss": 0.0096,
      "step": 14412
    },
    {
      "epoch": 2.020042046250876,
      "grad_norm": 0.5913904905319214,
      "learning_rate": 0.00010041138483616359,
      "loss": 0.0589,
      "step": 14413
    },
    {
      "epoch": 2.0201822004204626,
      "grad_norm": 0.26117458939552307,
      "learning_rate": 0.00010039703420234393,
      "loss": 0.0338,
      "step": 14414
    },
    {
      "epoch": 2.020322354590049,
      "grad_norm": 0.986532986164093,
      "learning_rate": 0.00010038268356852427,
      "loss": 0.1555,
      "step": 14415
    },
    {
      "epoch": 2.0204625087596355,
      "grad_norm": 0.24483701586723328,
      "learning_rate": 0.0001003683329347046,
      "loss": 0.0344,
      "step": 14416
    },
    {
      "epoch": 2.020602662929222,
      "grad_norm": 0.28337931632995605,
      "learning_rate": 0.00010035398230088495,
      "loss": 0.0268,
      "step": 14417
    },
    {
      "epoch": 2.020742817098809,
      "grad_norm": 0.7726407647132874,
      "learning_rate": 0.00010033963166706528,
      "loss": 0.038,
      "step": 14418
    },
    {
      "epoch": 2.0208829712683953,
      "grad_norm": 0.8181468844413757,
      "learning_rate": 0.00010032528103324562,
      "loss": 0.1987,
      "step": 14419
    },
    {
      "epoch": 2.021023125437982,
      "grad_norm": 1.5980054140090942,
      "learning_rate": 0.00010031093039942597,
      "loss": 0.2307,
      "step": 14420
    },
    {
      "epoch": 2.0211632796075683,
      "grad_norm": 0.11903064697980881,
      "learning_rate": 0.0001002965797656063,
      "loss": 0.0301,
      "step": 14421
    },
    {
      "epoch": 2.0213034337771547,
      "grad_norm": 0.11184151470661163,
      "learning_rate": 0.00010028222913178665,
      "loss": 0.0115,
      "step": 14422
    },
    {
      "epoch": 2.0214435879467416,
      "grad_norm": 0.182241752743721,
      "learning_rate": 0.00010026787849796699,
      "loss": 0.0173,
      "step": 14423
    },
    {
      "epoch": 2.021583742116328,
      "grad_norm": 0.22382840514183044,
      "learning_rate": 0.00010025352786414732,
      "loss": 0.065,
      "step": 14424
    },
    {
      "epoch": 2.0217238962859145,
      "grad_norm": 0.22317323088645935,
      "learning_rate": 0.00010023917723032766,
      "loss": 0.0296,
      "step": 14425
    },
    {
      "epoch": 2.021864050455501,
      "grad_norm": 0.3675713539123535,
      "learning_rate": 0.00010022482659650799,
      "loss": 0.0302,
      "step": 14426
    },
    {
      "epoch": 2.0220042046250875,
      "grad_norm": 0.22817359864711761,
      "learning_rate": 0.00010021047596268835,
      "loss": 0.0716,
      "step": 14427
    },
    {
      "epoch": 2.0221443587946744,
      "grad_norm": 0.2380255162715912,
      "learning_rate": 0.00010019612532886869,
      "loss": 0.0425,
      "step": 14428
    },
    {
      "epoch": 2.022284512964261,
      "grad_norm": 0.10376962274312973,
      "learning_rate": 0.00010018177469504902,
      "loss": 0.0111,
      "step": 14429
    },
    {
      "epoch": 2.0224246671338473,
      "grad_norm": 0.12821175158023834,
      "learning_rate": 0.00010016742406122936,
      "loss": 0.0331,
      "step": 14430
    },
    {
      "epoch": 2.0225648213034337,
      "grad_norm": 0.10619758814573288,
      "learning_rate": 0.0001001530734274097,
      "loss": 0.0196,
      "step": 14431
    },
    {
      "epoch": 2.02270497547302,
      "grad_norm": 0.29811498522758484,
      "learning_rate": 0.00010013872279359003,
      "loss": 0.0524,
      "step": 14432
    },
    {
      "epoch": 2.0228451296426067,
      "grad_norm": 0.28642040491104126,
      "learning_rate": 0.00010012437215977039,
      "loss": 0.0438,
      "step": 14433
    },
    {
      "epoch": 2.0229852838121936,
      "grad_norm": 0.36200645565986633,
      "learning_rate": 0.00010011002152595073,
      "loss": 0.0585,
      "step": 14434
    },
    {
      "epoch": 2.02312543798178,
      "grad_norm": 0.12036851048469543,
      "learning_rate": 0.00010009567089213106,
      "loss": 0.0101,
      "step": 14435
    },
    {
      "epoch": 2.0232655921513665,
      "grad_norm": 0.3903593420982361,
      "learning_rate": 0.0001000813202583114,
      "loss": 0.0319,
      "step": 14436
    },
    {
      "epoch": 2.023405746320953,
      "grad_norm": 0.21410390734672546,
      "learning_rate": 0.00010006696962449173,
      "loss": 0.0153,
      "step": 14437
    },
    {
      "epoch": 2.0235459004905394,
      "grad_norm": 0.1886194944381714,
      "learning_rate": 0.00010005261899067208,
      "loss": 0.017,
      "step": 14438
    },
    {
      "epoch": 2.0236860546601263,
      "grad_norm": 0.13052061200141907,
      "learning_rate": 0.00010003826835685243,
      "loss": 0.012,
      "step": 14439
    },
    {
      "epoch": 2.0238262088297128,
      "grad_norm": 0.09371787309646606,
      "learning_rate": 0.00010002391772303275,
      "loss": 0.0111,
      "step": 14440
    },
    {
      "epoch": 2.023966362999299,
      "grad_norm": 0.18195025622844696,
      "learning_rate": 0.0001000095670892131,
      "loss": 0.034,
      "step": 14441
    },
    {
      "epoch": 2.0241065171688857,
      "grad_norm": 0.0965365394949913,
      "learning_rate": 9.999521645539345e-05,
      "loss": 0.0111,
      "step": 14442
    },
    {
      "epoch": 2.024246671338472,
      "grad_norm": 0.09762502461671829,
      "learning_rate": 9.998086582157378e-05,
      "loss": 0.0125,
      "step": 14443
    },
    {
      "epoch": 2.024386825508059,
      "grad_norm": 0.1168842762708664,
      "learning_rate": 9.996651518775412e-05,
      "loss": 0.0132,
      "step": 14444
    },
    {
      "epoch": 2.0245269796776455,
      "grad_norm": 0.2688584327697754,
      "learning_rate": 9.995216455393445e-05,
      "loss": 0.0656,
      "step": 14445
    },
    {
      "epoch": 2.024667133847232,
      "grad_norm": 0.15135279297828674,
      "learning_rate": 9.99378139201148e-05,
      "loss": 0.0393,
      "step": 14446
    },
    {
      "epoch": 2.0248072880168184,
      "grad_norm": 0.18845799565315247,
      "learning_rate": 9.992346328629514e-05,
      "loss": 0.0143,
      "step": 14447
    },
    {
      "epoch": 2.024947442186405,
      "grad_norm": 0.20573319494724274,
      "learning_rate": 9.990911265247546e-05,
      "loss": 0.0325,
      "step": 14448
    },
    {
      "epoch": 2.025087596355992,
      "grad_norm": 0.16403938829898834,
      "learning_rate": 9.989476201865582e-05,
      "loss": 0.0239,
      "step": 14449
    },
    {
      "epoch": 2.0252277505255782,
      "grad_norm": 0.1833772212266922,
      "learning_rate": 9.988041138483616e-05,
      "loss": 0.0604,
      "step": 14450
    },
    {
      "epoch": 2.0253679046951647,
      "grad_norm": 0.07425392419099808,
      "learning_rate": 9.986606075101649e-05,
      "loss": 0.0049,
      "step": 14451
    },
    {
      "epoch": 2.025508058864751,
      "grad_norm": 0.4004087746143341,
      "learning_rate": 9.985171011719683e-05,
      "loss": 0.029,
      "step": 14452
    },
    {
      "epoch": 2.0256482130343376,
      "grad_norm": 1.3883413076400757,
      "learning_rate": 9.983735948337716e-05,
      "loss": 0.0555,
      "step": 14453
    },
    {
      "epoch": 2.0257883672039245,
      "grad_norm": 0.2745436131954193,
      "learning_rate": 9.982300884955752e-05,
      "loss": 0.0208,
      "step": 14454
    },
    {
      "epoch": 2.025928521373511,
      "grad_norm": 0.2581547796726227,
      "learning_rate": 9.980865821573786e-05,
      "loss": 0.047,
      "step": 14455
    },
    {
      "epoch": 2.0260686755430974,
      "grad_norm": 0.1561075747013092,
      "learning_rate": 9.979430758191819e-05,
      "loss": 0.0206,
      "step": 14456
    },
    {
      "epoch": 2.026208829712684,
      "grad_norm": 0.026349002495408058,
      "learning_rate": 9.977995694809853e-05,
      "loss": 0.0036,
      "step": 14457
    },
    {
      "epoch": 2.0263489838822704,
      "grad_norm": 0.1585310995578766,
      "learning_rate": 9.976560631427889e-05,
      "loss": 0.0176,
      "step": 14458
    },
    {
      "epoch": 2.026489138051857,
      "grad_norm": 0.1944943368434906,
      "learning_rate": 9.975125568045921e-05,
      "loss": 0.0181,
      "step": 14459
    },
    {
      "epoch": 2.0266292922214437,
      "grad_norm": 0.12261159718036652,
      "learning_rate": 9.973690504663956e-05,
      "loss": 0.0143,
      "step": 14460
    },
    {
      "epoch": 2.02676944639103,
      "grad_norm": 0.5750231742858887,
      "learning_rate": 9.972255441281988e-05,
      "loss": 0.0414,
      "step": 14461
    },
    {
      "epoch": 2.0269096005606166,
      "grad_norm": 0.3503555357456207,
      "learning_rate": 9.970820377900023e-05,
      "loss": 0.0275,
      "step": 14462
    },
    {
      "epoch": 2.027049754730203,
      "grad_norm": 0.5148195028305054,
      "learning_rate": 9.969385314518057e-05,
      "loss": 0.0185,
      "step": 14463
    },
    {
      "epoch": 2.0271899088997896,
      "grad_norm": 0.486939400434494,
      "learning_rate": 9.96795025113609e-05,
      "loss": 0.0452,
      "step": 14464
    },
    {
      "epoch": 2.0273300630693765,
      "grad_norm": 0.6523065567016602,
      "learning_rate": 9.966515187754125e-05,
      "loss": 0.0463,
      "step": 14465
    },
    {
      "epoch": 2.027470217238963,
      "grad_norm": 0.22304333746433258,
      "learning_rate": 9.96508012437216e-05,
      "loss": 0.0146,
      "step": 14466
    },
    {
      "epoch": 2.0276103714085494,
      "grad_norm": 0.3470122516155243,
      "learning_rate": 9.963645060990192e-05,
      "loss": 0.0391,
      "step": 14467
    },
    {
      "epoch": 2.027750525578136,
      "grad_norm": 0.20208820700645447,
      "learning_rate": 9.962209997608227e-05,
      "loss": 0.0185,
      "step": 14468
    },
    {
      "epoch": 2.0278906797477223,
      "grad_norm": 0.33971959352493286,
      "learning_rate": 9.960774934226262e-05,
      "loss": 0.0732,
      "step": 14469
    },
    {
      "epoch": 2.028030833917309,
      "grad_norm": 0.6886947751045227,
      "learning_rate": 9.959339870844295e-05,
      "loss": 0.0474,
      "step": 14470
    },
    {
      "epoch": 2.0281709880868957,
      "grad_norm": 0.23506073653697968,
      "learning_rate": 9.957904807462329e-05,
      "loss": 0.0516,
      "step": 14471
    },
    {
      "epoch": 2.028311142256482,
      "grad_norm": 0.18277232348918915,
      "learning_rate": 9.956469744080362e-05,
      "loss": 0.025,
      "step": 14472
    },
    {
      "epoch": 2.0284512964260686,
      "grad_norm": 0.08887793868780136,
      "learning_rate": 9.955034680698396e-05,
      "loss": 0.0083,
      "step": 14473
    },
    {
      "epoch": 2.028591450595655,
      "grad_norm": 0.29537466168403625,
      "learning_rate": 9.953599617316432e-05,
      "loss": 0.0232,
      "step": 14474
    },
    {
      "epoch": 2.028731604765242,
      "grad_norm": 0.1345331370830536,
      "learning_rate": 9.952164553934465e-05,
      "loss": 0.0137,
      "step": 14475
    },
    {
      "epoch": 2.0288717589348284,
      "grad_norm": 0.07308512926101685,
      "learning_rate": 9.950729490552499e-05,
      "loss": 0.0054,
      "step": 14476
    },
    {
      "epoch": 2.029011913104415,
      "grad_norm": 0.12065040320158005,
      "learning_rate": 9.949294427170533e-05,
      "loss": 0.0064,
      "step": 14477
    },
    {
      "epoch": 2.0291520672740013,
      "grad_norm": 0.183810755610466,
      "learning_rate": 9.947859363788566e-05,
      "loss": 0.0334,
      "step": 14478
    },
    {
      "epoch": 2.029292221443588,
      "grad_norm": 0.16208213567733765,
      "learning_rate": 9.9464243004066e-05,
      "loss": 0.0267,
      "step": 14479
    },
    {
      "epoch": 2.0294323756131747,
      "grad_norm": 0.5596699118614197,
      "learning_rate": 9.944989237024633e-05,
      "loss": 0.0951,
      "step": 14480
    },
    {
      "epoch": 2.029572529782761,
      "grad_norm": 0.23863954842090607,
      "learning_rate": 9.943554173642669e-05,
      "loss": 0.0454,
      "step": 14481
    },
    {
      "epoch": 2.0297126839523476,
      "grad_norm": 0.2928112745285034,
      "learning_rate": 9.942119110260703e-05,
      "loss": 0.0346,
      "step": 14482
    },
    {
      "epoch": 2.029852838121934,
      "grad_norm": 0.11333935707807541,
      "learning_rate": 9.940684046878736e-05,
      "loss": 0.0129,
      "step": 14483
    },
    {
      "epoch": 2.0299929922915205,
      "grad_norm": 0.3229149281978607,
      "learning_rate": 9.93924898349677e-05,
      "loss": 0.0386,
      "step": 14484
    },
    {
      "epoch": 2.0301331464611074,
      "grad_norm": 0.1360381543636322,
      "learning_rate": 9.937813920114805e-05,
      "loss": 0.021,
      "step": 14485
    },
    {
      "epoch": 2.030273300630694,
      "grad_norm": 0.08666659891605377,
      "learning_rate": 9.936378856732838e-05,
      "loss": 0.0065,
      "step": 14486
    },
    {
      "epoch": 2.0304134548002803,
      "grad_norm": 0.17681150138378143,
      "learning_rate": 9.934943793350873e-05,
      "loss": 0.0363,
      "step": 14487
    },
    {
      "epoch": 2.030553608969867,
      "grad_norm": 0.18216165900230408,
      "learning_rate": 9.933508729968905e-05,
      "loss": 0.0176,
      "step": 14488
    },
    {
      "epoch": 2.0306937631394533,
      "grad_norm": 0.12447208911180496,
      "learning_rate": 9.93207366658694e-05,
      "loss": 0.0142,
      "step": 14489
    },
    {
      "epoch": 2.0308339173090397,
      "grad_norm": 0.2375001162290573,
      "learning_rate": 9.930638603204975e-05,
      "loss": 0.032,
      "step": 14490
    },
    {
      "epoch": 2.0309740714786266,
      "grad_norm": 0.2855256497859955,
      "learning_rate": 9.929203539823008e-05,
      "loss": 0.037,
      "step": 14491
    },
    {
      "epoch": 2.031114225648213,
      "grad_norm": 0.15849518775939941,
      "learning_rate": 9.927768476441042e-05,
      "loss": 0.0345,
      "step": 14492
    },
    {
      "epoch": 2.0312543798177995,
      "grad_norm": 0.3493972718715668,
      "learning_rate": 9.926333413059076e-05,
      "loss": 0.0625,
      "step": 14493
    },
    {
      "epoch": 2.031394533987386,
      "grad_norm": 0.21555230021476746,
      "learning_rate": 9.924898349677109e-05,
      "loss": 0.037,
      "step": 14494
    },
    {
      "epoch": 2.0315346881569725,
      "grad_norm": 0.1917514204978943,
      "learning_rate": 9.923463286295144e-05,
      "loss": 0.0128,
      "step": 14495
    },
    {
      "epoch": 2.0316748423265594,
      "grad_norm": 0.11909850686788559,
      "learning_rate": 9.922028222913176e-05,
      "loss": 0.0242,
      "step": 14496
    },
    {
      "epoch": 2.031814996496146,
      "grad_norm": 0.39441365003585815,
      "learning_rate": 9.920593159531212e-05,
      "loss": 0.0448,
      "step": 14497
    },
    {
      "epoch": 2.0319551506657323,
      "grad_norm": 0.4820905327796936,
      "learning_rate": 9.919158096149246e-05,
      "loss": 0.0665,
      "step": 14498
    },
    {
      "epoch": 2.0320953048353187,
      "grad_norm": 0.22301989793777466,
      "learning_rate": 9.917723032767279e-05,
      "loss": 0.0143,
      "step": 14499
    },
    {
      "epoch": 2.032235459004905,
      "grad_norm": 0.23312315344810486,
      "learning_rate": 9.916287969385313e-05,
      "loss": 0.0121,
      "step": 14500
    },
    {
      "epoch": 2.032375613174492,
      "grad_norm": 0.24394792318344116,
      "learning_rate": 9.914852906003349e-05,
      "loss": 0.0134,
      "step": 14501
    },
    {
      "epoch": 2.0325157673440786,
      "grad_norm": 0.15797056257724762,
      "learning_rate": 9.913417842621382e-05,
      "loss": 0.0204,
      "step": 14502
    },
    {
      "epoch": 2.032655921513665,
      "grad_norm": 0.6717702150344849,
      "learning_rate": 9.911982779239416e-05,
      "loss": 0.0599,
      "step": 14503
    },
    {
      "epoch": 2.0327960756832515,
      "grad_norm": 0.21511223912239075,
      "learning_rate": 9.91054771585745e-05,
      "loss": 0.0228,
      "step": 14504
    },
    {
      "epoch": 2.032936229852838,
      "grad_norm": 0.3279041051864624,
      "learning_rate": 9.909112652475483e-05,
      "loss": 0.0383,
      "step": 14505
    },
    {
      "epoch": 2.033076384022425,
      "grad_norm": 0.035527553409338,
      "learning_rate": 9.907677589093518e-05,
      "loss": 0.0031,
      "step": 14506
    },
    {
      "epoch": 2.0332165381920113,
      "grad_norm": 0.36504456400871277,
      "learning_rate": 9.906242525711551e-05,
      "loss": 0.0546,
      "step": 14507
    },
    {
      "epoch": 2.0333566923615978,
      "grad_norm": 0.23326686024665833,
      "learning_rate": 9.904807462329586e-05,
      "loss": 0.0155,
      "step": 14508
    },
    {
      "epoch": 2.0334968465311842,
      "grad_norm": 1.6420146226882935,
      "learning_rate": 9.90337239894762e-05,
      "loss": 0.0572,
      "step": 14509
    },
    {
      "epoch": 2.0336370007007707,
      "grad_norm": 0.4718138873577118,
      "learning_rate": 9.901937335565653e-05,
      "loss": 0.0493,
      "step": 14510
    },
    {
      "epoch": 2.0337771548703576,
      "grad_norm": 0.26314330101013184,
      "learning_rate": 9.900502272183687e-05,
      "loss": 0.0593,
      "step": 14511
    },
    {
      "epoch": 2.033917309039944,
      "grad_norm": 0.24281050264835358,
      "learning_rate": 9.899067208801722e-05,
      "loss": 0.0139,
      "step": 14512
    },
    {
      "epoch": 2.0340574632095305,
      "grad_norm": 0.32817041873931885,
      "learning_rate": 9.897632145419755e-05,
      "loss": 0.0469,
      "step": 14513
    },
    {
      "epoch": 2.034197617379117,
      "grad_norm": 0.08950285613536835,
      "learning_rate": 9.89619708203779e-05,
      "loss": 0.0203,
      "step": 14514
    },
    {
      "epoch": 2.0343377715487034,
      "grad_norm": 0.38477644324302673,
      "learning_rate": 9.894762018655822e-05,
      "loss": 0.0386,
      "step": 14515
    },
    {
      "epoch": 2.0344779257182903,
      "grad_norm": 0.21968553960323334,
      "learning_rate": 9.893326955273857e-05,
      "loss": 0.06,
      "step": 14516
    },
    {
      "epoch": 2.034618079887877,
      "grad_norm": 0.15917927026748657,
      "learning_rate": 9.891891891891892e-05,
      "loss": 0.017,
      "step": 14517
    },
    {
      "epoch": 2.0347582340574633,
      "grad_norm": 0.03450698405504227,
      "learning_rate": 9.890456828509925e-05,
      "loss": 0.0027,
      "step": 14518
    },
    {
      "epoch": 2.0348983882270497,
      "grad_norm": 0.36688032746315,
      "learning_rate": 9.889021765127959e-05,
      "loss": 0.0706,
      "step": 14519
    },
    {
      "epoch": 2.035038542396636,
      "grad_norm": NaN,
      "learning_rate": 9.887586701745993e-05,
      "loss": 0.3196,
      "step": 14520
    },
    {
      "epoch": 2.0351786965662226,
      "grad_norm": 0.1534242033958435,
      "learning_rate": 9.887586701745993e-05,
      "loss": 0.0225,
      "step": 14521
    },
    {
      "epoch": 2.0353188507358095,
      "grad_norm": 0.17387579381465912,
      "learning_rate": 9.886151638364026e-05,
      "loss": 0.0303,
      "step": 14522
    },
    {
      "epoch": 2.035459004905396,
      "grad_norm": 0.15925176441669464,
      "learning_rate": 9.884716574982062e-05,
      "loss": 0.0145,
      "step": 14523
    },
    {
      "epoch": 2.0355991590749825,
      "grad_norm": 0.42877814173698425,
      "learning_rate": 9.883281511600095e-05,
      "loss": 0.0641,
      "step": 14524
    },
    {
      "epoch": 2.035739313244569,
      "grad_norm": 0.28471511602401733,
      "learning_rate": 9.881846448218129e-05,
      "loss": 0.0323,
      "step": 14525
    },
    {
      "epoch": 2.0358794674141554,
      "grad_norm": 0.09378957748413086,
      "learning_rate": 9.880411384836163e-05,
      "loss": 0.0102,
      "step": 14526
    },
    {
      "epoch": 2.0360196215837423,
      "grad_norm": 0.5060093998908997,
      "learning_rate": 9.878976321454196e-05,
      "loss": 0.0363,
      "step": 14527
    },
    {
      "epoch": 2.0361597757533287,
      "grad_norm": 0.13294915854930878,
      "learning_rate": 9.87754125807223e-05,
      "loss": 0.0117,
      "step": 14528
    },
    {
      "epoch": 2.036299929922915,
      "grad_norm": 1.1636812686920166,
      "learning_rate": 9.876106194690266e-05,
      "loss": 0.071,
      "step": 14529
    },
    {
      "epoch": 2.0364400840925017,
      "grad_norm": 0.08930788934230804,
      "learning_rate": 9.874671131308299e-05,
      "loss": 0.0055,
      "step": 14530
    },
    {
      "epoch": 2.036580238262088,
      "grad_norm": 0.5023823380470276,
      "learning_rate": 9.873236067926333e-05,
      "loss": 0.0447,
      "step": 14531
    },
    {
      "epoch": 2.036720392431675,
      "grad_norm": 0.2278163582086563,
      "learning_rate": 9.871801004544366e-05,
      "loss": 0.0267,
      "step": 14532
    },
    {
      "epoch": 2.0368605466012615,
      "grad_norm": 0.24701005220413208,
      "learning_rate": 9.8703659411624e-05,
      "loss": 0.0464,
      "step": 14533
    },
    {
      "epoch": 2.037000700770848,
      "grad_norm": 0.09374483674764633,
      "learning_rate": 9.868930877780435e-05,
      "loss": 0.0067,
      "step": 14534
    },
    {
      "epoch": 2.0371408549404344,
      "grad_norm": 0.19721589982509613,
      "learning_rate": 9.867495814398468e-05,
      "loss": 0.0455,
      "step": 14535
    },
    {
      "epoch": 2.037281009110021,
      "grad_norm": 0.20399923622608185,
      "learning_rate": 9.866060751016502e-05,
      "loss": 0.0069,
      "step": 14536
    },
    {
      "epoch": 2.0374211632796078,
      "grad_norm": 0.21088121831417084,
      "learning_rate": 9.864625687634537e-05,
      "loss": 0.0422,
      "step": 14537
    },
    {
      "epoch": 2.037561317449194,
      "grad_norm": 0.06076661869883537,
      "learning_rate": 9.86319062425257e-05,
      "loss": 0.0071,
      "step": 14538
    },
    {
      "epoch": 2.0377014716187807,
      "grad_norm": 0.1837327778339386,
      "learning_rate": 9.861755560870605e-05,
      "loss": 0.0062,
      "step": 14539
    },
    {
      "epoch": 2.037841625788367,
      "grad_norm": 0.365121066570282,
      "learning_rate": 9.860320497488638e-05,
      "loss": 0.0499,
      "step": 14540
    },
    {
      "epoch": 2.0379817799579536,
      "grad_norm": 0.13977009057998657,
      "learning_rate": 9.858885434106672e-05,
      "loss": 0.0083,
      "step": 14541
    },
    {
      "epoch": 2.0381219341275405,
      "grad_norm": 0.6147711873054504,
      "learning_rate": 9.857450370724706e-05,
      "loss": 0.0314,
      "step": 14542
    },
    {
      "epoch": 2.038262088297127,
      "grad_norm": 0.5251495838165283,
      "learning_rate": 9.856015307342739e-05,
      "loss": 0.0663,
      "step": 14543
    },
    {
      "epoch": 2.0384022424667134,
      "grad_norm": 0.12395856529474258,
      "learning_rate": 9.854580243960773e-05,
      "loss": 0.0226,
      "step": 14544
    },
    {
      "epoch": 2.0385423966363,
      "grad_norm": 0.02359222248196602,
      "learning_rate": 9.853145180578809e-05,
      "loss": 0.002,
      "step": 14545
    },
    {
      "epoch": 2.0386825508058863,
      "grad_norm": 0.18385474383831024,
      "learning_rate": 9.851710117196842e-05,
      "loss": 0.009,
      "step": 14546
    },
    {
      "epoch": 2.0388227049754732,
      "grad_norm": 0.0622151754796505,
      "learning_rate": 9.850275053814876e-05,
      "loss": 0.0042,
      "step": 14547
    },
    {
      "epoch": 2.0389628591450597,
      "grad_norm": 0.7372965812683105,
      "learning_rate": 9.84883999043291e-05,
      "loss": 0.0404,
      "step": 14548
    },
    {
      "epoch": 2.039103013314646,
      "grad_norm": 0.23335157334804535,
      "learning_rate": 9.847404927050943e-05,
      "loss": 0.0477,
      "step": 14549
    },
    {
      "epoch": 2.0392431674842326,
      "grad_norm": 0.06278901547193527,
      "learning_rate": 9.845969863668979e-05,
      "loss": 0.0051,
      "step": 14550
    },
    {
      "epoch": 2.039383321653819,
      "grad_norm": 0.09418494254350662,
      "learning_rate": 9.844534800287012e-05,
      "loss": 0.0128,
      "step": 14551
    },
    {
      "epoch": 2.0395234758234055,
      "grad_norm": 0.43041256070137024,
      "learning_rate": 9.843099736905046e-05,
      "loss": 0.0575,
      "step": 14552
    },
    {
      "epoch": 2.0396636299929924,
      "grad_norm": 0.4484850764274597,
      "learning_rate": 9.84166467352308e-05,
      "loss": 0.0583,
      "step": 14553
    },
    {
      "epoch": 2.039803784162579,
      "grad_norm": 0.25465598702430725,
      "learning_rate": 9.840229610141113e-05,
      "loss": 0.015,
      "step": 14554
    },
    {
      "epoch": 2.0399439383321654,
      "grad_norm": 0.17130139470100403,
      "learning_rate": 9.838794546759148e-05,
      "loss": 0.0269,
      "step": 14555
    },
    {
      "epoch": 2.040084092501752,
      "grad_norm": 0.29218029975891113,
      "learning_rate": 9.837359483377183e-05,
      "loss": 0.0145,
      "step": 14556
    },
    {
      "epoch": 2.0402242466713383,
      "grad_norm": 0.24774783849716187,
      "learning_rate": 9.835924419995215e-05,
      "loss": 0.0219,
      "step": 14557
    },
    {
      "epoch": 2.040364400840925,
      "grad_norm": 0.19440512359142303,
      "learning_rate": 9.83448935661325e-05,
      "loss": 0.013,
      "step": 14558
    },
    {
      "epoch": 2.0405045550105116,
      "grad_norm": 0.14187614619731903,
      "learning_rate": 9.833054293231283e-05,
      "loss": 0.0455,
      "step": 14559
    },
    {
      "epoch": 2.040644709180098,
      "grad_norm": 0.20071375370025635,
      "learning_rate": 9.831619229849317e-05,
      "loss": 0.0171,
      "step": 14560
    },
    {
      "epoch": 2.0407848633496846,
      "grad_norm": 0.5312033295631409,
      "learning_rate": 9.830184166467352e-05,
      "loss": 0.0422,
      "step": 14561
    },
    {
      "epoch": 2.040925017519271,
      "grad_norm": 0.40411654114723206,
      "learning_rate": 9.828749103085385e-05,
      "loss": 0.037,
      "step": 14562
    },
    {
      "epoch": 2.041065171688858,
      "grad_norm": 0.26766201853752136,
      "learning_rate": 9.82731403970342e-05,
      "loss": 0.0547,
      "step": 14563
    },
    {
      "epoch": 2.0412053258584444,
      "grad_norm": 0.7721509337425232,
      "learning_rate": 9.825878976321454e-05,
      "loss": 0.0823,
      "step": 14564
    },
    {
      "epoch": 2.041345480028031,
      "grad_norm": 0.7094186544418335,
      "learning_rate": 9.824443912939486e-05,
      "loss": 0.0415,
      "step": 14565
    },
    {
      "epoch": 2.0414856341976173,
      "grad_norm": 0.21951128542423248,
      "learning_rate": 9.823008849557522e-05,
      "loss": 0.0223,
      "step": 14566
    },
    {
      "epoch": 2.0416257883672038,
      "grad_norm": 0.13689817488193512,
      "learning_rate": 9.821573786175555e-05,
      "loss": 0.0049,
      "step": 14567
    },
    {
      "epoch": 2.0417659425367907,
      "grad_norm": 1.0043683052062988,
      "learning_rate": 9.820138722793589e-05,
      "loss": 0.0343,
      "step": 14568
    },
    {
      "epoch": 2.041906096706377,
      "grad_norm": 3.0330190658569336,
      "learning_rate": 9.818703659411623e-05,
      "loss": 0.3474,
      "step": 14569
    },
    {
      "epoch": 2.0420462508759636,
      "grad_norm": 0.044684477150440216,
      "learning_rate": 9.817268596029656e-05,
      "loss": 0.0023,
      "step": 14570
    },
    {
      "epoch": 2.04218640504555,
      "grad_norm": 0.29785287380218506,
      "learning_rate": 9.815833532647692e-05,
      "loss": 0.05,
      "step": 14571
    },
    {
      "epoch": 2.0423265592151365,
      "grad_norm": 0.14133098721504211,
      "learning_rate": 9.814398469265726e-05,
      "loss": 0.0239,
      "step": 14572
    },
    {
      "epoch": 2.0424667133847234,
      "grad_norm": 0.3275611102581024,
      "learning_rate": 9.812963405883759e-05,
      "loss": 0.0534,
      "step": 14573
    },
    {
      "epoch": 2.04260686755431,
      "grad_norm": 0.12047803401947021,
      "learning_rate": 9.811528342501793e-05,
      "loss": 0.0126,
      "step": 14574
    },
    {
      "epoch": 2.0427470217238963,
      "grad_norm": 0.09081151336431503,
      "learning_rate": 9.810093279119826e-05,
      "loss": 0.0106,
      "step": 14575
    },
    {
      "epoch": 2.042887175893483,
      "grad_norm": 0.24666573107242584,
      "learning_rate": 9.80865821573786e-05,
      "loss": 0.0419,
      "step": 14576
    },
    {
      "epoch": 2.0430273300630692,
      "grad_norm": 0.13887514173984528,
      "learning_rate": 9.807223152355896e-05,
      "loss": 0.0259,
      "step": 14577
    },
    {
      "epoch": 2.0431674842326557,
      "grad_norm": 0.3158944845199585,
      "learning_rate": 9.805788088973929e-05,
      "loss": 0.0288,
      "step": 14578
    },
    {
      "epoch": 2.0433076384022426,
      "grad_norm": 0.4252224266529083,
      "learning_rate": 9.804353025591963e-05,
      "loss": 0.0826,
      "step": 14579
    },
    {
      "epoch": 2.043447792571829,
      "grad_norm": 0.5137424468994141,
      "learning_rate": 9.802917962209997e-05,
      "loss": 0.0507,
      "step": 14580
    },
    {
      "epoch": 2.0435879467414155,
      "grad_norm": 0.20665203034877777,
      "learning_rate": 9.80148289882803e-05,
      "loss": 0.0206,
      "step": 14581
    },
    {
      "epoch": 2.043728100911002,
      "grad_norm": 0.48795753717422485,
      "learning_rate": 9.800047835446065e-05,
      "loss": 0.0782,
      "step": 14582
    },
    {
      "epoch": 2.0438682550805884,
      "grad_norm": 0.296686053276062,
      "learning_rate": 9.7986127720641e-05,
      "loss": 0.0736,
      "step": 14583
    },
    {
      "epoch": 2.0440084092501754,
      "grad_norm": 0.24494190514087677,
      "learning_rate": 9.797177708682132e-05,
      "loss": 0.0491,
      "step": 14584
    },
    {
      "epoch": 2.044148563419762,
      "grad_norm": 0.0630914568901062,
      "learning_rate": 9.795742645300167e-05,
      "loss": 0.0074,
      "step": 14585
    },
    {
      "epoch": 2.0442887175893483,
      "grad_norm": 0.423389196395874,
      "learning_rate": 9.7943075819182e-05,
      "loss": 0.1202,
      "step": 14586
    },
    {
      "epoch": 2.0444288717589347,
      "grad_norm": 0.23362980782985687,
      "learning_rate": 9.792872518536235e-05,
      "loss": 0.013,
      "step": 14587
    },
    {
      "epoch": 2.044569025928521,
      "grad_norm": 0.21208184957504272,
      "learning_rate": 9.791437455154269e-05,
      "loss": 0.0604,
      "step": 14588
    },
    {
      "epoch": 2.044709180098108,
      "grad_norm": 0.20589450001716614,
      "learning_rate": 9.790002391772302e-05,
      "loss": 0.0353,
      "step": 14589
    },
    {
      "epoch": 2.0448493342676946,
      "grad_norm": 0.3093615472316742,
      "learning_rate": 9.788567328390336e-05,
      "loss": 0.0306,
      "step": 14590
    },
    {
      "epoch": 2.044989488437281,
      "grad_norm": 0.11499712616205215,
      "learning_rate": 9.78713226500837e-05,
      "loss": 0.0156,
      "step": 14591
    },
    {
      "epoch": 2.0451296426068675,
      "grad_norm": 0.09062822163105011,
      "learning_rate": 9.785697201626403e-05,
      "loss": 0.0083,
      "step": 14592
    },
    {
      "epoch": 2.045269796776454,
      "grad_norm": 0.12232232838869095,
      "learning_rate": 9.784262138244439e-05,
      "loss": 0.0186,
      "step": 14593
    },
    {
      "epoch": 2.045409950946041,
      "grad_norm": 0.5004503130912781,
      "learning_rate": 9.782827074862472e-05,
      "loss": 0.0187,
      "step": 14594
    },
    {
      "epoch": 2.0455501051156273,
      "grad_norm": 0.05065590888261795,
      "learning_rate": 9.781392011480506e-05,
      "loss": 0.0036,
      "step": 14595
    },
    {
      "epoch": 2.0456902592852138,
      "grad_norm": 0.10655776411294937,
      "learning_rate": 9.77995694809854e-05,
      "loss": 0.0093,
      "step": 14596
    },
    {
      "epoch": 2.0458304134548,
      "grad_norm": 0.062482330948114395,
      "learning_rate": 9.778521884716573e-05,
      "loss": 0.0064,
      "step": 14597
    },
    {
      "epoch": 2.0459705676243867,
      "grad_norm": 0.4218922257423401,
      "learning_rate": 9.777086821334609e-05,
      "loss": 0.0551,
      "step": 14598
    },
    {
      "epoch": 2.0461107217939736,
      "grad_norm": 0.13488878309726715,
      "learning_rate": 9.775651757952643e-05,
      "loss": 0.011,
      "step": 14599
    },
    {
      "epoch": 2.04625087596356,
      "grad_norm": 0.2483682930469513,
      "learning_rate": 9.774216694570676e-05,
      "loss": 0.0135,
      "step": 14600
    },
    {
      "epoch": 2.0463910301331465,
      "grad_norm": 0.45724424719810486,
      "learning_rate": 9.77278163118871e-05,
      "loss": 0.0619,
      "step": 14601
    },
    {
      "epoch": 2.046531184302733,
      "grad_norm": 0.23024897277355194,
      "learning_rate": 9.771346567806743e-05,
      "loss": 0.0601,
      "step": 14602
    },
    {
      "epoch": 2.0466713384723194,
      "grad_norm": 0.39379122853279114,
      "learning_rate": 9.769911504424778e-05,
      "loss": 0.0251,
      "step": 14603
    },
    {
      "epoch": 2.046811492641906,
      "grad_norm": 0.4059712588787079,
      "learning_rate": 9.768476441042813e-05,
      "loss": 0.0268,
      "step": 14604
    },
    {
      "epoch": 2.0469516468114928,
      "grad_norm": 0.034140221774578094,
      "learning_rate": 9.767041377660845e-05,
      "loss": 0.0037,
      "step": 14605
    },
    {
      "epoch": 2.0470918009810792,
      "grad_norm": 0.17910324037075043,
      "learning_rate": 9.76560631427888e-05,
      "loss": 0.0092,
      "step": 14606
    },
    {
      "epoch": 2.0472319551506657,
      "grad_norm": 0.26883307099342346,
      "learning_rate": 9.764171250896914e-05,
      "loss": 0.0336,
      "step": 14607
    },
    {
      "epoch": 2.047372109320252,
      "grad_norm": 0.08821037411689758,
      "learning_rate": 9.762736187514947e-05,
      "loss": 0.0079,
      "step": 14608
    },
    {
      "epoch": 2.0475122634898386,
      "grad_norm": 0.27331221103668213,
      "learning_rate": 9.761301124132982e-05,
      "loss": 0.0581,
      "step": 14609
    },
    {
      "epoch": 2.0476524176594255,
      "grad_norm": 0.2804799973964691,
      "learning_rate": 9.759866060751015e-05,
      "loss": 0.0194,
      "step": 14610
    },
    {
      "epoch": 2.047792571829012,
      "grad_norm": 0.6483967900276184,
      "learning_rate": 9.75843099736905e-05,
      "loss": 0.1039,
      "step": 14611
    },
    {
      "epoch": 2.0479327259985984,
      "grad_norm": 0.31753939390182495,
      "learning_rate": 9.756995933987084e-05,
      "loss": 0.0273,
      "step": 14612
    },
    {
      "epoch": 2.048072880168185,
      "grad_norm": 0.5934249758720398,
      "learning_rate": 9.755560870605116e-05,
      "loss": 0.0464,
      "step": 14613
    },
    {
      "epoch": 2.0482130343377714,
      "grad_norm": 0.21168068051338196,
      "learning_rate": 9.754125807223152e-05,
      "loss": 0.0133,
      "step": 14614
    },
    {
      "epoch": 2.0483531885073583,
      "grad_norm": 0.3751913905143738,
      "learning_rate": 9.752690743841186e-05,
      "loss": 0.0291,
      "step": 14615
    },
    {
      "epoch": 2.0484933426769447,
      "grad_norm": 0.1160811111330986,
      "learning_rate": 9.751255680459219e-05,
      "loss": 0.011,
      "step": 14616
    },
    {
      "epoch": 2.048633496846531,
      "grad_norm": 0.6843565702438354,
      "learning_rate": 9.749820617077253e-05,
      "loss": 0.0332,
      "step": 14617
    },
    {
      "epoch": 2.0487736510161176,
      "grad_norm": 2.3835079669952393,
      "learning_rate": 9.748385553695289e-05,
      "loss": 0.2327,
      "step": 14618
    },
    {
      "epoch": 2.048913805185704,
      "grad_norm": 0.4588441252708435,
      "learning_rate": 9.746950490313322e-05,
      "loss": 0.0399,
      "step": 14619
    },
    {
      "epoch": 2.049053959355291,
      "grad_norm": 3.7697811126708984,
      "learning_rate": 9.745515426931356e-05,
      "loss": 0.0705,
      "step": 14620
    },
    {
      "epoch": 2.0491941135248775,
      "grad_norm": 0.2121492326259613,
      "learning_rate": 9.744080363549389e-05,
      "loss": 0.0215,
      "step": 14621
    },
    {
      "epoch": 2.049334267694464,
      "grad_norm": 0.09743131697177887,
      "learning_rate": 9.742645300167423e-05,
      "loss": 0.01,
      "step": 14622
    },
    {
      "epoch": 2.0494744218640504,
      "grad_norm": 0.09635625034570694,
      "learning_rate": 9.741210236785457e-05,
      "loss": 0.0075,
      "step": 14623
    },
    {
      "epoch": 2.049614576033637,
      "grad_norm": 0.7265884876251221,
      "learning_rate": 9.73977517340349e-05,
      "loss": 0.0291,
      "step": 14624
    },
    {
      "epoch": 2.0497547302032237,
      "grad_norm": 0.250246524810791,
      "learning_rate": 9.738340110021526e-05,
      "loss": 0.0466,
      "step": 14625
    },
    {
      "epoch": 2.04989488437281,
      "grad_norm": 0.051374293863773346,
      "learning_rate": 9.73690504663956e-05,
      "loss": 0.0041,
      "step": 14626
    },
    {
      "epoch": 2.0500350385423967,
      "grad_norm": 0.31546446681022644,
      "learning_rate": 9.735469983257593e-05,
      "loss": 0.0131,
      "step": 14627
    },
    {
      "epoch": 2.050175192711983,
      "grad_norm": 0.4238879680633545,
      "learning_rate": 9.734034919875627e-05,
      "loss": 0.0822,
      "step": 14628
    },
    {
      "epoch": 2.0503153468815696,
      "grad_norm": 0.1808554232120514,
      "learning_rate": 9.73259985649366e-05,
      "loss": 0.0124,
      "step": 14629
    },
    {
      "epoch": 2.0504555010511565,
      "grad_norm": 0.7445446252822876,
      "learning_rate": 9.731164793111695e-05,
      "loss": 0.0444,
      "step": 14630
    },
    {
      "epoch": 2.050595655220743,
      "grad_norm": 0.20793062448501587,
      "learning_rate": 9.72972972972973e-05,
      "loss": 0.0297,
      "step": 14631
    },
    {
      "epoch": 2.0507358093903294,
      "grad_norm": 0.2349175065755844,
      "learning_rate": 9.728294666347762e-05,
      "loss": 0.01,
      "step": 14632
    },
    {
      "epoch": 2.050875963559916,
      "grad_norm": 0.19442962110042572,
      "learning_rate": 9.726859602965797e-05,
      "loss": 0.0374,
      "step": 14633
    },
    {
      "epoch": 2.0510161177295023,
      "grad_norm": 0.23417486250400543,
      "learning_rate": 9.725424539583832e-05,
      "loss": 0.0326,
      "step": 14634
    },
    {
      "epoch": 2.051156271899089,
      "grad_norm": 0.791458785533905,
      "learning_rate": 9.723989476201865e-05,
      "loss": 0.0718,
      "step": 14635
    },
    {
      "epoch": 2.0512964260686757,
      "grad_norm": 0.15302234888076782,
      "learning_rate": 9.722554412819899e-05,
      "loss": 0.0272,
      "step": 14636
    },
    {
      "epoch": 2.051436580238262,
      "grad_norm": 0.07960764318704605,
      "learning_rate": 9.721119349437932e-05,
      "loss": 0.0049,
      "step": 14637
    },
    {
      "epoch": 2.0515767344078486,
      "grad_norm": 0.2141430675983429,
      "learning_rate": 9.719684286055966e-05,
      "loss": 0.0377,
      "step": 14638
    },
    {
      "epoch": 2.051716888577435,
      "grad_norm": 0.3910596966743469,
      "learning_rate": 9.718249222674e-05,
      "loss": 0.0565,
      "step": 14639
    },
    {
      "epoch": 2.0518570427470215,
      "grad_norm": 0.03645210340619087,
      "learning_rate": 9.716814159292033e-05,
      "loss": 0.0019,
      "step": 14640
    },
    {
      "epoch": 2.0519971969166084,
      "grad_norm": 1.0703109502792358,
      "learning_rate": 9.715379095910069e-05,
      "loss": 0.0793,
      "step": 14641
    },
    {
      "epoch": 2.052137351086195,
      "grad_norm": 0.07595302164554596,
      "learning_rate": 9.713944032528103e-05,
      "loss": 0.0142,
      "step": 14642
    },
    {
      "epoch": 2.0522775052557813,
      "grad_norm": 0.1973145455121994,
      "learning_rate": 9.712508969146136e-05,
      "loss": 0.0187,
      "step": 14643
    },
    {
      "epoch": 2.052417659425368,
      "grad_norm": 0.08789122849702835,
      "learning_rate": 9.71107390576417e-05,
      "loss": 0.0229,
      "step": 14644
    },
    {
      "epoch": 2.0525578135949543,
      "grad_norm": 0.29548490047454834,
      "learning_rate": 9.709638842382203e-05,
      "loss": 0.0267,
      "step": 14645
    },
    {
      "epoch": 2.052697967764541,
      "grad_norm": 0.24903497099876404,
      "learning_rate": 9.708203779000239e-05,
      "loss": 0.0304,
      "step": 14646
    },
    {
      "epoch": 2.0528381219341276,
      "grad_norm": 0.08784700185060501,
      "learning_rate": 9.706768715618273e-05,
      "loss": 0.0175,
      "step": 14647
    },
    {
      "epoch": 2.052978276103714,
      "grad_norm": 0.47082197666168213,
      "learning_rate": 9.705333652236306e-05,
      "loss": 0.0583,
      "step": 14648
    },
    {
      "epoch": 2.0531184302733005,
      "grad_norm": 0.2877300977706909,
      "learning_rate": 9.70389858885434e-05,
      "loss": 0.0363,
      "step": 14649
    },
    {
      "epoch": 2.053258584442887,
      "grad_norm": 0.5941381454467773,
      "learning_rate": 9.702463525472375e-05,
      "loss": 0.1078,
      "step": 14650
    },
    {
      "epoch": 2.053398738612474,
      "grad_norm": 0.5060108304023743,
      "learning_rate": 9.701028462090408e-05,
      "loss": 0.0312,
      "step": 14651
    },
    {
      "epoch": 2.0535388927820604,
      "grad_norm": 1.0659782886505127,
      "learning_rate": 9.699593398708443e-05,
      "loss": 0.172,
      "step": 14652
    },
    {
      "epoch": 2.053679046951647,
      "grad_norm": 0.25115126371383667,
      "learning_rate": 9.698158335326475e-05,
      "loss": 0.0192,
      "step": 14653
    },
    {
      "epoch": 2.0538192011212333,
      "grad_norm": 0.317066490650177,
      "learning_rate": 9.69672327194451e-05,
      "loss": 0.0852,
      "step": 14654
    },
    {
      "epoch": 2.0539593552908197,
      "grad_norm": 0.48709696531295776,
      "learning_rate": 9.695288208562544e-05,
      "loss": 0.0518,
      "step": 14655
    },
    {
      "epoch": 2.0540995094604066,
      "grad_norm": 0.4695848524570465,
      "learning_rate": 9.693853145180577e-05,
      "loss": 0.0801,
      "step": 14656
    },
    {
      "epoch": 2.054239663629993,
      "grad_norm": 0.24045182764530182,
      "learning_rate": 9.692418081798612e-05,
      "loss": 0.0228,
      "step": 14657
    },
    {
      "epoch": 2.0543798177995796,
      "grad_norm": 0.36782848834991455,
      "learning_rate": 9.690983018416646e-05,
      "loss": 0.0577,
      "step": 14658
    },
    {
      "epoch": 2.054519971969166,
      "grad_norm": 0.24183416366577148,
      "learning_rate": 9.689547955034679e-05,
      "loss": 0.081,
      "step": 14659
    },
    {
      "epoch": 2.0546601261387525,
      "grad_norm": 0.44207194447517395,
      "learning_rate": 9.688112891652713e-05,
      "loss": 0.0355,
      "step": 14660
    },
    {
      "epoch": 2.0548002803083394,
      "grad_norm": 0.3749368488788605,
      "learning_rate": 9.686677828270749e-05,
      "loss": 0.0294,
      "step": 14661
    },
    {
      "epoch": 2.054940434477926,
      "grad_norm": 0.4326549172401428,
      "learning_rate": 9.685242764888782e-05,
      "loss": 0.0348,
      "step": 14662
    },
    {
      "epoch": 2.0550805886475123,
      "grad_norm": 0.30109161138534546,
      "learning_rate": 9.683807701506816e-05,
      "loss": 0.0492,
      "step": 14663
    },
    {
      "epoch": 2.0552207428170988,
      "grad_norm": 0.01653483137488365,
      "learning_rate": 9.682372638124849e-05,
      "loss": 0.0018,
      "step": 14664
    },
    {
      "epoch": 2.0553608969866852,
      "grad_norm": 0.21044203639030457,
      "learning_rate": 9.680937574742883e-05,
      "loss": 0.0061,
      "step": 14665
    },
    {
      "epoch": 2.0555010511562717,
      "grad_norm": 0.8893194198608398,
      "learning_rate": 9.679502511360919e-05,
      "loss": 0.0775,
      "step": 14666
    },
    {
      "epoch": 2.0556412053258586,
      "grad_norm": 0.6652569770812988,
      "learning_rate": 9.678067447978952e-05,
      "loss": 0.0732,
      "step": 14667
    },
    {
      "epoch": 2.055781359495445,
      "grad_norm": 3.4368643760681152,
      "learning_rate": 9.676632384596986e-05,
      "loss": 0.1104,
      "step": 14668
    },
    {
      "epoch": 2.0559215136650315,
      "grad_norm": 0.638452410697937,
      "learning_rate": 9.67519732121502e-05,
      "loss": 0.0368,
      "step": 14669
    },
    {
      "epoch": 2.056061667834618,
      "grad_norm": 0.5533886551856995,
      "learning_rate": 9.673762257833053e-05,
      "loss": 0.0661,
      "step": 14670
    },
    {
      "epoch": 2.0562018220042044,
      "grad_norm": 0.12036539614200592,
      "learning_rate": 9.672327194451087e-05,
      "loss": 0.0248,
      "step": 14671
    },
    {
      "epoch": 2.0563419761737913,
      "grad_norm": 0.12919162213802338,
      "learning_rate": 9.67089213106912e-05,
      "loss": 0.0324,
      "step": 14672
    },
    {
      "epoch": 2.056482130343378,
      "grad_norm": 0.2486315816640854,
      "learning_rate": 9.669457067687156e-05,
      "loss": 0.0056,
      "step": 14673
    },
    {
      "epoch": 2.0566222845129642,
      "grad_norm": 0.4113796055316925,
      "learning_rate": 9.66802200430519e-05,
      "loss": 0.0394,
      "step": 14674
    },
    {
      "epoch": 2.0567624386825507,
      "grad_norm": 0.1997787356376648,
      "learning_rate": 9.666586940923223e-05,
      "loss": 0.0341,
      "step": 14675
    },
    {
      "epoch": 2.056902592852137,
      "grad_norm": 0.32701364159584045,
      "learning_rate": 9.665151877541257e-05,
      "loss": 0.032,
      "step": 14676
    },
    {
      "epoch": 2.057042747021724,
      "grad_norm": 0.09779198467731476,
      "learning_rate": 9.663716814159292e-05,
      "loss": 0.0114,
      "step": 14677
    },
    {
      "epoch": 2.0571829011913105,
      "grad_norm": 0.3206586241722107,
      "learning_rate": 9.662281750777325e-05,
      "loss": 0.0131,
      "step": 14678
    },
    {
      "epoch": 2.057323055360897,
      "grad_norm": 0.30883026123046875,
      "learning_rate": 9.66084668739536e-05,
      "loss": 0.0332,
      "step": 14679
    },
    {
      "epoch": 2.0574632095304835,
      "grad_norm": 0.3375200629234314,
      "learning_rate": 9.659411624013392e-05,
      "loss": 0.0615,
      "step": 14680
    },
    {
      "epoch": 2.05760336370007,
      "grad_norm": 0.16260966658592224,
      "learning_rate": 9.657976560631427e-05,
      "loss": 0.0162,
      "step": 14681
    },
    {
      "epoch": 2.057743517869657,
      "grad_norm": 0.171087846159935,
      "learning_rate": 9.656541497249462e-05,
      "loss": 0.0243,
      "step": 14682
    },
    {
      "epoch": 2.0578836720392433,
      "grad_norm": 0.24988307058811188,
      "learning_rate": 9.655106433867495e-05,
      "loss": 0.0658,
      "step": 14683
    },
    {
      "epoch": 2.0580238262088297,
      "grad_norm": 0.1413131207227707,
      "learning_rate": 9.653671370485529e-05,
      "loss": 0.038,
      "step": 14684
    },
    {
      "epoch": 2.058163980378416,
      "grad_norm": 0.4566271901130676,
      "learning_rate": 9.652236307103563e-05,
      "loss": 0.0287,
      "step": 14685
    },
    {
      "epoch": 2.0583041345480027,
      "grad_norm": 0.24705936014652252,
      "learning_rate": 9.650801243721596e-05,
      "loss": 0.0739,
      "step": 14686
    },
    {
      "epoch": 2.0584442887175896,
      "grad_norm": 0.20026229321956635,
      "learning_rate": 9.64936618033963e-05,
      "loss": 0.0092,
      "step": 14687
    },
    {
      "epoch": 2.058584442887176,
      "grad_norm": 0.07073967903852463,
      "learning_rate": 9.647931116957665e-05,
      "loss": 0.0094,
      "step": 14688
    },
    {
      "epoch": 2.0587245970567625,
      "grad_norm": 0.4652431905269623,
      "learning_rate": 9.646496053575699e-05,
      "loss": 0.0341,
      "step": 14689
    },
    {
      "epoch": 2.058864751226349,
      "grad_norm": 0.47174856066703796,
      "learning_rate": 9.645060990193733e-05,
      "loss": 0.0589,
      "step": 14690
    },
    {
      "epoch": 2.0590049053959354,
      "grad_norm": 0.12462644279003143,
      "learning_rate": 9.643625926811766e-05,
      "loss": 0.0205,
      "step": 14691
    },
    {
      "epoch": 2.0591450595655223,
      "grad_norm": 0.2781469225883484,
      "learning_rate": 9.6421908634298e-05,
      "loss": 0.0503,
      "step": 14692
    },
    {
      "epoch": 2.0592852137351088,
      "grad_norm": 0.46155428886413574,
      "learning_rate": 9.640755800047836e-05,
      "loss": 0.0246,
      "step": 14693
    },
    {
      "epoch": 2.059425367904695,
      "grad_norm": 0.4891853928565979,
      "learning_rate": 9.639320736665869e-05,
      "loss": 0.0386,
      "step": 14694
    },
    {
      "epoch": 2.0595655220742817,
      "grad_norm": 0.12988385558128357,
      "learning_rate": 9.637885673283903e-05,
      "loss": 0.0149,
      "step": 14695
    },
    {
      "epoch": 2.059705676243868,
      "grad_norm": 0.5600318908691406,
      "learning_rate": 9.636450609901937e-05,
      "loss": 0.0392,
      "step": 14696
    },
    {
      "epoch": 2.0598458304134546,
      "grad_norm": 0.4391777217388153,
      "learning_rate": 9.63501554651997e-05,
      "loss": 0.0482,
      "step": 14697
    },
    {
      "epoch": 2.0599859845830415,
      "grad_norm": 0.14108222723007202,
      "learning_rate": 9.633580483138005e-05,
      "loss": 0.0238,
      "step": 14698
    },
    {
      "epoch": 2.060126138752628,
      "grad_norm": 0.12062130123376846,
      "learning_rate": 9.632145419756038e-05,
      "loss": 0.0289,
      "step": 14699
    },
    {
      "epoch": 2.0602662929222144,
      "grad_norm": 0.07438750565052032,
      "learning_rate": 9.630710356374072e-05,
      "loss": 0.0092,
      "step": 14700
    },
    {
      "epoch": 2.060406447091801,
      "grad_norm": 0.1751435101032257,
      "learning_rate": 9.629275292992107e-05,
      "loss": 0.0184,
      "step": 14701
    },
    {
      "epoch": 2.0605466012613873,
      "grad_norm": 0.19144080579280853,
      "learning_rate": 9.62784022961014e-05,
      "loss": 0.0127,
      "step": 14702
    },
    {
      "epoch": 2.0606867554309742,
      "grad_norm": 0.24372050166130066,
      "learning_rate": 9.626405166228174e-05,
      "loss": 0.0428,
      "step": 14703
    },
    {
      "epoch": 2.0608269096005607,
      "grad_norm": 0.16367536783218384,
      "learning_rate": 9.624970102846209e-05,
      "loss": 0.0592,
      "step": 14704
    },
    {
      "epoch": 2.060967063770147,
      "grad_norm": 0.2811152935028076,
      "learning_rate": 9.623535039464242e-05,
      "loss": 0.0204,
      "step": 14705
    },
    {
      "epoch": 2.0611072179397336,
      "grad_norm": 0.21813584864139557,
      "learning_rate": 9.622099976082276e-05,
      "loss": 0.0209,
      "step": 14706
    },
    {
      "epoch": 2.06124737210932,
      "grad_norm": 0.09859484434127808,
      "learning_rate": 9.620664912700309e-05,
      "loss": 0.0113,
      "step": 14707
    },
    {
      "epoch": 2.061387526278907,
      "grad_norm": 1.1235848665237427,
      "learning_rate": 9.619229849318343e-05,
      "loss": 0.0476,
      "step": 14708
    },
    {
      "epoch": 2.0615276804484934,
      "grad_norm": 0.2925051748752594,
      "learning_rate": 9.617794785936379e-05,
      "loss": 0.0202,
      "step": 14709
    },
    {
      "epoch": 2.06166783461808,
      "grad_norm": 0.6838991641998291,
      "learning_rate": 9.616359722554412e-05,
      "loss": 0.0469,
      "step": 14710
    },
    {
      "epoch": 2.0618079887876664,
      "grad_norm": 0.32836249470710754,
      "learning_rate": 9.614924659172446e-05,
      "loss": 0.0305,
      "step": 14711
    },
    {
      "epoch": 2.061948142957253,
      "grad_norm": 0.059826552867889404,
      "learning_rate": 9.61348959579048e-05,
      "loss": 0.0037,
      "step": 14712
    },
    {
      "epoch": 2.0620882971268397,
      "grad_norm": 0.7304503917694092,
      "learning_rate": 9.612054532408513e-05,
      "loss": 0.0692,
      "step": 14713
    },
    {
      "epoch": 2.062228451296426,
      "grad_norm": 0.21271809935569763,
      "learning_rate": 9.610619469026549e-05,
      "loss": 0.0427,
      "step": 14714
    },
    {
      "epoch": 2.0623686054660126,
      "grad_norm": 0.15730740129947662,
      "learning_rate": 9.609184405644582e-05,
      "loss": 0.0081,
      "step": 14715
    },
    {
      "epoch": 2.062508759635599,
      "grad_norm": 0.3494764268398285,
      "learning_rate": 9.607749342262616e-05,
      "loss": 0.0434,
      "step": 14716
    },
    {
      "epoch": 2.0626489138051856,
      "grad_norm": 0.6698757410049438,
      "learning_rate": 9.60631427888065e-05,
      "loss": 0.0376,
      "step": 14717
    },
    {
      "epoch": 2.0627890679747725,
      "grad_norm": 0.9941021203994751,
      "learning_rate": 9.604879215498683e-05,
      "loss": 0.0556,
      "step": 14718
    },
    {
      "epoch": 2.062929222144359,
      "grad_norm": 0.3633747696876526,
      "learning_rate": 9.603444152116717e-05,
      "loss": 0.0298,
      "step": 14719
    },
    {
      "epoch": 2.0630693763139454,
      "grad_norm": 0.8286240100860596,
      "learning_rate": 9.602009088734753e-05,
      "loss": 0.0695,
      "step": 14720
    },
    {
      "epoch": 2.063209530483532,
      "grad_norm": 0.37841561436653137,
      "learning_rate": 9.600574025352785e-05,
      "loss": 0.0645,
      "step": 14721
    },
    {
      "epoch": 2.0633496846531183,
      "grad_norm": 0.35046958923339844,
      "learning_rate": 9.59913896197082e-05,
      "loss": 0.0805,
      "step": 14722
    },
    {
      "epoch": 2.063489838822705,
      "grad_norm": 0.16203956305980682,
      "learning_rate": 9.597703898588853e-05,
      "loss": 0.0206,
      "step": 14723
    },
    {
      "epoch": 2.0636299929922917,
      "grad_norm": 0.09663776308298111,
      "learning_rate": 9.596268835206887e-05,
      "loss": 0.0134,
      "step": 14724
    },
    {
      "epoch": 2.063770147161878,
      "grad_norm": 0.2519320249557495,
      "learning_rate": 9.594833771824922e-05,
      "loss": 0.0378,
      "step": 14725
    },
    {
      "epoch": 2.0639103013314646,
      "grad_norm": 0.20961010456085205,
      "learning_rate": 9.593398708442955e-05,
      "loss": 0.0233,
      "step": 14726
    },
    {
      "epoch": 2.064050455501051,
      "grad_norm": 0.18142196536064148,
      "learning_rate": 9.59196364506099e-05,
      "loss": 0.0468,
      "step": 14727
    },
    {
      "epoch": 2.0641906096706375,
      "grad_norm": 0.197112038731575,
      "learning_rate": 9.590528581679024e-05,
      "loss": 0.0152,
      "step": 14728
    },
    {
      "epoch": 2.0643307638402244,
      "grad_norm": 0.3717029094696045,
      "learning_rate": 9.589093518297056e-05,
      "loss": 0.0231,
      "step": 14729
    },
    {
      "epoch": 2.064470918009811,
      "grad_norm": 0.22763249278068542,
      "learning_rate": 9.587658454915092e-05,
      "loss": 0.032,
      "step": 14730
    },
    {
      "epoch": 2.0646110721793973,
      "grad_norm": 0.13348527252674103,
      "learning_rate": 9.586223391533126e-05,
      "loss": 0.0242,
      "step": 14731
    },
    {
      "epoch": 2.064751226348984,
      "grad_norm": 0.34499162435531616,
      "learning_rate": 9.584788328151159e-05,
      "loss": 0.0215,
      "step": 14732
    },
    {
      "epoch": 2.0648913805185702,
      "grad_norm": 0.23842300474643707,
      "learning_rate": 9.583353264769193e-05,
      "loss": 0.0812,
      "step": 14733
    },
    {
      "epoch": 2.065031534688157,
      "grad_norm": 0.07128602266311646,
      "learning_rate": 9.581918201387226e-05,
      "loss": 0.0062,
      "step": 14734
    },
    {
      "epoch": 2.0651716888577436,
      "grad_norm": 0.2558954954147339,
      "learning_rate": 9.58048313800526e-05,
      "loss": 0.0076,
      "step": 14735
    },
    {
      "epoch": 2.06531184302733,
      "grad_norm": 0.10760050266981125,
      "learning_rate": 9.579048074623296e-05,
      "loss": 0.0058,
      "step": 14736
    },
    {
      "epoch": 2.0654519971969165,
      "grad_norm": 0.15660718083381653,
      "learning_rate": 9.577613011241329e-05,
      "loss": 0.0352,
      "step": 14737
    },
    {
      "epoch": 2.065592151366503,
      "grad_norm": 0.1692039519548416,
      "learning_rate": 9.576177947859363e-05,
      "loss": 0.0176,
      "step": 14738
    },
    {
      "epoch": 2.06573230553609,
      "grad_norm": 0.11902142316102982,
      "learning_rate": 9.574742884477397e-05,
      "loss": 0.0123,
      "step": 14739
    },
    {
      "epoch": 2.0658724597056763,
      "grad_norm": 0.10024394094944,
      "learning_rate": 9.57330782109543e-05,
      "loss": 0.0124,
      "step": 14740
    },
    {
      "epoch": 2.066012613875263,
      "grad_norm": 0.08610425889492035,
      "learning_rate": 9.571872757713466e-05,
      "loss": 0.0156,
      "step": 14741
    },
    {
      "epoch": 2.0661527680448493,
      "grad_norm": 0.1398169994354248,
      "learning_rate": 9.570437694331498e-05,
      "loss": 0.0291,
      "step": 14742
    },
    {
      "epoch": 2.0662929222144357,
      "grad_norm": 0.1331656128168106,
      "learning_rate": 9.569002630949533e-05,
      "loss": 0.0313,
      "step": 14743
    },
    {
      "epoch": 2.0664330763840226,
      "grad_norm": 0.25851210951805115,
      "learning_rate": 9.567567567567567e-05,
      "loss": 0.0219,
      "step": 14744
    },
    {
      "epoch": 2.066573230553609,
      "grad_norm": 0.2615129351615906,
      "learning_rate": 9.5661325041856e-05,
      "loss": 0.0345,
      "step": 14745
    },
    {
      "epoch": 2.0667133847231955,
      "grad_norm": 0.14532554149627686,
      "learning_rate": 9.564697440803635e-05,
      "loss": 0.0141,
      "step": 14746
    },
    {
      "epoch": 2.066853538892782,
      "grad_norm": 0.0849614068865776,
      "learning_rate": 9.56326237742167e-05,
      "loss": 0.0051,
      "step": 14747
    },
    {
      "epoch": 2.0669936930623685,
      "grad_norm": 0.4805326759815216,
      "learning_rate": 9.561827314039702e-05,
      "loss": 0.0438,
      "step": 14748
    },
    {
      "epoch": 2.067133847231955,
      "grad_norm": 0.26844966411590576,
      "learning_rate": 9.560392250657737e-05,
      "loss": 0.0099,
      "step": 14749
    },
    {
      "epoch": 2.067274001401542,
      "grad_norm": 0.3314116299152374,
      "learning_rate": 9.55895718727577e-05,
      "loss": 0.0184,
      "step": 14750
    },
    {
      "epoch": 2.0674141555711283,
      "grad_norm": 0.44598299264907837,
      "learning_rate": 9.557522123893804e-05,
      "loss": 0.0258,
      "step": 14751
    },
    {
      "epoch": 2.0675543097407147,
      "grad_norm": 0.15765361487865448,
      "learning_rate": 9.556087060511839e-05,
      "loss": 0.0129,
      "step": 14752
    },
    {
      "epoch": 2.067694463910301,
      "grad_norm": 0.20319585502147675,
      "learning_rate": 9.554651997129872e-05,
      "loss": 0.0566,
      "step": 14753
    },
    {
      "epoch": 2.0678346180798877,
      "grad_norm": 0.1543562114238739,
      "learning_rate": 9.553216933747906e-05,
      "loss": 0.02,
      "step": 14754
    },
    {
      "epoch": 2.0679747722494746,
      "grad_norm": 0.495862752199173,
      "learning_rate": 9.55178187036594e-05,
      "loss": 0.0193,
      "step": 14755
    },
    {
      "epoch": 2.068114926419061,
      "grad_norm": 0.14246585965156555,
      "learning_rate": 9.550346806983973e-05,
      "loss": 0.0339,
      "step": 14756
    },
    {
      "epoch": 2.0682550805886475,
      "grad_norm": 0.23880068957805634,
      "learning_rate": 9.548911743602009e-05,
      "loss": 0.0331,
      "step": 14757
    },
    {
      "epoch": 2.068395234758234,
      "grad_norm": 0.1509748101234436,
      "learning_rate": 9.547476680220042e-05,
      "loss": 0.0137,
      "step": 14758
    },
    {
      "epoch": 2.0685353889278204,
      "grad_norm": 0.01984417624771595,
      "learning_rate": 9.546041616838076e-05,
      "loss": 0.002,
      "step": 14759
    },
    {
      "epoch": 2.0686755430974073,
      "grad_norm": 0.8755910396575928,
      "learning_rate": 9.54460655345611e-05,
      "loss": 0.1317,
      "step": 14760
    },
    {
      "epoch": 2.0688156972669938,
      "grad_norm": 0.5292233824729919,
      "learning_rate": 9.543171490074143e-05,
      "loss": 0.1019,
      "step": 14761
    },
    {
      "epoch": 2.0689558514365802,
      "grad_norm": 0.060510825365781784,
      "learning_rate": 9.541736426692179e-05,
      "loss": 0.0026,
      "step": 14762
    },
    {
      "epoch": 2.0690960056061667,
      "grad_norm": 0.18027576804161072,
      "learning_rate": 9.540301363310213e-05,
      "loss": 0.0334,
      "step": 14763
    },
    {
      "epoch": 2.069236159775753,
      "grad_norm": 0.30132341384887695,
      "learning_rate": 9.538866299928246e-05,
      "loss": 0.0309,
      "step": 14764
    },
    {
      "epoch": 2.06937631394534,
      "grad_norm": 0.5033591985702515,
      "learning_rate": 9.53743123654628e-05,
      "loss": 0.1201,
      "step": 14765
    },
    {
      "epoch": 2.0695164681149265,
      "grad_norm": 0.47116026282310486,
      "learning_rate": 9.535996173164314e-05,
      "loss": 0.0153,
      "step": 14766
    },
    {
      "epoch": 2.069656622284513,
      "grad_norm": 0.03919925540685654,
      "learning_rate": 9.534561109782347e-05,
      "loss": 0.0029,
      "step": 14767
    },
    {
      "epoch": 2.0697967764540994,
      "grad_norm": 0.2626688778400421,
      "learning_rate": 9.533126046400383e-05,
      "loss": 0.0083,
      "step": 14768
    },
    {
      "epoch": 2.069936930623686,
      "grad_norm": 0.8016400933265686,
      "learning_rate": 9.531690983018415e-05,
      "loss": 0.0521,
      "step": 14769
    },
    {
      "epoch": 2.070077084793273,
      "grad_norm": 0.025379212573170662,
      "learning_rate": 9.53025591963645e-05,
      "loss": 0.002,
      "step": 14770
    },
    {
      "epoch": 2.0702172389628593,
      "grad_norm": 0.20608901977539062,
      "learning_rate": 9.528820856254484e-05,
      "loss": 0.0235,
      "step": 14771
    },
    {
      "epoch": 2.0703573931324457,
      "grad_norm": 0.19838140904903412,
      "learning_rate": 9.527385792872517e-05,
      "loss": 0.0165,
      "step": 14772
    },
    {
      "epoch": 2.070497547302032,
      "grad_norm": 0.14500342309474945,
      "learning_rate": 9.525950729490552e-05,
      "loss": 0.0181,
      "step": 14773
    },
    {
      "epoch": 2.0706377014716186,
      "grad_norm": 0.2321852147579193,
      "learning_rate": 9.524515666108586e-05,
      "loss": 0.0242,
      "step": 14774
    },
    {
      "epoch": 2.0707778556412055,
      "grad_norm": 0.17350716888904572,
      "learning_rate": 9.52308060272662e-05,
      "loss": 0.0407,
      "step": 14775
    },
    {
      "epoch": 2.070918009810792,
      "grad_norm": 0.3438485860824585,
      "learning_rate": 9.521645539344654e-05,
      "loss": 0.0234,
      "step": 14776
    },
    {
      "epoch": 2.0710581639803785,
      "grad_norm": 0.18411746621131897,
      "learning_rate": 9.520210475962686e-05,
      "loss": 0.0454,
      "step": 14777
    },
    {
      "epoch": 2.071198318149965,
      "grad_norm": 0.2771705985069275,
      "learning_rate": 9.518775412580722e-05,
      "loss": 0.0169,
      "step": 14778
    },
    {
      "epoch": 2.0713384723195514,
      "grad_norm": 0.061035752296447754,
      "learning_rate": 9.517340349198756e-05,
      "loss": 0.0042,
      "step": 14779
    },
    {
      "epoch": 2.071478626489138,
      "grad_norm": 0.26582252979278564,
      "learning_rate": 9.515905285816789e-05,
      "loss": 0.034,
      "step": 14780
    },
    {
      "epoch": 2.0716187806587247,
      "grad_norm": 0.37238767743110657,
      "learning_rate": 9.514470222434823e-05,
      "loss": 0.0659,
      "step": 14781
    },
    {
      "epoch": 2.071758934828311,
      "grad_norm": 0.07462254911661148,
      "learning_rate": 9.513035159052857e-05,
      "loss": 0.0042,
      "step": 14782
    },
    {
      "epoch": 2.0718990889978977,
      "grad_norm": 0.2020999938249588,
      "learning_rate": 9.51160009567089e-05,
      "loss": 0.0062,
      "step": 14783
    },
    {
      "epoch": 2.072039243167484,
      "grad_norm": 0.20204313099384308,
      "learning_rate": 9.510165032288926e-05,
      "loss": 0.0144,
      "step": 14784
    },
    {
      "epoch": 2.0721793973370706,
      "grad_norm": 0.171986386179924,
      "learning_rate": 9.508729968906959e-05,
      "loss": 0.0307,
      "step": 14785
    },
    {
      "epoch": 2.0723195515066575,
      "grad_norm": 0.32277095317840576,
      "learning_rate": 9.507294905524993e-05,
      "loss": 0.0572,
      "step": 14786
    },
    {
      "epoch": 2.072459705676244,
      "grad_norm": 0.22186830639839172,
      "learning_rate": 9.505859842143027e-05,
      "loss": 0.0178,
      "step": 14787
    },
    {
      "epoch": 2.0725998598458304,
      "grad_norm": 0.22441157698631287,
      "learning_rate": 9.50442477876106e-05,
      "loss": 0.0086,
      "step": 14788
    },
    {
      "epoch": 2.072740014015417,
      "grad_norm": 0.17712929844856262,
      "learning_rate": 9.502989715379096e-05,
      "loss": 0.0327,
      "step": 14789
    },
    {
      "epoch": 2.0728801681850033,
      "grad_norm": 0.22668881714344025,
      "learning_rate": 9.50155465199713e-05,
      "loss": 0.0116,
      "step": 14790
    },
    {
      "epoch": 2.07302032235459,
      "grad_norm": 0.26879608631134033,
      "learning_rate": 9.500119588615163e-05,
      "loss": 0.0419,
      "step": 14791
    },
    {
      "epoch": 2.0731604765241767,
      "grad_norm": 0.1716337651014328,
      "learning_rate": 9.498684525233197e-05,
      "loss": 0.0225,
      "step": 14792
    },
    {
      "epoch": 2.073300630693763,
      "grad_norm": 0.14381234347820282,
      "learning_rate": 9.49724946185123e-05,
      "loss": 0.0182,
      "step": 14793
    },
    {
      "epoch": 2.0734407848633496,
      "grad_norm": 0.11770530790090561,
      "learning_rate": 9.495814398469265e-05,
      "loss": 0.0244,
      "step": 14794
    },
    {
      "epoch": 2.073580939032936,
      "grad_norm": 0.2196437567472458,
      "learning_rate": 9.4943793350873e-05,
      "loss": 0.0492,
      "step": 14795
    },
    {
      "epoch": 2.073721093202523,
      "grad_norm": 0.31392401456832886,
      "learning_rate": 9.492944271705332e-05,
      "loss": 0.044,
      "step": 14796
    },
    {
      "epoch": 2.0738612473721094,
      "grad_norm": 0.4719558656215668,
      "learning_rate": 9.491509208323367e-05,
      "loss": 0.0526,
      "step": 14797
    },
    {
      "epoch": 2.074001401541696,
      "grad_norm": 0.14595955610275269,
      "learning_rate": 9.490074144941401e-05,
      "loss": 0.0135,
      "step": 14798
    },
    {
      "epoch": 2.0741415557112823,
      "grad_norm": 0.09964407235383987,
      "learning_rate": 9.488639081559435e-05,
      "loss": 0.0044,
      "step": 14799
    },
    {
      "epoch": 2.074281709880869,
      "grad_norm": 0.3317849934101105,
      "learning_rate": 9.487204018177469e-05,
      "loss": 0.0454,
      "step": 14800
    },
    {
      "epoch": 2.0744218640504557,
      "grad_norm": 0.3520927131175995,
      "learning_rate": 9.485768954795502e-05,
      "loss": 0.02,
      "step": 14801
    },
    {
      "epoch": 2.074562018220042,
      "grad_norm": 0.18402360379695892,
      "learning_rate": 9.484333891413536e-05,
      "loss": 0.0139,
      "step": 14802
    },
    {
      "epoch": 2.0747021723896286,
      "grad_norm": 0.25146934390068054,
      "learning_rate": 9.48289882803157e-05,
      "loss": 0.0447,
      "step": 14803
    },
    {
      "epoch": 2.074842326559215,
      "grad_norm": 0.7736601233482361,
      "learning_rate": 9.481463764649603e-05,
      "loss": 0.0533,
      "step": 14804
    },
    {
      "epoch": 2.0749824807288015,
      "grad_norm": 0.13212797045707703,
      "learning_rate": 9.480028701267639e-05,
      "loss": 0.0405,
      "step": 14805
    },
    {
      "epoch": 2.0751226348983884,
      "grad_norm": 0.24164605140686035,
      "learning_rate": 9.478593637885673e-05,
      "loss": 0.0506,
      "step": 14806
    },
    {
      "epoch": 2.075262789067975,
      "grad_norm": 0.17525669932365417,
      "learning_rate": 9.477158574503706e-05,
      "loss": 0.0293,
      "step": 14807
    },
    {
      "epoch": 2.0754029432375614,
      "grad_norm": 0.2722138464450836,
      "learning_rate": 9.47572351112174e-05,
      "loss": 0.036,
      "step": 14808
    },
    {
      "epoch": 2.075543097407148,
      "grad_norm": 0.27481138706207275,
      "learning_rate": 9.474288447739776e-05,
      "loss": 0.0679,
      "step": 14809
    },
    {
      "epoch": 2.0756832515767343,
      "grad_norm": 0.16709543764591217,
      "learning_rate": 9.472853384357809e-05,
      "loss": 0.0326,
      "step": 14810
    },
    {
      "epoch": 2.0758234057463207,
      "grad_norm": 0.4637592136859894,
      "learning_rate": 9.471418320975843e-05,
      "loss": 0.0382,
      "step": 14811
    },
    {
      "epoch": 2.0759635599159076,
      "grad_norm": 0.3533687889575958,
      "learning_rate": 9.469983257593876e-05,
      "loss": 0.0362,
      "step": 14812
    },
    {
      "epoch": 2.076103714085494,
      "grad_norm": 0.15624204277992249,
      "learning_rate": 9.46854819421191e-05,
      "loss": 0.0531,
      "step": 14813
    },
    {
      "epoch": 2.0762438682550806,
      "grad_norm": 0.6235769987106323,
      "learning_rate": 9.467113130829944e-05,
      "loss": 0.016,
      "step": 14814
    },
    {
      "epoch": 2.076384022424667,
      "grad_norm": 0.5708448886871338,
      "learning_rate": 9.465678067447978e-05,
      "loss": 0.0708,
      "step": 14815
    },
    {
      "epoch": 2.0765241765942535,
      "grad_norm": 0.8833497762680054,
      "learning_rate": 9.464243004066013e-05,
      "loss": 0.1026,
      "step": 14816
    },
    {
      "epoch": 2.0766643307638404,
      "grad_norm": 0.5276862382888794,
      "learning_rate": 9.462807940684047e-05,
      "loss": 0.0206,
      "step": 14817
    },
    {
      "epoch": 2.076804484933427,
      "grad_norm": 0.05500069260597229,
      "learning_rate": 9.46137287730208e-05,
      "loss": 0.0028,
      "step": 14818
    },
    {
      "epoch": 2.0769446391030133,
      "grad_norm": 1.0682421922683716,
      "learning_rate": 9.459937813920114e-05,
      "loss": 0.037,
      "step": 14819
    },
    {
      "epoch": 2.0770847932725998,
      "grad_norm": 3.1182734966278076,
      "learning_rate": 9.458502750538147e-05,
      "loss": 0.0421,
      "step": 14820
    },
    {
      "epoch": 2.077224947442186,
      "grad_norm": 0.07194377481937408,
      "learning_rate": 9.457067687156182e-05,
      "loss": 0.0065,
      "step": 14821
    },
    {
      "epoch": 2.077365101611773,
      "grad_norm": 0.3269622027873993,
      "learning_rate": 9.455632623774216e-05,
      "loss": 0.0164,
      "step": 14822
    },
    {
      "epoch": 2.0775052557813596,
      "grad_norm": 0.5817004442214966,
      "learning_rate": 9.454197560392249e-05,
      "loss": 0.0766,
      "step": 14823
    },
    {
      "epoch": 2.077645409950946,
      "grad_norm": 0.09402894228696823,
      "learning_rate": 9.452762497010283e-05,
      "loss": 0.0181,
      "step": 14824
    },
    {
      "epoch": 2.0777855641205325,
      "grad_norm": 0.28523150086402893,
      "learning_rate": 9.451327433628319e-05,
      "loss": 0.0746,
      "step": 14825
    },
    {
      "epoch": 2.077925718290119,
      "grad_norm": 0.1748807430267334,
      "learning_rate": 9.449892370246352e-05,
      "loss": 0.0121,
      "step": 14826
    },
    {
      "epoch": 2.078065872459706,
      "grad_norm": 0.0679233968257904,
      "learning_rate": 9.448457306864386e-05,
      "loss": 0.0047,
      "step": 14827
    },
    {
      "epoch": 2.0782060266292923,
      "grad_norm": 0.13750289380550385,
      "learning_rate": 9.447022243482419e-05,
      "loss": 0.0448,
      "step": 14828
    },
    {
      "epoch": 2.078346180798879,
      "grad_norm": 0.4468323290348053,
      "learning_rate": 9.445587180100453e-05,
      "loss": 0.0615,
      "step": 14829
    },
    {
      "epoch": 2.0784863349684652,
      "grad_norm": 0.13197042047977448,
      "learning_rate": 9.444152116718487e-05,
      "loss": 0.0156,
      "step": 14830
    },
    {
      "epoch": 2.0786264891380517,
      "grad_norm": 0.3296675384044647,
      "learning_rate": 9.442717053336522e-05,
      "loss": 0.0473,
      "step": 14831
    },
    {
      "epoch": 2.0787666433076386,
      "grad_norm": 0.24824611842632294,
      "learning_rate": 9.441281989954556e-05,
      "loss": 0.0215,
      "step": 14832
    },
    {
      "epoch": 2.078906797477225,
      "grad_norm": 0.11044174432754517,
      "learning_rate": 9.43984692657259e-05,
      "loss": 0.0115,
      "step": 14833
    },
    {
      "epoch": 2.0790469516468115,
      "grad_norm": 0.34932875633239746,
      "learning_rate": 9.438411863190623e-05,
      "loss": 0.0189,
      "step": 14834
    },
    {
      "epoch": 2.079187105816398,
      "grad_norm": 0.1812528520822525,
      "learning_rate": 9.436976799808657e-05,
      "loss": 0.0153,
      "step": 14835
    },
    {
      "epoch": 2.0793272599859844,
      "grad_norm": 0.35823875665664673,
      "learning_rate": 9.43554173642669e-05,
      "loss": 0.0553,
      "step": 14836
    },
    {
      "epoch": 2.0794674141555713,
      "grad_norm": 0.3172191381454468,
      "learning_rate": 9.434106673044726e-05,
      "loss": 0.0371,
      "step": 14837
    },
    {
      "epoch": 2.079607568325158,
      "grad_norm": 0.46404993534088135,
      "learning_rate": 9.43267160966276e-05,
      "loss": 0.0778,
      "step": 14838
    },
    {
      "epoch": 2.0797477224947443,
      "grad_norm": 0.10121031105518341,
      "learning_rate": 9.431236546280793e-05,
      "loss": 0.0078,
      "step": 14839
    },
    {
      "epoch": 2.0798878766643307,
      "grad_norm": 0.1828089952468872,
      "learning_rate": 9.429801482898827e-05,
      "loss": 0.0269,
      "step": 14840
    },
    {
      "epoch": 2.080028030833917,
      "grad_norm": 0.2937658131122589,
      "learning_rate": 9.428366419516862e-05,
      "loss": 0.0156,
      "step": 14841
    },
    {
      "epoch": 2.0801681850035036,
      "grad_norm": 0.15324236452579498,
      "learning_rate": 9.426931356134895e-05,
      "loss": 0.0307,
      "step": 14842
    },
    {
      "epoch": 2.0803083391730905,
      "grad_norm": 0.06872966140508652,
      "learning_rate": 9.42549629275293e-05,
      "loss": 0.0044,
      "step": 14843
    },
    {
      "epoch": 2.080448493342677,
      "grad_norm": 0.3504643440246582,
      "learning_rate": 9.424061229370964e-05,
      "loss": 0.0516,
      "step": 14844
    },
    {
      "epoch": 2.0805886475122635,
      "grad_norm": 0.370417982339859,
      "learning_rate": 9.422626165988997e-05,
      "loss": 0.0282,
      "step": 14845
    },
    {
      "epoch": 2.08072880168185,
      "grad_norm": 0.444980263710022,
      "learning_rate": 9.421191102607031e-05,
      "loss": 0.0259,
      "step": 14846
    },
    {
      "epoch": 2.0808689558514364,
      "grad_norm": 0.029027914628386497,
      "learning_rate": 9.419756039225065e-05,
      "loss": 0.0026,
      "step": 14847
    },
    {
      "epoch": 2.0810091100210233,
      "grad_norm": 0.16119834780693054,
      "learning_rate": 9.418320975843099e-05,
      "loss": 0.0161,
      "step": 14848
    },
    {
      "epoch": 2.0811492641906097,
      "grad_norm": 0.022371480241417885,
      "learning_rate": 9.416885912461133e-05,
      "loss": 0.0026,
      "step": 14849
    },
    {
      "epoch": 2.081289418360196,
      "grad_norm": 0.6151967644691467,
      "learning_rate": 9.415450849079166e-05,
      "loss": 0.0567,
      "step": 14850
    },
    {
      "epoch": 2.0814295725297827,
      "grad_norm": 0.2203221172094345,
      "learning_rate": 9.4140157856972e-05,
      "loss": 0.0163,
      "step": 14851
    },
    {
      "epoch": 2.081569726699369,
      "grad_norm": 0.16146695613861084,
      "learning_rate": 9.412580722315236e-05,
      "loss": 0.0175,
      "step": 14852
    },
    {
      "epoch": 2.081709880868956,
      "grad_norm": 0.34382033348083496,
      "learning_rate": 9.411145658933269e-05,
      "loss": 0.0363,
      "step": 14853
    },
    {
      "epoch": 2.0818500350385425,
      "grad_norm": 0.14570355415344238,
      "learning_rate": 9.409710595551303e-05,
      "loss": 0.0126,
      "step": 14854
    },
    {
      "epoch": 2.081990189208129,
      "grad_norm": 0.1399497538805008,
      "learning_rate": 9.408275532169336e-05,
      "loss": 0.0029,
      "step": 14855
    },
    {
      "epoch": 2.0821303433777154,
      "grad_norm": 0.7136857509613037,
      "learning_rate": 9.40684046878737e-05,
      "loss": 0.1233,
      "step": 14856
    },
    {
      "epoch": 2.082270497547302,
      "grad_norm": 0.08985535055398941,
      "learning_rate": 9.405405405405406e-05,
      "loss": 0.0068,
      "step": 14857
    },
    {
      "epoch": 2.0824106517168888,
      "grad_norm": 0.7420704364776611,
      "learning_rate": 9.403970342023439e-05,
      "loss": 0.0596,
      "step": 14858
    },
    {
      "epoch": 2.0825508058864752,
      "grad_norm": 0.27444449067115784,
      "learning_rate": 9.402535278641473e-05,
      "loss": 0.0115,
      "step": 14859
    },
    {
      "epoch": 2.0826909600560617,
      "grad_norm": 0.02939487434923649,
      "learning_rate": 9.401100215259507e-05,
      "loss": 0.0025,
      "step": 14860
    },
    {
      "epoch": 2.082831114225648,
      "grad_norm": 0.3931616246700287,
      "learning_rate": 9.39966515187754e-05,
      "loss": 0.0309,
      "step": 14861
    },
    {
      "epoch": 2.0829712683952346,
      "grad_norm": 0.38206520676612854,
      "learning_rate": 9.398230088495574e-05,
      "loss": 0.057,
      "step": 14862
    },
    {
      "epoch": 2.0831114225648215,
      "grad_norm": 0.02849535085260868,
      "learning_rate": 9.396795025113608e-05,
      "loss": 0.0031,
      "step": 14863
    },
    {
      "epoch": 2.083251576734408,
      "grad_norm": 0.1993153840303421,
      "learning_rate": 9.395359961731642e-05,
      "loss": 0.0223,
      "step": 14864
    },
    {
      "epoch": 2.0833917309039944,
      "grad_norm": 0.5830625891685486,
      "learning_rate": 9.393924898349677e-05,
      "loss": 0.0331,
      "step": 14865
    },
    {
      "epoch": 2.083531885073581,
      "grad_norm": 0.39083006978034973,
      "learning_rate": 9.39248983496771e-05,
      "loss": 0.0347,
      "step": 14866
    },
    {
      "epoch": 2.0836720392431674,
      "grad_norm": 0.27150633931159973,
      "learning_rate": 9.391054771585744e-05,
      "loss": 0.0439,
      "step": 14867
    },
    {
      "epoch": 2.0838121934127543,
      "grad_norm": 0.4067094326019287,
      "learning_rate": 9.389619708203779e-05,
      "loss": 0.0573,
      "step": 14868
    },
    {
      "epoch": 2.0839523475823407,
      "grad_norm": 0.35810044407844543,
      "learning_rate": 9.388184644821812e-05,
      "loss": 0.0382,
      "step": 14869
    },
    {
      "epoch": 2.084092501751927,
      "grad_norm": 0.8136069178581238,
      "learning_rate": 9.386749581439846e-05,
      "loss": 0.0432,
      "step": 14870
    },
    {
      "epoch": 2.0842326559215136,
      "grad_norm": 0.06815282255411148,
      "learning_rate": 9.385314518057879e-05,
      "loss": 0.0075,
      "step": 14871
    },
    {
      "epoch": 2.0843728100911,
      "grad_norm": 0.1116667091846466,
      "learning_rate": 9.383879454675913e-05,
      "loss": 0.0085,
      "step": 14872
    },
    {
      "epoch": 2.0845129642606866,
      "grad_norm": 0.24363498389720917,
      "learning_rate": 9.382444391293949e-05,
      "loss": 0.0626,
      "step": 14873
    },
    {
      "epoch": 2.0846531184302735,
      "grad_norm": 0.17404122650623322,
      "learning_rate": 9.381009327911982e-05,
      "loss": 0.0119,
      "step": 14874
    },
    {
      "epoch": 2.08479327259986,
      "grad_norm": 0.25800928473472595,
      "learning_rate": 9.379574264530016e-05,
      "loss": 0.0028,
      "step": 14875
    },
    {
      "epoch": 2.0849334267694464,
      "grad_norm": 0.09687094390392303,
      "learning_rate": 9.37813920114805e-05,
      "loss": 0.0252,
      "step": 14876
    },
    {
      "epoch": 2.085073580939033,
      "grad_norm": 0.1450301706790924,
      "learning_rate": 9.376704137766083e-05,
      "loss": 0.0343,
      "step": 14877
    },
    {
      "epoch": 2.0852137351086193,
      "grad_norm": 0.14033378660678864,
      "learning_rate": 9.375269074384117e-05,
      "loss": 0.0139,
      "step": 14878
    },
    {
      "epoch": 2.085353889278206,
      "grad_norm": 0.19887270033359528,
      "learning_rate": 9.373834011002153e-05,
      "loss": 0.0286,
      "step": 14879
    },
    {
      "epoch": 2.0854940434477927,
      "grad_norm": 0.38860470056533813,
      "learning_rate": 9.372398947620186e-05,
      "loss": 0.0248,
      "step": 14880
    },
    {
      "epoch": 2.085634197617379,
      "grad_norm": 0.20430448651313782,
      "learning_rate": 9.37096388423822e-05,
      "loss": 0.0423,
      "step": 14881
    },
    {
      "epoch": 2.0857743517869656,
      "grad_norm": 0.11464152485132217,
      "learning_rate": 9.369528820856253e-05,
      "loss": 0.0208,
      "step": 14882
    },
    {
      "epoch": 2.085914505956552,
      "grad_norm": 0.06624356657266617,
      "learning_rate": 9.368093757474287e-05,
      "loss": 0.0032,
      "step": 14883
    },
    {
      "epoch": 2.086054660126139,
      "grad_norm": 0.24441885948181152,
      "learning_rate": 9.366658694092323e-05,
      "loss": 0.0275,
      "step": 14884
    },
    {
      "epoch": 2.0861948142957254,
      "grad_norm": 0.4828717112541199,
      "learning_rate": 9.365223630710355e-05,
      "loss": 0.1089,
      "step": 14885
    },
    {
      "epoch": 2.086334968465312,
      "grad_norm": 0.3373447060585022,
      "learning_rate": 9.36378856732839e-05,
      "loss": 0.0398,
      "step": 14886
    },
    {
      "epoch": 2.0864751226348983,
      "grad_norm": 0.23026028275489807,
      "learning_rate": 9.362353503946424e-05,
      "loss": 0.0349,
      "step": 14887
    },
    {
      "epoch": 2.0866152768044848,
      "grad_norm": 0.40679338574409485,
      "learning_rate": 9.360918440564457e-05,
      "loss": 0.0214,
      "step": 14888
    },
    {
      "epoch": 2.0867554309740717,
      "grad_norm": 0.09359004348516464,
      "learning_rate": 9.359483377182492e-05,
      "loss": 0.054,
      "step": 14889
    },
    {
      "epoch": 2.086895585143658,
      "grad_norm": 0.4207221567630768,
      "learning_rate": 9.358048313800525e-05,
      "loss": 0.0301,
      "step": 14890
    },
    {
      "epoch": 2.0870357393132446,
      "grad_norm": 0.056044742465019226,
      "learning_rate": 9.35661325041856e-05,
      "loss": 0.0055,
      "step": 14891
    },
    {
      "epoch": 2.087175893482831,
      "grad_norm": 0.14045846462249756,
      "learning_rate": 9.355178187036594e-05,
      "loss": 0.0221,
      "step": 14892
    },
    {
      "epoch": 2.0873160476524175,
      "grad_norm": 1.7049165964126587,
      "learning_rate": 9.353743123654626e-05,
      "loss": 0.0691,
      "step": 14893
    },
    {
      "epoch": 2.087456201822004,
      "grad_norm": 0.08491361886262894,
      "learning_rate": 9.35230806027266e-05,
      "loss": 0.0094,
      "step": 14894
    },
    {
      "epoch": 2.087596355991591,
      "grad_norm": 0.18862558901309967,
      "learning_rate": 9.350872996890696e-05,
      "loss": 0.0523,
      "step": 14895
    },
    {
      "epoch": 2.0877365101611773,
      "grad_norm": 0.4013764560222626,
      "learning_rate": 9.349437933508729e-05,
      "loss": 0.073,
      "step": 14896
    },
    {
      "epoch": 2.087876664330764,
      "grad_norm": 0.8161226511001587,
      "learning_rate": 9.348002870126763e-05,
      "loss": 0.0526,
      "step": 14897
    },
    {
      "epoch": 2.0880168185003503,
      "grad_norm": 0.24728426337242126,
      "learning_rate": 9.346567806744796e-05,
      "loss": 0.0491,
      "step": 14898
    },
    {
      "epoch": 2.0881569726699367,
      "grad_norm": 0.15324163436889648,
      "learning_rate": 9.34513274336283e-05,
      "loss": 0.0472,
      "step": 14899
    },
    {
      "epoch": 2.0882971268395236,
      "grad_norm": 0.048960451036691666,
      "learning_rate": 9.343697679980866e-05,
      "loss": 0.005,
      "step": 14900
    },
    {
      "epoch": 2.08843728100911,
      "grad_norm": 0.08626089245080948,
      "learning_rate": 9.342262616598899e-05,
      "loss": 0.0035,
      "step": 14901
    },
    {
      "epoch": 2.0885774351786965,
      "grad_norm": 0.31123363971710205,
      "learning_rate": 9.340827553216933e-05,
      "loss": 0.0281,
      "step": 14902
    },
    {
      "epoch": 2.088717589348283,
      "grad_norm": 0.19551149010658264,
      "learning_rate": 9.339392489834967e-05,
      "loss": 0.0169,
      "step": 14903
    },
    {
      "epoch": 2.0888577435178695,
      "grad_norm": 0.19941270351409912,
      "learning_rate": 9.337957426453e-05,
      "loss": 0.0093,
      "step": 14904
    },
    {
      "epoch": 2.0889978976874564,
      "grad_norm": 0.09357432276010513,
      "learning_rate": 9.336522363071036e-05,
      "loss": 0.0048,
      "step": 14905
    },
    {
      "epoch": 2.089138051857043,
      "grad_norm": 0.3881312608718872,
      "learning_rate": 9.335087299689068e-05,
      "loss": 0.0264,
      "step": 14906
    },
    {
      "epoch": 2.0892782060266293,
      "grad_norm": 0.46990731358528137,
      "learning_rate": 9.333652236307103e-05,
      "loss": 0.0462,
      "step": 14907
    },
    {
      "epoch": 2.0894183601962157,
      "grad_norm": 0.09395871311426163,
      "learning_rate": 9.332217172925137e-05,
      "loss": 0.0039,
      "step": 14908
    },
    {
      "epoch": 2.089558514365802,
      "grad_norm": 0.4471248984336853,
      "learning_rate": 9.33078210954317e-05,
      "loss": 0.0274,
      "step": 14909
    },
    {
      "epoch": 2.089698668535389,
      "grad_norm": 0.28431451320648193,
      "learning_rate": 9.329347046161205e-05,
      "loss": 0.0214,
      "step": 14910
    },
    {
      "epoch": 2.0898388227049756,
      "grad_norm": 0.769780695438385,
      "learning_rate": 9.32791198277924e-05,
      "loss": 0.0725,
      "step": 14911
    },
    {
      "epoch": 2.089978976874562,
      "grad_norm": 0.18135784566402435,
      "learning_rate": 9.326476919397272e-05,
      "loss": 0.0185,
      "step": 14912
    },
    {
      "epoch": 2.0901191310441485,
      "grad_norm": 0.20114558935165405,
      "learning_rate": 9.325041856015307e-05,
      "loss": 0.008,
      "step": 14913
    },
    {
      "epoch": 2.090259285213735,
      "grad_norm": 0.5058400630950928,
      "learning_rate": 9.323606792633341e-05,
      "loss": 0.0284,
      "step": 14914
    },
    {
      "epoch": 2.090399439383322,
      "grad_norm": 0.11433473974466324,
      "learning_rate": 9.322171729251374e-05,
      "loss": 0.0197,
      "step": 14915
    },
    {
      "epoch": 2.0905395935529083,
      "grad_norm": 0.22332864999771118,
      "learning_rate": 9.320736665869409e-05,
      "loss": 0.0287,
      "step": 14916
    },
    {
      "epoch": 2.0906797477224948,
      "grad_norm": 0.05541842058300972,
      "learning_rate": 9.319301602487442e-05,
      "loss": 0.0062,
      "step": 14917
    },
    {
      "epoch": 2.0908199018920812,
      "grad_norm": 0.20706160366535187,
      "learning_rate": 9.317866539105476e-05,
      "loss": 0.0071,
      "step": 14918
    },
    {
      "epoch": 2.0909600560616677,
      "grad_norm": 0.7946418523788452,
      "learning_rate": 9.31643147572351e-05,
      "loss": 0.2562,
      "step": 14919
    },
    {
      "epoch": 2.0911002102312546,
      "grad_norm": 0.3528929650783539,
      "learning_rate": 9.314996412341543e-05,
      "loss": 0.0479,
      "step": 14920
    },
    {
      "epoch": 2.091240364400841,
      "grad_norm": 0.5547553896903992,
      "learning_rate": 9.313561348959579e-05,
      "loss": 0.0528,
      "step": 14921
    },
    {
      "epoch": 2.0913805185704275,
      "grad_norm": 0.2563416063785553,
      "learning_rate": 9.312126285577613e-05,
      "loss": 0.0476,
      "step": 14922
    },
    {
      "epoch": 2.091520672740014,
      "grad_norm": 0.12070932239294052,
      "learning_rate": 9.310691222195646e-05,
      "loss": 0.0184,
      "step": 14923
    },
    {
      "epoch": 2.0916608269096004,
      "grad_norm": 0.22918175160884857,
      "learning_rate": 9.30925615881368e-05,
      "loss": 0.0347,
      "step": 14924
    },
    {
      "epoch": 2.091800981079187,
      "grad_norm": 0.4921470880508423,
      "learning_rate": 9.307821095431713e-05,
      "loss": 0.0564,
      "step": 14925
    },
    {
      "epoch": 2.091941135248774,
      "grad_norm": 0.3449842929840088,
      "learning_rate": 9.306386032049749e-05,
      "loss": 0.0686,
      "step": 14926
    },
    {
      "epoch": 2.0920812894183602,
      "grad_norm": 0.4262050688266754,
      "learning_rate": 9.304950968667783e-05,
      "loss": 0.0416,
      "step": 14927
    },
    {
      "epoch": 2.0922214435879467,
      "grad_norm": 0.3482522666454315,
      "learning_rate": 9.303515905285816e-05,
      "loss": 0.0443,
      "step": 14928
    },
    {
      "epoch": 2.092361597757533,
      "grad_norm": 0.35002216696739197,
      "learning_rate": 9.30208084190385e-05,
      "loss": 0.0447,
      "step": 14929
    },
    {
      "epoch": 2.0925017519271196,
      "grad_norm": 0.12243515253067017,
      "learning_rate": 9.300645778521884e-05,
      "loss": 0.0067,
      "step": 14930
    },
    {
      "epoch": 2.0926419060967065,
      "grad_norm": 0.23505719006061554,
      "learning_rate": 9.299210715139917e-05,
      "loss": 0.0247,
      "step": 14931
    },
    {
      "epoch": 2.092782060266293,
      "grad_norm": 0.7883654832839966,
      "learning_rate": 9.297775651757953e-05,
      "loss": 0.0863,
      "step": 14932
    },
    {
      "epoch": 2.0929222144358794,
      "grad_norm": 0.26290085911750793,
      "learning_rate": 9.296340588375985e-05,
      "loss": 0.0134,
      "step": 14933
    },
    {
      "epoch": 2.093062368605466,
      "grad_norm": 0.2892436981201172,
      "learning_rate": 9.29490552499402e-05,
      "loss": 0.022,
      "step": 14934
    },
    {
      "epoch": 2.0932025227750524,
      "grad_norm": 0.25163471698760986,
      "learning_rate": 9.293470461612054e-05,
      "loss": 0.0328,
      "step": 14935
    },
    {
      "epoch": 2.0933426769446393,
      "grad_norm": 0.047421135008335114,
      "learning_rate": 9.292035398230087e-05,
      "loss": 0.0041,
      "step": 14936
    },
    {
      "epoch": 2.0934828311142257,
      "grad_norm": 0.1715579628944397,
      "learning_rate": 9.290600334848122e-05,
      "loss": 0.0104,
      "step": 14937
    },
    {
      "epoch": 2.093622985283812,
      "grad_norm": 0.2589839994907379,
      "learning_rate": 9.289165271466156e-05,
      "loss": 0.0371,
      "step": 14938
    },
    {
      "epoch": 2.0937631394533986,
      "grad_norm": 0.18597784638404846,
      "learning_rate": 9.287730208084189e-05,
      "loss": 0.0228,
      "step": 14939
    },
    {
      "epoch": 2.093903293622985,
      "grad_norm": 0.2406667172908783,
      "learning_rate": 9.286295144702224e-05,
      "loss": 0.0158,
      "step": 14940
    },
    {
      "epoch": 2.094043447792572,
      "grad_norm": 0.05905997008085251,
      "learning_rate": 9.284860081320256e-05,
      "loss": 0.0051,
      "step": 14941
    },
    {
      "epoch": 2.0941836019621585,
      "grad_norm": 0.09497687965631485,
      "learning_rate": 9.283425017938292e-05,
      "loss": 0.0112,
      "step": 14942
    },
    {
      "epoch": 2.094323756131745,
      "grad_norm": 0.11059311777353287,
      "learning_rate": 9.281989954556326e-05,
      "loss": 0.0303,
      "step": 14943
    },
    {
      "epoch": 2.0944639103013314,
      "grad_norm": 0.25810706615448,
      "learning_rate": 9.280554891174359e-05,
      "loss": 0.0248,
      "step": 14944
    },
    {
      "epoch": 2.094604064470918,
      "grad_norm": 0.11542551964521408,
      "learning_rate": 9.279119827792393e-05,
      "loss": 0.0104,
      "step": 14945
    },
    {
      "epoch": 2.0947442186405048,
      "grad_norm": 0.3972245156764984,
      "learning_rate": 9.277684764410427e-05,
      "loss": 0.0486,
      "step": 14946
    },
    {
      "epoch": 2.094884372810091,
      "grad_norm": 0.4474865198135376,
      "learning_rate": 9.27624970102846e-05,
      "loss": 0.0626,
      "step": 14947
    },
    {
      "epoch": 2.0950245269796777,
      "grad_norm": 0.15198349952697754,
      "learning_rate": 9.274814637646496e-05,
      "loss": 0.0087,
      "step": 14948
    },
    {
      "epoch": 2.095164681149264,
      "grad_norm": 0.21712373197078705,
      "learning_rate": 9.273379574264529e-05,
      "loss": 0.0449,
      "step": 14949
    },
    {
      "epoch": 2.0953048353188506,
      "grad_norm": 0.3505094647407532,
      "learning_rate": 9.271944510882563e-05,
      "loss": 0.0591,
      "step": 14950
    },
    {
      "epoch": 2.0954449894884375,
      "grad_norm": 0.21627293527126312,
      "learning_rate": 9.270509447500597e-05,
      "loss": 0.0131,
      "step": 14951
    },
    {
      "epoch": 2.095585143658024,
      "grad_norm": 0.2086072862148285,
      "learning_rate": 9.26907438411863e-05,
      "loss": 0.0297,
      "step": 14952
    },
    {
      "epoch": 2.0957252978276104,
      "grad_norm": 0.04036533087491989,
      "learning_rate": 9.267639320736666e-05,
      "loss": 0.0042,
      "step": 14953
    },
    {
      "epoch": 2.095865451997197,
      "grad_norm": 0.4073203206062317,
      "learning_rate": 9.2662042573547e-05,
      "loss": 0.049,
      "step": 14954
    },
    {
      "epoch": 2.0960056061667833,
      "grad_norm": 0.32072335481643677,
      "learning_rate": 9.264769193972733e-05,
      "loss": 0.0179,
      "step": 14955
    },
    {
      "epoch": 2.09614576033637,
      "grad_norm": 0.21063660085201263,
      "learning_rate": 9.263334130590767e-05,
      "loss": 0.0112,
      "step": 14956
    },
    {
      "epoch": 2.0962859145059567,
      "grad_norm": 0.22807221114635468,
      "learning_rate": 9.261899067208801e-05,
      "loss": 0.0194,
      "step": 14957
    },
    {
      "epoch": 2.096426068675543,
      "grad_norm": 0.20837998390197754,
      "learning_rate": 9.260464003826835e-05,
      "loss": 0.0346,
      "step": 14958
    },
    {
      "epoch": 2.0965662228451296,
      "grad_norm": 0.20721688866615295,
      "learning_rate": 9.25902894044487e-05,
      "loss": 0.0247,
      "step": 14959
    },
    {
      "epoch": 2.096706377014716,
      "grad_norm": 0.18544557690620422,
      "learning_rate": 9.257593877062902e-05,
      "loss": 0.015,
      "step": 14960
    },
    {
      "epoch": 2.0968465311843025,
      "grad_norm": 0.15007969737052917,
      "learning_rate": 9.256158813680937e-05,
      "loss": 0.0247,
      "step": 14961
    },
    {
      "epoch": 2.0969866853538894,
      "grad_norm": 0.5063300728797913,
      "learning_rate": 9.254723750298971e-05,
      "loss": 0.03,
      "step": 14962
    },
    {
      "epoch": 2.097126839523476,
      "grad_norm": 0.12026919424533844,
      "learning_rate": 9.253288686917004e-05,
      "loss": 0.0187,
      "step": 14963
    },
    {
      "epoch": 2.0972669936930624,
      "grad_norm": 0.5496593713760376,
      "learning_rate": 9.251853623535039e-05,
      "loss": 0.0387,
      "step": 14964
    },
    {
      "epoch": 2.097407147862649,
      "grad_norm": 2.127610206604004,
      "learning_rate": 9.250418560153073e-05,
      "loss": 0.1238,
      "step": 14965
    },
    {
      "epoch": 2.0975473020322353,
      "grad_norm": 0.35284027457237244,
      "learning_rate": 9.248983496771106e-05,
      "loss": 0.0308,
      "step": 14966
    },
    {
      "epoch": 2.097687456201822,
      "grad_norm": 0.6448554396629333,
      "learning_rate": 9.24754843338914e-05,
      "loss": 0.0573,
      "step": 14967
    },
    {
      "epoch": 2.0978276103714086,
      "grad_norm": 0.1658954620361328,
      "learning_rate": 9.246113370007173e-05,
      "loss": 0.0177,
      "step": 14968
    },
    {
      "epoch": 2.097967764540995,
      "grad_norm": 0.1602509766817093,
      "learning_rate": 9.244678306625209e-05,
      "loss": 0.0134,
      "step": 14969
    },
    {
      "epoch": 2.0981079187105816,
      "grad_norm": 0.11166302114725113,
      "learning_rate": 9.243243243243243e-05,
      "loss": 0.0104,
      "step": 14970
    },
    {
      "epoch": 2.098248072880168,
      "grad_norm": 0.2130666822195053,
      "learning_rate": 9.241808179861276e-05,
      "loss": 0.0302,
      "step": 14971
    },
    {
      "epoch": 2.098388227049755,
      "grad_norm": 0.22285212576389313,
      "learning_rate": 9.24037311647931e-05,
      "loss": 0.0069,
      "step": 14972
    },
    {
      "epoch": 2.0985283812193414,
      "grad_norm": 0.19388644397258759,
      "learning_rate": 9.238938053097344e-05,
      "loss": 0.0305,
      "step": 14973
    },
    {
      "epoch": 2.098668535388928,
      "grad_norm": 0.471657931804657,
      "learning_rate": 9.237502989715379e-05,
      "loss": 0.0308,
      "step": 14974
    },
    {
      "epoch": 2.0988086895585143,
      "grad_norm": 0.2812153398990631,
      "learning_rate": 9.236067926333413e-05,
      "loss": 0.0582,
      "step": 14975
    },
    {
      "epoch": 2.0989488437281008,
      "grad_norm": 0.37519508600234985,
      "learning_rate": 9.234632862951446e-05,
      "loss": 0.0257,
      "step": 14976
    },
    {
      "epoch": 2.0990889978976877,
      "grad_norm": 0.26630720496177673,
      "learning_rate": 9.23319779956948e-05,
      "loss": 0.1327,
      "step": 14977
    },
    {
      "epoch": 2.099229152067274,
      "grad_norm": 0.17649459838867188,
      "learning_rate": 9.231762736187514e-05,
      "loss": 0.0081,
      "step": 14978
    },
    {
      "epoch": 2.0993693062368606,
      "grad_norm": 0.1197456419467926,
      "learning_rate": 9.230327672805547e-05,
      "loss": 0.0103,
      "step": 14979
    },
    {
      "epoch": 2.099509460406447,
      "grad_norm": 0.3109102249145508,
      "learning_rate": 9.228892609423582e-05,
      "loss": 0.0443,
      "step": 14980
    },
    {
      "epoch": 2.0996496145760335,
      "grad_norm": 0.37403836846351624,
      "learning_rate": 9.227457546041617e-05,
      "loss": 0.037,
      "step": 14981
    },
    {
      "epoch": 2.0997897687456204,
      "grad_norm": 0.15053027868270874,
      "learning_rate": 9.22602248265965e-05,
      "loss": 0.0114,
      "step": 14982
    },
    {
      "epoch": 2.099929922915207,
      "grad_norm": 0.22350580990314484,
      "learning_rate": 9.224587419277684e-05,
      "loss": 0.0824,
      "step": 14983
    },
    {
      "epoch": 2.1000700770847933,
      "grad_norm": 0.23979859054088593,
      "learning_rate": 9.223152355895717e-05,
      "loss": 0.0443,
      "step": 14984
    },
    {
      "epoch": 2.10021023125438,
      "grad_norm": 0.06921961903572083,
      "learning_rate": 9.221717292513752e-05,
      "loss": 0.0047,
      "step": 14985
    },
    {
      "epoch": 2.1003503854239662,
      "grad_norm": 0.5286343693733215,
      "learning_rate": 9.220282229131786e-05,
      "loss": 0.0571,
      "step": 14986
    },
    {
      "epoch": 2.1004905395935527,
      "grad_norm": 0.43653401732444763,
      "learning_rate": 9.218847165749819e-05,
      "loss": 0.0439,
      "step": 14987
    },
    {
      "epoch": 2.1006306937631396,
      "grad_norm": 0.26151660084724426,
      "learning_rate": 9.217412102367853e-05,
      "loss": 0.0322,
      "step": 14988
    },
    {
      "epoch": 2.100770847932726,
      "grad_norm": 0.2586711049079895,
      "learning_rate": 9.215977038985888e-05,
      "loss": 0.0626,
      "step": 14989
    },
    {
      "epoch": 2.1009110021023125,
      "grad_norm": 0.18462680280208588,
      "learning_rate": 9.214541975603922e-05,
      "loss": 0.0431,
      "step": 14990
    },
    {
      "epoch": 2.101051156271899,
      "grad_norm": 0.19644387066364288,
      "learning_rate": 9.213106912221956e-05,
      "loss": 0.015,
      "step": 14991
    },
    {
      "epoch": 2.1011913104414854,
      "grad_norm": 0.16139784455299377,
      "learning_rate": 9.21167184883999e-05,
      "loss": 0.0326,
      "step": 14992
    },
    {
      "epoch": 2.1013314646110723,
      "grad_norm": 0.3132253885269165,
      "learning_rate": 9.210236785458023e-05,
      "loss": 0.0495,
      "step": 14993
    },
    {
      "epoch": 2.101471618780659,
      "grad_norm": 0.16853821277618408,
      "learning_rate": 9.208801722076057e-05,
      "loss": 0.0165,
      "step": 14994
    },
    {
      "epoch": 2.1016117729502453,
      "grad_norm": 0.1404504030942917,
      "learning_rate": 9.20736665869409e-05,
      "loss": 0.0313,
      "step": 14995
    },
    {
      "epoch": 2.1017519271198317,
      "grad_norm": 0.49739542603492737,
      "learning_rate": 9.205931595312126e-05,
      "loss": 0.0471,
      "step": 14996
    },
    {
      "epoch": 2.101892081289418,
      "grad_norm": 1.9007058143615723,
      "learning_rate": 9.20449653193016e-05,
      "loss": 0.0628,
      "step": 14997
    },
    {
      "epoch": 2.102032235459005,
      "grad_norm": 0.13976016640663147,
      "learning_rate": 9.203061468548193e-05,
      "loss": 0.0125,
      "step": 14998
    },
    {
      "epoch": 2.1021723896285915,
      "grad_norm": 0.1881859302520752,
      "learning_rate": 9.201626405166227e-05,
      "loss": 0.0152,
      "step": 14999
    },
    {
      "epoch": 2.102312543798178,
      "grad_norm": 0.3632969260215759,
      "learning_rate": 9.200191341784263e-05,
      "loss": 0.0191,
      "step": 15000
    },
    {
      "epoch": 2.1024526979677645,
      "grad_norm": 0.05354613810777664,
      "learning_rate": 9.198756278402296e-05,
      "loss": 0.0047,
      "step": 15001
    },
    {
      "epoch": 2.102592852137351,
      "grad_norm": 0.47835275530815125,
      "learning_rate": 9.19732121502033e-05,
      "loss": 0.0455,
      "step": 15002
    },
    {
      "epoch": 2.102733006306938,
      "grad_norm": 0.0926915854215622,
      "learning_rate": 9.195886151638363e-05,
      "loss": 0.0127,
      "step": 15003
    },
    {
      "epoch": 2.1028731604765243,
      "grad_norm": 0.3747750222682953,
      "learning_rate": 9.194451088256397e-05,
      "loss": 0.0142,
      "step": 15004
    },
    {
      "epoch": 2.1030133146461107,
      "grad_norm": 0.2441720962524414,
      "learning_rate": 9.193016024874431e-05,
      "loss": 0.0454,
      "step": 15005
    },
    {
      "epoch": 2.103153468815697,
      "grad_norm": 0.22082845866680145,
      "learning_rate": 9.191580961492465e-05,
      "loss": 0.0097,
      "step": 15006
    },
    {
      "epoch": 2.1032936229852837,
      "grad_norm": 0.23242922127246857,
      "learning_rate": 9.1901458981105e-05,
      "loss": 0.0357,
      "step": 15007
    },
    {
      "epoch": 2.1034337771548706,
      "grad_norm": 0.04633573070168495,
      "learning_rate": 9.188710834728534e-05,
      "loss": 0.0047,
      "step": 15008
    },
    {
      "epoch": 2.103573931324457,
      "grad_norm": 0.1291520744562149,
      "learning_rate": 9.187275771346566e-05,
      "loss": 0.0128,
      "step": 15009
    },
    {
      "epoch": 2.1037140854940435,
      "grad_norm": 0.34169691801071167,
      "learning_rate": 9.185840707964601e-05,
      "loss": 0.0442,
      "step": 15010
    },
    {
      "epoch": 2.10385423966363,
      "grad_norm": 0.1102093979716301,
      "learning_rate": 9.184405644582634e-05,
      "loss": 0.0195,
      "step": 15011
    },
    {
      "epoch": 2.1039943938332164,
      "grad_norm": 0.07528748363256454,
      "learning_rate": 9.182970581200669e-05,
      "loss": 0.0054,
      "step": 15012
    },
    {
      "epoch": 2.1041345480028033,
      "grad_norm": 0.28213754296302795,
      "learning_rate": 9.181535517818703e-05,
      "loss": 0.0441,
      "step": 15013
    },
    {
      "epoch": 2.1042747021723898,
      "grad_norm": 0.5728062987327576,
      "learning_rate": 9.180100454436736e-05,
      "loss": 0.0743,
      "step": 15014
    },
    {
      "epoch": 2.1044148563419762,
      "grad_norm": 0.12452620267868042,
      "learning_rate": 9.17866539105477e-05,
      "loss": 0.011,
      "step": 15015
    },
    {
      "epoch": 2.1045550105115627,
      "grad_norm": 0.1997743397951126,
      "learning_rate": 9.177230327672806e-05,
      "loss": 0.0293,
      "step": 15016
    },
    {
      "epoch": 2.104695164681149,
      "grad_norm": 0.06323877722024918,
      "learning_rate": 9.175795264290839e-05,
      "loss": 0.0062,
      "step": 15017
    },
    {
      "epoch": 2.1048353188507356,
      "grad_norm": 0.9490754008293152,
      "learning_rate": 9.174360200908873e-05,
      "loss": 0.1493,
      "step": 15018
    },
    {
      "epoch": 2.1049754730203225,
      "grad_norm": 0.09940283745527267,
      "learning_rate": 9.172925137526906e-05,
      "loss": 0.0032,
      "step": 15019
    },
    {
      "epoch": 2.105115627189909,
      "grad_norm": 0.7202023267745972,
      "learning_rate": 9.17149007414494e-05,
      "loss": 0.0641,
      "step": 15020
    },
    {
      "epoch": 2.1052557813594954,
      "grad_norm": 0.4888520836830139,
      "learning_rate": 9.170055010762976e-05,
      "loss": 0.0873,
      "step": 15021
    },
    {
      "epoch": 2.105395935529082,
      "grad_norm": 0.076435387134552,
      "learning_rate": 9.168619947381009e-05,
      "loss": 0.0105,
      "step": 15022
    },
    {
      "epoch": 2.1055360896986683,
      "grad_norm": 0.1431753784418106,
      "learning_rate": 9.167184883999043e-05,
      "loss": 0.0173,
      "step": 15023
    },
    {
      "epoch": 2.1056762438682552,
      "grad_norm": 0.3891960680484772,
      "learning_rate": 9.165749820617077e-05,
      "loss": 0.0149,
      "step": 15024
    },
    {
      "epoch": 2.1058163980378417,
      "grad_norm": 0.13248807191848755,
      "learning_rate": 9.16431475723511e-05,
      "loss": 0.0327,
      "step": 15025
    },
    {
      "epoch": 2.105956552207428,
      "grad_norm": 0.20844857394695282,
      "learning_rate": 9.162879693853144e-05,
      "loss": 0.0522,
      "step": 15026
    },
    {
      "epoch": 2.1060967063770146,
      "grad_norm": 0.14371460676193237,
      "learning_rate": 9.16144463047118e-05,
      "loss": 0.0128,
      "step": 15027
    },
    {
      "epoch": 2.106236860546601,
      "grad_norm": 0.5110690593719482,
      "learning_rate": 9.160009567089212e-05,
      "loss": 0.0422,
      "step": 15028
    },
    {
      "epoch": 2.106377014716188,
      "grad_norm": 0.5287502408027649,
      "learning_rate": 9.158574503707247e-05,
      "loss": 0.0326,
      "step": 15029
    },
    {
      "epoch": 2.1065171688857744,
      "grad_norm": 0.2070641815662384,
      "learning_rate": 9.15713944032528e-05,
      "loss": 0.0427,
      "step": 15030
    },
    {
      "epoch": 2.106657323055361,
      "grad_norm": 0.2262621521949768,
      "learning_rate": 9.155704376943314e-05,
      "loss": 0.0228,
      "step": 15031
    },
    {
      "epoch": 2.1067974772249474,
      "grad_norm": 0.3216087222099304,
      "learning_rate": 9.154269313561349e-05,
      "loss": 0.0424,
      "step": 15032
    },
    {
      "epoch": 2.106937631394534,
      "grad_norm": 0.33272409439086914,
      "learning_rate": 9.152834250179382e-05,
      "loss": 0.0527,
      "step": 15033
    },
    {
      "epoch": 2.1070777855641207,
      "grad_norm": 0.2307039499282837,
      "learning_rate": 9.151399186797416e-05,
      "loss": 0.035,
      "step": 15034
    },
    {
      "epoch": 2.107217939733707,
      "grad_norm": 0.23774205148220062,
      "learning_rate": 9.14996412341545e-05,
      "loss": 0.0088,
      "step": 15035
    },
    {
      "epoch": 2.1073580939032936,
      "grad_norm": 0.21409811079502106,
      "learning_rate": 9.148529060033483e-05,
      "loss": 0.0376,
      "step": 15036
    },
    {
      "epoch": 2.10749824807288,
      "grad_norm": 0.12241832911968231,
      "learning_rate": 9.147093996651519e-05,
      "loss": 0.0375,
      "step": 15037
    },
    {
      "epoch": 2.1076384022424666,
      "grad_norm": 0.1866355538368225,
      "learning_rate": 9.145658933269552e-05,
      "loss": 0.0221,
      "step": 15038
    },
    {
      "epoch": 2.1077785564120535,
      "grad_norm": 0.30886805057525635,
      "learning_rate": 9.144223869887586e-05,
      "loss": 0.0213,
      "step": 15039
    },
    {
      "epoch": 2.10791871058164,
      "grad_norm": 0.22603009641170502,
      "learning_rate": 9.14278880650562e-05,
      "loss": 0.0253,
      "step": 15040
    },
    {
      "epoch": 2.1080588647512264,
      "grad_norm": 0.23343555629253387,
      "learning_rate": 9.141353743123653e-05,
      "loss": 0.0256,
      "step": 15041
    },
    {
      "epoch": 2.108199018920813,
      "grad_norm": 0.2893286943435669,
      "learning_rate": 9.139918679741687e-05,
      "loss": 0.0338,
      "step": 15042
    },
    {
      "epoch": 2.1083391730903993,
      "grad_norm": 0.21442104876041412,
      "learning_rate": 9.138483616359723e-05,
      "loss": 0.0343,
      "step": 15043
    },
    {
      "epoch": 2.108479327259986,
      "grad_norm": 0.43006017804145813,
      "learning_rate": 9.137048552977756e-05,
      "loss": 0.0487,
      "step": 15044
    },
    {
      "epoch": 2.1086194814295727,
      "grad_norm": 0.18558984994888306,
      "learning_rate": 9.13561348959579e-05,
      "loss": 0.0116,
      "step": 15045
    },
    {
      "epoch": 2.108759635599159,
      "grad_norm": 0.3185671269893646,
      "learning_rate": 9.134178426213823e-05,
      "loss": 0.0109,
      "step": 15046
    },
    {
      "epoch": 2.1088997897687456,
      "grad_norm": 0.13591139018535614,
      "learning_rate": 9.132743362831857e-05,
      "loss": 0.0282,
      "step": 15047
    },
    {
      "epoch": 2.109039943938332,
      "grad_norm": 0.48032113909721375,
      "learning_rate": 9.131308299449893e-05,
      "loss": 0.062,
      "step": 15048
    },
    {
      "epoch": 2.1091800981079185,
      "grad_norm": 0.44369369745254517,
      "learning_rate": 9.129873236067925e-05,
      "loss": 0.0596,
      "step": 15049
    },
    {
      "epoch": 2.1093202522775054,
      "grad_norm": 0.20268294215202332,
      "learning_rate": 9.12843817268596e-05,
      "loss": 0.066,
      "step": 15050
    },
    {
      "epoch": 2.109460406447092,
      "grad_norm": 0.33319535851478577,
      "learning_rate": 9.127003109303994e-05,
      "loss": 0.0703,
      "step": 15051
    },
    {
      "epoch": 2.1096005606166783,
      "grad_norm": 0.0852564200758934,
      "learning_rate": 9.125568045922027e-05,
      "loss": 0.0221,
      "step": 15052
    },
    {
      "epoch": 2.109740714786265,
      "grad_norm": 0.7165330052375793,
      "learning_rate": 9.124132982540062e-05,
      "loss": 0.0214,
      "step": 15053
    },
    {
      "epoch": 2.1098808689558513,
      "grad_norm": 0.3211999833583832,
      "learning_rate": 9.122697919158095e-05,
      "loss": 0.0176,
      "step": 15054
    },
    {
      "epoch": 2.110021023125438,
      "grad_norm": 0.2102336287498474,
      "learning_rate": 9.12126285577613e-05,
      "loss": 0.0153,
      "step": 15055
    },
    {
      "epoch": 2.1101611772950246,
      "grad_norm": 0.28420835733413696,
      "learning_rate": 9.119827792394164e-05,
      "loss": 0.0416,
      "step": 15056
    },
    {
      "epoch": 2.110301331464611,
      "grad_norm": 0.2846992313861847,
      "learning_rate": 9.118392729012196e-05,
      "loss": 0.0673,
      "step": 15057
    },
    {
      "epoch": 2.1104414856341975,
      "grad_norm": 0.12391500174999237,
      "learning_rate": 9.11695766563023e-05,
      "loss": 0.009,
      "step": 15058
    },
    {
      "epoch": 2.110581639803784,
      "grad_norm": 0.16295009851455688,
      "learning_rate": 9.115522602248266e-05,
      "loss": 0.0362,
      "step": 15059
    },
    {
      "epoch": 2.110721793973371,
      "grad_norm": 0.32742732763290405,
      "learning_rate": 9.114087538866299e-05,
      "loss": 0.0157,
      "step": 15060
    },
    {
      "epoch": 2.1108619481429574,
      "grad_norm": 0.24940289556980133,
      "learning_rate": 9.112652475484333e-05,
      "loss": 0.0164,
      "step": 15061
    },
    {
      "epoch": 2.111002102312544,
      "grad_norm": 0.47123095393180847,
      "learning_rate": 9.111217412102367e-05,
      "loss": 0.0217,
      "step": 15062
    },
    {
      "epoch": 2.1111422564821303,
      "grad_norm": 0.0580204539000988,
      "learning_rate": 9.1097823487204e-05,
      "loss": 0.0057,
      "step": 15063
    },
    {
      "epoch": 2.1112824106517167,
      "grad_norm": 0.2690834403038025,
      "learning_rate": 9.108347285338436e-05,
      "loss": 0.0394,
      "step": 15064
    },
    {
      "epoch": 2.1114225648213036,
      "grad_norm": 0.43335792422294617,
      "learning_rate": 9.106912221956469e-05,
      "loss": 0.0839,
      "step": 15065
    },
    {
      "epoch": 2.11156271899089,
      "grad_norm": 0.4414302408695221,
      "learning_rate": 9.105477158574503e-05,
      "loss": 0.0187,
      "step": 15066
    },
    {
      "epoch": 2.1117028731604766,
      "grad_norm": 0.7357553839683533,
      "learning_rate": 9.104042095192537e-05,
      "loss": 0.0398,
      "step": 15067
    },
    {
      "epoch": 2.111843027330063,
      "grad_norm": 0.436136394739151,
      "learning_rate": 9.10260703181057e-05,
      "loss": 0.0594,
      "step": 15068
    },
    {
      "epoch": 2.1119831814996495,
      "grad_norm": 1.2286555767059326,
      "learning_rate": 9.101171968428606e-05,
      "loss": 0.0562,
      "step": 15069
    },
    {
      "epoch": 2.112123335669236,
      "grad_norm": 2.225698232650757,
      "learning_rate": 9.09973690504664e-05,
      "loss": 0.1456,
      "step": 15070
    },
    {
      "epoch": 2.112263489838823,
      "grad_norm": 0.4084653854370117,
      "learning_rate": 9.098301841664673e-05,
      "loss": 0.0312,
      "step": 15071
    },
    {
      "epoch": 2.1124036440084093,
      "grad_norm": 0.09670858085155487,
      "learning_rate": 9.096866778282707e-05,
      "loss": 0.0166,
      "step": 15072
    },
    {
      "epoch": 2.1125437981779958,
      "grad_norm": 0.30542707443237305,
      "learning_rate": 9.09543171490074e-05,
      "loss": 0.0362,
      "step": 15073
    },
    {
      "epoch": 2.112683952347582,
      "grad_norm": 0.1805487424135208,
      "learning_rate": 9.093996651518774e-05,
      "loss": 0.0277,
      "step": 15074
    },
    {
      "epoch": 2.1128241065171687,
      "grad_norm": 0.1850481629371643,
      "learning_rate": 9.09256158813681e-05,
      "loss": 0.0099,
      "step": 15075
    },
    {
      "epoch": 2.1129642606867556,
      "grad_norm": 0.1688794046640396,
      "learning_rate": 9.091126524754842e-05,
      "loss": 0.0567,
      "step": 15076
    },
    {
      "epoch": 2.113104414856342,
      "grad_norm": 0.09091239422559738,
      "learning_rate": 9.089691461372877e-05,
      "loss": 0.0226,
      "step": 15077
    },
    {
      "epoch": 2.1132445690259285,
      "grad_norm": 0.22825609147548676,
      "learning_rate": 9.088256397990911e-05,
      "loss": 0.0372,
      "step": 15078
    },
    {
      "epoch": 2.113384723195515,
      "grad_norm": 0.6003784537315369,
      "learning_rate": 9.086821334608944e-05,
      "loss": 0.0778,
      "step": 15079
    },
    {
      "epoch": 2.1135248773651014,
      "grad_norm": 0.18894918262958527,
      "learning_rate": 9.085386271226979e-05,
      "loss": 0.0235,
      "step": 15080
    },
    {
      "epoch": 2.1136650315346883,
      "grad_norm": 0.5138563513755798,
      "learning_rate": 9.083951207845012e-05,
      "loss": 0.0473,
      "step": 15081
    },
    {
      "epoch": 2.113805185704275,
      "grad_norm": 0.21134959161281586,
      "learning_rate": 9.082516144463046e-05,
      "loss": 0.041,
      "step": 15082
    },
    {
      "epoch": 2.1139453398738612,
      "grad_norm": 0.20066098868846893,
      "learning_rate": 9.08108108108108e-05,
      "loss": 0.0197,
      "step": 15083
    },
    {
      "epoch": 2.1140854940434477,
      "grad_norm": 0.5675198435783386,
      "learning_rate": 9.079646017699113e-05,
      "loss": 0.0218,
      "step": 15084
    },
    {
      "epoch": 2.114225648213034,
      "grad_norm": 0.20292043685913086,
      "learning_rate": 9.078210954317149e-05,
      "loss": 0.0226,
      "step": 15085
    },
    {
      "epoch": 2.114365802382621,
      "grad_norm": 0.14065830409526825,
      "learning_rate": 9.076775890935183e-05,
      "loss": 0.0261,
      "step": 15086
    },
    {
      "epoch": 2.1145059565522075,
      "grad_norm": 0.21567001938819885,
      "learning_rate": 9.075340827553216e-05,
      "loss": 0.0085,
      "step": 15087
    },
    {
      "epoch": 2.114646110721794,
      "grad_norm": 0.28315526247024536,
      "learning_rate": 9.07390576417125e-05,
      "loss": 0.071,
      "step": 15088
    },
    {
      "epoch": 2.1147862648913804,
      "grad_norm": 0.3099398612976074,
      "learning_rate": 9.072470700789283e-05,
      "loss": 0.0318,
      "step": 15089
    },
    {
      "epoch": 2.114926419060967,
      "grad_norm": 0.036153245717287064,
      "learning_rate": 9.071035637407317e-05,
      "loss": 0.0038,
      "step": 15090
    },
    {
      "epoch": 2.115066573230554,
      "grad_norm": 0.23041552305221558,
      "learning_rate": 9.069600574025353e-05,
      "loss": 0.0487,
      "step": 15091
    },
    {
      "epoch": 2.1152067274001403,
      "grad_norm": 0.5008238554000854,
      "learning_rate": 9.068165510643386e-05,
      "loss": 0.0296,
      "step": 15092
    },
    {
      "epoch": 2.1153468815697267,
      "grad_norm": 0.04281735047698021,
      "learning_rate": 9.06673044726142e-05,
      "loss": 0.0046,
      "step": 15093
    },
    {
      "epoch": 2.115487035739313,
      "grad_norm": 0.4874298572540283,
      "learning_rate": 9.065295383879454e-05,
      "loss": 0.0169,
      "step": 15094
    },
    {
      "epoch": 2.1156271899088996,
      "grad_norm": 0.1225438192486763,
      "learning_rate": 9.063860320497487e-05,
      "loss": 0.0048,
      "step": 15095
    },
    {
      "epoch": 2.1157673440784865,
      "grad_norm": 0.16113333404064178,
      "learning_rate": 9.062425257115523e-05,
      "loss": 0.0195,
      "step": 15096
    },
    {
      "epoch": 2.115907498248073,
      "grad_norm": 0.5520235896110535,
      "learning_rate": 9.060990193733555e-05,
      "loss": 0.0225,
      "step": 15097
    },
    {
      "epoch": 2.1160476524176595,
      "grad_norm": 0.9119699001312256,
      "learning_rate": 9.05955513035159e-05,
      "loss": 0.0232,
      "step": 15098
    },
    {
      "epoch": 2.116187806587246,
      "grad_norm": 0.2690444588661194,
      "learning_rate": 9.058120066969624e-05,
      "loss": 0.065,
      "step": 15099
    },
    {
      "epoch": 2.1163279607568324,
      "grad_norm": 0.5533698201179504,
      "learning_rate": 9.056685003587657e-05,
      "loss": 0.0323,
      "step": 15100
    },
    {
      "epoch": 2.116468114926419,
      "grad_norm": 0.49332866072654724,
      "learning_rate": 9.055249940205692e-05,
      "loss": 0.039,
      "step": 15101
    },
    {
      "epoch": 2.1166082690960057,
      "grad_norm": 0.0938776284456253,
      "learning_rate": 9.053814876823726e-05,
      "loss": 0.0111,
      "step": 15102
    },
    {
      "epoch": 2.116748423265592,
      "grad_norm": 0.4235430359840393,
      "learning_rate": 9.052379813441759e-05,
      "loss": 0.0745,
      "step": 15103
    },
    {
      "epoch": 2.1168885774351787,
      "grad_norm": 0.09138573706150055,
      "learning_rate": 9.050944750059794e-05,
      "loss": 0.0091,
      "step": 15104
    },
    {
      "epoch": 2.117028731604765,
      "grad_norm": 0.4912731945514679,
      "learning_rate": 9.049509686677828e-05,
      "loss": 0.0201,
      "step": 15105
    },
    {
      "epoch": 2.1171688857743516,
      "grad_norm": 0.3563465476036072,
      "learning_rate": 9.04807462329586e-05,
      "loss": 0.0196,
      "step": 15106
    },
    {
      "epoch": 2.1173090399439385,
      "grad_norm": 0.053236689418554306,
      "learning_rate": 9.046639559913896e-05,
      "loss": 0.0028,
      "step": 15107
    },
    {
      "epoch": 2.117449194113525,
      "grad_norm": 0.37504300475120544,
      "learning_rate": 9.045204496531929e-05,
      "loss": 0.0361,
      "step": 15108
    },
    {
      "epoch": 2.1175893482831114,
      "grad_norm": 0.14413096010684967,
      "learning_rate": 9.043769433149963e-05,
      "loss": 0.0082,
      "step": 15109
    },
    {
      "epoch": 2.117729502452698,
      "grad_norm": 0.2738886773586273,
      "learning_rate": 9.042334369767997e-05,
      "loss": 0.035,
      "step": 15110
    },
    {
      "epoch": 2.1178696566222843,
      "grad_norm": 0.2167007476091385,
      "learning_rate": 9.04089930638603e-05,
      "loss": 0.0231,
      "step": 15111
    },
    {
      "epoch": 2.1180098107918712,
      "grad_norm": 0.117019884288311,
      "learning_rate": 9.039464243004066e-05,
      "loss": 0.0117,
      "step": 15112
    },
    {
      "epoch": 2.1181499649614577,
      "grad_norm": 0.1578192412853241,
      "learning_rate": 9.0380291796221e-05,
      "loss": 0.009,
      "step": 15113
    },
    {
      "epoch": 2.118290119131044,
      "grad_norm": 0.25520068407058716,
      "learning_rate": 9.036594116240133e-05,
      "loss": 0.0287,
      "step": 15114
    },
    {
      "epoch": 2.1184302733006306,
      "grad_norm": 0.8332095146179199,
      "learning_rate": 9.035159052858167e-05,
      "loss": 0.0544,
      "step": 15115
    },
    {
      "epoch": 2.118570427470217,
      "grad_norm": 0.9073244333267212,
      "learning_rate": 9.0337239894762e-05,
      "loss": 0.0597,
      "step": 15116
    },
    {
      "epoch": 2.118710581639804,
      "grad_norm": 1.3589686155319214,
      "learning_rate": 9.032288926094236e-05,
      "loss": 0.0506,
      "step": 15117
    },
    {
      "epoch": 2.1188507358093904,
      "grad_norm": 0.18314681947231293,
      "learning_rate": 9.03085386271227e-05,
      "loss": 0.0026,
      "step": 15118
    },
    {
      "epoch": 2.118990889978977,
      "grad_norm": 1.2299740314483643,
      "learning_rate": 9.029418799330303e-05,
      "loss": 0.0446,
      "step": 15119
    },
    {
      "epoch": 2.1191310441485633,
      "grad_norm": 3.954184055328369,
      "learning_rate": 9.027983735948337e-05,
      "loss": 0.15,
      "step": 15120
    },
    {
      "epoch": 2.11927119831815,
      "grad_norm": 0.33408287167549133,
      "learning_rate": 9.026548672566371e-05,
      "loss": 0.0796,
      "step": 15121
    },
    {
      "epoch": 2.1194113524877367,
      "grad_norm": 0.23975008726119995,
      "learning_rate": 9.025113609184404e-05,
      "loss": 0.034,
      "step": 15122
    },
    {
      "epoch": 2.119551506657323,
      "grad_norm": 0.24978220462799072,
      "learning_rate": 9.02367854580244e-05,
      "loss": 0.0057,
      "step": 15123
    },
    {
      "epoch": 2.1196916608269096,
      "grad_norm": 0.6574159264564514,
      "learning_rate": 9.022243482420472e-05,
      "loss": 0.0353,
      "step": 15124
    },
    {
      "epoch": 2.119831814996496,
      "grad_norm": 0.1471230834722519,
      "learning_rate": 9.020808419038507e-05,
      "loss": 0.0215,
      "step": 15125
    },
    {
      "epoch": 2.1199719691660825,
      "grad_norm": 0.1987726241350174,
      "learning_rate": 9.019373355656541e-05,
      "loss": 0.0405,
      "step": 15126
    },
    {
      "epoch": 2.1201121233356695,
      "grad_norm": 0.3172414004802704,
      "learning_rate": 9.017938292274574e-05,
      "loss": 0.0706,
      "step": 15127
    },
    {
      "epoch": 2.120252277505256,
      "grad_norm": 0.6078309416770935,
      "learning_rate": 9.016503228892609e-05,
      "loss": 0.0444,
      "step": 15128
    },
    {
      "epoch": 2.1203924316748424,
      "grad_norm": 0.11829888820648193,
      "learning_rate": 9.015068165510643e-05,
      "loss": 0.0146,
      "step": 15129
    },
    {
      "epoch": 2.120532585844429,
      "grad_norm": 0.11326898634433746,
      "learning_rate": 9.013633102128676e-05,
      "loss": 0.0081,
      "step": 15130
    },
    {
      "epoch": 2.1206727400140153,
      "grad_norm": 0.18263469636440277,
      "learning_rate": 9.01219803874671e-05,
      "loss": 0.0124,
      "step": 15131
    },
    {
      "epoch": 2.1208128941836017,
      "grad_norm": 0.21754388511180878,
      "learning_rate": 9.010762975364743e-05,
      "loss": 0.0731,
      "step": 15132
    },
    {
      "epoch": 2.1209530483531887,
      "grad_norm": 0.21932834386825562,
      "learning_rate": 9.009327911982779e-05,
      "loss": 0.0151,
      "step": 15133
    },
    {
      "epoch": 2.121093202522775,
      "grad_norm": 0.5655601620674133,
      "learning_rate": 9.007892848600813e-05,
      "loss": 0.0856,
      "step": 15134
    },
    {
      "epoch": 2.1212333566923616,
      "grad_norm": 0.1506335437297821,
      "learning_rate": 9.006457785218846e-05,
      "loss": 0.0145,
      "step": 15135
    },
    {
      "epoch": 2.121373510861948,
      "grad_norm": 0.38737690448760986,
      "learning_rate": 9.00502272183688e-05,
      "loss": 0.0291,
      "step": 15136
    },
    {
      "epoch": 2.1215136650315345,
      "grad_norm": 0.05948127061128616,
      "learning_rate": 9.003587658454914e-05,
      "loss": 0.0034,
      "step": 15137
    },
    {
      "epoch": 2.1216538192011214,
      "grad_norm": 0.4119667112827301,
      "learning_rate": 9.002152595072947e-05,
      "loss": 0.0484,
      "step": 15138
    },
    {
      "epoch": 2.121793973370708,
      "grad_norm": 0.08403030782938004,
      "learning_rate": 9.000717531690983e-05,
      "loss": 0.0144,
      "step": 15139
    },
    {
      "epoch": 2.1219341275402943,
      "grad_norm": 0.29257088899612427,
      "learning_rate": 8.999282468309017e-05,
      "loss": 0.0643,
      "step": 15140
    },
    {
      "epoch": 2.1220742817098808,
      "grad_norm": 0.40619707107543945,
      "learning_rate": 8.99784740492705e-05,
      "loss": 0.0295,
      "step": 15141
    },
    {
      "epoch": 2.1222144358794672,
      "grad_norm": 0.1887200027704239,
      "learning_rate": 8.996412341545084e-05,
      "loss": 0.0379,
      "step": 15142
    },
    {
      "epoch": 2.122354590049054,
      "grad_norm": 0.7990462779998779,
      "learning_rate": 8.994977278163117e-05,
      "loss": 0.0617,
      "step": 15143
    },
    {
      "epoch": 2.1224947442186406,
      "grad_norm": 0.15633563697338104,
      "learning_rate": 8.993542214781152e-05,
      "loss": 0.0071,
      "step": 15144
    },
    {
      "epoch": 2.122634898388227,
      "grad_norm": 0.34932953119277954,
      "learning_rate": 8.992107151399187e-05,
      "loss": 0.0391,
      "step": 15145
    },
    {
      "epoch": 2.1227750525578135,
      "grad_norm": 0.3217383027076721,
      "learning_rate": 8.99067208801722e-05,
      "loss": 0.0109,
      "step": 15146
    },
    {
      "epoch": 2.1229152067274,
      "grad_norm": 0.07806293666362762,
      "learning_rate": 8.989237024635254e-05,
      "loss": 0.045,
      "step": 15147
    },
    {
      "epoch": 2.123055360896987,
      "grad_norm": 0.5690058469772339,
      "learning_rate": 8.98780196125329e-05,
      "loss": 0.0592,
      "step": 15148
    },
    {
      "epoch": 2.1231955150665733,
      "grad_norm": 0.2966855466365814,
      "learning_rate": 8.986366897871322e-05,
      "loss": 0.0454,
      "step": 15149
    },
    {
      "epoch": 2.12333566923616,
      "grad_norm": 0.11907083541154861,
      "learning_rate": 8.984931834489356e-05,
      "loss": 0.0096,
      "step": 15150
    },
    {
      "epoch": 2.1234758234057463,
      "grad_norm": 0.2653689682483673,
      "learning_rate": 8.983496771107389e-05,
      "loss": 0.0471,
      "step": 15151
    },
    {
      "epoch": 2.1236159775753327,
      "grad_norm": 0.10507310926914215,
      "learning_rate": 8.982061707725423e-05,
      "loss": 0.0079,
      "step": 15152
    },
    {
      "epoch": 2.1237561317449196,
      "grad_norm": 0.27919918298721313,
      "learning_rate": 8.980626644343458e-05,
      "loss": 0.0534,
      "step": 15153
    },
    {
      "epoch": 2.123896285914506,
      "grad_norm": 0.8085997700691223,
      "learning_rate": 8.97919158096149e-05,
      "loss": 0.0247,
      "step": 15154
    },
    {
      "epoch": 2.1240364400840925,
      "grad_norm": 0.15061944723129272,
      "learning_rate": 8.977756517579526e-05,
      "loss": 0.0149,
      "step": 15155
    },
    {
      "epoch": 2.124176594253679,
      "grad_norm": 0.16149917244911194,
      "learning_rate": 8.97632145419756e-05,
      "loss": 0.0103,
      "step": 15156
    },
    {
      "epoch": 2.1243167484232655,
      "grad_norm": 0.5081571340560913,
      "learning_rate": 8.974886390815593e-05,
      "loss": 0.0163,
      "step": 15157
    },
    {
      "epoch": 2.1244569025928524,
      "grad_norm": 0.11678512394428253,
      "learning_rate": 8.973451327433627e-05,
      "loss": 0.0106,
      "step": 15158
    },
    {
      "epoch": 2.124597056762439,
      "grad_norm": 1.1179828643798828,
      "learning_rate": 8.97201626405166e-05,
      "loss": 0.0461,
      "step": 15159
    },
    {
      "epoch": 2.1247372109320253,
      "grad_norm": 0.20063628256320953,
      "learning_rate": 8.970581200669696e-05,
      "loss": 0.0157,
      "step": 15160
    },
    {
      "epoch": 2.1248773651016117,
      "grad_norm": 0.16870984435081482,
      "learning_rate": 8.96914613728773e-05,
      "loss": 0.0277,
      "step": 15161
    },
    {
      "epoch": 2.125017519271198,
      "grad_norm": 0.3032510280609131,
      "learning_rate": 8.967711073905763e-05,
      "loss": 0.0578,
      "step": 15162
    },
    {
      "epoch": 2.1251576734407847,
      "grad_norm": 0.10628536343574524,
      "learning_rate": 8.966276010523797e-05,
      "loss": 0.0089,
      "step": 15163
    },
    {
      "epoch": 2.1252978276103716,
      "grad_norm": 0.08259095251560211,
      "learning_rate": 8.964840947141833e-05,
      "loss": 0.0089,
      "step": 15164
    },
    {
      "epoch": 2.125437981779958,
      "grad_norm": 1.2813059091567993,
      "learning_rate": 8.963405883759866e-05,
      "loss": 0.0433,
      "step": 15165
    },
    {
      "epoch": 2.1255781359495445,
      "grad_norm": 0.190083846449852,
      "learning_rate": 8.9619708203779e-05,
      "loss": 0.029,
      "step": 15166
    },
    {
      "epoch": 2.125718290119131,
      "grad_norm": 1.1281046867370605,
      "learning_rate": 8.960535756995933e-05,
      "loss": 0.1216,
      "step": 15167
    },
    {
      "epoch": 2.1258584442887174,
      "grad_norm": 0.8072153925895691,
      "learning_rate": 8.959100693613967e-05,
      "loss": 0.112,
      "step": 15168
    },
    {
      "epoch": 2.1259985984583043,
      "grad_norm": 0.4067913591861725,
      "learning_rate": 8.957665630232001e-05,
      "loss": 0.1216,
      "step": 15169
    },
    {
      "epoch": 2.1261387526278908,
      "grad_norm": 0.6728463768959045,
      "learning_rate": 8.956230566850034e-05,
      "loss": 0.0463,
      "step": 15170
    },
    {
      "epoch": 2.126278906797477,
      "grad_norm": 0.2107124924659729,
      "learning_rate": 8.95479550346807e-05,
      "loss": 0.0496,
      "step": 15171
    },
    {
      "epoch": 2.1264190609670637,
      "grad_norm": 0.22387650609016418,
      "learning_rate": 8.953360440086104e-05,
      "loss": 0.0189,
      "step": 15172
    },
    {
      "epoch": 2.12655921513665,
      "grad_norm": 0.3885575830936432,
      "learning_rate": 8.951925376704136e-05,
      "loss": 0.0483,
      "step": 15173
    },
    {
      "epoch": 2.126699369306237,
      "grad_norm": 0.3481922447681427,
      "learning_rate": 8.950490313322171e-05,
      "loss": 0.0536,
      "step": 15174
    },
    {
      "epoch": 2.1268395234758235,
      "grad_norm": 0.347377210855484,
      "learning_rate": 8.949055249940206e-05,
      "loss": 0.0228,
      "step": 15175
    },
    {
      "epoch": 2.12697967764541,
      "grad_norm": 0.15950915217399597,
      "learning_rate": 8.947620186558239e-05,
      "loss": 0.0302,
      "step": 15176
    },
    {
      "epoch": 2.1271198318149964,
      "grad_norm": 0.19280372560024261,
      "learning_rate": 8.946185123176273e-05,
      "loss": 0.0274,
      "step": 15177
    },
    {
      "epoch": 2.127259985984583,
      "grad_norm": 0.19810117781162262,
      "learning_rate": 8.944750059794306e-05,
      "loss": 0.0305,
      "step": 15178
    },
    {
      "epoch": 2.12740014015417,
      "grad_norm": 0.24144794046878815,
      "learning_rate": 8.94331499641234e-05,
      "loss": 0.0564,
      "step": 15179
    },
    {
      "epoch": 2.1275402943237562,
      "grad_norm": 0.48928698897361755,
      "learning_rate": 8.941879933030376e-05,
      "loss": 0.0456,
      "step": 15180
    },
    {
      "epoch": 2.1276804484933427,
      "grad_norm": 0.19243639707565308,
      "learning_rate": 8.940444869648409e-05,
      "loss": 0.0239,
      "step": 15181
    },
    {
      "epoch": 2.127820602662929,
      "grad_norm": 0.2557934820652008,
      "learning_rate": 8.939009806266443e-05,
      "loss": 0.0271,
      "step": 15182
    },
    {
      "epoch": 2.1279607568325156,
      "grad_norm": 0.2183423936367035,
      "learning_rate": 8.937574742884477e-05,
      "loss": 0.0374,
      "step": 15183
    },
    {
      "epoch": 2.128100911002102,
      "grad_norm": 0.20925457775592804,
      "learning_rate": 8.93613967950251e-05,
      "loss": 0.0196,
      "step": 15184
    },
    {
      "epoch": 2.128241065171689,
      "grad_norm": 0.25236693024635315,
      "learning_rate": 8.934704616120544e-05,
      "loss": 0.0328,
      "step": 15185
    },
    {
      "epoch": 2.1283812193412754,
      "grad_norm": 0.18731079995632172,
      "learning_rate": 8.933269552738577e-05,
      "loss": 0.0138,
      "step": 15186
    },
    {
      "epoch": 2.128521373510862,
      "grad_norm": 0.16091123223304749,
      "learning_rate": 8.931834489356613e-05,
      "loss": 0.0096,
      "step": 15187
    },
    {
      "epoch": 2.1286615276804484,
      "grad_norm": 0.26221275329589844,
      "learning_rate": 8.930399425974647e-05,
      "loss": 0.0563,
      "step": 15188
    },
    {
      "epoch": 2.1288016818500353,
      "grad_norm": 0.09023147076368332,
      "learning_rate": 8.92896436259268e-05,
      "loss": 0.016,
      "step": 15189
    },
    {
      "epoch": 2.1289418360196217,
      "grad_norm": 0.38218751549720764,
      "learning_rate": 8.927529299210714e-05,
      "loss": 0.0449,
      "step": 15190
    },
    {
      "epoch": 2.129081990189208,
      "grad_norm": 0.34352391958236694,
      "learning_rate": 8.92609423582875e-05,
      "loss": 0.0313,
      "step": 15191
    },
    {
      "epoch": 2.1292221443587946,
      "grad_norm": 0.21111008524894714,
      "learning_rate": 8.924659172446782e-05,
      "loss": 0.0317,
      "step": 15192
    },
    {
      "epoch": 2.129362298528381,
      "grad_norm": 0.2605690360069275,
      "learning_rate": 8.923224109064817e-05,
      "loss": 0.0435,
      "step": 15193
    },
    {
      "epoch": 2.1295024526979676,
      "grad_norm": 0.5971900224685669,
      "learning_rate": 8.92178904568285e-05,
      "loss": 0.0203,
      "step": 15194
    },
    {
      "epoch": 2.1296426068675545,
      "grad_norm": 0.2440394163131714,
      "learning_rate": 8.920353982300884e-05,
      "loss": 0.0185,
      "step": 15195
    },
    {
      "epoch": 2.129782761037141,
      "grad_norm": 0.24238969385623932,
      "learning_rate": 8.918918918918919e-05,
      "loss": 0.0547,
      "step": 15196
    },
    {
      "epoch": 2.1299229152067274,
      "grad_norm": 0.1616920679807663,
      "learning_rate": 8.917483855536952e-05,
      "loss": 0.0152,
      "step": 15197
    },
    {
      "epoch": 2.130063069376314,
      "grad_norm": 0.3849319517612457,
      "learning_rate": 8.916048792154986e-05,
      "loss": 0.0604,
      "step": 15198
    },
    {
      "epoch": 2.1302032235459003,
      "grad_norm": 0.29612505435943604,
      "learning_rate": 8.91461372877302e-05,
      "loss": 0.0125,
      "step": 15199
    },
    {
      "epoch": 2.130343377715487,
      "grad_norm": 0.16791734099388123,
      "learning_rate": 8.913178665391053e-05,
      "loss": 0.0335,
      "step": 15200
    },
    {
      "epoch": 2.1304835318850737,
      "grad_norm": 0.06513180583715439,
      "learning_rate": 8.911743602009088e-05,
      "loss": 0.0042,
      "step": 15201
    },
    {
      "epoch": 2.13062368605466,
      "grad_norm": 0.19055478274822235,
      "learning_rate": 8.91030853862712e-05,
      "loss": 0.0128,
      "step": 15202
    },
    {
      "epoch": 2.1307638402242466,
      "grad_norm": 0.23441746830940247,
      "learning_rate": 8.908873475245156e-05,
      "loss": 0.0171,
      "step": 15203
    },
    {
      "epoch": 2.130903994393833,
      "grad_norm": 0.1360238939523697,
      "learning_rate": 8.90743841186319e-05,
      "loss": 0.0111,
      "step": 15204
    },
    {
      "epoch": 2.13104414856342,
      "grad_norm": 0.43195340037345886,
      "learning_rate": 8.906003348481223e-05,
      "loss": 0.0571,
      "step": 15205
    },
    {
      "epoch": 2.1311843027330064,
      "grad_norm": 0.05040917918086052,
      "learning_rate": 8.904568285099257e-05,
      "loss": 0.0034,
      "step": 15206
    },
    {
      "epoch": 2.131324456902593,
      "grad_norm": 0.18566788733005524,
      "learning_rate": 8.903133221717293e-05,
      "loss": 0.0085,
      "step": 15207
    },
    {
      "epoch": 2.1314646110721793,
      "grad_norm": 0.4247599244117737,
      "learning_rate": 8.901698158335326e-05,
      "loss": 0.0326,
      "step": 15208
    },
    {
      "epoch": 2.131604765241766,
      "grad_norm": 0.44142937660217285,
      "learning_rate": 8.90026309495336e-05,
      "loss": 0.0197,
      "step": 15209
    },
    {
      "epoch": 2.1317449194113527,
      "grad_norm": 0.27340027689933777,
      "learning_rate": 8.898828031571393e-05,
      "loss": 0.0215,
      "step": 15210
    },
    {
      "epoch": 2.131885073580939,
      "grad_norm": 0.13507801294326782,
      "learning_rate": 8.897392968189427e-05,
      "loss": 0.0084,
      "step": 15211
    },
    {
      "epoch": 2.1320252277505256,
      "grad_norm": 0.4074319303035736,
      "learning_rate": 8.895957904807463e-05,
      "loss": 0.0418,
      "step": 15212
    },
    {
      "epoch": 2.132165381920112,
      "grad_norm": 0.3392559885978699,
      "learning_rate": 8.894522841425495e-05,
      "loss": 0.0343,
      "step": 15213
    },
    {
      "epoch": 2.1323055360896985,
      "grad_norm": 0.19033478200435638,
      "learning_rate": 8.89308777804353e-05,
      "loss": 0.0117,
      "step": 15214
    },
    {
      "epoch": 2.132445690259285,
      "grad_norm": 0.4595630466938019,
      "learning_rate": 8.891652714661564e-05,
      "loss": 0.0301,
      "step": 15215
    },
    {
      "epoch": 2.132585844428872,
      "grad_norm": 0.04427970573306084,
      "learning_rate": 8.890217651279597e-05,
      "loss": 0.0017,
      "step": 15216
    },
    {
      "epoch": 2.1327259985984584,
      "grad_norm": 0.5045440196990967,
      "learning_rate": 8.888782587897631e-05,
      "loss": 0.0104,
      "step": 15217
    },
    {
      "epoch": 2.132866152768045,
      "grad_norm": 0.4157669246196747,
      "learning_rate": 8.887347524515666e-05,
      "loss": 0.0553,
      "step": 15218
    },
    {
      "epoch": 2.1330063069376313,
      "grad_norm": 2.893181800842285,
      "learning_rate": 8.8859124611337e-05,
      "loss": 0.1899,
      "step": 15219
    },
    {
      "epoch": 2.133146461107218,
      "grad_norm": 1.4312331676483154,
      "learning_rate": 8.884477397751734e-05,
      "loss": 0.0328,
      "step": 15220
    },
    {
      "epoch": 2.1332866152768046,
      "grad_norm": 0.40497875213623047,
      "learning_rate": 8.883042334369766e-05,
      "loss": 0.0109,
      "step": 15221
    },
    {
      "epoch": 2.133426769446391,
      "grad_norm": 0.3215682804584503,
      "learning_rate": 8.8816072709878e-05,
      "loss": 0.0271,
      "step": 15222
    },
    {
      "epoch": 2.1335669236159776,
      "grad_norm": 0.2847515046596527,
      "learning_rate": 8.880172207605836e-05,
      "loss": 0.0351,
      "step": 15223
    },
    {
      "epoch": 2.133707077785564,
      "grad_norm": 0.11130450665950775,
      "learning_rate": 8.878737144223869e-05,
      "loss": 0.0094,
      "step": 15224
    },
    {
      "epoch": 2.1338472319551505,
      "grad_norm": 0.3143520653247833,
      "learning_rate": 8.877302080841903e-05,
      "loss": 0.0302,
      "step": 15225
    },
    {
      "epoch": 2.1339873861247374,
      "grad_norm": 0.9587177038192749,
      "learning_rate": 8.875867017459937e-05,
      "loss": 0.1074,
      "step": 15226
    },
    {
      "epoch": 2.134127540294324,
      "grad_norm": 0.3057488799095154,
      "learning_rate": 8.87443195407797e-05,
      "loss": 0.0134,
      "step": 15227
    },
    {
      "epoch": 2.1342676944639103,
      "grad_norm": 0.11165688931941986,
      "learning_rate": 8.872996890696006e-05,
      "loss": 0.0053,
      "step": 15228
    },
    {
      "epoch": 2.1344078486334968,
      "grad_norm": 0.29063186049461365,
      "learning_rate": 8.871561827314039e-05,
      "loss": 0.0579,
      "step": 15229
    },
    {
      "epoch": 2.134548002803083,
      "grad_norm": 0.21145744621753693,
      "learning_rate": 8.870126763932073e-05,
      "loss": 0.0217,
      "step": 15230
    },
    {
      "epoch": 2.13468815697267,
      "grad_norm": 0.1780807226896286,
      "learning_rate": 8.868691700550107e-05,
      "loss": 0.0136,
      "step": 15231
    },
    {
      "epoch": 2.1348283111422566,
      "grad_norm": 0.22877776622772217,
      "learning_rate": 8.86725663716814e-05,
      "loss": 0.0293,
      "step": 15232
    },
    {
      "epoch": 2.134968465311843,
      "grad_norm": 0.2548004686832428,
      "learning_rate": 8.865821573786174e-05,
      "loss": 0.0552,
      "step": 15233
    },
    {
      "epoch": 2.1351086194814295,
      "grad_norm": 0.10363692790269852,
      "learning_rate": 8.86438651040421e-05,
      "loss": 0.0107,
      "step": 15234
    },
    {
      "epoch": 2.135248773651016,
      "grad_norm": 0.1902381181716919,
      "learning_rate": 8.862951447022243e-05,
      "loss": 0.0241,
      "step": 15235
    },
    {
      "epoch": 2.135388927820603,
      "grad_norm": 0.1303856521844864,
      "learning_rate": 8.861516383640277e-05,
      "loss": 0.0335,
      "step": 15236
    },
    {
      "epoch": 2.1355290819901893,
      "grad_norm": 0.157906174659729,
      "learning_rate": 8.86008132025831e-05,
      "loss": 0.0157,
      "step": 15237
    },
    {
      "epoch": 2.1356692361597758,
      "grad_norm": 0.03608357161283493,
      "learning_rate": 8.858646256876344e-05,
      "loss": 0.0047,
      "step": 15238
    },
    {
      "epoch": 2.1358093903293622,
      "grad_norm": 0.16708041727542877,
      "learning_rate": 8.85721119349438e-05,
      "loss": 0.0197,
      "step": 15239
    },
    {
      "epoch": 2.1359495444989487,
      "grad_norm": 0.47102996706962585,
      "learning_rate": 8.855776130112412e-05,
      "loss": 0.04,
      "step": 15240
    },
    {
      "epoch": 2.1360896986685356,
      "grad_norm": 0.2384759485721588,
      "learning_rate": 8.854341066730447e-05,
      "loss": 0.0538,
      "step": 15241
    },
    {
      "epoch": 2.136229852838122,
      "grad_norm": 0.06157780811190605,
      "learning_rate": 8.852906003348481e-05,
      "loss": 0.0058,
      "step": 15242
    },
    {
      "epoch": 2.1363700070077085,
      "grad_norm": 0.48096176981925964,
      "learning_rate": 8.851470939966514e-05,
      "loss": 0.0614,
      "step": 15243
    },
    {
      "epoch": 2.136510161177295,
      "grad_norm": 0.26099517941474915,
      "learning_rate": 8.850035876584549e-05,
      "loss": 0.0267,
      "step": 15244
    },
    {
      "epoch": 2.1366503153468814,
      "grad_norm": 0.21684157848358154,
      "learning_rate": 8.848600813202582e-05,
      "loss": 0.0143,
      "step": 15245
    },
    {
      "epoch": 2.136790469516468,
      "grad_norm": 0.3699253499507904,
      "learning_rate": 8.847165749820616e-05,
      "loss": 0.0482,
      "step": 15246
    },
    {
      "epoch": 2.136930623686055,
      "grad_norm": 0.04588789492845535,
      "learning_rate": 8.84573068643865e-05,
      "loss": 0.0036,
      "step": 15247
    },
    {
      "epoch": 2.1370707778556413,
      "grad_norm": 0.2779466509819031,
      "learning_rate": 8.844295623056683e-05,
      "loss": 0.0731,
      "step": 15248
    },
    {
      "epoch": 2.1372109320252277,
      "grad_norm": 0.3055035471916199,
      "learning_rate": 8.842860559674718e-05,
      "loss": 0.0313,
      "step": 15249
    },
    {
      "epoch": 2.137351086194814,
      "grad_norm": 0.22759056091308594,
      "learning_rate": 8.841425496292753e-05,
      "loss": 0.0273,
      "step": 15250
    },
    {
      "epoch": 2.1374912403644006,
      "grad_norm": 0.15930531919002533,
      "learning_rate": 8.839990432910786e-05,
      "loss": 0.0248,
      "step": 15251
    },
    {
      "epoch": 2.1376313945339875,
      "grad_norm": 0.1530465930700302,
      "learning_rate": 8.83855536952882e-05,
      "loss": 0.0271,
      "step": 15252
    },
    {
      "epoch": 2.137771548703574,
      "grad_norm": 0.20927803218364716,
      "learning_rate": 8.837120306146854e-05,
      "loss": 0.0224,
      "step": 15253
    },
    {
      "epoch": 2.1379117028731605,
      "grad_norm": 0.7085014581680298,
      "learning_rate": 8.835685242764887e-05,
      "loss": 0.106,
      "step": 15254
    },
    {
      "epoch": 2.138051857042747,
      "grad_norm": 0.2739074230194092,
      "learning_rate": 8.834250179382923e-05,
      "loss": 0.0553,
      "step": 15255
    },
    {
      "epoch": 2.1381920112123334,
      "grad_norm": 0.632444441318512,
      "learning_rate": 8.832815116000956e-05,
      "loss": 0.1238,
      "step": 15256
    },
    {
      "epoch": 2.1383321653819203,
      "grad_norm": 0.13395020365715027,
      "learning_rate": 8.83138005261899e-05,
      "loss": 0.0218,
      "step": 15257
    },
    {
      "epoch": 2.1384723195515067,
      "grad_norm": 0.14127908647060394,
      "learning_rate": 8.829944989237024e-05,
      "loss": 0.0059,
      "step": 15258
    },
    {
      "epoch": 2.138612473721093,
      "grad_norm": 0.12099134176969528,
      "learning_rate": 8.828509925855057e-05,
      "loss": 0.0065,
      "step": 15259
    },
    {
      "epoch": 2.1387526278906797,
      "grad_norm": 0.3147258162498474,
      "learning_rate": 8.827074862473093e-05,
      "loss": 0.0198,
      "step": 15260
    },
    {
      "epoch": 2.138892782060266,
      "grad_norm": 0.1141904890537262,
      "learning_rate": 8.825639799091127e-05,
      "loss": 0.0366,
      "step": 15261
    },
    {
      "epoch": 2.139032936229853,
      "grad_norm": 0.5908268094062805,
      "learning_rate": 8.82420473570916e-05,
      "loss": 0.0356,
      "step": 15262
    },
    {
      "epoch": 2.1391730903994395,
      "grad_norm": 0.5669394135475159,
      "learning_rate": 8.822769672327194e-05,
      "loss": 0.1031,
      "step": 15263
    },
    {
      "epoch": 2.139313244569026,
      "grad_norm": 0.3832843005657196,
      "learning_rate": 8.821334608945227e-05,
      "loss": 0.0256,
      "step": 15264
    },
    {
      "epoch": 2.1394533987386124,
      "grad_norm": 0.20154747366905212,
      "learning_rate": 8.819899545563261e-05,
      "loss": 0.0185,
      "step": 15265
    },
    {
      "epoch": 2.139593552908199,
      "grad_norm": 0.3565092086791992,
      "learning_rate": 8.818464482181296e-05,
      "loss": 0.0158,
      "step": 15266
    },
    {
      "epoch": 2.1397337070777858,
      "grad_norm": 0.3719462752342224,
      "learning_rate": 8.817029418799329e-05,
      "loss": 0.0125,
      "step": 15267
    },
    {
      "epoch": 2.139873861247372,
      "grad_norm": 0.8984256386756897,
      "learning_rate": 8.815594355417364e-05,
      "loss": 0.058,
      "step": 15268
    },
    {
      "epoch": 2.1400140154169587,
      "grad_norm": 0.5621334314346313,
      "learning_rate": 8.814159292035398e-05,
      "loss": 0.0299,
      "step": 15269
    },
    {
      "epoch": 2.140154169586545,
      "grad_norm": 0.08213845640420914,
      "learning_rate": 8.81272422865343e-05,
      "loss": 0.0057,
      "step": 15270
    },
    {
      "epoch": 2.1402943237561316,
      "grad_norm": 0.32144880294799805,
      "learning_rate": 8.811289165271466e-05,
      "loss": 0.055,
      "step": 15271
    },
    {
      "epoch": 2.1404344779257185,
      "grad_norm": 0.28303712606430054,
      "learning_rate": 8.809854101889499e-05,
      "loss": 0.035,
      "step": 15272
    },
    {
      "epoch": 2.140574632095305,
      "grad_norm": 0.22838906943798065,
      "learning_rate": 8.808419038507533e-05,
      "loss": 0.0313,
      "step": 15273
    },
    {
      "epoch": 2.1407147862648914,
      "grad_norm": 0.18559999763965607,
      "learning_rate": 8.806983975125567e-05,
      "loss": 0.0391,
      "step": 15274
    },
    {
      "epoch": 2.140854940434478,
      "grad_norm": 0.41581666469573975,
      "learning_rate": 8.8055489117436e-05,
      "loss": 0.0709,
      "step": 15275
    },
    {
      "epoch": 2.1409950946040643,
      "grad_norm": 0.42859184741973877,
      "learning_rate": 8.804113848361636e-05,
      "loss": 0.0133,
      "step": 15276
    },
    {
      "epoch": 2.141135248773651,
      "grad_norm": 0.6941805481910706,
      "learning_rate": 8.80267878497967e-05,
      "loss": 0.0725,
      "step": 15277
    },
    {
      "epoch": 2.1412754029432377,
      "grad_norm": 0.35639214515686035,
      "learning_rate": 8.801243721597703e-05,
      "loss": 0.0428,
      "step": 15278
    },
    {
      "epoch": 2.141415557112824,
      "grad_norm": 0.141020268201828,
      "learning_rate": 8.799808658215737e-05,
      "loss": 0.0185,
      "step": 15279
    },
    {
      "epoch": 2.1415557112824106,
      "grad_norm": 0.29653400182724,
      "learning_rate": 8.79837359483377e-05,
      "loss": 0.034,
      "step": 15280
    },
    {
      "epoch": 2.141695865451997,
      "grad_norm": 0.1827317178249359,
      "learning_rate": 8.796938531451804e-05,
      "loss": 0.0265,
      "step": 15281
    },
    {
      "epoch": 2.1418360196215835,
      "grad_norm": 0.4067576229572296,
      "learning_rate": 8.79550346806984e-05,
      "loss": 0.0295,
      "step": 15282
    },
    {
      "epoch": 2.1419761737911704,
      "grad_norm": 0.2345951795578003,
      "learning_rate": 8.794068404687873e-05,
      "loss": 0.0674,
      "step": 15283
    },
    {
      "epoch": 2.142116327960757,
      "grad_norm": 0.11867791414260864,
      "learning_rate": 8.792633341305907e-05,
      "loss": 0.0168,
      "step": 15284
    },
    {
      "epoch": 2.1422564821303434,
      "grad_norm": 0.24366700649261475,
      "learning_rate": 8.791198277923941e-05,
      "loss": 0.0286,
      "step": 15285
    },
    {
      "epoch": 2.14239663629993,
      "grad_norm": 0.10818212479352951,
      "learning_rate": 8.789763214541974e-05,
      "loss": 0.0088,
      "step": 15286
    },
    {
      "epoch": 2.1425367904695163,
      "grad_norm": 0.339449018239975,
      "learning_rate": 8.78832815116001e-05,
      "loss": 0.0168,
      "step": 15287
    },
    {
      "epoch": 2.142676944639103,
      "grad_norm": 0.23022769391536713,
      "learning_rate": 8.786893087778044e-05,
      "loss": 0.0418,
      "step": 15288
    },
    {
      "epoch": 2.1428170988086896,
      "grad_norm": 0.3154343068599701,
      "learning_rate": 8.785458024396077e-05,
      "loss": 0.0152,
      "step": 15289
    },
    {
      "epoch": 2.142957252978276,
      "grad_norm": 0.08846689760684967,
      "learning_rate": 8.784022961014111e-05,
      "loss": 0.0072,
      "step": 15290
    },
    {
      "epoch": 2.1430974071478626,
      "grad_norm": 0.3567538559436798,
      "learning_rate": 8.782587897632144e-05,
      "loss": 0.0251,
      "step": 15291
    },
    {
      "epoch": 2.143237561317449,
      "grad_norm": 0.09792426228523254,
      "learning_rate": 8.781152834250179e-05,
      "loss": 0.0089,
      "step": 15292
    },
    {
      "epoch": 2.143377715487036,
      "grad_norm": 0.5313330292701721,
      "learning_rate": 8.779717770868213e-05,
      "loss": 0.0109,
      "step": 15293
    },
    {
      "epoch": 2.1435178696566224,
      "grad_norm": 0.506056547164917,
      "learning_rate": 8.778282707486246e-05,
      "loss": 0.0285,
      "step": 15294
    },
    {
      "epoch": 2.143658023826209,
      "grad_norm": 0.2221793234348297,
      "learning_rate": 8.77684764410428e-05,
      "loss": 0.0301,
      "step": 15295
    },
    {
      "epoch": 2.1437981779957953,
      "grad_norm": 0.2336333990097046,
      "learning_rate": 8.775412580722315e-05,
      "loss": 0.0284,
      "step": 15296
    },
    {
      "epoch": 2.1439383321653818,
      "grad_norm": 0.4865369498729706,
      "learning_rate": 8.773977517340348e-05,
      "loss": 0.0342,
      "step": 15297
    },
    {
      "epoch": 2.1440784863349687,
      "grad_norm": 0.13347235321998596,
      "learning_rate": 8.772542453958383e-05,
      "loss": 0.0124,
      "step": 15298
    },
    {
      "epoch": 2.144218640504555,
      "grad_norm": 0.12443598359823227,
      "learning_rate": 8.771107390576416e-05,
      "loss": 0.0133,
      "step": 15299
    },
    {
      "epoch": 2.1443587946741416,
      "grad_norm": 0.1774871051311493,
      "learning_rate": 8.76967232719445e-05,
      "loss": 0.0181,
      "step": 15300
    },
    {
      "epoch": 2.144498948843728,
      "grad_norm": 0.17956632375717163,
      "learning_rate": 8.768237263812484e-05,
      "loss": 0.02,
      "step": 15301
    },
    {
      "epoch": 2.1446391030133145,
      "grad_norm": 0.35058730840682983,
      "learning_rate": 8.766802200430517e-05,
      "loss": 0.0416,
      "step": 15302
    },
    {
      "epoch": 2.1447792571829014,
      "grad_norm": 0.07622545212507248,
      "learning_rate": 8.765367137048553e-05,
      "loss": 0.0069,
      "step": 15303
    },
    {
      "epoch": 2.144919411352488,
      "grad_norm": 0.5079339146614075,
      "learning_rate": 8.763932073666587e-05,
      "loss": 0.0464,
      "step": 15304
    },
    {
      "epoch": 2.1450595655220743,
      "grad_norm": 0.3639276623725891,
      "learning_rate": 8.76249701028462e-05,
      "loss": 0.0487,
      "step": 15305
    },
    {
      "epoch": 2.145199719691661,
      "grad_norm": 0.07134179025888443,
      "learning_rate": 8.761061946902654e-05,
      "loss": 0.0073,
      "step": 15306
    },
    {
      "epoch": 2.1453398738612472,
      "grad_norm": 0.15557405352592468,
      "learning_rate": 8.759626883520687e-05,
      "loss": 0.0283,
      "step": 15307
    },
    {
      "epoch": 2.1454800280308337,
      "grad_norm": 0.26771560311317444,
      "learning_rate": 8.758191820138722e-05,
      "loss": 0.0107,
      "step": 15308
    },
    {
      "epoch": 2.1456201822004206,
      "grad_norm": 0.3347119092941284,
      "learning_rate": 8.756756756756757e-05,
      "loss": 0.0366,
      "step": 15309
    },
    {
      "epoch": 2.145760336370007,
      "grad_norm": 0.7270070314407349,
      "learning_rate": 8.75532169337479e-05,
      "loss": 0.0109,
      "step": 15310
    },
    {
      "epoch": 2.1459004905395935,
      "grad_norm": 0.2638106644153595,
      "learning_rate": 8.753886629992824e-05,
      "loss": 0.0114,
      "step": 15311
    },
    {
      "epoch": 2.14604064470918,
      "grad_norm": 0.19569070637226105,
      "learning_rate": 8.752451566610858e-05,
      "loss": 0.0203,
      "step": 15312
    },
    {
      "epoch": 2.1461807988787664,
      "grad_norm": 0.3968261480331421,
      "learning_rate": 8.751016503228891e-05,
      "loss": 0.0219,
      "step": 15313
    },
    {
      "epoch": 2.1463209530483534,
      "grad_norm": 0.08780604600906372,
      "learning_rate": 8.749581439846926e-05,
      "loss": 0.0289,
      "step": 15314
    },
    {
      "epoch": 2.14646110721794,
      "grad_norm": 0.24469269812107086,
      "learning_rate": 8.748146376464959e-05,
      "loss": 0.0381,
      "step": 15315
    },
    {
      "epoch": 2.1466012613875263,
      "grad_norm": 0.8661059737205505,
      "learning_rate": 8.746711313082993e-05,
      "loss": 0.0577,
      "step": 15316
    },
    {
      "epoch": 2.1467414155571127,
      "grad_norm": 0.166100412607193,
      "learning_rate": 8.745276249701028e-05,
      "loss": 0.0089,
      "step": 15317
    },
    {
      "epoch": 2.146881569726699,
      "grad_norm": 0.39557480812072754,
      "learning_rate": 8.74384118631906e-05,
      "loss": 0.0062,
      "step": 15318
    },
    {
      "epoch": 2.147021723896286,
      "grad_norm": 1.7046048641204834,
      "learning_rate": 8.742406122937096e-05,
      "loss": 0.0849,
      "step": 15319
    },
    {
      "epoch": 2.1471618780658726,
      "grad_norm": 1.7429933547973633,
      "learning_rate": 8.74097105955513e-05,
      "loss": 0.2424,
      "step": 15320
    },
    {
      "epoch": 2.147302032235459,
      "grad_norm": 0.08225884288549423,
      "learning_rate": 8.739535996173163e-05,
      "loss": 0.0109,
      "step": 15321
    },
    {
      "epoch": 2.1474421864050455,
      "grad_norm": 0.29446908831596375,
      "learning_rate": 8.738100932791197e-05,
      "loss": 0.0648,
      "step": 15322
    },
    {
      "epoch": 2.147582340574632,
      "grad_norm": 0.13351160287857056,
      "learning_rate": 8.736665869409233e-05,
      "loss": 0.0096,
      "step": 15323
    },
    {
      "epoch": 2.147722494744219,
      "grad_norm": 0.14831215143203735,
      "learning_rate": 8.735230806027266e-05,
      "loss": 0.0148,
      "step": 15324
    },
    {
      "epoch": 2.1478626489138053,
      "grad_norm": 0.45169252157211304,
      "learning_rate": 8.7337957426453e-05,
      "loss": 0.1339,
      "step": 15325
    },
    {
      "epoch": 2.1480028030833918,
      "grad_norm": 0.11727157235145569,
      "learning_rate": 8.732360679263333e-05,
      "loss": 0.0115,
      "step": 15326
    },
    {
      "epoch": 2.148142957252978,
      "grad_norm": 0.09567470848560333,
      "learning_rate": 8.730925615881367e-05,
      "loss": 0.0072,
      "step": 15327
    },
    {
      "epoch": 2.1482831114225647,
      "grad_norm": 0.3257153034210205,
      "learning_rate": 8.729490552499401e-05,
      "loss": 0.0163,
      "step": 15328
    },
    {
      "epoch": 2.148423265592151,
      "grad_norm": 0.12457440793514252,
      "learning_rate": 8.728055489117434e-05,
      "loss": 0.0082,
      "step": 15329
    },
    {
      "epoch": 2.148563419761738,
      "grad_norm": 0.34792935848236084,
      "learning_rate": 8.72662042573547e-05,
      "loss": 0.0743,
      "step": 15330
    },
    {
      "epoch": 2.1487035739313245,
      "grad_norm": 0.16119292378425598,
      "learning_rate": 8.725185362353504e-05,
      "loss": 0.0182,
      "step": 15331
    },
    {
      "epoch": 2.148843728100911,
      "grad_norm": 0.24097059667110443,
      "learning_rate": 8.723750298971537e-05,
      "loss": 0.0262,
      "step": 15332
    },
    {
      "epoch": 2.1489838822704974,
      "grad_norm": 0.4506450295448303,
      "learning_rate": 8.722315235589571e-05,
      "loss": 0.042,
      "step": 15333
    },
    {
      "epoch": 2.1491240364400843,
      "grad_norm": 0.03821880370378494,
      "learning_rate": 8.720880172207604e-05,
      "loss": 0.0029,
      "step": 15334
    },
    {
      "epoch": 2.1492641906096708,
      "grad_norm": 0.26305267214775085,
      "learning_rate": 8.71944510882564e-05,
      "loss": 0.0341,
      "step": 15335
    },
    {
      "epoch": 2.1494043447792572,
      "grad_norm": 0.21345479786396027,
      "learning_rate": 8.718010045443674e-05,
      "loss": 0.014,
      "step": 15336
    },
    {
      "epoch": 2.1495444989488437,
      "grad_norm": 0.26840320229530334,
      "learning_rate": 8.716574982061706e-05,
      "loss": 0.0426,
      "step": 15337
    },
    {
      "epoch": 2.14968465311843,
      "grad_norm": 1.1672320365905762,
      "learning_rate": 8.715139918679741e-05,
      "loss": 0.0295,
      "step": 15338
    },
    {
      "epoch": 2.1498248072880166,
      "grad_norm": 0.2941757142543793,
      "learning_rate": 8.713704855297776e-05,
      "loss": 0.0188,
      "step": 15339
    },
    {
      "epoch": 2.1499649614576035,
      "grad_norm": 0.19164547324180603,
      "learning_rate": 8.712269791915809e-05,
      "loss": 0.0158,
      "step": 15340
    },
    {
      "epoch": 2.15010511562719,
      "grad_norm": 0.314678817987442,
      "learning_rate": 8.710834728533843e-05,
      "loss": 0.0448,
      "step": 15341
    },
    {
      "epoch": 2.1502452697967764,
      "grad_norm": 0.13238593935966492,
      "learning_rate": 8.709399665151876e-05,
      "loss": 0.0263,
      "step": 15342
    },
    {
      "epoch": 2.150385423966363,
      "grad_norm": 0.0877959132194519,
      "learning_rate": 8.70796460176991e-05,
      "loss": 0.0051,
      "step": 15343
    },
    {
      "epoch": 2.1505255781359494,
      "grad_norm": 0.18056292831897736,
      "learning_rate": 8.706529538387945e-05,
      "loss": 0.0102,
      "step": 15344
    },
    {
      "epoch": 2.1506657323055363,
      "grad_norm": 0.14252761006355286,
      "learning_rate": 8.705094475005977e-05,
      "loss": 0.0338,
      "step": 15345
    },
    {
      "epoch": 2.1508058864751227,
      "grad_norm": 0.43279486894607544,
      "learning_rate": 8.703659411624013e-05,
      "loss": 0.0356,
      "step": 15346
    },
    {
      "epoch": 2.150946040644709,
      "grad_norm": 0.1459585577249527,
      "learning_rate": 8.702224348242047e-05,
      "loss": 0.0161,
      "step": 15347
    },
    {
      "epoch": 2.1510861948142956,
      "grad_norm": 0.31226831674575806,
      "learning_rate": 8.70078928486008e-05,
      "loss": 0.0317,
      "step": 15348
    },
    {
      "epoch": 2.151226348983882,
      "grad_norm": 0.2940607964992523,
      "learning_rate": 8.699354221478114e-05,
      "loss": 0.0328,
      "step": 15349
    },
    {
      "epoch": 2.151366503153469,
      "grad_norm": 0.07008252292871475,
      "learning_rate": 8.697919158096147e-05,
      "loss": 0.0049,
      "step": 15350
    },
    {
      "epoch": 2.1515066573230555,
      "grad_norm": 0.19552765786647797,
      "learning_rate": 8.696484094714183e-05,
      "loss": 0.0345,
      "step": 15351
    },
    {
      "epoch": 2.151646811492642,
      "grad_norm": 0.08969911187887192,
      "learning_rate": 8.695049031332217e-05,
      "loss": 0.0077,
      "step": 15352
    },
    {
      "epoch": 2.1517869656622284,
      "grad_norm": 0.10951448231935501,
      "learning_rate": 8.69361396795025e-05,
      "loss": 0.0096,
      "step": 15353
    },
    {
      "epoch": 2.151927119831815,
      "grad_norm": 0.14570145308971405,
      "learning_rate": 8.692178904568284e-05,
      "loss": 0.0361,
      "step": 15354
    },
    {
      "epoch": 2.1520672740014017,
      "grad_norm": 0.253081351518631,
      "learning_rate": 8.69074384118632e-05,
      "loss": 0.0231,
      "step": 15355
    },
    {
      "epoch": 2.152207428170988,
      "grad_norm": 0.16390496492385864,
      "learning_rate": 8.689308777804352e-05,
      "loss": 0.0126,
      "step": 15356
    },
    {
      "epoch": 2.1523475823405747,
      "grad_norm": 0.07381082326173782,
      "learning_rate": 8.687873714422387e-05,
      "loss": 0.0136,
      "step": 15357
    },
    {
      "epoch": 2.152487736510161,
      "grad_norm": 0.22054754197597504,
      "learning_rate": 8.68643865104042e-05,
      "loss": 0.0166,
      "step": 15358
    },
    {
      "epoch": 2.1526278906797476,
      "grad_norm": 0.8149431347846985,
      "learning_rate": 8.685003587658454e-05,
      "loss": 0.0539,
      "step": 15359
    },
    {
      "epoch": 2.152768044849334,
      "grad_norm": 0.356865257024765,
      "learning_rate": 8.683568524276488e-05,
      "loss": 0.0333,
      "step": 15360
    },
    {
      "epoch": 2.152908199018921,
      "grad_norm": 0.1602364480495453,
      "learning_rate": 8.682133460894522e-05,
      "loss": 0.0114,
      "step": 15361
    },
    {
      "epoch": 2.1530483531885074,
      "grad_norm": 0.17978474497795105,
      "learning_rate": 8.680698397512556e-05,
      "loss": 0.0082,
      "step": 15362
    },
    {
      "epoch": 2.153188507358094,
      "grad_norm": 0.22223439812660217,
      "learning_rate": 8.67926333413059e-05,
      "loss": 0.055,
      "step": 15363
    },
    {
      "epoch": 2.1533286615276803,
      "grad_norm": 0.3857043385505676,
      "learning_rate": 8.677828270748623e-05,
      "loss": 0.0196,
      "step": 15364
    },
    {
      "epoch": 2.1534688156972672,
      "grad_norm": 0.2708699703216553,
      "learning_rate": 8.676393207366658e-05,
      "loss": 0.0055,
      "step": 15365
    },
    {
      "epoch": 2.1536089698668537,
      "grad_norm": 0.9954331517219543,
      "learning_rate": 8.674958143984693e-05,
      "loss": 0.1388,
      "step": 15366
    },
    {
      "epoch": 2.15374912403644,
      "grad_norm": 0.49957525730133057,
      "learning_rate": 8.673523080602726e-05,
      "loss": 0.0211,
      "step": 15367
    },
    {
      "epoch": 2.1538892782060266,
      "grad_norm": 0.19946284592151642,
      "learning_rate": 8.67208801722076e-05,
      "loss": 0.0122,
      "step": 15368
    },
    {
      "epoch": 2.154029432375613,
      "grad_norm": 0.13076874613761902,
      "learning_rate": 8.670652953838793e-05,
      "loss": 0.0041,
      "step": 15369
    },
    {
      "epoch": 2.1541695865451995,
      "grad_norm": 0.3796508312225342,
      "learning_rate": 8.669217890456827e-05,
      "loss": 0.0102,
      "step": 15370
    },
    {
      "epoch": 2.1543097407147864,
      "grad_norm": 0.13304169476032257,
      "learning_rate": 8.667782827074863e-05,
      "loss": 0.0183,
      "step": 15371
    },
    {
      "epoch": 2.154449894884373,
      "grad_norm": 0.21116164326667786,
      "learning_rate": 8.666347763692896e-05,
      "loss": 0.0166,
      "step": 15372
    },
    {
      "epoch": 2.1545900490539593,
      "grad_norm": 0.12565453350543976,
      "learning_rate": 8.66491270031093e-05,
      "loss": 0.0197,
      "step": 15373
    },
    {
      "epoch": 2.154730203223546,
      "grad_norm": 0.23317596316337585,
      "learning_rate": 8.663477636928964e-05,
      "loss": 0.0178,
      "step": 15374
    },
    {
      "epoch": 2.1548703573931323,
      "grad_norm": 0.16248784959316254,
      "learning_rate": 8.662042573546997e-05,
      "loss": 0.0247,
      "step": 15375
    },
    {
      "epoch": 2.155010511562719,
      "grad_norm": 0.1887175291776657,
      "learning_rate": 8.660607510165031e-05,
      "loss": 0.0347,
      "step": 15376
    },
    {
      "epoch": 2.1551506657323056,
      "grad_norm": 0.2740532457828522,
      "learning_rate": 8.659172446783065e-05,
      "loss": 0.0208,
      "step": 15377
    },
    {
      "epoch": 2.155290819901892,
      "grad_norm": 0.19722363352775574,
      "learning_rate": 8.6577373834011e-05,
      "loss": 0.0154,
      "step": 15378
    },
    {
      "epoch": 2.1554309740714785,
      "grad_norm": 0.98005211353302,
      "learning_rate": 8.656302320019134e-05,
      "loss": 0.0789,
      "step": 15379
    },
    {
      "epoch": 2.155571128241065,
      "grad_norm": 0.1699998825788498,
      "learning_rate": 8.654867256637167e-05,
      "loss": 0.0195,
      "step": 15380
    },
    {
      "epoch": 2.155711282410652,
      "grad_norm": 0.11828425526618958,
      "learning_rate": 8.653432193255201e-05,
      "loss": 0.0123,
      "step": 15381
    },
    {
      "epoch": 2.1558514365802384,
      "grad_norm": 0.1831325888633728,
      "learning_rate": 8.651997129873236e-05,
      "loss": 0.0145,
      "step": 15382
    },
    {
      "epoch": 2.155991590749825,
      "grad_norm": 0.11076951771974564,
      "learning_rate": 8.65056206649127e-05,
      "loss": 0.0104,
      "step": 15383
    },
    {
      "epoch": 2.1561317449194113,
      "grad_norm": 0.07718174904584885,
      "learning_rate": 8.649127003109304e-05,
      "loss": 0.0141,
      "step": 15384
    },
    {
      "epoch": 2.1562718990889977,
      "grad_norm": 0.07783882319927216,
      "learning_rate": 8.647691939727336e-05,
      "loss": 0.0051,
      "step": 15385
    },
    {
      "epoch": 2.1564120532585846,
      "grad_norm": 0.23227113485336304,
      "learning_rate": 8.64625687634537e-05,
      "loss": 0.0461,
      "step": 15386
    },
    {
      "epoch": 2.156552207428171,
      "grad_norm": 0.808319628238678,
      "learning_rate": 8.644821812963406e-05,
      "loss": 0.0329,
      "step": 15387
    },
    {
      "epoch": 2.1566923615977576,
      "grad_norm": 0.1746961921453476,
      "learning_rate": 8.643386749581439e-05,
      "loss": 0.0596,
      "step": 15388
    },
    {
      "epoch": 2.156832515767344,
      "grad_norm": 0.17402112483978271,
      "learning_rate": 8.641951686199473e-05,
      "loss": 0.0674,
      "step": 15389
    },
    {
      "epoch": 2.1569726699369305,
      "grad_norm": 0.28075429797172546,
      "learning_rate": 8.640516622817507e-05,
      "loss": 0.0364,
      "step": 15390
    },
    {
      "epoch": 2.157112824106517,
      "grad_norm": 0.05913713946938515,
      "learning_rate": 8.63908155943554e-05,
      "loss": 0.0186,
      "step": 15391
    },
    {
      "epoch": 2.157252978276104,
      "grad_norm": 0.3385704457759857,
      "learning_rate": 8.637646496053575e-05,
      "loss": 0.0693,
      "step": 15392
    },
    {
      "epoch": 2.1573931324456903,
      "grad_norm": 0.19813212752342224,
      "learning_rate": 8.636211432671609e-05,
      "loss": 0.03,
      "step": 15393
    },
    {
      "epoch": 2.1575332866152768,
      "grad_norm": 0.37917935848236084,
      "learning_rate": 8.634776369289643e-05,
      "loss": 0.054,
      "step": 15394
    },
    {
      "epoch": 2.1576734407848632,
      "grad_norm": 0.12380579113960266,
      "learning_rate": 8.633341305907677e-05,
      "loss": 0.0077,
      "step": 15395
    },
    {
      "epoch": 2.1578135949544497,
      "grad_norm": 0.4453830122947693,
      "learning_rate": 8.63190624252571e-05,
      "loss": 0.0278,
      "step": 15396
    },
    {
      "epoch": 2.1579537491240366,
      "grad_norm": 0.5230569243431091,
      "learning_rate": 8.630471179143744e-05,
      "loss": 0.0948,
      "step": 15397
    },
    {
      "epoch": 2.158093903293623,
      "grad_norm": 0.06244026869535446,
      "learning_rate": 8.62903611576178e-05,
      "loss": 0.0061,
      "step": 15398
    },
    {
      "epoch": 2.1582340574632095,
      "grad_norm": 0.16651076078414917,
      "learning_rate": 8.627601052379813e-05,
      "loss": 0.0118,
      "step": 15399
    },
    {
      "epoch": 2.158374211632796,
      "grad_norm": 0.1521030217409134,
      "learning_rate": 8.626165988997847e-05,
      "loss": 0.011,
      "step": 15400
    },
    {
      "epoch": 2.1585143658023824,
      "grad_norm": 0.27640479803085327,
      "learning_rate": 8.624730925615881e-05,
      "loss": 0.0716,
      "step": 15401
    },
    {
      "epoch": 2.1586545199719693,
      "grad_norm": 0.08490371704101562,
      "learning_rate": 8.623295862233914e-05,
      "loss": 0.0097,
      "step": 15402
    },
    {
      "epoch": 2.158794674141556,
      "grad_norm": 0.22089050710201263,
      "learning_rate": 8.62186079885195e-05,
      "loss": 0.0102,
      "step": 15403
    },
    {
      "epoch": 2.1589348283111423,
      "grad_norm": 0.24452009797096252,
      "learning_rate": 8.620425735469982e-05,
      "loss": 0.0395,
      "step": 15404
    },
    {
      "epoch": 2.1590749824807287,
      "grad_norm": 0.23416975140571594,
      "learning_rate": 8.618990672088017e-05,
      "loss": 0.0421,
      "step": 15405
    },
    {
      "epoch": 2.159215136650315,
      "grad_norm": 0.4060458242893219,
      "learning_rate": 8.617555608706051e-05,
      "loss": 0.029,
      "step": 15406
    },
    {
      "epoch": 2.159355290819902,
      "grad_norm": 0.10219886898994446,
      "learning_rate": 8.616120545324084e-05,
      "loss": 0.0052,
      "step": 15407
    },
    {
      "epoch": 2.1594954449894885,
      "grad_norm": 0.2730812430381775,
      "learning_rate": 8.614685481942118e-05,
      "loss": 0.0319,
      "step": 15408
    },
    {
      "epoch": 2.159635599159075,
      "grad_norm": 0.08153347671031952,
      "learning_rate": 8.613250418560153e-05,
      "loss": 0.0088,
      "step": 15409
    },
    {
      "epoch": 2.1597757533286615,
      "grad_norm": 0.4057573974132538,
      "learning_rate": 8.611815355178186e-05,
      "loss": 0.0823,
      "step": 15410
    },
    {
      "epoch": 2.159915907498248,
      "grad_norm": 0.4863761365413666,
      "learning_rate": 8.61038029179622e-05,
      "loss": 0.0524,
      "step": 15411
    },
    {
      "epoch": 2.160056061667835,
      "grad_norm": 0.03333592787384987,
      "learning_rate": 8.608945228414253e-05,
      "loss": 0.0019,
      "step": 15412
    },
    {
      "epoch": 2.1601962158374213,
      "grad_norm": 0.48775067925453186,
      "learning_rate": 8.607510165032288e-05,
      "loss": 0.0567,
      "step": 15413
    },
    {
      "epoch": 2.1603363700070077,
      "grad_norm": 0.47304537892341614,
      "learning_rate": 8.606075101650323e-05,
      "loss": 0.039,
      "step": 15414
    },
    {
      "epoch": 2.160476524176594,
      "grad_norm": 0.22962334752082825,
      "learning_rate": 8.604640038268356e-05,
      "loss": 0.0068,
      "step": 15415
    },
    {
      "epoch": 2.1606166783461807,
      "grad_norm": 0.06080048531293869,
      "learning_rate": 8.60320497488639e-05,
      "loss": 0.0058,
      "step": 15416
    },
    {
      "epoch": 2.1607568325157676,
      "grad_norm": 0.3105883300304413,
      "learning_rate": 8.601769911504424e-05,
      "loss": 0.0632,
      "step": 15417
    },
    {
      "epoch": 2.160896986685354,
      "grad_norm": 1.8197739124298096,
      "learning_rate": 8.600334848122457e-05,
      "loss": 0.0529,
      "step": 15418
    },
    {
      "epoch": 2.1610371408549405,
      "grad_norm": 0.23264367878437042,
      "learning_rate": 8.598899784740493e-05,
      "loss": 0.0782,
      "step": 15419
    },
    {
      "epoch": 2.161177295024527,
      "grad_norm": 0.9086876511573792,
      "learning_rate": 8.597464721358526e-05,
      "loss": 0.0526,
      "step": 15420
    },
    {
      "epoch": 2.1613174491941134,
      "grad_norm": 0.11605122685432434,
      "learning_rate": 8.59602965797656e-05,
      "loss": 0.0241,
      "step": 15421
    },
    {
      "epoch": 2.1614576033637,
      "grad_norm": 0.5580950379371643,
      "learning_rate": 8.594594594594594e-05,
      "loss": 0.0593,
      "step": 15422
    },
    {
      "epoch": 2.1615977575332868,
      "grad_norm": 0.21459150314331055,
      "learning_rate": 8.593159531212627e-05,
      "loss": 0.0211,
      "step": 15423
    },
    {
      "epoch": 2.161737911702873,
      "grad_norm": 0.15788954496383667,
      "learning_rate": 8.591724467830661e-05,
      "loss": 0.0386,
      "step": 15424
    },
    {
      "epoch": 2.1618780658724597,
      "grad_norm": 0.058036163449287415,
      "learning_rate": 8.590289404448697e-05,
      "loss": 0.0053,
      "step": 15425
    },
    {
      "epoch": 2.162018220042046,
      "grad_norm": 0.19580991566181183,
      "learning_rate": 8.58885434106673e-05,
      "loss": 0.0558,
      "step": 15426
    },
    {
      "epoch": 2.1621583742116326,
      "grad_norm": 0.28299856185913086,
      "learning_rate": 8.587419277684764e-05,
      "loss": 0.053,
      "step": 15427
    },
    {
      "epoch": 2.1622985283812195,
      "grad_norm": 0.1504969596862793,
      "learning_rate": 8.585984214302797e-05,
      "loss": 0.0192,
      "step": 15428
    },
    {
      "epoch": 2.162438682550806,
      "grad_norm": 0.1376238316297531,
      "learning_rate": 8.584549150920831e-05,
      "loss": 0.0073,
      "step": 15429
    },
    {
      "epoch": 2.1625788367203924,
      "grad_norm": 0.17269179224967957,
      "learning_rate": 8.583114087538866e-05,
      "loss": 0.0224,
      "step": 15430
    },
    {
      "epoch": 2.162718990889979,
      "grad_norm": 0.06337358802556992,
      "learning_rate": 8.581679024156899e-05,
      "loss": 0.0033,
      "step": 15431
    },
    {
      "epoch": 2.1628591450595653,
      "grad_norm": 0.3600456714630127,
      "learning_rate": 8.580243960774934e-05,
      "loss": 0.0311,
      "step": 15432
    },
    {
      "epoch": 2.1629992992291522,
      "grad_norm": 0.09824953973293304,
      "learning_rate": 8.578808897392968e-05,
      "loss": 0.0081,
      "step": 15433
    },
    {
      "epoch": 2.1631394533987387,
      "grad_norm": 1.3486517667770386,
      "learning_rate": 8.577373834011e-05,
      "loss": 0.0131,
      "step": 15434
    },
    {
      "epoch": 2.163279607568325,
      "grad_norm": 0.2690047323703766,
      "learning_rate": 8.575938770629036e-05,
      "loss": 0.0305,
      "step": 15435
    },
    {
      "epoch": 2.1634197617379116,
      "grad_norm": 0.11086538434028625,
      "learning_rate": 8.57450370724707e-05,
      "loss": 0.0125,
      "step": 15436
    },
    {
      "epoch": 2.163559915907498,
      "grad_norm": 0.19321806728839874,
      "learning_rate": 8.573068643865103e-05,
      "loss": 0.0196,
      "step": 15437
    },
    {
      "epoch": 2.163700070077085,
      "grad_norm": 0.1602705717086792,
      "learning_rate": 8.571633580483137e-05,
      "loss": 0.0298,
      "step": 15438
    },
    {
      "epoch": 2.1638402242466714,
      "grad_norm": 0.5952815413475037,
      "learning_rate": 8.57019851710117e-05,
      "loss": 0.0649,
      "step": 15439
    },
    {
      "epoch": 2.163980378416258,
      "grad_norm": 0.22576525807380676,
      "learning_rate": 8.568763453719204e-05,
      "loss": 0.0163,
      "step": 15440
    },
    {
      "epoch": 2.1641205325858444,
      "grad_norm": 0.2509418725967407,
      "learning_rate": 8.56732839033724e-05,
      "loss": 0.0355,
      "step": 15441
    },
    {
      "epoch": 2.164260686755431,
      "grad_norm": 0.22534427046775818,
      "learning_rate": 8.565893326955273e-05,
      "loss": 0.0178,
      "step": 15442
    },
    {
      "epoch": 2.1644008409250177,
      "grad_norm": 0.11104210466146469,
      "learning_rate": 8.564458263573307e-05,
      "loss": 0.0068,
      "step": 15443
    },
    {
      "epoch": 2.164540995094604,
      "grad_norm": 0.02992108464241028,
      "learning_rate": 8.563023200191341e-05,
      "loss": 0.0021,
      "step": 15444
    },
    {
      "epoch": 2.1646811492641906,
      "grad_norm": 0.3460654020309448,
      "learning_rate": 8.561588136809374e-05,
      "loss": 0.0303,
      "step": 15445
    },
    {
      "epoch": 2.164821303433777,
      "grad_norm": 0.18774725496768951,
      "learning_rate": 8.56015307342741e-05,
      "loss": 0.0119,
      "step": 15446
    },
    {
      "epoch": 2.1649614576033636,
      "grad_norm": 0.22597059607505798,
      "learning_rate": 8.558718010045443e-05,
      "loss": 0.0459,
      "step": 15447
    },
    {
      "epoch": 2.1651016117729505,
      "grad_norm": 0.2159491777420044,
      "learning_rate": 8.557282946663477e-05,
      "loss": 0.0349,
      "step": 15448
    },
    {
      "epoch": 2.165241765942537,
      "grad_norm": 0.2804139256477356,
      "learning_rate": 8.555847883281511e-05,
      "loss": 0.0308,
      "step": 15449
    },
    {
      "epoch": 2.1653819201121234,
      "grad_norm": 0.1893835961818695,
      "learning_rate": 8.554412819899544e-05,
      "loss": 0.0756,
      "step": 15450
    },
    {
      "epoch": 2.16552207428171,
      "grad_norm": 0.2896615266799927,
      "learning_rate": 8.55297775651758e-05,
      "loss": 0.0885,
      "step": 15451
    },
    {
      "epoch": 2.1656622284512963,
      "grad_norm": 0.12295167148113251,
      "learning_rate": 8.551542693135614e-05,
      "loss": 0.0073,
      "step": 15452
    },
    {
      "epoch": 2.1658023826208828,
      "grad_norm": 0.18107888102531433,
      "learning_rate": 8.550107629753647e-05,
      "loss": 0.0083,
      "step": 15453
    },
    {
      "epoch": 2.1659425367904697,
      "grad_norm": 0.11750922352075577,
      "learning_rate": 8.548672566371681e-05,
      "loss": 0.0128,
      "step": 15454
    },
    {
      "epoch": 2.166082690960056,
      "grad_norm": 0.2867882549762726,
      "learning_rate": 8.547237502989714e-05,
      "loss": 0.056,
      "step": 15455
    },
    {
      "epoch": 2.1662228451296426,
      "grad_norm": 0.3362749218940735,
      "learning_rate": 8.545802439607748e-05,
      "loss": 0.049,
      "step": 15456
    },
    {
      "epoch": 2.166362999299229,
      "grad_norm": 0.1076216995716095,
      "learning_rate": 8.544367376225783e-05,
      "loss": 0.034,
      "step": 15457
    },
    {
      "epoch": 2.1665031534688155,
      "grad_norm": 0.22457194328308105,
      "learning_rate": 8.542932312843816e-05,
      "loss": 0.0179,
      "step": 15458
    },
    {
      "epoch": 2.1666433076384024,
      "grad_norm": 0.19430334866046906,
      "learning_rate": 8.54149724946185e-05,
      "loss": 0.0279,
      "step": 15459
    },
    {
      "epoch": 2.166783461807989,
      "grad_norm": 0.028072109445929527,
      "learning_rate": 8.540062186079885e-05,
      "loss": 0.002,
      "step": 15460
    },
    {
      "epoch": 2.1669236159775753,
      "grad_norm": 0.9777933359146118,
      "learning_rate": 8.538627122697917e-05,
      "loss": 0.0389,
      "step": 15461
    },
    {
      "epoch": 2.167063770147162,
      "grad_norm": 0.3590082824230194,
      "learning_rate": 8.537192059315953e-05,
      "loss": 0.0063,
      "step": 15462
    },
    {
      "epoch": 2.1672039243167482,
      "grad_norm": 0.05881037190556526,
      "learning_rate": 8.535756995933986e-05,
      "loss": 0.0054,
      "step": 15463
    },
    {
      "epoch": 2.167344078486335,
      "grad_norm": 0.4571841061115265,
      "learning_rate": 8.53432193255202e-05,
      "loss": 0.093,
      "step": 15464
    },
    {
      "epoch": 2.1674842326559216,
      "grad_norm": 1.7408326864242554,
      "learning_rate": 8.532886869170054e-05,
      "loss": 0.1378,
      "step": 15465
    },
    {
      "epoch": 2.167624386825508,
      "grad_norm": 0.3223438858985901,
      "learning_rate": 8.531451805788087e-05,
      "loss": 0.0082,
      "step": 15466
    },
    {
      "epoch": 2.1677645409950945,
      "grad_norm": 1.9041938781738281,
      "learning_rate": 8.530016742406123e-05,
      "loss": 0.1508,
      "step": 15467
    },
    {
      "epoch": 2.167904695164681,
      "grad_norm": 0.07710761576890945,
      "learning_rate": 8.528581679024157e-05,
      "loss": 0.0057,
      "step": 15468
    },
    {
      "epoch": 2.168044849334268,
      "grad_norm": 0.3869759142398834,
      "learning_rate": 8.52714661564219e-05,
      "loss": 0.0278,
      "step": 15469
    },
    {
      "epoch": 2.1681850035038543,
      "grad_norm": 0.8428178429603577,
      "learning_rate": 8.525711552260224e-05,
      "loss": 0.1272,
      "step": 15470
    },
    {
      "epoch": 2.168325157673441,
      "grad_norm": 0.3046847879886627,
      "learning_rate": 8.524276488878258e-05,
      "loss": 0.0419,
      "step": 15471
    },
    {
      "epoch": 2.1684653118430273,
      "grad_norm": 0.1405702531337738,
      "learning_rate": 8.522841425496292e-05,
      "loss": 0.0146,
      "step": 15472
    },
    {
      "epoch": 2.1686054660126137,
      "grad_norm": 0.30379512906074524,
      "learning_rate": 8.521406362114327e-05,
      "loss": 0.052,
      "step": 15473
    },
    {
      "epoch": 2.1687456201822006,
      "grad_norm": 0.09665139019489288,
      "learning_rate": 8.51997129873236e-05,
      "loss": 0.0256,
      "step": 15474
    },
    {
      "epoch": 2.168885774351787,
      "grad_norm": 0.26153433322906494,
      "learning_rate": 8.518536235350394e-05,
      "loss": 0.0582,
      "step": 15475
    },
    {
      "epoch": 2.1690259285213735,
      "grad_norm": 0.2731797397136688,
      "learning_rate": 8.517101171968428e-05,
      "loss": 0.0348,
      "step": 15476
    },
    {
      "epoch": 2.16916608269096,
      "grad_norm": 0.1394374966621399,
      "learning_rate": 8.515666108586461e-05,
      "loss": 0.0161,
      "step": 15477
    },
    {
      "epoch": 2.1693062368605465,
      "grad_norm": 0.16301877796649933,
      "learning_rate": 8.514231045204496e-05,
      "loss": 0.018,
      "step": 15478
    },
    {
      "epoch": 2.1694463910301334,
      "grad_norm": 0.18953461945056915,
      "learning_rate": 8.51279598182253e-05,
      "loss": 0.0321,
      "step": 15479
    },
    {
      "epoch": 2.16958654519972,
      "grad_norm": 0.25635281205177307,
      "learning_rate": 8.511360918440563e-05,
      "loss": 0.0215,
      "step": 15480
    },
    {
      "epoch": 2.1697266993693063,
      "grad_norm": 0.12357616424560547,
      "learning_rate": 8.509925855058598e-05,
      "loss": 0.0194,
      "step": 15481
    },
    {
      "epoch": 2.1698668535388927,
      "grad_norm": 0.1335647702217102,
      "learning_rate": 8.50849079167663e-05,
      "loss": 0.0151,
      "step": 15482
    },
    {
      "epoch": 2.170007007708479,
      "grad_norm": 0.5593230128288269,
      "learning_rate": 8.507055728294666e-05,
      "loss": 0.1031,
      "step": 15483
    },
    {
      "epoch": 2.1701471618780657,
      "grad_norm": 0.26036587357521057,
      "learning_rate": 8.5056206649127e-05,
      "loss": 0.0102,
      "step": 15484
    },
    {
      "epoch": 2.1702873160476526,
      "grad_norm": 0.26807889342308044,
      "learning_rate": 8.504185601530733e-05,
      "loss": 0.0237,
      "step": 15485
    },
    {
      "epoch": 2.170427470217239,
      "grad_norm": 0.1733667403459549,
      "learning_rate": 8.502750538148767e-05,
      "loss": 0.0445,
      "step": 15486
    },
    {
      "epoch": 2.1705676243868255,
      "grad_norm": 0.37354567646980286,
      "learning_rate": 8.501315474766802e-05,
      "loss": 0.0953,
      "step": 15487
    },
    {
      "epoch": 2.170707778556412,
      "grad_norm": 0.7670668363571167,
      "learning_rate": 8.499880411384836e-05,
      "loss": 0.0467,
      "step": 15488
    },
    {
      "epoch": 2.1708479327259984,
      "grad_norm": 0.25219079852104187,
      "learning_rate": 8.49844534800287e-05,
      "loss": 0.0088,
      "step": 15489
    },
    {
      "epoch": 2.1709880868955853,
      "grad_norm": 0.48107919096946716,
      "learning_rate": 8.497010284620903e-05,
      "loss": 0.0167,
      "step": 15490
    },
    {
      "epoch": 2.1711282410651718,
      "grad_norm": 0.06198776140809059,
      "learning_rate": 8.495575221238937e-05,
      "loss": 0.0114,
      "step": 15491
    },
    {
      "epoch": 2.1712683952347582,
      "grad_norm": 0.3005569875240326,
      "learning_rate": 8.494140157856971e-05,
      "loss": 0.0915,
      "step": 15492
    },
    {
      "epoch": 2.1714085494043447,
      "grad_norm": 0.25535961985588074,
      "learning_rate": 8.492705094475004e-05,
      "loss": 0.1006,
      "step": 15493
    },
    {
      "epoch": 2.171548703573931,
      "grad_norm": 0.2022262066602707,
      "learning_rate": 8.49127003109304e-05,
      "loss": 0.0369,
      "step": 15494
    },
    {
      "epoch": 2.171688857743518,
      "grad_norm": 0.11930090188980103,
      "learning_rate": 8.489834967711074e-05,
      "loss": 0.0061,
      "step": 15495
    },
    {
      "epoch": 2.1718290119131045,
      "grad_norm": 0.8684285879135132,
      "learning_rate": 8.488399904329107e-05,
      "loss": 0.059,
      "step": 15496
    },
    {
      "epoch": 2.171969166082691,
      "grad_norm": 0.3775840699672699,
      "learning_rate": 8.486964840947141e-05,
      "loss": 0.0123,
      "step": 15497
    },
    {
      "epoch": 2.1721093202522774,
      "grad_norm": 0.22777603566646576,
      "learning_rate": 8.485529777565174e-05,
      "loss": 0.026,
      "step": 15498
    },
    {
      "epoch": 2.172249474421864,
      "grad_norm": 1.125455379486084,
      "learning_rate": 8.48409471418321e-05,
      "loss": 0.0287,
      "step": 15499
    },
    {
      "epoch": 2.172389628591451,
      "grad_norm": 0.14573828876018524,
      "learning_rate": 8.482659650801244e-05,
      "loss": 0.0257,
      "step": 15500
    },
    {
      "epoch": 2.1725297827610373,
      "grad_norm": 0.2938830554485321,
      "learning_rate": 8.481224587419276e-05,
      "loss": 0.0168,
      "step": 15501
    },
    {
      "epoch": 2.1726699369306237,
      "grad_norm": 0.4528128504753113,
      "learning_rate": 8.47978952403731e-05,
      "loss": 0.0535,
      "step": 15502
    },
    {
      "epoch": 2.17281009110021,
      "grad_norm": 0.17684079706668854,
      "learning_rate": 8.478354460655345e-05,
      "loss": 0.0129,
      "step": 15503
    },
    {
      "epoch": 2.1729502452697966,
      "grad_norm": 0.1797565370798111,
      "learning_rate": 8.476919397273379e-05,
      "loss": 0.0173,
      "step": 15504
    },
    {
      "epoch": 2.173090399439383,
      "grad_norm": 0.02430335059762001,
      "learning_rate": 8.475484333891413e-05,
      "loss": 0.0023,
      "step": 15505
    },
    {
      "epoch": 2.17323055360897,
      "grad_norm": 1.0244117975234985,
      "learning_rate": 8.474049270509446e-05,
      "loss": 0.0593,
      "step": 15506
    },
    {
      "epoch": 2.1733707077785565,
      "grad_norm": 0.4036937654018402,
      "learning_rate": 8.47261420712748e-05,
      "loss": 0.0653,
      "step": 15507
    },
    {
      "epoch": 2.173510861948143,
      "grad_norm": 0.6878527998924255,
      "learning_rate": 8.471179143745515e-05,
      "loss": 0.0787,
      "step": 15508
    },
    {
      "epoch": 2.1736510161177294,
      "grad_norm": 0.10806319117546082,
      "learning_rate": 8.469744080363547e-05,
      "loss": 0.0055,
      "step": 15509
    },
    {
      "epoch": 2.1737911702873163,
      "grad_norm": 0.26832789182662964,
      "learning_rate": 8.468309016981583e-05,
      "loss": 0.052,
      "step": 15510
    },
    {
      "epoch": 2.1739313244569027,
      "grad_norm": 0.13309180736541748,
      "learning_rate": 8.466873953599617e-05,
      "loss": 0.0656,
      "step": 15511
    },
    {
      "epoch": 2.174071478626489,
      "grad_norm": 0.013673707842826843,
      "learning_rate": 8.46543889021765e-05,
      "loss": 0.0012,
      "step": 15512
    },
    {
      "epoch": 2.1742116327960757,
      "grad_norm": 0.1314159482717514,
      "learning_rate": 8.464003826835684e-05,
      "loss": 0.0125,
      "step": 15513
    },
    {
      "epoch": 2.174351786965662,
      "grad_norm": 0.3085721433162689,
      "learning_rate": 8.46256876345372e-05,
      "loss": 0.104,
      "step": 15514
    },
    {
      "epoch": 2.1744919411352486,
      "grad_norm": 0.16569912433624268,
      "learning_rate": 8.461133700071753e-05,
      "loss": 0.0211,
      "step": 15515
    },
    {
      "epoch": 2.1746320953048355,
      "grad_norm": 0.22588637471199036,
      "learning_rate": 8.459698636689787e-05,
      "loss": 0.0415,
      "step": 15516
    },
    {
      "epoch": 2.174772249474422,
      "grad_norm": 0.29579490423202515,
      "learning_rate": 8.45826357330782e-05,
      "loss": 0.0557,
      "step": 15517
    },
    {
      "epoch": 2.1749124036440084,
      "grad_norm": 0.30834218859672546,
      "learning_rate": 8.456828509925854e-05,
      "loss": 0.0173,
      "step": 15518
    },
    {
      "epoch": 2.175052557813595,
      "grad_norm": 0.730586588382721,
      "learning_rate": 8.455393446543888e-05,
      "loss": 0.0328,
      "step": 15519
    },
    {
      "epoch": 2.1751927119831813,
      "grad_norm": 0.08423329889774323,
      "learning_rate": 8.453958383161922e-05,
      "loss": 0.0047,
      "step": 15520
    },
    {
      "epoch": 2.175332866152768,
      "grad_norm": 0.3593636751174927,
      "learning_rate": 8.452523319779957e-05,
      "loss": 0.0284,
      "step": 15521
    },
    {
      "epoch": 2.1754730203223547,
      "grad_norm": 0.2500229775905609,
      "learning_rate": 8.451088256397991e-05,
      "loss": 0.0389,
      "step": 15522
    },
    {
      "epoch": 2.175613174491941,
      "grad_norm": 0.15188990533351898,
      "learning_rate": 8.449653193016024e-05,
      "loss": 0.0214,
      "step": 15523
    },
    {
      "epoch": 2.1757533286615276,
      "grad_norm": 0.09120609611272812,
      "learning_rate": 8.448218129634058e-05,
      "loss": 0.0139,
      "step": 15524
    },
    {
      "epoch": 2.175893482831114,
      "grad_norm": 0.22486819326877594,
      "learning_rate": 8.446783066252091e-05,
      "loss": 0.0278,
      "step": 15525
    },
    {
      "epoch": 2.176033637000701,
      "grad_norm": 0.0872679054737091,
      "learning_rate": 8.445348002870126e-05,
      "loss": 0.01,
      "step": 15526
    },
    {
      "epoch": 2.1761737911702874,
      "grad_norm": 0.09633751958608627,
      "learning_rate": 8.44391293948816e-05,
      "loss": 0.0246,
      "step": 15527
    },
    {
      "epoch": 2.176313945339874,
      "grad_norm": 0.11577687412500381,
      "learning_rate": 8.442477876106193e-05,
      "loss": 0.0062,
      "step": 15528
    },
    {
      "epoch": 2.1764540995094603,
      "grad_norm": 0.10755651444196701,
      "learning_rate": 8.441042812724228e-05,
      "loss": 0.0137,
      "step": 15529
    },
    {
      "epoch": 2.176594253679047,
      "grad_norm": 0.38401877880096436,
      "learning_rate": 8.439607749342263e-05,
      "loss": 0.0142,
      "step": 15530
    },
    {
      "epoch": 2.1767344078486337,
      "grad_norm": 0.07044266909360886,
      "learning_rate": 8.438172685960296e-05,
      "loss": 0.0125,
      "step": 15531
    },
    {
      "epoch": 2.17687456201822,
      "grad_norm": 0.40811052918434143,
      "learning_rate": 8.43673762257833e-05,
      "loss": 0.0388,
      "step": 15532
    },
    {
      "epoch": 2.1770147161878066,
      "grad_norm": 0.21889354288578033,
      "learning_rate": 8.435302559196363e-05,
      "loss": 0.0526,
      "step": 15533
    },
    {
      "epoch": 2.177154870357393,
      "grad_norm": 0.21658968925476074,
      "learning_rate": 8.433867495814397e-05,
      "loss": 0.0376,
      "step": 15534
    },
    {
      "epoch": 2.1772950245269795,
      "grad_norm": 0.06853201240301132,
      "learning_rate": 8.432432432432432e-05,
      "loss": 0.0059,
      "step": 15535
    },
    {
      "epoch": 2.177435178696566,
      "grad_norm": 0.28299546241760254,
      "learning_rate": 8.430997369050466e-05,
      "loss": 0.0249,
      "step": 15536
    },
    {
      "epoch": 2.177575332866153,
      "grad_norm": 0.1991221308708191,
      "learning_rate": 8.4295623056685e-05,
      "loss": 0.0798,
      "step": 15537
    },
    {
      "epoch": 2.1777154870357394,
      "grad_norm": 0.22990986704826355,
      "learning_rate": 8.428127242286534e-05,
      "loss": 0.0081,
      "step": 15538
    },
    {
      "epoch": 2.177855641205326,
      "grad_norm": 0.11797018349170685,
      "learning_rate": 8.426692178904567e-05,
      "loss": 0.0296,
      "step": 15539
    },
    {
      "epoch": 2.1779957953749123,
      "grad_norm": 0.11003970354795456,
      "learning_rate": 8.425257115522601e-05,
      "loss": 0.0145,
      "step": 15540
    },
    {
      "epoch": 2.178135949544499,
      "grad_norm": 0.05072799697518349,
      "learning_rate": 8.423822052140634e-05,
      "loss": 0.0028,
      "step": 15541
    },
    {
      "epoch": 2.1782761037140856,
      "grad_norm": 0.3402263820171356,
      "learning_rate": 8.42238698875867e-05,
      "loss": 0.0314,
      "step": 15542
    },
    {
      "epoch": 2.178416257883672,
      "grad_norm": 0.22324246168136597,
      "learning_rate": 8.420951925376704e-05,
      "loss": 0.0286,
      "step": 15543
    },
    {
      "epoch": 2.1785564120532586,
      "grad_norm": 0.27035585045814514,
      "learning_rate": 8.419516861994737e-05,
      "loss": 0.0195,
      "step": 15544
    },
    {
      "epoch": 2.178696566222845,
      "grad_norm": 0.2956833839416504,
      "learning_rate": 8.418081798612771e-05,
      "loss": 0.034,
      "step": 15545
    },
    {
      "epoch": 2.1788367203924315,
      "grad_norm": 0.23110932111740112,
      "learning_rate": 8.416646735230806e-05,
      "loss": 0.026,
      "step": 15546
    },
    {
      "epoch": 2.1789768745620184,
      "grad_norm": 0.15697066485881805,
      "learning_rate": 8.41521167184884e-05,
      "loss": 0.0132,
      "step": 15547
    },
    {
      "epoch": 2.179117028731605,
      "grad_norm": 0.25031179189682007,
      "learning_rate": 8.413776608466874e-05,
      "loss": 0.0132,
      "step": 15548
    },
    {
      "epoch": 2.1792571829011913,
      "grad_norm": 0.6249184012413025,
      "learning_rate": 8.412341545084908e-05,
      "loss": 0.0316,
      "step": 15549
    },
    {
      "epoch": 2.1793973370707778,
      "grad_norm": 0.1307736337184906,
      "learning_rate": 8.41090648170294e-05,
      "loss": 0.0211,
      "step": 15550
    },
    {
      "epoch": 2.179537491240364,
      "grad_norm": 0.3390008509159088,
      "learning_rate": 8.409471418320975e-05,
      "loss": 0.0416,
      "step": 15551
    },
    {
      "epoch": 2.179677645409951,
      "grad_norm": 0.6477555632591248,
      "learning_rate": 8.408036354939009e-05,
      "loss": 0.034,
      "step": 15552
    },
    {
      "epoch": 2.1798177995795376,
      "grad_norm": 0.6356512308120728,
      "learning_rate": 8.406601291557043e-05,
      "loss": 0.0368,
      "step": 15553
    },
    {
      "epoch": 2.179957953749124,
      "grad_norm": 0.22738708555698395,
      "learning_rate": 8.405166228175077e-05,
      "loss": 0.0224,
      "step": 15554
    },
    {
      "epoch": 2.1800981079187105,
      "grad_norm": 0.05572545900940895,
      "learning_rate": 8.40373116479311e-05,
      "loss": 0.0026,
      "step": 15555
    },
    {
      "epoch": 2.180238262088297,
      "grad_norm": 0.11022742092609406,
      "learning_rate": 8.402296101411145e-05,
      "loss": 0.0113,
      "step": 15556
    },
    {
      "epoch": 2.180378416257884,
      "grad_norm": 0.32729431986808777,
      "learning_rate": 8.40086103802918e-05,
      "loss": 0.0207,
      "step": 15557
    },
    {
      "epoch": 2.1805185704274703,
      "grad_norm": 0.16148781776428223,
      "learning_rate": 8.399425974647213e-05,
      "loss": 0.0285,
      "step": 15558
    },
    {
      "epoch": 2.180658724597057,
      "grad_norm": 0.14963307976722717,
      "learning_rate": 8.397990911265247e-05,
      "loss": 0.0314,
      "step": 15559
    },
    {
      "epoch": 2.1807988787666432,
      "grad_norm": 0.23668617010116577,
      "learning_rate": 8.39655584788328e-05,
      "loss": 0.05,
      "step": 15560
    },
    {
      "epoch": 2.1809390329362297,
      "grad_norm": 1.4304167032241821,
      "learning_rate": 8.395120784501314e-05,
      "loss": 0.1111,
      "step": 15561
    },
    {
      "epoch": 2.1810791871058166,
      "grad_norm": 0.5803403854370117,
      "learning_rate": 8.39368572111935e-05,
      "loss": 0.0838,
      "step": 15562
    },
    {
      "epoch": 2.181219341275403,
      "grad_norm": 0.42928269505500793,
      "learning_rate": 8.392250657737383e-05,
      "loss": 0.0095,
      "step": 15563
    },
    {
      "epoch": 2.1813594954449895,
      "grad_norm": 0.08319581300020218,
      "learning_rate": 8.390815594355417e-05,
      "loss": 0.0033,
      "step": 15564
    },
    {
      "epoch": 2.181499649614576,
      "grad_norm": 0.23862707614898682,
      "learning_rate": 8.389380530973451e-05,
      "loss": 0.0365,
      "step": 15565
    },
    {
      "epoch": 2.1816398037841624,
      "grad_norm": 2.4348604679107666,
      "learning_rate": 8.387945467591484e-05,
      "loss": 0.0827,
      "step": 15566
    },
    {
      "epoch": 2.181779957953749,
      "grad_norm": 0.10237549245357513,
      "learning_rate": 8.386510404209518e-05,
      "loss": 0.0042,
      "step": 15567
    },
    {
      "epoch": 2.181920112123336,
      "grad_norm": 0.39557740092277527,
      "learning_rate": 8.385075340827552e-05,
      "loss": 0.0294,
      "step": 15568
    },
    {
      "epoch": 2.1820602662929223,
      "grad_norm": 1.4514493942260742,
      "learning_rate": 8.383640277445587e-05,
      "loss": 0.2599,
      "step": 15569
    },
    {
      "epoch": 2.1822004204625087,
      "grad_norm": 0.22535893321037292,
      "learning_rate": 8.382205214063621e-05,
      "loss": 0.0943,
      "step": 15570
    },
    {
      "epoch": 2.182340574632095,
      "grad_norm": 0.20261456072330475,
      "learning_rate": 8.380770150681654e-05,
      "loss": 0.0234,
      "step": 15571
    },
    {
      "epoch": 2.1824807288016816,
      "grad_norm": 0.1964482069015503,
      "learning_rate": 8.379335087299688e-05,
      "loss": 0.0082,
      "step": 15572
    },
    {
      "epoch": 2.1826208829712685,
      "grad_norm": 0.18574251234531403,
      "learning_rate": 8.377900023917723e-05,
      "loss": 0.0169,
      "step": 15573
    },
    {
      "epoch": 2.182761037140855,
      "grad_norm": 0.1273166686296463,
      "learning_rate": 8.376464960535756e-05,
      "loss": 0.0182,
      "step": 15574
    },
    {
      "epoch": 2.1829011913104415,
      "grad_norm": 0.2799573540687561,
      "learning_rate": 8.37502989715379e-05,
      "loss": 0.0314,
      "step": 15575
    },
    {
      "epoch": 2.183041345480028,
      "grad_norm": 0.29510653018951416,
      "learning_rate": 8.373594833771823e-05,
      "loss": 0.0646,
      "step": 15576
    },
    {
      "epoch": 2.1831814996496144,
      "grad_norm": 0.21985645592212677,
      "learning_rate": 8.372159770389858e-05,
      "loss": 0.0562,
      "step": 15577
    },
    {
      "epoch": 2.1833216538192013,
      "grad_norm": 0.1172732561826706,
      "learning_rate": 8.370724707007893e-05,
      "loss": 0.011,
      "step": 15578
    },
    {
      "epoch": 2.1834618079887878,
      "grad_norm": 0.23802527785301208,
      "learning_rate": 8.369289643625926e-05,
      "loss": 0.0518,
      "step": 15579
    },
    {
      "epoch": 2.183601962158374,
      "grad_norm": 0.12701694667339325,
      "learning_rate": 8.36785458024396e-05,
      "loss": 0.023,
      "step": 15580
    },
    {
      "epoch": 2.1837421163279607,
      "grad_norm": 0.3812354803085327,
      "learning_rate": 8.366419516861994e-05,
      "loss": 0.0321,
      "step": 15581
    },
    {
      "epoch": 2.183882270497547,
      "grad_norm": 0.3682289719581604,
      "learning_rate": 8.364984453480027e-05,
      "loss": 0.0544,
      "step": 15582
    },
    {
      "epoch": 2.184022424667134,
      "grad_norm": 0.3134566843509674,
      "learning_rate": 8.363549390098061e-05,
      "loss": 0.0543,
      "step": 15583
    },
    {
      "epoch": 2.1841625788367205,
      "grad_norm": 0.14092233777046204,
      "learning_rate": 8.362114326716097e-05,
      "loss": 0.0174,
      "step": 15584
    },
    {
      "epoch": 2.184302733006307,
      "grad_norm": 0.11824993044137955,
      "learning_rate": 8.36067926333413e-05,
      "loss": 0.0128,
      "step": 15585
    },
    {
      "epoch": 2.1844428871758934,
      "grad_norm": 0.190822571516037,
      "learning_rate": 8.359244199952164e-05,
      "loss": 0.0436,
      "step": 15586
    },
    {
      "epoch": 2.18458304134548,
      "grad_norm": 0.05926097184419632,
      "learning_rate": 8.357809136570197e-05,
      "loss": 0.0093,
      "step": 15587
    },
    {
      "epoch": 2.1847231955150668,
      "grad_norm": 0.16819730401039124,
      "learning_rate": 8.356374073188231e-05,
      "loss": 0.0459,
      "step": 15588
    },
    {
      "epoch": 2.1848633496846532,
      "grad_norm": 1.3574039936065674,
      "learning_rate": 8.354939009806267e-05,
      "loss": 0.0848,
      "step": 15589
    },
    {
      "epoch": 2.1850035038542397,
      "grad_norm": 0.20953278243541718,
      "learning_rate": 8.3535039464243e-05,
      "loss": 0.0247,
      "step": 15590
    },
    {
      "epoch": 2.185143658023826,
      "grad_norm": 0.1937340348958969,
      "learning_rate": 8.352068883042334e-05,
      "loss": 0.0201,
      "step": 15591
    },
    {
      "epoch": 2.1852838121934126,
      "grad_norm": 0.31552377343177795,
      "learning_rate": 8.350633819660368e-05,
      "loss": 0.0179,
      "step": 15592
    },
    {
      "epoch": 2.1854239663629995,
      "grad_norm": 0.1680138260126114,
      "learning_rate": 8.349198756278401e-05,
      "loss": 0.0064,
      "step": 15593
    },
    {
      "epoch": 2.185564120532586,
      "grad_norm": 0.30566340684890747,
      "learning_rate": 8.347763692896436e-05,
      "loss": 0.0245,
      "step": 15594
    },
    {
      "epoch": 2.1857042747021724,
      "grad_norm": 0.18651804327964783,
      "learning_rate": 8.346328629514469e-05,
      "loss": 0.0393,
      "step": 15595
    },
    {
      "epoch": 2.185844428871759,
      "grad_norm": 0.24722754955291748,
      "learning_rate": 8.344893566132503e-05,
      "loss": 0.0273,
      "step": 15596
    },
    {
      "epoch": 2.1859845830413454,
      "grad_norm": 0.048305630683898926,
      "learning_rate": 8.343458502750538e-05,
      "loss": 0.0057,
      "step": 15597
    },
    {
      "epoch": 2.186124737210932,
      "grad_norm": 0.17607168853282928,
      "learning_rate": 8.34202343936857e-05,
      "loss": 0.032,
      "step": 15598
    },
    {
      "epoch": 2.1862648913805187,
      "grad_norm": 0.2770044505596161,
      "learning_rate": 8.340588375986606e-05,
      "loss": 0.0181,
      "step": 15599
    },
    {
      "epoch": 2.186405045550105,
      "grad_norm": 0.13178180158138275,
      "learning_rate": 8.33915331260464e-05,
      "loss": 0.008,
      "step": 15600
    },
    {
      "epoch": 2.1865451997196916,
      "grad_norm": 0.20559416711330414,
      "learning_rate": 8.337718249222673e-05,
      "loss": 0.0358,
      "step": 15601
    },
    {
      "epoch": 2.186685353889278,
      "grad_norm": 0.08401753008365631,
      "learning_rate": 8.336283185840707e-05,
      "loss": 0.0064,
      "step": 15602
    },
    {
      "epoch": 2.1868255080588646,
      "grad_norm": 0.39797934889793396,
      "learning_rate": 8.33484812245874e-05,
      "loss": 0.0704,
      "step": 15603
    },
    {
      "epoch": 2.1869656622284515,
      "grad_norm": 0.3423841595649719,
      "learning_rate": 8.333413059076774e-05,
      "loss": 0.0291,
      "step": 15604
    },
    {
      "epoch": 2.187105816398038,
      "grad_norm": 0.4315163195133209,
      "learning_rate": 8.33197799569481e-05,
      "loss": 0.0302,
      "step": 15605
    },
    {
      "epoch": 2.1872459705676244,
      "grad_norm": 0.12608914077281952,
      "learning_rate": 8.330542932312843e-05,
      "loss": 0.0153,
      "step": 15606
    },
    {
      "epoch": 2.187386124737211,
      "grad_norm": 0.1963818520307541,
      "learning_rate": 8.329107868930877e-05,
      "loss": 0.0285,
      "step": 15607
    },
    {
      "epoch": 2.1875262789067973,
      "grad_norm": 0.3715837001800537,
      "learning_rate": 8.327672805548911e-05,
      "loss": 0.0143,
      "step": 15608
    },
    {
      "epoch": 2.187666433076384,
      "grad_norm": 0.1704009473323822,
      "learning_rate": 8.326237742166944e-05,
      "loss": 0.0239,
      "step": 15609
    },
    {
      "epoch": 2.1878065872459707,
      "grad_norm": 0.1322927176952362,
      "learning_rate": 8.32480267878498e-05,
      "loss": 0.014,
      "step": 15610
    },
    {
      "epoch": 2.187946741415557,
      "grad_norm": 0.041757725179195404,
      "learning_rate": 8.323367615403013e-05,
      "loss": 0.0049,
      "step": 15611
    },
    {
      "epoch": 2.1880868955851436,
      "grad_norm": 0.1920960247516632,
      "learning_rate": 8.321932552021047e-05,
      "loss": 0.0141,
      "step": 15612
    },
    {
      "epoch": 2.18822704975473,
      "grad_norm": 0.39365094900131226,
      "learning_rate": 8.320497488639081e-05,
      "loss": 0.0131,
      "step": 15613
    },
    {
      "epoch": 2.188367203924317,
      "grad_norm": 0.610433042049408,
      "learning_rate": 8.319062425257114e-05,
      "loss": 0.0352,
      "step": 15614
    },
    {
      "epoch": 2.1885073580939034,
      "grad_norm": 0.1152632012963295,
      "learning_rate": 8.31762736187515e-05,
      "loss": 0.0087,
      "step": 15615
    },
    {
      "epoch": 2.18864751226349,
      "grad_norm": 0.40525224804878235,
      "learning_rate": 8.316192298493184e-05,
      "loss": 0.0312,
      "step": 15616
    },
    {
      "epoch": 2.1887876664330763,
      "grad_norm": 0.3027915060520172,
      "learning_rate": 8.314757235111217e-05,
      "loss": 0.0117,
      "step": 15617
    },
    {
      "epoch": 2.1889278206026628,
      "grad_norm": 0.38697054982185364,
      "learning_rate": 8.313322171729251e-05,
      "loss": 0.1625,
      "step": 15618
    },
    {
      "epoch": 2.1890679747722497,
      "grad_norm": 0.5098830461502075,
      "learning_rate": 8.311887108347285e-05,
      "loss": 0.0333,
      "step": 15619
    },
    {
      "epoch": 2.189208128941836,
      "grad_norm": 0.20754776895046234,
      "learning_rate": 8.310452044965318e-05,
      "loss": 0.0043,
      "step": 15620
    },
    {
      "epoch": 2.1893482831114226,
      "grad_norm": 0.4233125150203705,
      "learning_rate": 8.309016981583353e-05,
      "loss": 0.0401,
      "step": 15621
    },
    {
      "epoch": 2.189488437281009,
      "grad_norm": 0.5059199333190918,
      "learning_rate": 8.307581918201386e-05,
      "loss": 0.0302,
      "step": 15622
    },
    {
      "epoch": 2.1896285914505955,
      "grad_norm": 0.22175323963165283,
      "learning_rate": 8.30614685481942e-05,
      "loss": 0.0175,
      "step": 15623
    },
    {
      "epoch": 2.1897687456201824,
      "grad_norm": 0.2992059290409088,
      "learning_rate": 8.304711791437455e-05,
      "loss": 0.0599,
      "step": 15624
    },
    {
      "epoch": 2.189908899789769,
      "grad_norm": 0.08164261281490326,
      "learning_rate": 8.303276728055487e-05,
      "loss": 0.0108,
      "step": 15625
    },
    {
      "epoch": 2.1900490539593553,
      "grad_norm": 0.14064110815525055,
      "learning_rate": 8.301841664673523e-05,
      "loss": 0.0245,
      "step": 15626
    },
    {
      "epoch": 2.190189208128942,
      "grad_norm": 0.04325925558805466,
      "learning_rate": 8.300406601291557e-05,
      "loss": 0.0034,
      "step": 15627
    },
    {
      "epoch": 2.1903293622985283,
      "grad_norm": 0.3913405239582062,
      "learning_rate": 8.29897153790959e-05,
      "loss": 0.0661,
      "step": 15628
    },
    {
      "epoch": 2.1904695164681147,
      "grad_norm": 0.172748863697052,
      "learning_rate": 8.297536474527624e-05,
      "loss": 0.0116,
      "step": 15629
    },
    {
      "epoch": 2.1906096706377016,
      "grad_norm": 0.30456429719924927,
      "learning_rate": 8.296101411145657e-05,
      "loss": 0.0753,
      "step": 15630
    },
    {
      "epoch": 2.190749824807288,
      "grad_norm": 0.09865356236696243,
      "learning_rate": 8.294666347763693e-05,
      "loss": 0.0086,
      "step": 15631
    },
    {
      "epoch": 2.1908899789768745,
      "grad_norm": 0.13714629411697388,
      "learning_rate": 8.293231284381727e-05,
      "loss": 0.0131,
      "step": 15632
    },
    {
      "epoch": 2.191030133146461,
      "grad_norm": 0.3552764356136322,
      "learning_rate": 8.29179622099976e-05,
      "loss": 0.0402,
      "step": 15633
    },
    {
      "epoch": 2.1911702873160475,
      "grad_norm": 0.32466045022010803,
      "learning_rate": 8.290361157617794e-05,
      "loss": 0.0261,
      "step": 15634
    },
    {
      "epoch": 2.1913104414856344,
      "grad_norm": 0.2198207676410675,
      "learning_rate": 8.288926094235828e-05,
      "loss": 0.0158,
      "step": 15635
    },
    {
      "epoch": 2.191450595655221,
      "grad_norm": 0.22316282987594604,
      "learning_rate": 8.287491030853861e-05,
      "loss": 0.0305,
      "step": 15636
    },
    {
      "epoch": 2.1915907498248073,
      "grad_norm": 0.21982179582118988,
      "learning_rate": 8.286055967471897e-05,
      "loss": 0.0502,
      "step": 15637
    },
    {
      "epoch": 2.1917309039943937,
      "grad_norm": 0.4248678386211395,
      "learning_rate": 8.28462090408993e-05,
      "loss": 0.0421,
      "step": 15638
    },
    {
      "epoch": 2.19187105816398,
      "grad_norm": 0.10520629584789276,
      "learning_rate": 8.283185840707964e-05,
      "loss": 0.0035,
      "step": 15639
    },
    {
      "epoch": 2.192011212333567,
      "grad_norm": 0.08738252520561218,
      "learning_rate": 8.281750777325998e-05,
      "loss": 0.0077,
      "step": 15640
    },
    {
      "epoch": 2.1921513665031536,
      "grad_norm": 0.2180115133523941,
      "learning_rate": 8.280315713944031e-05,
      "loss": 0.0336,
      "step": 15641
    },
    {
      "epoch": 2.19229152067274,
      "grad_norm": 0.1201363056898117,
      "learning_rate": 8.278880650562066e-05,
      "loss": 0.0144,
      "step": 15642
    },
    {
      "epoch": 2.1924316748423265,
      "grad_norm": 0.1581735908985138,
      "learning_rate": 8.2774455871801e-05,
      "loss": 0.0251,
      "step": 15643
    },
    {
      "epoch": 2.192571829011913,
      "grad_norm": 0.24820972979068756,
      "learning_rate": 8.276010523798133e-05,
      "loss": 0.0229,
      "step": 15644
    },
    {
      "epoch": 2.1927119831815,
      "grad_norm": 0.5092044472694397,
      "learning_rate": 8.274575460416168e-05,
      "loss": 0.1268,
      "step": 15645
    },
    {
      "epoch": 2.1928521373510863,
      "grad_norm": 0.39710351824760437,
      "learning_rate": 8.2731403970342e-05,
      "loss": 0.0382,
      "step": 15646
    },
    {
      "epoch": 2.1929922915206728,
      "grad_norm": 0.08613835275173187,
      "learning_rate": 8.271705333652236e-05,
      "loss": 0.0091,
      "step": 15647
    },
    {
      "epoch": 2.1931324456902592,
      "grad_norm": 0.08021494746208191,
      "learning_rate": 8.27027027027027e-05,
      "loss": 0.0249,
      "step": 15648
    },
    {
      "epoch": 2.1932725998598457,
      "grad_norm": 0.08379749208688736,
      "learning_rate": 8.268835206888303e-05,
      "loss": 0.0048,
      "step": 15649
    },
    {
      "epoch": 2.193412754029432,
      "grad_norm": 0.35445496439933777,
      "learning_rate": 8.267400143506337e-05,
      "loss": 0.0933,
      "step": 15650
    },
    {
      "epoch": 2.193552908199019,
      "grad_norm": 0.1973295956850052,
      "learning_rate": 8.265965080124372e-05,
      "loss": 0.0758,
      "step": 15651
    },
    {
      "epoch": 2.1936930623686055,
      "grad_norm": 0.3184235095977783,
      "learning_rate": 8.264530016742404e-05,
      "loss": 0.0551,
      "step": 15652
    },
    {
      "epoch": 2.193833216538192,
      "grad_norm": 0.6656478643417358,
      "learning_rate": 8.26309495336044e-05,
      "loss": 0.0574,
      "step": 15653
    },
    {
      "epoch": 2.1939733707077784,
      "grad_norm": 0.2157331109046936,
      "learning_rate": 8.261659889978473e-05,
      "loss": 0.0266,
      "step": 15654
    },
    {
      "epoch": 2.1941135248773653,
      "grad_norm": 0.18208567798137665,
      "learning_rate": 8.260224826596507e-05,
      "loss": 0.0152,
      "step": 15655
    },
    {
      "epoch": 2.194253679046952,
      "grad_norm": 0.3098605275154114,
      "learning_rate": 8.258789763214541e-05,
      "loss": 0.0664,
      "step": 15656
    },
    {
      "epoch": 2.1943938332165382,
      "grad_norm": 0.14590348303318024,
      "learning_rate": 8.257354699832574e-05,
      "loss": 0.0181,
      "step": 15657
    },
    {
      "epoch": 2.1945339873861247,
      "grad_norm": 0.16141259670257568,
      "learning_rate": 8.25591963645061e-05,
      "loss": 0.036,
      "step": 15658
    },
    {
      "epoch": 2.194674141555711,
      "grad_norm": 0.8787152171134949,
      "learning_rate": 8.254484573068644e-05,
      "loss": 0.0519,
      "step": 15659
    },
    {
      "epoch": 2.1948142957252976,
      "grad_norm": 0.10097132623195648,
      "learning_rate": 8.253049509686677e-05,
      "loss": 0.0114,
      "step": 15660
    },
    {
      "epoch": 2.1949544498948845,
      "grad_norm": 1.190528392791748,
      "learning_rate": 8.251614446304711e-05,
      "loss": 0.0448,
      "step": 15661
    },
    {
      "epoch": 2.195094604064471,
      "grad_norm": 0.12969990074634552,
      "learning_rate": 8.250179382922745e-05,
      "loss": 0.0046,
      "step": 15662
    },
    {
      "epoch": 2.1952347582340574,
      "grad_norm": 0.4736093580722809,
      "learning_rate": 8.24874431954078e-05,
      "loss": 0.0601,
      "step": 15663
    },
    {
      "epoch": 2.195374912403644,
      "grad_norm": 0.2785087525844574,
      "learning_rate": 8.247309256158814e-05,
      "loss": 0.0082,
      "step": 15664
    },
    {
      "epoch": 2.1955150665732304,
      "grad_norm": 0.2481694370508194,
      "learning_rate": 8.245874192776846e-05,
      "loss": 0.0149,
      "step": 15665
    },
    {
      "epoch": 2.1956552207428173,
      "grad_norm": 1.168166995048523,
      "learning_rate": 8.24443912939488e-05,
      "loss": 0.0847,
      "step": 15666
    },
    {
      "epoch": 2.1957953749124037,
      "grad_norm": 0.646602213382721,
      "learning_rate": 8.243004066012915e-05,
      "loss": 0.0665,
      "step": 15667
    },
    {
      "epoch": 2.19593552908199,
      "grad_norm": 0.4012528657913208,
      "learning_rate": 8.241569002630948e-05,
      "loss": 0.0777,
      "step": 15668
    },
    {
      "epoch": 2.1960756832515766,
      "grad_norm": 0.6125938892364502,
      "learning_rate": 8.240133939248983e-05,
      "loss": 0.1572,
      "step": 15669
    },
    {
      "epoch": 2.196215837421163,
      "grad_norm": 4.760125637054443,
      "learning_rate": 8.238698875867018e-05,
      "loss": 0.1638,
      "step": 15670
    },
    {
      "epoch": 2.19635599159075,
      "grad_norm": 0.1384824812412262,
      "learning_rate": 8.23726381248505e-05,
      "loss": 0.0214,
      "step": 15671
    },
    {
      "epoch": 2.1964961457603365,
      "grad_norm": 0.276643842458725,
      "learning_rate": 8.235828749103085e-05,
      "loss": 0.0246,
      "step": 15672
    },
    {
      "epoch": 2.196636299929923,
      "grad_norm": 0.3067857325077057,
      "learning_rate": 8.234393685721117e-05,
      "loss": 0.0462,
      "step": 15673
    },
    {
      "epoch": 2.1967764540995094,
      "grad_norm": 0.2647033929824829,
      "learning_rate": 8.232958622339153e-05,
      "loss": 0.02,
      "step": 15674
    },
    {
      "epoch": 2.196916608269096,
      "grad_norm": 0.08739818632602692,
      "learning_rate": 8.231523558957187e-05,
      "loss": 0.0128,
      "step": 15675
    },
    {
      "epoch": 2.1970567624386828,
      "grad_norm": 0.40206077694892883,
      "learning_rate": 8.23008849557522e-05,
      "loss": 0.0448,
      "step": 15676
    },
    {
      "epoch": 2.197196916608269,
      "grad_norm": 0.1920454055070877,
      "learning_rate": 8.228653432193254e-05,
      "loss": 0.0423,
      "step": 15677
    },
    {
      "epoch": 2.1973370707778557,
      "grad_norm": 0.3084172308444977,
      "learning_rate": 8.227218368811288e-05,
      "loss": 0.036,
      "step": 15678
    },
    {
      "epoch": 2.197477224947442,
      "grad_norm": 0.27525025606155396,
      "learning_rate": 8.225783305429323e-05,
      "loss": 0.0446,
      "step": 15679
    },
    {
      "epoch": 2.1976173791170286,
      "grad_norm": 0.12222567945718765,
      "learning_rate": 8.224348242047357e-05,
      "loss": 0.0378,
      "step": 15680
    },
    {
      "epoch": 2.197757533286615,
      "grad_norm": 0.46290692687034607,
      "learning_rate": 8.22291317866539e-05,
      "loss": 0.0686,
      "step": 15681
    },
    {
      "epoch": 2.197897687456202,
      "grad_norm": 0.34350353479385376,
      "learning_rate": 8.221478115283424e-05,
      "loss": 0.0634,
      "step": 15682
    },
    {
      "epoch": 2.1980378416257884,
      "grad_norm": 0.07873933762311935,
      "learning_rate": 8.220043051901458e-05,
      "loss": 0.0123,
      "step": 15683
    },
    {
      "epoch": 2.198177995795375,
      "grad_norm": 0.6873977184295654,
      "learning_rate": 8.218607988519491e-05,
      "loss": 0.0334,
      "step": 15684
    },
    {
      "epoch": 2.1983181499649613,
      "grad_norm": 0.25200486183166504,
      "learning_rate": 8.217172925137527e-05,
      "loss": 0.0489,
      "step": 15685
    },
    {
      "epoch": 2.1984583041345482,
      "grad_norm": 0.10532114654779434,
      "learning_rate": 8.215737861755561e-05,
      "loss": 0.0134,
      "step": 15686
    },
    {
      "epoch": 2.1985984583041347,
      "grad_norm": 0.2552756369113922,
      "learning_rate": 8.214302798373594e-05,
      "loss": 0.0315,
      "step": 15687
    },
    {
      "epoch": 2.198738612473721,
      "grad_norm": 0.1742653101682663,
      "learning_rate": 8.212867734991628e-05,
      "loss": 0.0186,
      "step": 15688
    },
    {
      "epoch": 2.1988787666433076,
      "grad_norm": 0.3285054564476013,
      "learning_rate": 8.211432671609661e-05,
      "loss": 0.0207,
      "step": 15689
    },
    {
      "epoch": 2.199018920812894,
      "grad_norm": 0.2602458894252777,
      "learning_rate": 8.209997608227696e-05,
      "loss": 0.0257,
      "step": 15690
    },
    {
      "epoch": 2.1991590749824805,
      "grad_norm": 0.06654641032218933,
      "learning_rate": 8.20856254484573e-05,
      "loss": 0.008,
      "step": 15691
    },
    {
      "epoch": 2.1992992291520674,
      "grad_norm": 0.1489337831735611,
      "learning_rate": 8.207127481463763e-05,
      "loss": 0.0259,
      "step": 15692
    },
    {
      "epoch": 2.199439383321654,
      "grad_norm": 0.19741222262382507,
      "learning_rate": 8.205692418081798e-05,
      "loss": 0.0253,
      "step": 15693
    },
    {
      "epoch": 2.1995795374912404,
      "grad_norm": 0.28719857335090637,
      "learning_rate": 8.204257354699832e-05,
      "loss": 0.0116,
      "step": 15694
    },
    {
      "epoch": 2.199719691660827,
      "grad_norm": 0.548724353313446,
      "learning_rate": 8.202822291317866e-05,
      "loss": 0.0288,
      "step": 15695
    },
    {
      "epoch": 2.1998598458304133,
      "grad_norm": 0.17500430345535278,
      "learning_rate": 8.2013872279359e-05,
      "loss": 0.0143,
      "step": 15696
    },
    {
      "epoch": 2.2,
      "grad_norm": 0.21776443719863892,
      "learning_rate": 8.199952164553934e-05,
      "loss": 0.1035,
      "step": 15697
    },
    {
      "epoch": 2.2001401541695866,
      "grad_norm": 0.3577345609664917,
      "learning_rate": 8.198517101171967e-05,
      "loss": 0.0669,
      "step": 15698
    },
    {
      "epoch": 2.200280308339173,
      "grad_norm": 0.2582548260688782,
      "learning_rate": 8.197082037790002e-05,
      "loss": 0.0967,
      "step": 15699
    },
    {
      "epoch": 2.2004204625087596,
      "grad_norm": 0.2810274064540863,
      "learning_rate": 8.195646974408034e-05,
      "loss": 0.033,
      "step": 15700
    },
    {
      "epoch": 2.200560616678346,
      "grad_norm": 0.09076671302318573,
      "learning_rate": 8.19421191102607e-05,
      "loss": 0.0093,
      "step": 15701
    },
    {
      "epoch": 2.200700770847933,
      "grad_norm": 0.1197870597243309,
      "learning_rate": 8.192776847644104e-05,
      "loss": 0.0294,
      "step": 15702
    },
    {
      "epoch": 2.2008409250175194,
      "grad_norm": 0.5010340809822083,
      "learning_rate": 8.191341784262137e-05,
      "loss": 0.039,
      "step": 15703
    },
    {
      "epoch": 2.200981079187106,
      "grad_norm": 0.1756926327943802,
      "learning_rate": 8.189906720880171e-05,
      "loss": 0.0605,
      "step": 15704
    },
    {
      "epoch": 2.2011212333566923,
      "grad_norm": 0.20833875238895416,
      "learning_rate": 8.188471657498207e-05,
      "loss": 0.0113,
      "step": 15705
    },
    {
      "epoch": 2.2012613875262788,
      "grad_norm": 0.4840003550052643,
      "learning_rate": 8.18703659411624e-05,
      "loss": 0.0371,
      "step": 15706
    },
    {
      "epoch": 2.2014015416958657,
      "grad_norm": 0.3240581154823303,
      "learning_rate": 8.185601530734274e-05,
      "loss": 0.0395,
      "step": 15707
    },
    {
      "epoch": 2.201541695865452,
      "grad_norm": 0.12794287502765656,
      "learning_rate": 8.184166467352307e-05,
      "loss": 0.017,
      "step": 15708
    },
    {
      "epoch": 2.2016818500350386,
      "grad_norm": 0.7090755105018616,
      "learning_rate": 8.182731403970341e-05,
      "loss": 0.0775,
      "step": 15709
    },
    {
      "epoch": 2.201822004204625,
      "grad_norm": 0.13308900594711304,
      "learning_rate": 8.181296340588376e-05,
      "loss": 0.0218,
      "step": 15710
    },
    {
      "epoch": 2.2019621583742115,
      "grad_norm": 0.15461736917495728,
      "learning_rate": 8.17986127720641e-05,
      "loss": 0.0443,
      "step": 15711
    },
    {
      "epoch": 2.202102312543798,
      "grad_norm": 0.13474911451339722,
      "learning_rate": 8.178426213824444e-05,
      "loss": 0.0191,
      "step": 15712
    },
    {
      "epoch": 2.202242466713385,
      "grad_norm": 0.293152391910553,
      "learning_rate": 8.176991150442478e-05,
      "loss": 0.0422,
      "step": 15713
    },
    {
      "epoch": 2.2023826208829713,
      "grad_norm": 1.1336601972579956,
      "learning_rate": 8.17555608706051e-05,
      "loss": 0.0359,
      "step": 15714
    },
    {
      "epoch": 2.202522775052558,
      "grad_norm": 0.024906359612941742,
      "learning_rate": 8.174121023678545e-05,
      "loss": 0.0022,
      "step": 15715
    },
    {
      "epoch": 2.2026629292221442,
      "grad_norm": 0.12857244908809662,
      "learning_rate": 8.172685960296578e-05,
      "loss": 0.0075,
      "step": 15716
    },
    {
      "epoch": 2.2028030833917307,
      "grad_norm": 0.28347164392471313,
      "learning_rate": 8.171250896914613e-05,
      "loss": 0.0088,
      "step": 15717
    },
    {
      "epoch": 2.2029432375613176,
      "grad_norm": 0.19016562402248383,
      "learning_rate": 8.169815833532647e-05,
      "loss": 0.009,
      "step": 15718
    },
    {
      "epoch": 2.203083391730904,
      "grad_norm": 1.0748003721237183,
      "learning_rate": 8.16838077015068e-05,
      "loss": 0.113,
      "step": 15719
    },
    {
      "epoch": 2.2032235459004905,
      "grad_norm": 0.5496826171875,
      "learning_rate": 8.166945706768715e-05,
      "loss": 0.0278,
      "step": 15720
    },
    {
      "epoch": 2.203363700070077,
      "grad_norm": 0.12921671569347382,
      "learning_rate": 8.16551064338675e-05,
      "loss": 0.0136,
      "step": 15721
    },
    {
      "epoch": 2.2035038542396634,
      "grad_norm": 0.20940911769866943,
      "learning_rate": 8.164075580004783e-05,
      "loss": 0.0318,
      "step": 15722
    },
    {
      "epoch": 2.2036440084092503,
      "grad_norm": 0.21101483702659607,
      "learning_rate": 8.162640516622817e-05,
      "loss": 0.045,
      "step": 15723
    },
    {
      "epoch": 2.203784162578837,
      "grad_norm": 0.5724600553512573,
      "learning_rate": 8.16120545324085e-05,
      "loss": 0.0215,
      "step": 15724
    },
    {
      "epoch": 2.2039243167484233,
      "grad_norm": 0.07508917897939682,
      "learning_rate": 8.159770389858884e-05,
      "loss": 0.0055,
      "step": 15725
    },
    {
      "epoch": 2.2040644709180097,
      "grad_norm": 0.09718918055295944,
      "learning_rate": 8.15833532647692e-05,
      "loss": 0.0223,
      "step": 15726
    },
    {
      "epoch": 2.204204625087596,
      "grad_norm": 0.025699010118842125,
      "learning_rate": 8.156900263094953e-05,
      "loss": 0.0028,
      "step": 15727
    },
    {
      "epoch": 2.204344779257183,
      "grad_norm": 0.1400269865989685,
      "learning_rate": 8.155465199712987e-05,
      "loss": 0.021,
      "step": 15728
    },
    {
      "epoch": 2.2044849334267695,
      "grad_norm": 0.09716742485761642,
      "learning_rate": 8.154030136331021e-05,
      "loss": 0.0119,
      "step": 15729
    },
    {
      "epoch": 2.204625087596356,
      "grad_norm": 0.13582508265972137,
      "learning_rate": 8.152595072949054e-05,
      "loss": 0.0066,
      "step": 15730
    },
    {
      "epoch": 2.2047652417659425,
      "grad_norm": 0.34170815348625183,
      "learning_rate": 8.151160009567088e-05,
      "loss": 0.034,
      "step": 15731
    },
    {
      "epoch": 2.204905395935529,
      "grad_norm": 0.08761975914239883,
      "learning_rate": 8.149724946185124e-05,
      "loss": 0.006,
      "step": 15732
    },
    {
      "epoch": 2.205045550105116,
      "grad_norm": 0.06848367303609848,
      "learning_rate": 8.148289882803157e-05,
      "loss": 0.0084,
      "step": 15733
    },
    {
      "epoch": 2.2051857042747023,
      "grad_norm": 0.24451403319835663,
      "learning_rate": 8.146854819421191e-05,
      "loss": 0.0271,
      "step": 15734
    },
    {
      "epoch": 2.2053258584442887,
      "grad_norm": 0.1860520839691162,
      "learning_rate": 8.145419756039224e-05,
      "loss": 0.0252,
      "step": 15735
    },
    {
      "epoch": 2.205466012613875,
      "grad_norm": 0.0633758008480072,
      "learning_rate": 8.143984692657258e-05,
      "loss": 0.0064,
      "step": 15736
    },
    {
      "epoch": 2.2056061667834617,
      "grad_norm": 0.27101773023605347,
      "learning_rate": 8.142549629275293e-05,
      "loss": 0.0291,
      "step": 15737
    },
    {
      "epoch": 2.2057463209530486,
      "grad_norm": 0.1298302710056305,
      "learning_rate": 8.141114565893326e-05,
      "loss": 0.0087,
      "step": 15738
    },
    {
      "epoch": 2.205886475122635,
      "grad_norm": 0.21071410179138184,
      "learning_rate": 8.13967950251136e-05,
      "loss": 0.0222,
      "step": 15739
    },
    {
      "epoch": 2.2060266292922215,
      "grad_norm": 0.3418385982513428,
      "learning_rate": 8.138244439129395e-05,
      "loss": 0.0479,
      "step": 15740
    },
    {
      "epoch": 2.206166783461808,
      "grad_norm": 0.13273079693317413,
      "learning_rate": 8.136809375747428e-05,
      "loss": 0.023,
      "step": 15741
    },
    {
      "epoch": 2.2063069376313944,
      "grad_norm": 0.22827519476413727,
      "learning_rate": 8.135374312365463e-05,
      "loss": 0.0238,
      "step": 15742
    },
    {
      "epoch": 2.206447091800981,
      "grad_norm": 0.7551180720329285,
      "learning_rate": 8.133939248983496e-05,
      "loss": 0.0267,
      "step": 15743
    },
    {
      "epoch": 2.2065872459705678,
      "grad_norm": 0.14162230491638184,
      "learning_rate": 8.13250418560153e-05,
      "loss": 0.0214,
      "step": 15744
    },
    {
      "epoch": 2.2067274001401542,
      "grad_norm": 0.1592104136943817,
      "learning_rate": 8.131069122219564e-05,
      "loss": 0.006,
      "step": 15745
    },
    {
      "epoch": 2.2068675543097407,
      "grad_norm": 0.27113738656044006,
      "learning_rate": 8.129634058837597e-05,
      "loss": 0.0191,
      "step": 15746
    },
    {
      "epoch": 2.207007708479327,
      "grad_norm": 0.10886117815971375,
      "learning_rate": 8.128198995455631e-05,
      "loss": 0.0144,
      "step": 15747
    },
    {
      "epoch": 2.2071478626489136,
      "grad_norm": 0.09191807359457016,
      "learning_rate": 8.126763932073667e-05,
      "loss": 0.0119,
      "step": 15748
    },
    {
      "epoch": 2.2072880168185005,
      "grad_norm": 0.46607664227485657,
      "learning_rate": 8.1253288686917e-05,
      "loss": 0.0641,
      "step": 15749
    },
    {
      "epoch": 2.207428170988087,
      "grad_norm": 0.339548796415329,
      "learning_rate": 8.123893805309734e-05,
      "loss": 0.0773,
      "step": 15750
    },
    {
      "epoch": 2.2075683251576734,
      "grad_norm": 0.2656480073928833,
      "learning_rate": 8.122458741927767e-05,
      "loss": 0.0838,
      "step": 15751
    },
    {
      "epoch": 2.20770847932726,
      "grad_norm": 0.12845660746097565,
      "learning_rate": 8.121023678545801e-05,
      "loss": 0.0152,
      "step": 15752
    },
    {
      "epoch": 2.2078486334968463,
      "grad_norm": 0.2567709684371948,
      "learning_rate": 8.119588615163837e-05,
      "loss": 0.0149,
      "step": 15753
    },
    {
      "epoch": 2.2079887876664333,
      "grad_norm": 0.09469053894281387,
      "learning_rate": 8.11815355178187e-05,
      "loss": 0.0092,
      "step": 15754
    },
    {
      "epoch": 2.2081289418360197,
      "grad_norm": 0.16169095039367676,
      "learning_rate": 8.116718488399904e-05,
      "loss": 0.0151,
      "step": 15755
    },
    {
      "epoch": 2.208269096005606,
      "grad_norm": 0.11135906726121902,
      "learning_rate": 8.115283425017938e-05,
      "loss": 0.0162,
      "step": 15756
    },
    {
      "epoch": 2.2084092501751926,
      "grad_norm": 0.23499958217144012,
      "learning_rate": 8.113848361635971e-05,
      "loss": 0.0463,
      "step": 15757
    },
    {
      "epoch": 2.208549404344779,
      "grad_norm": 0.3057786822319031,
      "learning_rate": 8.112413298254006e-05,
      "loss": 0.0277,
      "step": 15758
    },
    {
      "epoch": 2.208689558514366,
      "grad_norm": 0.21030275523662567,
      "learning_rate": 8.110978234872039e-05,
      "loss": 0.0326,
      "step": 15759
    },
    {
      "epoch": 2.2088297126839525,
      "grad_norm": 0.20218226313591003,
      "learning_rate": 8.109543171490073e-05,
      "loss": 0.0202,
      "step": 15760
    },
    {
      "epoch": 2.208969866853539,
      "grad_norm": 0.15559740364551544,
      "learning_rate": 8.108108108108108e-05,
      "loss": 0.024,
      "step": 15761
    },
    {
      "epoch": 2.2091100210231254,
      "grad_norm": 0.3975464105606079,
      "learning_rate": 8.10667304472614e-05,
      "loss": 0.0312,
      "step": 15762
    },
    {
      "epoch": 2.209250175192712,
      "grad_norm": 0.48109114170074463,
      "learning_rate": 8.105237981344175e-05,
      "loss": 0.0119,
      "step": 15763
    },
    {
      "epoch": 2.2093903293622987,
      "grad_norm": 0.7652236223220825,
      "learning_rate": 8.10380291796221e-05,
      "loss": 0.132,
      "step": 15764
    },
    {
      "epoch": 2.209530483531885,
      "grad_norm": 0.3879028260707855,
      "learning_rate": 8.102367854580243e-05,
      "loss": 0.0238,
      "step": 15765
    },
    {
      "epoch": 2.2096706377014717,
      "grad_norm": 0.18922802805900574,
      "learning_rate": 8.100932791198277e-05,
      "loss": 0.0454,
      "step": 15766
    },
    {
      "epoch": 2.209810791871058,
      "grad_norm": 0.3922787010669708,
      "learning_rate": 8.09949772781631e-05,
      "loss": 0.0189,
      "step": 15767
    },
    {
      "epoch": 2.2099509460406446,
      "grad_norm": 0.7028340101242065,
      "learning_rate": 8.098062664434344e-05,
      "loss": 0.1249,
      "step": 15768
    },
    {
      "epoch": 2.2100911002102315,
      "grad_norm": 0.9105939865112305,
      "learning_rate": 8.09662760105238e-05,
      "loss": 0.096,
      "step": 15769
    },
    {
      "epoch": 2.210231254379818,
      "grad_norm": 0.07654590159654617,
      "learning_rate": 8.095192537670413e-05,
      "loss": 0.0025,
      "step": 15770
    },
    {
      "epoch": 2.2103714085494044,
      "grad_norm": 0.11136972904205322,
      "learning_rate": 8.093757474288447e-05,
      "loss": 0.018,
      "step": 15771
    },
    {
      "epoch": 2.210511562718991,
      "grad_norm": 0.16092923283576965,
      "learning_rate": 8.092322410906481e-05,
      "loss": 0.05,
      "step": 15772
    },
    {
      "epoch": 2.2106517168885773,
      "grad_norm": 0.1688794642686844,
      "learning_rate": 8.090887347524514e-05,
      "loss": 0.0455,
      "step": 15773
    },
    {
      "epoch": 2.2107918710581638,
      "grad_norm": 0.25758206844329834,
      "learning_rate": 8.08945228414255e-05,
      "loss": 0.0384,
      "step": 15774
    },
    {
      "epoch": 2.2109320252277507,
      "grad_norm": 0.16360634565353394,
      "learning_rate": 8.088017220760584e-05,
      "loss": 0.021,
      "step": 15775
    },
    {
      "epoch": 2.211072179397337,
      "grad_norm": 0.12428990006446838,
      "learning_rate": 8.086582157378617e-05,
      "loss": 0.0392,
      "step": 15776
    },
    {
      "epoch": 2.2112123335669236,
      "grad_norm": 0.1947309970855713,
      "learning_rate": 8.085147093996651e-05,
      "loss": 0.0322,
      "step": 15777
    },
    {
      "epoch": 2.21135248773651,
      "grad_norm": 0.3968620300292969,
      "learning_rate": 8.083712030614684e-05,
      "loss": 0.0617,
      "step": 15778
    },
    {
      "epoch": 2.2114926419060965,
      "grad_norm": 0.2640931308269501,
      "learning_rate": 8.082276967232718e-05,
      "loss": 0.0371,
      "step": 15779
    },
    {
      "epoch": 2.2116327960756834,
      "grad_norm": 0.2555704414844513,
      "learning_rate": 8.080841903850754e-05,
      "loss": 0.0356,
      "step": 15780
    },
    {
      "epoch": 2.21177295024527,
      "grad_norm": 0.23118658363819122,
      "learning_rate": 8.079406840468786e-05,
      "loss": 0.0389,
      "step": 15781
    },
    {
      "epoch": 2.2119131044148563,
      "grad_norm": 0.19575774669647217,
      "learning_rate": 8.077971777086821e-05,
      "loss": 0.0244,
      "step": 15782
    },
    {
      "epoch": 2.212053258584443,
      "grad_norm": 0.31075602769851685,
      "learning_rate": 8.076536713704855e-05,
      "loss": 0.0345,
      "step": 15783
    },
    {
      "epoch": 2.2121934127540293,
      "grad_norm": 0.1856757402420044,
      "learning_rate": 8.075101650322888e-05,
      "loss": 0.0393,
      "step": 15784
    },
    {
      "epoch": 2.212333566923616,
      "grad_norm": 0.29064038395881653,
      "learning_rate": 8.073666586940923e-05,
      "loss": 0.0327,
      "step": 15785
    },
    {
      "epoch": 2.2124737210932026,
      "grad_norm": 0.2372765988111496,
      "learning_rate": 8.072231523558956e-05,
      "loss": 0.0602,
      "step": 15786
    },
    {
      "epoch": 2.212613875262789,
      "grad_norm": 0.16154570877552032,
      "learning_rate": 8.07079646017699e-05,
      "loss": 0.0359,
      "step": 15787
    },
    {
      "epoch": 2.2127540294323755,
      "grad_norm": 0.13458600640296936,
      "learning_rate": 8.069361396795025e-05,
      "loss": 0.0081,
      "step": 15788
    },
    {
      "epoch": 2.212894183601962,
      "grad_norm": 0.16281646490097046,
      "learning_rate": 8.067926333413057e-05,
      "loss": 0.0257,
      "step": 15789
    },
    {
      "epoch": 2.213034337771549,
      "grad_norm": 0.2414088100194931,
      "learning_rate": 8.066491270031093e-05,
      "loss": 0.0286,
      "step": 15790
    },
    {
      "epoch": 2.2131744919411354,
      "grad_norm": 0.11083386093378067,
      "learning_rate": 8.065056206649127e-05,
      "loss": 0.0092,
      "step": 15791
    },
    {
      "epoch": 2.213314646110722,
      "grad_norm": 0.26579955220222473,
      "learning_rate": 8.06362114326716e-05,
      "loss": 0.0507,
      "step": 15792
    },
    {
      "epoch": 2.2134548002803083,
      "grad_norm": 0.20172646641731262,
      "learning_rate": 8.062186079885194e-05,
      "loss": 0.0143,
      "step": 15793
    },
    {
      "epoch": 2.2135949544498947,
      "grad_norm": 0.24762548506259918,
      "learning_rate": 8.060751016503227e-05,
      "loss": 0.0601,
      "step": 15794
    },
    {
      "epoch": 2.213735108619481,
      "grad_norm": 0.3366411626338959,
      "learning_rate": 8.059315953121261e-05,
      "loss": 0.0344,
      "step": 15795
    },
    {
      "epoch": 2.213875262789068,
      "grad_norm": 0.26526540517807007,
      "learning_rate": 8.057880889739297e-05,
      "loss": 0.0535,
      "step": 15796
    },
    {
      "epoch": 2.2140154169586546,
      "grad_norm": 0.22602668404579163,
      "learning_rate": 8.05644582635733e-05,
      "loss": 0.0369,
      "step": 15797
    },
    {
      "epoch": 2.214155571128241,
      "grad_norm": 0.4422999620437622,
      "learning_rate": 8.055010762975364e-05,
      "loss": 0.0552,
      "step": 15798
    },
    {
      "epoch": 2.2142957252978275,
      "grad_norm": 0.2021181881427765,
      "learning_rate": 8.053575699593398e-05,
      "loss": 0.0341,
      "step": 15799
    },
    {
      "epoch": 2.2144358794674144,
      "grad_norm": 0.23665854334831238,
      "learning_rate": 8.052140636211431e-05,
      "loss": 0.0544,
      "step": 15800
    },
    {
      "epoch": 2.214576033637001,
      "grad_norm": 0.3430425822734833,
      "learning_rate": 8.050705572829467e-05,
      "loss": 0.0386,
      "step": 15801
    },
    {
      "epoch": 2.2147161878065873,
      "grad_norm": 0.14774034917354584,
      "learning_rate": 8.0492705094475e-05,
      "loss": 0.0317,
      "step": 15802
    },
    {
      "epoch": 2.2148563419761738,
      "grad_norm": 0.08194131404161453,
      "learning_rate": 8.047835446065534e-05,
      "loss": 0.0067,
      "step": 15803
    },
    {
      "epoch": 2.21499649614576,
      "grad_norm": 0.11670569330453873,
      "learning_rate": 8.046400382683568e-05,
      "loss": 0.0069,
      "step": 15804
    },
    {
      "epoch": 2.2151366503153467,
      "grad_norm": 0.39882898330688477,
      "learning_rate": 8.044965319301601e-05,
      "loss": 0.0218,
      "step": 15805
    },
    {
      "epoch": 2.2152768044849336,
      "grad_norm": 0.16679386794567108,
      "learning_rate": 8.043530255919636e-05,
      "loss": 0.0117,
      "step": 15806
    },
    {
      "epoch": 2.21541695865452,
      "grad_norm": 0.8383575677871704,
      "learning_rate": 8.04209519253767e-05,
      "loss": 0.0658,
      "step": 15807
    },
    {
      "epoch": 2.2155571128241065,
      "grad_norm": 0.11720184236764908,
      "learning_rate": 8.040660129155703e-05,
      "loss": 0.0085,
      "step": 15808
    },
    {
      "epoch": 2.215697266993693,
      "grad_norm": 0.17931196093559265,
      "learning_rate": 8.039225065773738e-05,
      "loss": 0.0147,
      "step": 15809
    },
    {
      "epoch": 2.2158374211632794,
      "grad_norm": 0.06549227982759476,
      "learning_rate": 8.037790002391772e-05,
      "loss": 0.0057,
      "step": 15810
    },
    {
      "epoch": 2.2159775753328663,
      "grad_norm": 0.08017376810312271,
      "learning_rate": 8.036354939009805e-05,
      "loss": 0.0174,
      "step": 15811
    },
    {
      "epoch": 2.216117729502453,
      "grad_norm": 0.036735810339450836,
      "learning_rate": 8.03491987562784e-05,
      "loss": 0.0031,
      "step": 15812
    },
    {
      "epoch": 2.2162578836720392,
      "grad_norm": 0.6916406750679016,
      "learning_rate": 8.033484812245873e-05,
      "loss": 0.0385,
      "step": 15813
    },
    {
      "epoch": 2.2163980378416257,
      "grad_norm": 0.3367035388946533,
      "learning_rate": 8.032049748863907e-05,
      "loss": 0.0082,
      "step": 15814
    },
    {
      "epoch": 2.216538192011212,
      "grad_norm": 0.19987551867961884,
      "learning_rate": 8.030614685481942e-05,
      "loss": 0.0185,
      "step": 15815
    },
    {
      "epoch": 2.216678346180799,
      "grad_norm": 0.30841147899627686,
      "learning_rate": 8.029179622099974e-05,
      "loss": 0.024,
      "step": 15816
    },
    {
      "epoch": 2.2168185003503855,
      "grad_norm": 4.956757068634033,
      "learning_rate": 8.02774455871801e-05,
      "loss": 0.2366,
      "step": 15817
    },
    {
      "epoch": 2.216958654519972,
      "grad_norm": 0.4238620400428772,
      "learning_rate": 8.026309495336044e-05,
      "loss": 0.0473,
      "step": 15818
    },
    {
      "epoch": 2.2170988086895584,
      "grad_norm": 0.304696649312973,
      "learning_rate": 8.024874431954077e-05,
      "loss": 0.0131,
      "step": 15819
    },
    {
      "epoch": 2.217238962859145,
      "grad_norm": 1.752905011177063,
      "learning_rate": 8.023439368572111e-05,
      "loss": 0.1664,
      "step": 15820
    },
    {
      "epoch": 2.217379117028732,
      "grad_norm": 0.26391544938087463,
      "learning_rate": 8.022004305190144e-05,
      "loss": 0.0296,
      "step": 15821
    },
    {
      "epoch": 2.2175192711983183,
      "grad_norm": 0.3294784128665924,
      "learning_rate": 8.02056924180818e-05,
      "loss": 0.0588,
      "step": 15822
    },
    {
      "epoch": 2.2176594253679047,
      "grad_norm": 0.7349872589111328,
      "learning_rate": 8.019134178426214e-05,
      "loss": 0.0425,
      "step": 15823
    },
    {
      "epoch": 2.217799579537491,
      "grad_norm": 0.1430329531431198,
      "learning_rate": 8.017699115044247e-05,
      "loss": 0.0083,
      "step": 15824
    },
    {
      "epoch": 2.2179397337070776,
      "grad_norm": 0.3361467719078064,
      "learning_rate": 8.016264051662281e-05,
      "loss": 0.0264,
      "step": 15825
    },
    {
      "epoch": 2.218079887876664,
      "grad_norm": 0.21544459462165833,
      "learning_rate": 8.014828988280315e-05,
      "loss": 0.0619,
      "step": 15826
    },
    {
      "epoch": 2.218220042046251,
      "grad_norm": 0.16525933146476746,
      "learning_rate": 8.013393924898348e-05,
      "loss": 0.0151,
      "step": 15827
    },
    {
      "epoch": 2.2183601962158375,
      "grad_norm": 0.15224115550518036,
      "learning_rate": 8.011958861516384e-05,
      "loss": 0.0171,
      "step": 15828
    },
    {
      "epoch": 2.218500350385424,
      "grad_norm": 0.39868131279945374,
      "learning_rate": 8.010523798134416e-05,
      "loss": 0.0486,
      "step": 15829
    },
    {
      "epoch": 2.2186405045550104,
      "grad_norm": 0.4792543947696686,
      "learning_rate": 8.00908873475245e-05,
      "loss": 0.0305,
      "step": 15830
    },
    {
      "epoch": 2.2187806587245973,
      "grad_norm": 0.14432522654533386,
      "learning_rate": 8.007653671370485e-05,
      "loss": 0.0398,
      "step": 15831
    },
    {
      "epoch": 2.2189208128941837,
      "grad_norm": 0.3087961971759796,
      "learning_rate": 8.006218607988518e-05,
      "loss": 0.0279,
      "step": 15832
    },
    {
      "epoch": 2.21906096706377,
      "grad_norm": 0.10566693544387817,
      "learning_rate": 8.004783544606553e-05,
      "loss": 0.0084,
      "step": 15833
    },
    {
      "epoch": 2.2192011212333567,
      "grad_norm": 0.348856121301651,
      "learning_rate": 8.003348481224587e-05,
      "loss": 0.0187,
      "step": 15834
    },
    {
      "epoch": 2.219341275402943,
      "grad_norm": 0.07867445796728134,
      "learning_rate": 8.00191341784262e-05,
      "loss": 0.0112,
      "step": 15835
    },
    {
      "epoch": 2.2194814295725296,
      "grad_norm": 0.4268884062767029,
      "learning_rate": 8.000478354460655e-05,
      "loss": 0.069,
      "step": 15836
    },
    {
      "epoch": 2.2196215837421165,
      "grad_norm": 0.1312306672334671,
      "learning_rate": 7.999043291078687e-05,
      "loss": 0.024,
      "step": 15837
    },
    {
      "epoch": 2.219761737911703,
      "grad_norm": 0.13145339488983154,
      "learning_rate": 7.997608227696723e-05,
      "loss": 0.0108,
      "step": 15838
    },
    {
      "epoch": 2.2199018920812894,
      "grad_norm": 0.38580700755119324,
      "learning_rate": 7.996173164314757e-05,
      "loss": 0.0387,
      "step": 15839
    },
    {
      "epoch": 2.220042046250876,
      "grad_norm": 0.2187608778476715,
      "learning_rate": 7.99473810093279e-05,
      "loss": 0.0289,
      "step": 15840
    },
    {
      "epoch": 2.2201822004204623,
      "grad_norm": 0.4201809763908386,
      "learning_rate": 7.993303037550824e-05,
      "loss": 0.0349,
      "step": 15841
    },
    {
      "epoch": 2.2203223545900492,
      "grad_norm": 0.13052545487880707,
      "learning_rate": 7.991867974168858e-05,
      "loss": 0.0141,
      "step": 15842
    },
    {
      "epoch": 2.2204625087596357,
      "grad_norm": 0.25117334723472595,
      "learning_rate": 7.990432910786891e-05,
      "loss": 0.0269,
      "step": 15843
    },
    {
      "epoch": 2.220602662929222,
      "grad_norm": 0.17811857163906097,
      "learning_rate": 7.988997847404927e-05,
      "loss": 0.0431,
      "step": 15844
    },
    {
      "epoch": 2.2207428170988086,
      "grad_norm": 0.11061608046293259,
      "learning_rate": 7.987562784022961e-05,
      "loss": 0.006,
      "step": 15845
    },
    {
      "epoch": 2.220882971268395,
      "grad_norm": 0.13096849620342255,
      "learning_rate": 7.986127720640994e-05,
      "loss": 0.0135,
      "step": 15846
    },
    {
      "epoch": 2.221023125437982,
      "grad_norm": 0.1101568415760994,
      "learning_rate": 7.984692657259028e-05,
      "loss": 0.0121,
      "step": 15847
    },
    {
      "epoch": 2.2211632796075684,
      "grad_norm": 0.21730497479438782,
      "learning_rate": 7.983257593877061e-05,
      "loss": 0.0318,
      "step": 15848
    },
    {
      "epoch": 2.221303433777155,
      "grad_norm": 0.23728030920028687,
      "learning_rate": 7.981822530495097e-05,
      "loss": 0.0437,
      "step": 15849
    },
    {
      "epoch": 2.2214435879467413,
      "grad_norm": 0.30696719884872437,
      "learning_rate": 7.980387467113131e-05,
      "loss": 0.0508,
      "step": 15850
    },
    {
      "epoch": 2.221583742116328,
      "grad_norm": 0.06760215014219284,
      "learning_rate": 7.978952403731164e-05,
      "loss": 0.0059,
      "step": 15851
    },
    {
      "epoch": 2.2217238962859147,
      "grad_norm": 0.1362176388502121,
      "learning_rate": 7.977517340349198e-05,
      "loss": 0.0134,
      "step": 15852
    },
    {
      "epoch": 2.221864050455501,
      "grad_norm": 0.18390010297298431,
      "learning_rate": 7.976082276967233e-05,
      "loss": 0.0084,
      "step": 15853
    },
    {
      "epoch": 2.2220042046250876,
      "grad_norm": 0.17166583240032196,
      "learning_rate": 7.974647213585266e-05,
      "loss": 0.0114,
      "step": 15854
    },
    {
      "epoch": 2.222144358794674,
      "grad_norm": 0.0382232666015625,
      "learning_rate": 7.9732121502033e-05,
      "loss": 0.0043,
      "step": 15855
    },
    {
      "epoch": 2.2222845129642605,
      "grad_norm": 0.2232348918914795,
      "learning_rate": 7.971777086821333e-05,
      "loss": 0.0189,
      "step": 15856
    },
    {
      "epoch": 2.222424667133847,
      "grad_norm": 0.5808573365211487,
      "learning_rate": 7.970342023439368e-05,
      "loss": 0.0944,
      "step": 15857
    },
    {
      "epoch": 2.222564821303434,
      "grad_norm": 0.1688917577266693,
      "learning_rate": 7.968906960057402e-05,
      "loss": 0.0159,
      "step": 15858
    },
    {
      "epoch": 2.2227049754730204,
      "grad_norm": 0.22208845615386963,
      "learning_rate": 7.967471896675435e-05,
      "loss": 0.0186,
      "step": 15859
    },
    {
      "epoch": 2.222845129642607,
      "grad_norm": 0.30864429473876953,
      "learning_rate": 7.96603683329347e-05,
      "loss": 0.0315,
      "step": 15860
    },
    {
      "epoch": 2.2229852838121933,
      "grad_norm": 0.19438521564006805,
      "learning_rate": 7.964601769911504e-05,
      "loss": 0.0119,
      "step": 15861
    },
    {
      "epoch": 2.2231254379817798,
      "grad_norm": 0.396528422832489,
      "learning_rate": 7.963166706529537e-05,
      "loss": 0.02,
      "step": 15862
    },
    {
      "epoch": 2.2232655921513667,
      "grad_norm": 0.14854362607002258,
      "learning_rate": 7.961731643147571e-05,
      "loss": 0.0281,
      "step": 15863
    },
    {
      "epoch": 2.223405746320953,
      "grad_norm": 0.12330492585897446,
      "learning_rate": 7.960296579765604e-05,
      "loss": 0.0104,
      "step": 15864
    },
    {
      "epoch": 2.2235459004905396,
      "grad_norm": 0.8328088521957397,
      "learning_rate": 7.95886151638364e-05,
      "loss": 0.0498,
      "step": 15865
    },
    {
      "epoch": 2.223686054660126,
      "grad_norm": 0.2999665439128876,
      "learning_rate": 7.957426453001674e-05,
      "loss": 0.0348,
      "step": 15866
    },
    {
      "epoch": 2.2238262088297125,
      "grad_norm": 0.47874000668525696,
      "learning_rate": 7.955991389619707e-05,
      "loss": 0.074,
      "step": 15867
    },
    {
      "epoch": 2.2239663629992994,
      "grad_norm": 0.06939361244440079,
      "learning_rate": 7.954556326237741e-05,
      "loss": 0.0056,
      "step": 15868
    },
    {
      "epoch": 2.224106517168886,
      "grad_norm": 0.11657527089118958,
      "learning_rate": 7.953121262855777e-05,
      "loss": 0.008,
      "step": 15869
    },
    {
      "epoch": 2.2242466713384723,
      "grad_norm": 0.1638582944869995,
      "learning_rate": 7.95168619947381e-05,
      "loss": 0.0046,
      "step": 15870
    },
    {
      "epoch": 2.2243868255080588,
      "grad_norm": 0.20034876465797424,
      "learning_rate": 7.950251136091844e-05,
      "loss": 0.0532,
      "step": 15871
    },
    {
      "epoch": 2.2245269796776452,
      "grad_norm": 0.2525147497653961,
      "learning_rate": 7.948816072709877e-05,
      "loss": 0.0242,
      "step": 15872
    },
    {
      "epoch": 2.224667133847232,
      "grad_norm": 0.29534441232681274,
      "learning_rate": 7.947381009327911e-05,
      "loss": 0.0933,
      "step": 15873
    },
    {
      "epoch": 2.2248072880168186,
      "grad_norm": 0.4660206139087677,
      "learning_rate": 7.945945945945945e-05,
      "loss": 0.0504,
      "step": 15874
    },
    {
      "epoch": 2.224947442186405,
      "grad_norm": 0.15314942598342896,
      "learning_rate": 7.944510882563978e-05,
      "loss": 0.0432,
      "step": 15875
    },
    {
      "epoch": 2.2250875963559915,
      "grad_norm": 0.1253141462802887,
      "learning_rate": 7.943075819182014e-05,
      "loss": 0.0149,
      "step": 15876
    },
    {
      "epoch": 2.225227750525578,
      "grad_norm": 0.3300476372241974,
      "learning_rate": 7.941640755800048e-05,
      "loss": 0.0387,
      "step": 15877
    },
    {
      "epoch": 2.225367904695165,
      "grad_norm": 0.19265851378440857,
      "learning_rate": 7.94020569241808e-05,
      "loss": 0.0321,
      "step": 15878
    },
    {
      "epoch": 2.2255080588647513,
      "grad_norm": 0.1389550119638443,
      "learning_rate": 7.938770629036115e-05,
      "loss": 0.0119,
      "step": 15879
    },
    {
      "epoch": 2.225648213034338,
      "grad_norm": 0.06762897223234177,
      "learning_rate": 7.93733556565415e-05,
      "loss": 0.0057,
      "step": 15880
    },
    {
      "epoch": 2.2257883672039243,
      "grad_norm": 0.367824912071228,
      "learning_rate": 7.935900502272183e-05,
      "loss": 0.0511,
      "step": 15881
    },
    {
      "epoch": 2.2259285213735107,
      "grad_norm": 0.1511528044939041,
      "learning_rate": 7.934465438890217e-05,
      "loss": 0.0191,
      "step": 15882
    },
    {
      "epoch": 2.2260686755430976,
      "grad_norm": 0.2962946593761444,
      "learning_rate": 7.93303037550825e-05,
      "loss": 0.0525,
      "step": 15883
    },
    {
      "epoch": 2.226208829712684,
      "grad_norm": 0.15286998450756073,
      "learning_rate": 7.931595312126285e-05,
      "loss": 0.0134,
      "step": 15884
    },
    {
      "epoch": 2.2263489838822705,
      "grad_norm": 0.2257305085659027,
      "learning_rate": 7.93016024874432e-05,
      "loss": 0.0209,
      "step": 15885
    },
    {
      "epoch": 2.226489138051857,
      "grad_norm": 0.498569130897522,
      "learning_rate": 7.928725185362353e-05,
      "loss": 0.0445,
      "step": 15886
    },
    {
      "epoch": 2.2266292922214435,
      "grad_norm": 0.2954264283180237,
      "learning_rate": 7.927290121980387e-05,
      "loss": 0.0221,
      "step": 15887
    },
    {
      "epoch": 2.22676944639103,
      "grad_norm": 0.06533745676279068,
      "learning_rate": 7.925855058598421e-05,
      "loss": 0.0088,
      "step": 15888
    },
    {
      "epoch": 2.226909600560617,
      "grad_norm": 0.11143027245998383,
      "learning_rate": 7.924419995216454e-05,
      "loss": 0.0295,
      "step": 15889
    },
    {
      "epoch": 2.2270497547302033,
      "grad_norm": 0.3494460880756378,
      "learning_rate": 7.922984931834488e-05,
      "loss": 0.0806,
      "step": 15890
    },
    {
      "epoch": 2.2271899088997897,
      "grad_norm": 0.19088922441005707,
      "learning_rate": 7.921549868452521e-05,
      "loss": 0.0514,
      "step": 15891
    },
    {
      "epoch": 2.227330063069376,
      "grad_norm": 0.6574686765670776,
      "learning_rate": 7.920114805070557e-05,
      "loss": 0.081,
      "step": 15892
    },
    {
      "epoch": 2.2274702172389627,
      "grad_norm": 0.16093890368938446,
      "learning_rate": 7.918679741688591e-05,
      "loss": 0.0206,
      "step": 15893
    },
    {
      "epoch": 2.2276103714085496,
      "grad_norm": 0.07079556584358215,
      "learning_rate": 7.917244678306624e-05,
      "loss": 0.005,
      "step": 15894
    },
    {
      "epoch": 2.227750525578136,
      "grad_norm": 0.24389594793319702,
      "learning_rate": 7.915809614924658e-05,
      "loss": 0.0214,
      "step": 15895
    },
    {
      "epoch": 2.2278906797477225,
      "grad_norm": 0.230647012591362,
      "learning_rate": 7.914374551542694e-05,
      "loss": 0.0759,
      "step": 15896
    },
    {
      "epoch": 2.228030833917309,
      "grad_norm": 0.22221429646015167,
      "learning_rate": 7.912939488160727e-05,
      "loss": 0.0466,
      "step": 15897
    },
    {
      "epoch": 2.2281709880868954,
      "grad_norm": 0.2491951584815979,
      "learning_rate": 7.911504424778761e-05,
      "loss": 0.0463,
      "step": 15898
    },
    {
      "epoch": 2.2283111422564823,
      "grad_norm": 0.14288195967674255,
      "learning_rate": 7.910069361396794e-05,
      "loss": 0.0075,
      "step": 15899
    },
    {
      "epoch": 2.2284512964260688,
      "grad_norm": 0.1017732098698616,
      "learning_rate": 7.908634298014828e-05,
      "loss": 0.0084,
      "step": 15900
    },
    {
      "epoch": 2.228591450595655,
      "grad_norm": 0.368023544549942,
      "learning_rate": 7.907199234632863e-05,
      "loss": 0.0691,
      "step": 15901
    },
    {
      "epoch": 2.2287316047652417,
      "grad_norm": 0.06449741870164871,
      "learning_rate": 7.905764171250896e-05,
      "loss": 0.0132,
      "step": 15902
    },
    {
      "epoch": 2.228871758934828,
      "grad_norm": 0.22720414400100708,
      "learning_rate": 7.90432910786893e-05,
      "loss": 0.05,
      "step": 15903
    },
    {
      "epoch": 2.229011913104415,
      "grad_norm": 0.49815794825553894,
      "learning_rate": 7.902894044486965e-05,
      "loss": 0.0426,
      "step": 15904
    },
    {
      "epoch": 2.2291520672740015,
      "grad_norm": 0.21913887560367584,
      "learning_rate": 7.901458981104998e-05,
      "loss": 0.0913,
      "step": 15905
    },
    {
      "epoch": 2.229292221443588,
      "grad_norm": 0.21141712367534637,
      "learning_rate": 7.900023917723032e-05,
      "loss": 0.0543,
      "step": 15906
    },
    {
      "epoch": 2.2294323756131744,
      "grad_norm": 0.6616555452346802,
      "learning_rate": 7.898588854341065e-05,
      "loss": 0.0465,
      "step": 15907
    },
    {
      "epoch": 2.229572529782761,
      "grad_norm": 0.10346011072397232,
      "learning_rate": 7.8971537909591e-05,
      "loss": 0.0096,
      "step": 15908
    },
    {
      "epoch": 2.229712683952348,
      "grad_norm": 0.22836457192897797,
      "learning_rate": 7.895718727577134e-05,
      "loss": 0.0279,
      "step": 15909
    },
    {
      "epoch": 2.2298528381219342,
      "grad_norm": 0.2049017697572708,
      "learning_rate": 7.894283664195167e-05,
      "loss": 0.0188,
      "step": 15910
    },
    {
      "epoch": 2.2299929922915207,
      "grad_norm": 0.273671954870224,
      "learning_rate": 7.892848600813201e-05,
      "loss": 0.0312,
      "step": 15911
    },
    {
      "epoch": 2.230133146461107,
      "grad_norm": 0.32447123527526855,
      "learning_rate": 7.891413537431237e-05,
      "loss": 0.0443,
      "step": 15912
    },
    {
      "epoch": 2.2302733006306936,
      "grad_norm": 0.21466729044914246,
      "learning_rate": 7.88997847404927e-05,
      "loss": 0.0458,
      "step": 15913
    },
    {
      "epoch": 2.2304134548002805,
      "grad_norm": 0.18691864609718323,
      "learning_rate": 7.888543410667304e-05,
      "loss": 0.0153,
      "step": 15914
    },
    {
      "epoch": 2.230553608969867,
      "grad_norm": 0.32756713032722473,
      "learning_rate": 7.887108347285337e-05,
      "loss": 0.0262,
      "step": 15915
    },
    {
      "epoch": 2.2306937631394534,
      "grad_norm": 0.9787201285362244,
      "learning_rate": 7.885673283903371e-05,
      "loss": 0.0381,
      "step": 15916
    },
    {
      "epoch": 2.23083391730904,
      "grad_norm": 0.7889458537101746,
      "learning_rate": 7.884238220521407e-05,
      "loss": 0.0416,
      "step": 15917
    },
    {
      "epoch": 2.2309740714786264,
      "grad_norm": 0.6555795073509216,
      "learning_rate": 7.88280315713944e-05,
      "loss": 0.0272,
      "step": 15918
    },
    {
      "epoch": 2.231114225648213,
      "grad_norm": 0.9492841958999634,
      "learning_rate": 7.881368093757474e-05,
      "loss": 0.0525,
      "step": 15919
    },
    {
      "epoch": 2.2312543798177997,
      "grad_norm": 1.0203248262405396,
      "learning_rate": 7.879933030375508e-05,
      "loss": 0.0271,
      "step": 15920
    },
    {
      "epoch": 2.231394533987386,
      "grad_norm": 0.10743717849254608,
      "learning_rate": 7.878497966993541e-05,
      "loss": 0.0213,
      "step": 15921
    },
    {
      "epoch": 2.2315346881569726,
      "grad_norm": 0.13191774487495422,
      "learning_rate": 7.877062903611575e-05,
      "loss": 0.0105,
      "step": 15922
    },
    {
      "epoch": 2.231674842326559,
      "grad_norm": 0.1725078821182251,
      "learning_rate": 7.87562784022961e-05,
      "loss": 0.0367,
      "step": 15923
    },
    {
      "epoch": 2.2318149964961456,
      "grad_norm": 0.1473742574453354,
      "learning_rate": 7.874192776847643e-05,
      "loss": 0.0163,
      "step": 15924
    },
    {
      "epoch": 2.2319551506657325,
      "grad_norm": 0.20539367198944092,
      "learning_rate": 7.872757713465678e-05,
      "loss": 0.053,
      "step": 15925
    },
    {
      "epoch": 2.232095304835319,
      "grad_norm": 0.21103744208812714,
      "learning_rate": 7.87132265008371e-05,
      "loss": 0.0108,
      "step": 15926
    },
    {
      "epoch": 2.2322354590049054,
      "grad_norm": 0.14602236449718475,
      "learning_rate": 7.869887586701745e-05,
      "loss": 0.0106,
      "step": 15927
    },
    {
      "epoch": 2.232375613174492,
      "grad_norm": 0.1512695848941803,
      "learning_rate": 7.86845252331978e-05,
      "loss": 0.0145,
      "step": 15928
    },
    {
      "epoch": 2.2325157673440783,
      "grad_norm": 0.22045272588729858,
      "learning_rate": 7.867017459937813e-05,
      "loss": 0.0213,
      "step": 15929
    },
    {
      "epoch": 2.232655921513665,
      "grad_norm": 0.2753868103027344,
      "learning_rate": 7.865582396555847e-05,
      "loss": 0.0419,
      "step": 15930
    },
    {
      "epoch": 2.2327960756832517,
      "grad_norm": 0.19521088898181915,
      "learning_rate": 7.864147333173882e-05,
      "loss": 0.0236,
      "step": 15931
    },
    {
      "epoch": 2.232936229852838,
      "grad_norm": 0.3316512405872345,
      "learning_rate": 7.862712269791914e-05,
      "loss": 0.052,
      "step": 15932
    },
    {
      "epoch": 2.2330763840224246,
      "grad_norm": 0.20682649314403534,
      "learning_rate": 7.86127720640995e-05,
      "loss": 0.0182,
      "step": 15933
    },
    {
      "epoch": 2.233216538192011,
      "grad_norm": 0.08356979489326477,
      "learning_rate": 7.859842143027983e-05,
      "loss": 0.0101,
      "step": 15934
    },
    {
      "epoch": 2.233356692361598,
      "grad_norm": 0.07834531366825104,
      "learning_rate": 7.858407079646017e-05,
      "loss": 0.0082,
      "step": 15935
    },
    {
      "epoch": 2.2334968465311844,
      "grad_norm": 0.17835354804992676,
      "learning_rate": 7.856972016264051e-05,
      "loss": 0.0302,
      "step": 15936
    },
    {
      "epoch": 2.233637000700771,
      "grad_norm": 0.3725316822528839,
      "learning_rate": 7.855536952882084e-05,
      "loss": 0.0324,
      "step": 15937
    },
    {
      "epoch": 2.2337771548703573,
      "grad_norm": 0.09457622468471527,
      "learning_rate": 7.854101889500118e-05,
      "loss": 0.0156,
      "step": 15938
    },
    {
      "epoch": 2.233917309039944,
      "grad_norm": 0.217263862490654,
      "learning_rate": 7.852666826118154e-05,
      "loss": 0.0171,
      "step": 15939
    },
    {
      "epoch": 2.2340574632095302,
      "grad_norm": 0.2612225115299225,
      "learning_rate": 7.851231762736187e-05,
      "loss": 0.0466,
      "step": 15940
    },
    {
      "epoch": 2.234197617379117,
      "grad_norm": 0.34249433875083923,
      "learning_rate": 7.849796699354221e-05,
      "loss": 0.0455,
      "step": 15941
    },
    {
      "epoch": 2.2343377715487036,
      "grad_norm": 0.08639832586050034,
      "learning_rate": 7.848361635972254e-05,
      "loss": 0.0148,
      "step": 15942
    },
    {
      "epoch": 2.23447792571829,
      "grad_norm": 0.23127956688404083,
      "learning_rate": 7.846926572590288e-05,
      "loss": 0.0462,
      "step": 15943
    },
    {
      "epoch": 2.2346180798878765,
      "grad_norm": 0.48839306831359863,
      "learning_rate": 7.845491509208324e-05,
      "loss": 0.0429,
      "step": 15944
    },
    {
      "epoch": 2.2347582340574634,
      "grad_norm": 0.35168054699897766,
      "learning_rate": 7.844056445826356e-05,
      "loss": 0.0487,
      "step": 15945
    },
    {
      "epoch": 2.23489838822705,
      "grad_norm": 0.2229284942150116,
      "learning_rate": 7.842621382444391e-05,
      "loss": 0.034,
      "step": 15946
    },
    {
      "epoch": 2.2350385423966364,
      "grad_norm": 0.07867902517318726,
      "learning_rate": 7.841186319062425e-05,
      "loss": 0.0072,
      "step": 15947
    },
    {
      "epoch": 2.235178696566223,
      "grad_norm": 0.1597619652748108,
      "learning_rate": 7.839751255680458e-05,
      "loss": 0.0238,
      "step": 15948
    },
    {
      "epoch": 2.2353188507358093,
      "grad_norm": 0.1586986631155014,
      "learning_rate": 7.838316192298493e-05,
      "loss": 0.0116,
      "step": 15949
    },
    {
      "epoch": 2.2354590049053957,
      "grad_norm": 0.4441637694835663,
      "learning_rate": 7.836881128916526e-05,
      "loss": 0.0452,
      "step": 15950
    },
    {
      "epoch": 2.2355991590749826,
      "grad_norm": 0.21711398661136627,
      "learning_rate": 7.83544606553456e-05,
      "loss": 0.0362,
      "step": 15951
    },
    {
      "epoch": 2.235739313244569,
      "grad_norm": 0.09045593440532684,
      "learning_rate": 7.834011002152595e-05,
      "loss": 0.0063,
      "step": 15952
    },
    {
      "epoch": 2.2358794674141556,
      "grad_norm": 0.08355707675218582,
      "learning_rate": 7.832575938770627e-05,
      "loss": 0.0126,
      "step": 15953
    },
    {
      "epoch": 2.236019621583742,
      "grad_norm": 0.08078756183385849,
      "learning_rate": 7.831140875388662e-05,
      "loss": 0.0074,
      "step": 15954
    },
    {
      "epoch": 2.2361597757533285,
      "grad_norm": 0.12156230956315994,
      "learning_rate": 7.829705812006697e-05,
      "loss": 0.0197,
      "step": 15955
    },
    {
      "epoch": 2.2362999299229154,
      "grad_norm": 0.2588076591491699,
      "learning_rate": 7.82827074862473e-05,
      "loss": 0.0686,
      "step": 15956
    },
    {
      "epoch": 2.236440084092502,
      "grad_norm": 0.2731691002845764,
      "learning_rate": 7.826835685242764e-05,
      "loss": 0.0842,
      "step": 15957
    },
    {
      "epoch": 2.2365802382620883,
      "grad_norm": 0.2036910355091095,
      "learning_rate": 7.825400621860799e-05,
      "loss": 0.0341,
      "step": 15958
    },
    {
      "epoch": 2.2367203924316748,
      "grad_norm": 0.16906343400478363,
      "learning_rate": 7.823965558478831e-05,
      "loss": 0.0249,
      "step": 15959
    },
    {
      "epoch": 2.236860546601261,
      "grad_norm": 0.8929670453071594,
      "learning_rate": 7.822530495096867e-05,
      "loss": 0.0885,
      "step": 15960
    },
    {
      "epoch": 2.237000700770848,
      "grad_norm": 0.32786768674850464,
      "learning_rate": 7.8210954317149e-05,
      "loss": 0.0584,
      "step": 15961
    },
    {
      "epoch": 2.2371408549404346,
      "grad_norm": 0.9510080814361572,
      "learning_rate": 7.819660368332934e-05,
      "loss": 0.0295,
      "step": 15962
    },
    {
      "epoch": 2.237281009110021,
      "grad_norm": 0.543478786945343,
      "learning_rate": 7.818225304950968e-05,
      "loss": 0.0879,
      "step": 15963
    },
    {
      "epoch": 2.2374211632796075,
      "grad_norm": 0.34720319509506226,
      "learning_rate": 7.816790241569001e-05,
      "loss": 0.1185,
      "step": 15964
    },
    {
      "epoch": 2.237561317449194,
      "grad_norm": 0.5739341974258423,
      "learning_rate": 7.815355178187037e-05,
      "loss": 0.0742,
      "step": 15965
    },
    {
      "epoch": 2.237701471618781,
      "grad_norm": 0.8970751762390137,
      "learning_rate": 7.813920114805071e-05,
      "loss": 0.2244,
      "step": 15966
    },
    {
      "epoch": 2.2378416257883673,
      "grad_norm": 0.23464936017990112,
      "learning_rate": 7.812485051423104e-05,
      "loss": 0.0156,
      "step": 15967
    },
    {
      "epoch": 2.2379817799579538,
      "grad_norm": 0.13832317292690277,
      "learning_rate": 7.811049988041138e-05,
      "loss": 0.0189,
      "step": 15968
    },
    {
      "epoch": 2.2381219341275402,
      "grad_norm": 0.24305181205272675,
      "learning_rate": 7.809614924659171e-05,
      "loss": 0.0141,
      "step": 15969
    },
    {
      "epoch": 2.2382620882971267,
      "grad_norm": 0.18162932991981506,
      "learning_rate": 7.808179861277205e-05,
      "loss": 0.013,
      "step": 15970
    },
    {
      "epoch": 2.238402242466713,
      "grad_norm": 0.11373086273670197,
      "learning_rate": 7.80674479789524e-05,
      "loss": 0.0146,
      "step": 15971
    },
    {
      "epoch": 2.2385423966363,
      "grad_norm": 0.17261861264705658,
      "learning_rate": 7.805309734513273e-05,
      "loss": 0.0086,
      "step": 15972
    },
    {
      "epoch": 2.2386825508058865,
      "grad_norm": 0.07054518163204193,
      "learning_rate": 7.803874671131308e-05,
      "loss": 0.0047,
      "step": 15973
    },
    {
      "epoch": 2.238822704975473,
      "grad_norm": 0.16624481976032257,
      "learning_rate": 7.802439607749342e-05,
      "loss": 0.0372,
      "step": 15974
    },
    {
      "epoch": 2.2389628591450594,
      "grad_norm": 0.18570061028003693,
      "learning_rate": 7.801004544367375e-05,
      "loss": 0.0422,
      "step": 15975
    },
    {
      "epoch": 2.2391030133146463,
      "grad_norm": 0.1790638267993927,
      "learning_rate": 7.79956948098541e-05,
      "loss": 0.0153,
      "step": 15976
    },
    {
      "epoch": 2.239243167484233,
      "grad_norm": 0.7532731294631958,
      "learning_rate": 7.798134417603443e-05,
      "loss": 0.0849,
      "step": 15977
    },
    {
      "epoch": 2.2393833216538193,
      "grad_norm": 0.10992532223463058,
      "learning_rate": 7.796699354221477e-05,
      "loss": 0.0308,
      "step": 15978
    },
    {
      "epoch": 2.2395234758234057,
      "grad_norm": 0.3545118570327759,
      "learning_rate": 7.795264290839512e-05,
      "loss": 0.0908,
      "step": 15979
    },
    {
      "epoch": 2.239663629992992,
      "grad_norm": 0.14106042683124542,
      "learning_rate": 7.793829227457544e-05,
      "loss": 0.0236,
      "step": 15980
    },
    {
      "epoch": 2.2398037841625786,
      "grad_norm": 0.243631973862648,
      "learning_rate": 7.79239416407558e-05,
      "loss": 0.0381,
      "step": 15981
    },
    {
      "epoch": 2.2399439383321655,
      "grad_norm": 0.12529917061328888,
      "learning_rate": 7.790959100693614e-05,
      "loss": 0.0425,
      "step": 15982
    },
    {
      "epoch": 2.240084092501752,
      "grad_norm": 0.07214927673339844,
      "learning_rate": 7.789524037311647e-05,
      "loss": 0.0078,
      "step": 15983
    },
    {
      "epoch": 2.2402242466713385,
      "grad_norm": 0.08505829423666,
      "learning_rate": 7.788088973929681e-05,
      "loss": 0.0098,
      "step": 15984
    },
    {
      "epoch": 2.240364400840925,
      "grad_norm": 0.3662014603614807,
      "learning_rate": 7.786653910547714e-05,
      "loss": 0.0205,
      "step": 15985
    },
    {
      "epoch": 2.2405045550105114,
      "grad_norm": 0.24449719488620758,
      "learning_rate": 7.785218847165748e-05,
      "loss": 0.0334,
      "step": 15986
    },
    {
      "epoch": 2.2406447091800983,
      "grad_norm": 0.2470383644104004,
      "learning_rate": 7.783783783783784e-05,
      "loss": 0.0375,
      "step": 15987
    },
    {
      "epoch": 2.2407848633496847,
      "grad_norm": 0.13602149486541748,
      "learning_rate": 7.782348720401817e-05,
      "loss": 0.0132,
      "step": 15988
    },
    {
      "epoch": 2.240925017519271,
      "grad_norm": 0.27616575360298157,
      "learning_rate": 7.780913657019851e-05,
      "loss": 0.0423,
      "step": 15989
    },
    {
      "epoch": 2.2410651716888577,
      "grad_norm": 0.5360847115516663,
      "learning_rate": 7.779478593637885e-05,
      "loss": 0.0418,
      "step": 15990
    },
    {
      "epoch": 2.241205325858444,
      "grad_norm": 0.07039465010166168,
      "learning_rate": 7.778043530255918e-05,
      "loss": 0.0066,
      "step": 15991
    },
    {
      "epoch": 2.241345480028031,
      "grad_norm": 0.16954585909843445,
      "learning_rate": 7.776608466873954e-05,
      "loss": 0.0206,
      "step": 15992
    },
    {
      "epoch": 2.2414856341976175,
      "grad_norm": 0.491550087928772,
      "learning_rate": 7.775173403491988e-05,
      "loss": 0.0898,
      "step": 15993
    },
    {
      "epoch": 2.241625788367204,
      "grad_norm": 0.23907309770584106,
      "learning_rate": 7.77373834011002e-05,
      "loss": 0.065,
      "step": 15994
    },
    {
      "epoch": 2.2417659425367904,
      "grad_norm": 0.20439140498638153,
      "learning_rate": 7.772303276728055e-05,
      "loss": 0.0257,
      "step": 15995
    },
    {
      "epoch": 2.241906096706377,
      "grad_norm": 0.17895087599754333,
      "learning_rate": 7.770868213346088e-05,
      "loss": 0.0209,
      "step": 15996
    },
    {
      "epoch": 2.2420462508759638,
      "grad_norm": 0.2104882299900055,
      "learning_rate": 7.769433149964123e-05,
      "loss": 0.0167,
      "step": 15997
    },
    {
      "epoch": 2.2421864050455502,
      "grad_norm": 0.02608693577349186,
      "learning_rate": 7.767998086582157e-05,
      "loss": 0.0032,
      "step": 15998
    },
    {
      "epoch": 2.2423265592151367,
      "grad_norm": 0.3028233051300049,
      "learning_rate": 7.76656302320019e-05,
      "loss": 0.0414,
      "step": 15999
    },
    {
      "epoch": 2.242466713384723,
      "grad_norm": 0.13197112083435059,
      "learning_rate": 7.765127959818225e-05,
      "loss": 0.0323,
      "step": 16000
    },
    {
      "epoch": 2.2426068675543096,
      "grad_norm": 0.3226237893104553,
      "learning_rate": 7.763692896436259e-05,
      "loss": 0.0402,
      "step": 16001
    },
    {
      "epoch": 2.242747021723896,
      "grad_norm": 0.1994910091161728,
      "learning_rate": 7.762257833054292e-05,
      "loss": 0.0211,
      "step": 16002
    },
    {
      "epoch": 2.242887175893483,
      "grad_norm": 0.16609889268875122,
      "learning_rate": 7.760822769672327e-05,
      "loss": 0.0531,
      "step": 16003
    },
    {
      "epoch": 2.2430273300630694,
      "grad_norm": 0.14180219173431396,
      "learning_rate": 7.75938770629036e-05,
      "loss": 0.0312,
      "step": 16004
    },
    {
      "epoch": 2.243167484232656,
      "grad_norm": 0.055453915148973465,
      "learning_rate": 7.757952642908394e-05,
      "loss": 0.0045,
      "step": 16005
    },
    {
      "epoch": 2.2433076384022423,
      "grad_norm": 0.24018704891204834,
      "learning_rate": 7.756517579526428e-05,
      "loss": 0.0192,
      "step": 16006
    },
    {
      "epoch": 2.2434477925718292,
      "grad_norm": 0.16401568055152893,
      "learning_rate": 7.755082516144461e-05,
      "loss": 0.0334,
      "step": 16007
    },
    {
      "epoch": 2.2435879467414157,
      "grad_norm": 0.05596894398331642,
      "learning_rate": 7.753647452762497e-05,
      "loss": 0.0065,
      "step": 16008
    },
    {
      "epoch": 2.243728100911002,
      "grad_norm": 0.2165966033935547,
      "learning_rate": 7.752212389380531e-05,
      "loss": 0.0503,
      "step": 16009
    },
    {
      "epoch": 2.2438682550805886,
      "grad_norm": 0.2890777885913849,
      "learning_rate": 7.750777325998564e-05,
      "loss": 0.0209,
      "step": 16010
    },
    {
      "epoch": 2.244008409250175,
      "grad_norm": 0.5852856636047363,
      "learning_rate": 7.749342262616598e-05,
      "loss": 0.0822,
      "step": 16011
    },
    {
      "epoch": 2.2441485634197615,
      "grad_norm": 0.2690792381763458,
      "learning_rate": 7.747907199234631e-05,
      "loss": 0.0509,
      "step": 16012
    },
    {
      "epoch": 2.2442887175893484,
      "grad_norm": 0.43139123916625977,
      "learning_rate": 7.746472135852667e-05,
      "loss": 0.0388,
      "step": 16013
    },
    {
      "epoch": 2.244428871758935,
      "grad_norm": 0.309966504573822,
      "learning_rate": 7.745037072470701e-05,
      "loss": 0.0397,
      "step": 16014
    },
    {
      "epoch": 2.2445690259285214,
      "grad_norm": 0.4398465156555176,
      "learning_rate": 7.743602009088734e-05,
      "loss": 0.0234,
      "step": 16015
    },
    {
      "epoch": 2.244709180098108,
      "grad_norm": 0.10380081832408905,
      "learning_rate": 7.742166945706768e-05,
      "loss": 0.0176,
      "step": 16016
    },
    {
      "epoch": 2.2448493342676943,
      "grad_norm": 0.16911178827285767,
      "learning_rate": 7.740731882324802e-05,
      "loss": 0.0143,
      "step": 16017
    },
    {
      "epoch": 2.244989488437281,
      "grad_norm": 0.23687949776649475,
      "learning_rate": 7.739296818942835e-05,
      "loss": 0.0065,
      "step": 16018
    },
    {
      "epoch": 2.2451296426068676,
      "grad_norm": 0.4006432890892029,
      "learning_rate": 7.73786175556087e-05,
      "loss": 0.0238,
      "step": 16019
    },
    {
      "epoch": 2.245269796776454,
      "grad_norm": 3.2815606594085693,
      "learning_rate": 7.736426692178903e-05,
      "loss": 0.1635,
      "step": 16020
    },
    {
      "epoch": 2.2454099509460406,
      "grad_norm": 0.18410198390483856,
      "learning_rate": 7.734991628796938e-05,
      "loss": 0.0163,
      "step": 16021
    },
    {
      "epoch": 2.245550105115627,
      "grad_norm": 0.27747076749801636,
      "learning_rate": 7.733556565414972e-05,
      "loss": 0.0347,
      "step": 16022
    },
    {
      "epoch": 2.245690259285214,
      "grad_norm": 0.15299734473228455,
      "learning_rate": 7.732121502033005e-05,
      "loss": 0.0177,
      "step": 16023
    },
    {
      "epoch": 2.2458304134548004,
      "grad_norm": 0.21046869456768036,
      "learning_rate": 7.73068643865104e-05,
      "loss": 0.016,
      "step": 16024
    },
    {
      "epoch": 2.245970567624387,
      "grad_norm": 0.23049715161323547,
      "learning_rate": 7.729251375269074e-05,
      "loss": 0.0358,
      "step": 16025
    },
    {
      "epoch": 2.2461107217939733,
      "grad_norm": 0.06531044840812683,
      "learning_rate": 7.727816311887107e-05,
      "loss": 0.0071,
      "step": 16026
    },
    {
      "epoch": 2.2462508759635598,
      "grad_norm": 0.1953677088022232,
      "learning_rate": 7.726381248505141e-05,
      "loss": 0.0215,
      "step": 16027
    },
    {
      "epoch": 2.2463910301331467,
      "grad_norm": 0.09398777037858963,
      "learning_rate": 7.724946185123177e-05,
      "loss": 0.0127,
      "step": 16028
    },
    {
      "epoch": 2.246531184302733,
      "grad_norm": 0.09099038690328598,
      "learning_rate": 7.72351112174121e-05,
      "loss": 0.0115,
      "step": 16029
    },
    {
      "epoch": 2.2466713384723196,
      "grad_norm": 0.05430116131901741,
      "learning_rate": 7.722076058359244e-05,
      "loss": 0.0043,
      "step": 16030
    },
    {
      "epoch": 2.246811492641906,
      "grad_norm": 0.2957955002784729,
      "learning_rate": 7.720640994977277e-05,
      "loss": 0.0087,
      "step": 16031
    },
    {
      "epoch": 2.2469516468114925,
      "grad_norm": 0.2726699411869049,
      "learning_rate": 7.719205931595311e-05,
      "loss": 0.0725,
      "step": 16032
    },
    {
      "epoch": 2.247091800981079,
      "grad_norm": 0.09464208781719208,
      "learning_rate": 7.717770868213345e-05,
      "loss": 0.0151,
      "step": 16033
    },
    {
      "epoch": 2.247231955150666,
      "grad_norm": 0.23228485882282257,
      "learning_rate": 7.716335804831378e-05,
      "loss": 0.0347,
      "step": 16034
    },
    {
      "epoch": 2.2473721093202523,
      "grad_norm": 0.3284781575202942,
      "learning_rate": 7.714900741449414e-05,
      "loss": 0.0217,
      "step": 16035
    },
    {
      "epoch": 2.247512263489839,
      "grad_norm": 0.2675741910934448,
      "learning_rate": 7.713465678067448e-05,
      "loss": 0.068,
      "step": 16036
    },
    {
      "epoch": 2.2476524176594253,
      "grad_norm": 0.18760760128498077,
      "learning_rate": 7.712030614685481e-05,
      "loss": 0.0291,
      "step": 16037
    },
    {
      "epoch": 2.2477925718290117,
      "grad_norm": 0.09009896963834763,
      "learning_rate": 7.710595551303515e-05,
      "loss": 0.0098,
      "step": 16038
    },
    {
      "epoch": 2.2479327259985986,
      "grad_norm": 0.2564639449119568,
      "learning_rate": 7.709160487921548e-05,
      "loss": 0.0481,
      "step": 16039
    },
    {
      "epoch": 2.248072880168185,
      "grad_norm": 0.10603039711713791,
      "learning_rate": 7.707725424539584e-05,
      "loss": 0.0073,
      "step": 16040
    },
    {
      "epoch": 2.2482130343377715,
      "grad_norm": 0.2566912770271301,
      "learning_rate": 7.706290361157618e-05,
      "loss": 0.0396,
      "step": 16041
    },
    {
      "epoch": 2.248353188507358,
      "grad_norm": 0.17767763137817383,
      "learning_rate": 7.70485529777565e-05,
      "loss": 0.0147,
      "step": 16042
    },
    {
      "epoch": 2.2484933426769445,
      "grad_norm": 0.20038369297981262,
      "learning_rate": 7.703420234393685e-05,
      "loss": 0.0529,
      "step": 16043
    },
    {
      "epoch": 2.2486334968465314,
      "grad_norm": 0.7695234417915344,
      "learning_rate": 7.70198517101172e-05,
      "loss": 0.0411,
      "step": 16044
    },
    {
      "epoch": 2.248773651016118,
      "grad_norm": 0.2092326432466507,
      "learning_rate": 7.700550107629753e-05,
      "loss": 0.0088,
      "step": 16045
    },
    {
      "epoch": 2.2489138051857043,
      "grad_norm": 0.14405342936515808,
      "learning_rate": 7.699115044247787e-05,
      "loss": 0.0194,
      "step": 16046
    },
    {
      "epoch": 2.2490539593552907,
      "grad_norm": 0.22511768341064453,
      "learning_rate": 7.69767998086582e-05,
      "loss": 0.0166,
      "step": 16047
    },
    {
      "epoch": 2.249194113524877,
      "grad_norm": 0.10174588114023209,
      "learning_rate": 7.696244917483854e-05,
      "loss": 0.0174,
      "step": 16048
    },
    {
      "epoch": 2.249334267694464,
      "grad_norm": 0.2544974088668823,
      "learning_rate": 7.694809854101889e-05,
      "loss": 0.0274,
      "step": 16049
    },
    {
      "epoch": 2.2494744218640506,
      "grad_norm": 0.47817814350128174,
      "learning_rate": 7.693374790719923e-05,
      "loss": 0.0906,
      "step": 16050
    },
    {
      "epoch": 2.249614576033637,
      "grad_norm": 0.08470392972230911,
      "learning_rate": 7.691939727337957e-05,
      "loss": 0.0106,
      "step": 16051
    },
    {
      "epoch": 2.2497547302032235,
      "grad_norm": 0.16570419073104858,
      "learning_rate": 7.690504663955991e-05,
      "loss": 0.0057,
      "step": 16052
    },
    {
      "epoch": 2.24989488437281,
      "grad_norm": 0.3965643346309662,
      "learning_rate": 7.689069600574024e-05,
      "loss": 0.0748,
      "step": 16053
    },
    {
      "epoch": 2.2500350385423964,
      "grad_norm": 0.1252543330192566,
      "learning_rate": 7.687634537192058e-05,
      "loss": 0.0134,
      "step": 16054
    },
    {
      "epoch": 2.2501751927119833,
      "grad_norm": 0.22512929141521454,
      "learning_rate": 7.686199473810091e-05,
      "loss": 0.0176,
      "step": 16055
    },
    {
      "epoch": 2.2503153468815698,
      "grad_norm": 0.36529284715652466,
      "learning_rate": 7.684764410428127e-05,
      "loss": 0.0248,
      "step": 16056
    },
    {
      "epoch": 2.250455501051156,
      "grad_norm": 0.2669682800769806,
      "learning_rate": 7.683329347046161e-05,
      "loss": 0.0144,
      "step": 16057
    },
    {
      "epoch": 2.2505956552207427,
      "grad_norm": 0.2966409921646118,
      "learning_rate": 7.681894283664194e-05,
      "loss": 0.0178,
      "step": 16058
    },
    {
      "epoch": 2.2507358093903296,
      "grad_norm": 0.34525495767593384,
      "learning_rate": 7.680459220282228e-05,
      "loss": 0.0209,
      "step": 16059
    },
    {
      "epoch": 2.250875963559916,
      "grad_norm": 0.4665195345878601,
      "learning_rate": 7.679024156900264e-05,
      "loss": 0.0245,
      "step": 16060
    },
    {
      "epoch": 2.2510161177295025,
      "grad_norm": 0.10393421351909637,
      "learning_rate": 7.677589093518297e-05,
      "loss": 0.0099,
      "step": 16061
    },
    {
      "epoch": 2.251156271899089,
      "grad_norm": 0.6053932905197144,
      "learning_rate": 7.676154030136331e-05,
      "loss": 0.0339,
      "step": 16062
    },
    {
      "epoch": 2.2512964260686754,
      "grad_norm": 0.25282928347587585,
      "learning_rate": 7.674718966754364e-05,
      "loss": 0.0192,
      "step": 16063
    },
    {
      "epoch": 2.251436580238262,
      "grad_norm": 0.09686914831399918,
      "learning_rate": 7.673283903372398e-05,
      "loss": 0.0046,
      "step": 16064
    },
    {
      "epoch": 2.251576734407849,
      "grad_norm": 0.2713959217071533,
      "learning_rate": 7.671848839990432e-05,
      "loss": 0.0351,
      "step": 16065
    },
    {
      "epoch": 2.2517168885774352,
      "grad_norm": 0.0872144103050232,
      "learning_rate": 7.670413776608466e-05,
      "loss": 0.0072,
      "step": 16066
    },
    {
      "epoch": 2.2518570427470217,
      "grad_norm": 2.0572173595428467,
      "learning_rate": 7.6689787132265e-05,
      "loss": 0.1727,
      "step": 16067
    },
    {
      "epoch": 2.251997196916608,
      "grad_norm": 0.7735179662704468,
      "learning_rate": 7.667543649844535e-05,
      "loss": 0.1503,
      "step": 16068
    },
    {
      "epoch": 2.252137351086195,
      "grad_norm": 0.5997719764709473,
      "learning_rate": 7.666108586462568e-05,
      "loss": 0.0158,
      "step": 16069
    },
    {
      "epoch": 2.2522775052557815,
      "grad_norm": 2.608621597290039,
      "learning_rate": 7.664673523080602e-05,
      "loss": 0.2723,
      "step": 16070
    },
    {
      "epoch": 2.252417659425368,
      "grad_norm": 0.2662675380706787,
      "learning_rate": 7.663238459698637e-05,
      "loss": 0.046,
      "step": 16071
    },
    {
      "epoch": 2.2525578135949544,
      "grad_norm": 0.307176411151886,
      "learning_rate": 7.66180339631667e-05,
      "loss": 0.0306,
      "step": 16072
    },
    {
      "epoch": 2.252697967764541,
      "grad_norm": 0.08651146292686462,
      "learning_rate": 7.660368332934704e-05,
      "loss": 0.0125,
      "step": 16073
    },
    {
      "epoch": 2.2528381219341274,
      "grad_norm": 0.22952255606651306,
      "learning_rate": 7.658933269552737e-05,
      "loss": 0.0322,
      "step": 16074
    },
    {
      "epoch": 2.2529782761037143,
      "grad_norm": 0.21150943636894226,
      "learning_rate": 7.657498206170771e-05,
      "loss": 0.0235,
      "step": 16075
    },
    {
      "epoch": 2.2531184302733007,
      "grad_norm": 0.1866176277399063,
      "learning_rate": 7.656063142788807e-05,
      "loss": 0.0192,
      "step": 16076
    },
    {
      "epoch": 2.253258584442887,
      "grad_norm": 0.2756612300872803,
      "learning_rate": 7.65462807940684e-05,
      "loss": 0.0309,
      "step": 16077
    },
    {
      "epoch": 2.2533987386124736,
      "grad_norm": 0.1412435621023178,
      "learning_rate": 7.653193016024874e-05,
      "loss": 0.026,
      "step": 16078
    },
    {
      "epoch": 2.25353889278206,
      "grad_norm": 0.25307074189186096,
      "learning_rate": 7.651757952642908e-05,
      "loss": 0.0254,
      "step": 16079
    },
    {
      "epoch": 2.253679046951647,
      "grad_norm": 0.1415809541940689,
      "learning_rate": 7.650322889260941e-05,
      "loss": 0.0106,
      "step": 16080
    },
    {
      "epoch": 2.2538192011212335,
      "grad_norm": 0.10863583534955978,
      "learning_rate": 7.648887825878975e-05,
      "loss": 0.0085,
      "step": 16081
    },
    {
      "epoch": 2.25395935529082,
      "grad_norm": 0.4418480396270752,
      "learning_rate": 7.64745276249701e-05,
      "loss": 0.0246,
      "step": 16082
    },
    {
      "epoch": 2.2540995094604064,
      "grad_norm": 0.42418670654296875,
      "learning_rate": 7.646017699115044e-05,
      "loss": 0.0483,
      "step": 16083
    },
    {
      "epoch": 2.254239663629993,
      "grad_norm": 0.14411413669586182,
      "learning_rate": 7.644582635733078e-05,
      "loss": 0.0125,
      "step": 16084
    },
    {
      "epoch": 2.2543798177995793,
      "grad_norm": 0.2587570250034332,
      "learning_rate": 7.643147572351111e-05,
      "loss": 0.0237,
      "step": 16085
    },
    {
      "epoch": 2.254519971969166,
      "grad_norm": 0.2747219204902649,
      "learning_rate": 7.641712508969145e-05,
      "loss": 0.0212,
      "step": 16086
    },
    {
      "epoch": 2.2546601261387527,
      "grad_norm": 0.20177651941776276,
      "learning_rate": 7.64027744558718e-05,
      "loss": 0.0405,
      "step": 16087
    },
    {
      "epoch": 2.254800280308339,
      "grad_norm": 0.29879772663116455,
      "learning_rate": 7.638842382205213e-05,
      "loss": 0.0432,
      "step": 16088
    },
    {
      "epoch": 2.2549404344779256,
      "grad_norm": 0.22972798347473145,
      "learning_rate": 7.637407318823248e-05,
      "loss": 0.0171,
      "step": 16089
    },
    {
      "epoch": 2.2550805886475125,
      "grad_norm": 0.2076868861913681,
      "learning_rate": 7.63597225544128e-05,
      "loss": 0.0285,
      "step": 16090
    },
    {
      "epoch": 2.255220742817099,
      "grad_norm": 0.1993919312953949,
      "learning_rate": 7.634537192059315e-05,
      "loss": 0.0241,
      "step": 16091
    },
    {
      "epoch": 2.2553608969866854,
      "grad_norm": 0.03470282629132271,
      "learning_rate": 7.63310212867735e-05,
      "loss": 0.0015,
      "step": 16092
    },
    {
      "epoch": 2.255501051156272,
      "grad_norm": 0.2611018419265747,
      "learning_rate": 7.631667065295383e-05,
      "loss": 0.0267,
      "step": 16093
    },
    {
      "epoch": 2.2556412053258583,
      "grad_norm": 0.17524974048137665,
      "learning_rate": 7.630232001913417e-05,
      "loss": 0.0086,
      "step": 16094
    },
    {
      "epoch": 2.255781359495445,
      "grad_norm": 0.26347264647483826,
      "learning_rate": 7.628796938531452e-05,
      "loss": 0.0216,
      "step": 16095
    },
    {
      "epoch": 2.2559215136650317,
      "grad_norm": 0.07756421715021133,
      "learning_rate": 7.627361875149484e-05,
      "loss": 0.0106,
      "step": 16096
    },
    {
      "epoch": 2.256061667834618,
      "grad_norm": 0.3247242569923401,
      "learning_rate": 7.625926811767519e-05,
      "loss": 0.0172,
      "step": 16097
    },
    {
      "epoch": 2.2562018220042046,
      "grad_norm": 0.2472013235092163,
      "learning_rate": 7.624491748385553e-05,
      "loss": 0.0317,
      "step": 16098
    },
    {
      "epoch": 2.256341976173791,
      "grad_norm": 0.14462490379810333,
      "learning_rate": 7.623056685003587e-05,
      "loss": 0.0346,
      "step": 16099
    },
    {
      "epoch": 2.2564821303433775,
      "grad_norm": 0.26738011837005615,
      "learning_rate": 7.621621621621621e-05,
      "loss": 0.0248,
      "step": 16100
    },
    {
      "epoch": 2.2566222845129644,
      "grad_norm": 0.5486509203910828,
      "learning_rate": 7.620186558239654e-05,
      "loss": 0.1267,
      "step": 16101
    },
    {
      "epoch": 2.256762438682551,
      "grad_norm": 0.3519798219203949,
      "learning_rate": 7.618751494857688e-05,
      "loss": 0.0141,
      "step": 16102
    },
    {
      "epoch": 2.2569025928521373,
      "grad_norm": 0.5758402943611145,
      "learning_rate": 7.617316431475724e-05,
      "loss": 0.0553,
      "step": 16103
    },
    {
      "epoch": 2.257042747021724,
      "grad_norm": 0.13290636241436005,
      "learning_rate": 7.615881368093757e-05,
      "loss": 0.0077,
      "step": 16104
    },
    {
      "epoch": 2.2571829011913103,
      "grad_norm": 0.16322965919971466,
      "learning_rate": 7.614446304711791e-05,
      "loss": 0.0212,
      "step": 16105
    },
    {
      "epoch": 2.257323055360897,
      "grad_norm": 0.46359366178512573,
      "learning_rate": 7.613011241329825e-05,
      "loss": 0.051,
      "step": 16106
    },
    {
      "epoch": 2.2574632095304836,
      "grad_norm": 0.43224605917930603,
      "learning_rate": 7.611576177947858e-05,
      "loss": 0.0202,
      "step": 16107
    },
    {
      "epoch": 2.25760336370007,
      "grad_norm": 0.04966854304075241,
      "learning_rate": 7.610141114565894e-05,
      "loss": 0.0023,
      "step": 16108
    },
    {
      "epoch": 2.2577435178696565,
      "grad_norm": 0.06300931423902512,
      "learning_rate": 7.608706051183926e-05,
      "loss": 0.0038,
      "step": 16109
    },
    {
      "epoch": 2.257883672039243,
      "grad_norm": 0.18797099590301514,
      "learning_rate": 7.607270987801961e-05,
      "loss": 0.0144,
      "step": 16110
    },
    {
      "epoch": 2.25802382620883,
      "grad_norm": 0.34303364157676697,
      "learning_rate": 7.605835924419995e-05,
      "loss": 0.0534,
      "step": 16111
    },
    {
      "epoch": 2.2581639803784164,
      "grad_norm": 0.24372737109661102,
      "learning_rate": 7.604400861038028e-05,
      "loss": 0.0282,
      "step": 16112
    },
    {
      "epoch": 2.258304134548003,
      "grad_norm": 0.20383049547672272,
      "learning_rate": 7.602965797656062e-05,
      "loss": 0.012,
      "step": 16113
    },
    {
      "epoch": 2.2584442887175893,
      "grad_norm": 0.311422199010849,
      "learning_rate": 7.601530734274098e-05,
      "loss": 0.0506,
      "step": 16114
    },
    {
      "epoch": 2.2585844428871757,
      "grad_norm": 0.15760605037212372,
      "learning_rate": 7.60009567089213e-05,
      "loss": 0.0128,
      "step": 16115
    },
    {
      "epoch": 2.258724597056762,
      "grad_norm": 0.49481332302093506,
      "learning_rate": 7.598660607510165e-05,
      "loss": 0.0366,
      "step": 16116
    },
    {
      "epoch": 2.258864751226349,
      "grad_norm": 0.14541682600975037,
      "learning_rate": 7.597225544128197e-05,
      "loss": 0.008,
      "step": 16117
    },
    {
      "epoch": 2.2590049053959356,
      "grad_norm": 0.47736671566963196,
      "learning_rate": 7.595790480746232e-05,
      "loss": 0.0372,
      "step": 16118
    },
    {
      "epoch": 2.259145059565522,
      "grad_norm": 0.4966481924057007,
      "learning_rate": 7.594355417364267e-05,
      "loss": 0.0194,
      "step": 16119
    },
    {
      "epoch": 2.2592852137351085,
      "grad_norm": 3.5510356426239014,
      "learning_rate": 7.5929203539823e-05,
      "loss": 0.1012,
      "step": 16120
    },
    {
      "epoch": 2.2594253679046954,
      "grad_norm": 0.2877337336540222,
      "learning_rate": 7.591485290600334e-05,
      "loss": 0.0432,
      "step": 16121
    },
    {
      "epoch": 2.259565522074282,
      "grad_norm": 0.27166762948036194,
      "learning_rate": 7.590050227218369e-05,
      "loss": 0.0482,
      "step": 16122
    },
    {
      "epoch": 2.2597056762438683,
      "grad_norm": 0.3315783441066742,
      "learning_rate": 7.588615163836401e-05,
      "loss": 0.0573,
      "step": 16123
    },
    {
      "epoch": 2.2598458304134548,
      "grad_norm": 0.396234929561615,
      "learning_rate": 7.587180100454437e-05,
      "loss": 0.0317,
      "step": 16124
    },
    {
      "epoch": 2.2599859845830412,
      "grad_norm": 0.3639105260372162,
      "learning_rate": 7.58574503707247e-05,
      "loss": 0.0306,
      "step": 16125
    },
    {
      "epoch": 2.2601261387526277,
      "grad_norm": 0.1023806780576706,
      "learning_rate": 7.584309973690504e-05,
      "loss": 0.0105,
      "step": 16126
    },
    {
      "epoch": 2.2602662929222146,
      "grad_norm": 0.25488799810409546,
      "learning_rate": 7.582874910308538e-05,
      "loss": 0.0243,
      "step": 16127
    },
    {
      "epoch": 2.260406447091801,
      "grad_norm": 0.15481606125831604,
      "learning_rate": 7.581439846926571e-05,
      "loss": 0.0096,
      "step": 16128
    },
    {
      "epoch": 2.2605466012613875,
      "grad_norm": 0.18601356446743011,
      "learning_rate": 7.580004783544605e-05,
      "loss": 0.0327,
      "step": 16129
    },
    {
      "epoch": 2.260686755430974,
      "grad_norm": 0.47852209210395813,
      "learning_rate": 7.578569720162641e-05,
      "loss": 0.0293,
      "step": 16130
    },
    {
      "epoch": 2.2608269096005604,
      "grad_norm": 0.19744735956192017,
      "learning_rate": 7.577134656780674e-05,
      "loss": 0.0232,
      "step": 16131
    },
    {
      "epoch": 2.2609670637701473,
      "grad_norm": 0.2642534077167511,
      "learning_rate": 7.575699593398708e-05,
      "loss": 0.0123,
      "step": 16132
    },
    {
      "epoch": 2.261107217939734,
      "grad_norm": 0.3753826320171356,
      "learning_rate": 7.574264530016741e-05,
      "loss": 0.0154,
      "step": 16133
    },
    {
      "epoch": 2.2612473721093203,
      "grad_norm": 0.3004189729690552,
      "learning_rate": 7.572829466634775e-05,
      "loss": 0.0578,
      "step": 16134
    },
    {
      "epoch": 2.2613875262789067,
      "grad_norm": 0.1730934977531433,
      "learning_rate": 7.57139440325281e-05,
      "loss": 0.0105,
      "step": 16135
    },
    {
      "epoch": 2.261527680448493,
      "grad_norm": 0.21417799592018127,
      "learning_rate": 7.569959339870843e-05,
      "loss": 0.0193,
      "step": 16136
    },
    {
      "epoch": 2.26166783461808,
      "grad_norm": 0.1666230410337448,
      "learning_rate": 7.568524276488878e-05,
      "loss": 0.0135,
      "step": 16137
    },
    {
      "epoch": 2.2618079887876665,
      "grad_norm": 0.22300715744495392,
      "learning_rate": 7.567089213106912e-05,
      "loss": 0.0158,
      "step": 16138
    },
    {
      "epoch": 2.261948142957253,
      "grad_norm": 0.08547382801771164,
      "learning_rate": 7.565654149724945e-05,
      "loss": 0.0045,
      "step": 16139
    },
    {
      "epoch": 2.2620882971268395,
      "grad_norm": 0.3082025349140167,
      "learning_rate": 7.56421908634298e-05,
      "loss": 0.0749,
      "step": 16140
    },
    {
      "epoch": 2.262228451296426,
      "grad_norm": 0.2884487807750702,
      "learning_rate": 7.562784022961014e-05,
      "loss": 0.0107,
      "step": 16141
    },
    {
      "epoch": 2.262368605466013,
      "grad_norm": 0.11946040391921997,
      "learning_rate": 7.561348959579047e-05,
      "loss": 0.0065,
      "step": 16142
    },
    {
      "epoch": 2.2625087596355993,
      "grad_norm": 0.5854082703590393,
      "learning_rate": 7.559913896197082e-05,
      "loss": 0.0187,
      "step": 16143
    },
    {
      "epoch": 2.2626489138051857,
      "grad_norm": 0.3553685247898102,
      "learning_rate": 7.558478832815114e-05,
      "loss": 0.0357,
      "step": 16144
    },
    {
      "epoch": 2.262789067974772,
      "grad_norm": 0.2248820662498474,
      "learning_rate": 7.557043769433149e-05,
      "loss": 0.03,
      "step": 16145
    },
    {
      "epoch": 2.2629292221443587,
      "grad_norm": 0.07247384637594223,
      "learning_rate": 7.555608706051184e-05,
      "loss": 0.0033,
      "step": 16146
    },
    {
      "epoch": 2.263069376313945,
      "grad_norm": 0.7017240524291992,
      "learning_rate": 7.554173642669217e-05,
      "loss": 0.0143,
      "step": 16147
    },
    {
      "epoch": 2.263209530483532,
      "grad_norm": 0.35459059476852417,
      "learning_rate": 7.552738579287251e-05,
      "loss": 0.0545,
      "step": 16148
    },
    {
      "epoch": 2.2633496846531185,
      "grad_norm": 0.9143280386924744,
      "learning_rate": 7.551303515905285e-05,
      "loss": 0.0349,
      "step": 16149
    },
    {
      "epoch": 2.263489838822705,
      "grad_norm": 0.12702693045139313,
      "learning_rate": 7.549868452523318e-05,
      "loss": 0.0186,
      "step": 16150
    },
    {
      "epoch": 2.2636299929922914,
      "grad_norm": 0.3911602199077606,
      "learning_rate": 7.548433389141354e-05,
      "loss": 0.0273,
      "step": 16151
    },
    {
      "epoch": 2.2637701471618783,
      "grad_norm": 0.17827431857585907,
      "learning_rate": 7.546998325759387e-05,
      "loss": 0.0231,
      "step": 16152
    },
    {
      "epoch": 2.2639103013314648,
      "grad_norm": 0.09816229343414307,
      "learning_rate": 7.545563262377421e-05,
      "loss": 0.0051,
      "step": 16153
    },
    {
      "epoch": 2.264050455501051,
      "grad_norm": 0.1798473596572876,
      "learning_rate": 7.544128198995455e-05,
      "loss": 0.0123,
      "step": 16154
    },
    {
      "epoch": 2.2641906096706377,
      "grad_norm": 0.215614914894104,
      "learning_rate": 7.542693135613488e-05,
      "loss": 0.0109,
      "step": 16155
    },
    {
      "epoch": 2.264330763840224,
      "grad_norm": 0.503149151802063,
      "learning_rate": 7.541258072231524e-05,
      "loss": 0.0283,
      "step": 16156
    },
    {
      "epoch": 2.2644709180098106,
      "grad_norm": 0.6698678135871887,
      "learning_rate": 7.539823008849558e-05,
      "loss": 0.0381,
      "step": 16157
    },
    {
      "epoch": 2.2646110721793975,
      "grad_norm": 0.29192933440208435,
      "learning_rate": 7.53838794546759e-05,
      "loss": 0.0331,
      "step": 16158
    },
    {
      "epoch": 2.264751226348984,
      "grad_norm": 0.1856585294008255,
      "learning_rate": 7.536952882085625e-05,
      "loss": 0.0413,
      "step": 16159
    },
    {
      "epoch": 2.2648913805185704,
      "grad_norm": 0.12938711047172546,
      "learning_rate": 7.535517818703658e-05,
      "loss": 0.0167,
      "step": 16160
    },
    {
      "epoch": 2.265031534688157,
      "grad_norm": 0.10505757480859756,
      "learning_rate": 7.534082755321693e-05,
      "loss": 0.0324,
      "step": 16161
    },
    {
      "epoch": 2.2651716888577433,
      "grad_norm": 0.5922462344169617,
      "learning_rate": 7.532647691939727e-05,
      "loss": 0.0323,
      "step": 16162
    },
    {
      "epoch": 2.2653118430273302,
      "grad_norm": 0.18770937621593475,
      "learning_rate": 7.53121262855776e-05,
      "loss": 0.0343,
      "step": 16163
    },
    {
      "epoch": 2.2654519971969167,
      "grad_norm": 0.06127657741308212,
      "learning_rate": 7.529777565175795e-05,
      "loss": 0.0041,
      "step": 16164
    },
    {
      "epoch": 2.265592151366503,
      "grad_norm": 0.3119257688522339,
      "learning_rate": 7.528342501793829e-05,
      "loss": 0.0599,
      "step": 16165
    },
    {
      "epoch": 2.2657323055360896,
      "grad_norm": 0.9948211312294006,
      "learning_rate": 7.526907438411862e-05,
      "loss": 0.0591,
      "step": 16166
    },
    {
      "epoch": 2.265872459705676,
      "grad_norm": 0.3948303759098053,
      "learning_rate": 7.525472375029897e-05,
      "loss": 0.0331,
      "step": 16167
    },
    {
      "epoch": 2.266012613875263,
      "grad_norm": 0.7139501571655273,
      "learning_rate": 7.52403731164793e-05,
      "loss": 0.0334,
      "step": 16168
    },
    {
      "epoch": 2.2661527680448494,
      "grad_norm": 0.5135822296142578,
      "learning_rate": 7.522602248265964e-05,
      "loss": 0.0252,
      "step": 16169
    },
    {
      "epoch": 2.266292922214436,
      "grad_norm": 0.724551796913147,
      "learning_rate": 7.521167184883998e-05,
      "loss": 0.0818,
      "step": 16170
    },
    {
      "epoch": 2.2664330763840224,
      "grad_norm": 0.1179722398519516,
      "learning_rate": 7.519732121502031e-05,
      "loss": 0.0134,
      "step": 16171
    },
    {
      "epoch": 2.266573230553609,
      "grad_norm": 0.12055621296167374,
      "learning_rate": 7.518297058120067e-05,
      "loss": 0.0155,
      "step": 16172
    },
    {
      "epoch": 2.2667133847231957,
      "grad_norm": 0.22950193285942078,
      "learning_rate": 7.516861994738101e-05,
      "loss": 0.0118,
      "step": 16173
    },
    {
      "epoch": 2.266853538892782,
      "grad_norm": 0.3248205780982971,
      "learning_rate": 7.515426931356134e-05,
      "loss": 0.034,
      "step": 16174
    },
    {
      "epoch": 2.2669936930623686,
      "grad_norm": 0.13271230459213257,
      "learning_rate": 7.513991867974168e-05,
      "loss": 0.0245,
      "step": 16175
    },
    {
      "epoch": 2.267133847231955,
      "grad_norm": 0.1510247141122818,
      "learning_rate": 7.512556804592202e-05,
      "loss": 0.0148,
      "step": 16176
    },
    {
      "epoch": 2.2672740014015416,
      "grad_norm": 0.33316341042518616,
      "learning_rate": 7.511121741210237e-05,
      "loss": 0.027,
      "step": 16177
    },
    {
      "epoch": 2.267414155571128,
      "grad_norm": 0.3815443515777588,
      "learning_rate": 7.509686677828271e-05,
      "loss": 0.0219,
      "step": 16178
    },
    {
      "epoch": 2.267554309740715,
      "grad_norm": 0.614569365978241,
      "learning_rate": 7.508251614446304e-05,
      "loss": 0.0486,
      "step": 16179
    },
    {
      "epoch": 2.2676944639103014,
      "grad_norm": 0.08508642762899399,
      "learning_rate": 7.506816551064338e-05,
      "loss": 0.0172,
      "step": 16180
    },
    {
      "epoch": 2.267834618079888,
      "grad_norm": 0.45303839445114136,
      "learning_rate": 7.505381487682372e-05,
      "loss": 0.0399,
      "step": 16181
    },
    {
      "epoch": 2.2679747722494743,
      "grad_norm": 0.31105920672416687,
      "learning_rate": 7.503946424300405e-05,
      "loss": 0.0438,
      "step": 16182
    },
    {
      "epoch": 2.268114926419061,
      "grad_norm": 0.16513462364673615,
      "learning_rate": 7.50251136091844e-05,
      "loss": 0.012,
      "step": 16183
    },
    {
      "epoch": 2.2682550805886477,
      "grad_norm": 0.21608415246009827,
      "learning_rate": 7.501076297536475e-05,
      "loss": 0.0267,
      "step": 16184
    },
    {
      "epoch": 2.268395234758234,
      "grad_norm": 0.13387833535671234,
      "learning_rate": 7.499641234154508e-05,
      "loss": 0.0172,
      "step": 16185
    },
    {
      "epoch": 2.2685353889278206,
      "grad_norm": 0.10963825136423111,
      "learning_rate": 7.498206170772542e-05,
      "loss": 0.0136,
      "step": 16186
    },
    {
      "epoch": 2.268675543097407,
      "grad_norm": 0.19476504623889923,
      "learning_rate": 7.496771107390576e-05,
      "loss": 0.0236,
      "step": 16187
    },
    {
      "epoch": 2.2688156972669935,
      "grad_norm": 0.136983260512352,
      "learning_rate": 7.49533604400861e-05,
      "loss": 0.0141,
      "step": 16188
    },
    {
      "epoch": 2.2689558514365804,
      "grad_norm": 0.12143999338150024,
      "learning_rate": 7.493900980626643e-05,
      "loss": 0.0031,
      "step": 16189
    },
    {
      "epoch": 2.269096005606167,
      "grad_norm": 0.14770960807800293,
      "learning_rate": 7.492465917244679e-05,
      "loss": 0.0112,
      "step": 16190
    },
    {
      "epoch": 2.2692361597757533,
      "grad_norm": 0.09765385836362839,
      "learning_rate": 7.491030853862711e-05,
      "loss": 0.0179,
      "step": 16191
    },
    {
      "epoch": 2.26937631394534,
      "grad_norm": 0.10464373975992203,
      "learning_rate": 7.489595790480746e-05,
      "loss": 0.0078,
      "step": 16192
    },
    {
      "epoch": 2.2695164681149262,
      "grad_norm": 0.49572744965553284,
      "learning_rate": 7.48816072709878e-05,
      "loss": 0.054,
      "step": 16193
    },
    {
      "epoch": 2.269656622284513,
      "grad_norm": 0.14305707812309265,
      "learning_rate": 7.486725663716814e-05,
      "loss": 0.0137,
      "step": 16194
    },
    {
      "epoch": 2.2697967764540996,
      "grad_norm": 0.5051373839378357,
      "learning_rate": 7.485290600334847e-05,
      "loss": 0.0297,
      "step": 16195
    },
    {
      "epoch": 2.269936930623686,
      "grad_norm": 0.31745806336402893,
      "learning_rate": 7.483855536952881e-05,
      "loss": 0.0292,
      "step": 16196
    },
    {
      "epoch": 2.2700770847932725,
      "grad_norm": 0.23002316057682037,
      "learning_rate": 7.482420473570915e-05,
      "loss": 0.0342,
      "step": 16197
    },
    {
      "epoch": 2.270217238962859,
      "grad_norm": 0.4532015919685364,
      "learning_rate": 7.48098541018895e-05,
      "loss": 0.0519,
      "step": 16198
    },
    {
      "epoch": 2.2703573931324454,
      "grad_norm": 0.18719851970672607,
      "learning_rate": 7.479550346806984e-05,
      "loss": 0.0075,
      "step": 16199
    },
    {
      "epoch": 2.2704975473020323,
      "grad_norm": 0.19947662949562073,
      "learning_rate": 7.478115283425017e-05,
      "loss": 0.0252,
      "step": 16200
    },
    {
      "epoch": 2.270637701471619,
      "grad_norm": 0.42769742012023926,
      "learning_rate": 7.476680220043051e-05,
      "loss": 0.0301,
      "step": 16201
    },
    {
      "epoch": 2.2707778556412053,
      "grad_norm": 0.5155138969421387,
      "learning_rate": 7.475245156661085e-05,
      "loss": 0.0414,
      "step": 16202
    },
    {
      "epoch": 2.2709180098107917,
      "grad_norm": 0.42418935894966125,
      "learning_rate": 7.473810093279119e-05,
      "loss": 0.0323,
      "step": 16203
    },
    {
      "epoch": 2.2710581639803786,
      "grad_norm": 0.15725037455558777,
      "learning_rate": 7.472375029897154e-05,
      "loss": 0.0108,
      "step": 16204
    },
    {
      "epoch": 2.271198318149965,
      "grad_norm": 0.15566274523735046,
      "learning_rate": 7.470939966515186e-05,
      "loss": 0.0129,
      "step": 16205
    },
    {
      "epoch": 2.2713384723195515,
      "grad_norm": 0.17007608711719513,
      "learning_rate": 7.469504903133222e-05,
      "loss": 0.0146,
      "step": 16206
    },
    {
      "epoch": 2.271478626489138,
      "grad_norm": 0.28880584239959717,
      "learning_rate": 7.468069839751255e-05,
      "loss": 0.0584,
      "step": 16207
    },
    {
      "epoch": 2.2716187806587245,
      "grad_norm": 0.4530080258846283,
      "learning_rate": 7.466634776369289e-05,
      "loss": 0.0447,
      "step": 16208
    },
    {
      "epoch": 2.271758934828311,
      "grad_norm": 0.09833940863609314,
      "learning_rate": 7.465199712987323e-05,
      "loss": 0.0067,
      "step": 16209
    },
    {
      "epoch": 2.271899088997898,
      "grad_norm": 0.19690768420696259,
      "learning_rate": 7.463764649605357e-05,
      "loss": 0.0233,
      "step": 16210
    },
    {
      "epoch": 2.2720392431674843,
      "grad_norm": 0.04336480423808098,
      "learning_rate": 7.46232958622339e-05,
      "loss": 0.0022,
      "step": 16211
    },
    {
      "epoch": 2.2721793973370707,
      "grad_norm": 0.06939428299665451,
      "learning_rate": 7.460894522841424e-05,
      "loss": 0.0143,
      "step": 16212
    },
    {
      "epoch": 2.272319551506657,
      "grad_norm": 0.1259625256061554,
      "learning_rate": 7.459459459459459e-05,
      "loss": 0.0082,
      "step": 16213
    },
    {
      "epoch": 2.272459705676244,
      "grad_norm": 0.05249089375138283,
      "learning_rate": 7.458024396077493e-05,
      "loss": 0.0053,
      "step": 16214
    },
    {
      "epoch": 2.2725998598458306,
      "grad_norm": 0.4364691376686096,
      "learning_rate": 7.456589332695527e-05,
      "loss": 0.0227,
      "step": 16215
    },
    {
      "epoch": 2.272740014015417,
      "grad_norm": 0.3146170377731323,
      "learning_rate": 7.45515426931356e-05,
      "loss": 0.0534,
      "step": 16216
    },
    {
      "epoch": 2.2728801681850035,
      "grad_norm": 0.4993615746498108,
      "learning_rate": 7.453719205931596e-05,
      "loss": 0.1025,
      "step": 16217
    },
    {
      "epoch": 2.27302032235459,
      "grad_norm": 0.1917387843132019,
      "learning_rate": 7.452284142549628e-05,
      "loss": 0.0235,
      "step": 16218
    },
    {
      "epoch": 2.2731604765241764,
      "grad_norm": 1.1244992017745972,
      "learning_rate": 7.450849079167663e-05,
      "loss": 0.1698,
      "step": 16219
    },
    {
      "epoch": 2.2733006306937633,
      "grad_norm": 1.2356315851211548,
      "learning_rate": 7.449414015785697e-05,
      "loss": 0.0946,
      "step": 16220
    },
    {
      "epoch": 2.2734407848633498,
      "grad_norm": 0.18650762736797333,
      "learning_rate": 7.447978952403731e-05,
      "loss": 0.0213,
      "step": 16221
    },
    {
      "epoch": 2.2735809390329362,
      "grad_norm": 0.3772331774234772,
      "learning_rate": 7.446543889021765e-05,
      "loss": 0.0249,
      "step": 16222
    },
    {
      "epoch": 2.2737210932025227,
      "grad_norm": 0.18132229149341583,
      "learning_rate": 7.445108825639798e-05,
      "loss": 0.0068,
      "step": 16223
    },
    {
      "epoch": 2.273861247372109,
      "grad_norm": 0.06925246864557266,
      "learning_rate": 7.443673762257832e-05,
      "loss": 0.0054,
      "step": 16224
    },
    {
      "epoch": 2.274001401541696,
      "grad_norm": 0.1610412895679474,
      "learning_rate": 7.442238698875867e-05,
      "loss": 0.0267,
      "step": 16225
    },
    {
      "epoch": 2.2741415557112825,
      "grad_norm": 0.23465511202812195,
      "learning_rate": 7.440803635493901e-05,
      "loss": 0.0323,
      "step": 16226
    },
    {
      "epoch": 2.274281709880869,
      "grad_norm": 0.18110215663909912,
      "learning_rate": 7.439368572111934e-05,
      "loss": 0.0254,
      "step": 16227
    },
    {
      "epoch": 2.2744218640504554,
      "grad_norm": 0.018187547102570534,
      "learning_rate": 7.437933508729968e-05,
      "loss": 0.0015,
      "step": 16228
    },
    {
      "epoch": 2.274562018220042,
      "grad_norm": 0.2320176512002945,
      "learning_rate": 7.436498445348002e-05,
      "loss": 0.0791,
      "step": 16229
    },
    {
      "epoch": 2.2747021723896284,
      "grad_norm": 0.3378334641456604,
      "learning_rate": 7.435063381966036e-05,
      "loss": 0.008,
      "step": 16230
    },
    {
      "epoch": 2.2748423265592153,
      "grad_norm": 0.18843097984790802,
      "learning_rate": 7.43362831858407e-05,
      "loss": 0.0235,
      "step": 16231
    },
    {
      "epoch": 2.2749824807288017,
      "grad_norm": 0.1092878207564354,
      "learning_rate": 7.432193255202103e-05,
      "loss": 0.0112,
      "step": 16232
    },
    {
      "epoch": 2.275122634898388,
      "grad_norm": 0.4976484179496765,
      "learning_rate": 7.430758191820139e-05,
      "loss": 0.048,
      "step": 16233
    },
    {
      "epoch": 2.2752627890679746,
      "grad_norm": 0.2779671549797058,
      "learning_rate": 7.429323128438172e-05,
      "loss": 0.0705,
      "step": 16234
    },
    {
      "epoch": 2.2754029432375615,
      "grad_norm": 0.3394249975681305,
      "learning_rate": 7.427888065056206e-05,
      "loss": 0.0597,
      "step": 16235
    },
    {
      "epoch": 2.275543097407148,
      "grad_norm": 0.044666409492492676,
      "learning_rate": 7.42645300167424e-05,
      "loss": 0.0041,
      "step": 16236
    },
    {
      "epoch": 2.2756832515767345,
      "grad_norm": 0.15766453742980957,
      "learning_rate": 7.425017938292274e-05,
      "loss": 0.0391,
      "step": 16237
    },
    {
      "epoch": 2.275823405746321,
      "grad_norm": 0.5730196833610535,
      "learning_rate": 7.423582874910309e-05,
      "loss": 0.0676,
      "step": 16238
    },
    {
      "epoch": 2.2759635599159074,
      "grad_norm": 0.42103469371795654,
      "learning_rate": 7.422147811528341e-05,
      "loss": 0.0387,
      "step": 16239
    },
    {
      "epoch": 2.276103714085494,
      "grad_norm": 0.7110700607299805,
      "learning_rate": 7.420712748146376e-05,
      "loss": 0.0169,
      "step": 16240
    },
    {
      "epoch": 2.2762438682550807,
      "grad_norm": 0.22068460285663605,
      "learning_rate": 7.41927768476441e-05,
      "loss": 0.0225,
      "step": 16241
    },
    {
      "epoch": 2.276384022424667,
      "grad_norm": 0.1457379162311554,
      "learning_rate": 7.417842621382444e-05,
      "loss": 0.0213,
      "step": 16242
    },
    {
      "epoch": 2.2765241765942537,
      "grad_norm": 0.21809007227420807,
      "learning_rate": 7.416407558000477e-05,
      "loss": 0.0644,
      "step": 16243
    },
    {
      "epoch": 2.27666433076384,
      "grad_norm": 0.5167080163955688,
      "learning_rate": 7.414972494618511e-05,
      "loss": 0.0506,
      "step": 16244
    },
    {
      "epoch": 2.2768044849334266,
      "grad_norm": 0.3382079303264618,
      "learning_rate": 7.413537431236545e-05,
      "loss": 0.0261,
      "step": 16245
    },
    {
      "epoch": 2.2769446391030135,
      "grad_norm": 0.6171927452087402,
      "learning_rate": 7.41210236785458e-05,
      "loss": 0.0565,
      "step": 16246
    },
    {
      "epoch": 2.2770847932726,
      "grad_norm": 0.26119115948677063,
      "learning_rate": 7.410667304472614e-05,
      "loss": 0.0197,
      "step": 16247
    },
    {
      "epoch": 2.2772249474421864,
      "grad_norm": 0.4102140963077545,
      "learning_rate": 7.409232241090647e-05,
      "loss": 0.0512,
      "step": 16248
    },
    {
      "epoch": 2.277365101611773,
      "grad_norm": 0.05564238131046295,
      "learning_rate": 7.407797177708682e-05,
      "loss": 0.0061,
      "step": 16249
    },
    {
      "epoch": 2.2775052557813593,
      "grad_norm": 0.2707335948944092,
      "learning_rate": 7.406362114326715e-05,
      "loss": 0.0798,
      "step": 16250
    },
    {
      "epoch": 2.277645409950946,
      "grad_norm": 0.2930961549282074,
      "learning_rate": 7.404927050944749e-05,
      "loss": 0.0259,
      "step": 16251
    },
    {
      "epoch": 2.2777855641205327,
      "grad_norm": 0.1461946666240692,
      "learning_rate": 7.403491987562783e-05,
      "loss": 0.0183,
      "step": 16252
    },
    {
      "epoch": 2.277925718290119,
      "grad_norm": 0.08901000022888184,
      "learning_rate": 7.402056924180818e-05,
      "loss": 0.0108,
      "step": 16253
    },
    {
      "epoch": 2.2780658724597056,
      "grad_norm": 0.8977463841438293,
      "learning_rate": 7.400621860798852e-05,
      "loss": 0.0281,
      "step": 16254
    },
    {
      "epoch": 2.278206026629292,
      "grad_norm": 0.36621615290641785,
      "learning_rate": 7.399186797416885e-05,
      "loss": 0.0397,
      "step": 16255
    },
    {
      "epoch": 2.278346180798879,
      "grad_norm": 0.185032457113266,
      "learning_rate": 7.397751734034919e-05,
      "loss": 0.0104,
      "step": 16256
    },
    {
      "epoch": 2.2784863349684654,
      "grad_norm": 0.391226202249527,
      "learning_rate": 7.396316670652953e-05,
      "loss": 0.0147,
      "step": 16257
    },
    {
      "epoch": 2.278626489138052,
      "grad_norm": 0.1199955940246582,
      "learning_rate": 7.394881607270987e-05,
      "loss": 0.0192,
      "step": 16258
    },
    {
      "epoch": 2.2787666433076383,
      "grad_norm": 0.12605716288089752,
      "learning_rate": 7.39344654388902e-05,
      "loss": 0.0052,
      "step": 16259
    },
    {
      "epoch": 2.278906797477225,
      "grad_norm": 0.1453598141670227,
      "learning_rate": 7.392011480507056e-05,
      "loss": 0.0147,
      "step": 16260
    },
    {
      "epoch": 2.2790469516468113,
      "grad_norm": 0.20824269950389862,
      "learning_rate": 7.390576417125089e-05,
      "loss": 0.0097,
      "step": 16261
    },
    {
      "epoch": 2.279187105816398,
      "grad_norm": 0.29285532236099243,
      "learning_rate": 7.389141353743123e-05,
      "loss": 0.0183,
      "step": 16262
    },
    {
      "epoch": 2.2793272599859846,
      "grad_norm": 0.2977740466594696,
      "learning_rate": 7.387706290361157e-05,
      "loss": 0.0475,
      "step": 16263
    },
    {
      "epoch": 2.279467414155571,
      "grad_norm": 0.32792624831199646,
      "learning_rate": 7.386271226979191e-05,
      "loss": 0.0275,
      "step": 16264
    },
    {
      "epoch": 2.2796075683251575,
      "grad_norm": 1.1770472526550293,
      "learning_rate": 7.384836163597225e-05,
      "loss": 0.1272,
      "step": 16265
    },
    {
      "epoch": 2.2797477224947444,
      "grad_norm": 0.22878652811050415,
      "learning_rate": 7.383401100215258e-05,
      "loss": 0.0127,
      "step": 16266
    },
    {
      "epoch": 2.279887876664331,
      "grad_norm": 0.6441991329193115,
      "learning_rate": 7.381966036833293e-05,
      "loss": 0.0255,
      "step": 16267
    },
    {
      "epoch": 2.2800280308339174,
      "grad_norm": 0.4638057053089142,
      "learning_rate": 7.380530973451327e-05,
      "loss": 0.0321,
      "step": 16268
    },
    {
      "epoch": 2.280168185003504,
      "grad_norm": 1.3254001140594482,
      "learning_rate": 7.379095910069361e-05,
      "loss": 0.1844,
      "step": 16269
    },
    {
      "epoch": 2.2803083391730903,
      "grad_norm": 1.0455617904663086,
      "learning_rate": 7.377660846687395e-05,
      "loss": 0.0733,
      "step": 16270
    },
    {
      "epoch": 2.2804484933426767,
      "grad_norm": 0.3305984437465668,
      "learning_rate": 7.376225783305428e-05,
      "loss": 0.0525,
      "step": 16271
    },
    {
      "epoch": 2.2805886475122636,
      "grad_norm": 0.12932036817073822,
      "learning_rate": 7.374790719923464e-05,
      "loss": 0.0126,
      "step": 16272
    },
    {
      "epoch": 2.28072880168185,
      "grad_norm": 0.1050378680229187,
      "learning_rate": 7.373355656541496e-05,
      "loss": 0.0111,
      "step": 16273
    },
    {
      "epoch": 2.2808689558514366,
      "grad_norm": 0.1309729963541031,
      "learning_rate": 7.37192059315953e-05,
      "loss": 0.0075,
      "step": 16274
    },
    {
      "epoch": 2.281009110021023,
      "grad_norm": 0.3628782331943512,
      "learning_rate": 7.370485529777564e-05,
      "loss": 0.0314,
      "step": 16275
    },
    {
      "epoch": 2.2811492641906095,
      "grad_norm": 0.10706055164337158,
      "learning_rate": 7.369050466395599e-05,
      "loss": 0.0133,
      "step": 16276
    },
    {
      "epoch": 2.2812894183601964,
      "grad_norm": 0.08899001032114029,
      "learning_rate": 7.367615403013632e-05,
      "loss": 0.0122,
      "step": 16277
    },
    {
      "epoch": 2.281429572529783,
      "grad_norm": 0.28892695903778076,
      "learning_rate": 7.366180339631666e-05,
      "loss": 0.0502,
      "step": 16278
    },
    {
      "epoch": 2.2815697266993693,
      "grad_norm": 0.2373223453760147,
      "learning_rate": 7.3647452762497e-05,
      "loss": 0.0252,
      "step": 16279
    },
    {
      "epoch": 2.2817098808689558,
      "grad_norm": 0.08226356655359268,
      "learning_rate": 7.363310212867735e-05,
      "loss": 0.0115,
      "step": 16280
    },
    {
      "epoch": 2.2818500350385422,
      "grad_norm": 0.1792638748884201,
      "learning_rate": 7.361875149485769e-05,
      "loss": 0.0246,
      "step": 16281
    },
    {
      "epoch": 2.281990189208129,
      "grad_norm": 0.08150475472211838,
      "learning_rate": 7.360440086103802e-05,
      "loss": 0.0062,
      "step": 16282
    },
    {
      "epoch": 2.2821303433777156,
      "grad_norm": 0.1887059211730957,
      "learning_rate": 7.359005022721836e-05,
      "loss": 0.009,
      "step": 16283
    },
    {
      "epoch": 2.282270497547302,
      "grad_norm": 0.05382366105914116,
      "learning_rate": 7.35756995933987e-05,
      "loss": 0.0022,
      "step": 16284
    },
    {
      "epoch": 2.2824106517168885,
      "grad_norm": 0.4411434233188629,
      "learning_rate": 7.356134895957904e-05,
      "loss": 0.024,
      "step": 16285
    },
    {
      "epoch": 2.282550805886475,
      "grad_norm": 0.13121947646141052,
      "learning_rate": 7.354699832575939e-05,
      "loss": 0.0256,
      "step": 16286
    },
    {
      "epoch": 2.282690960056062,
      "grad_norm": 0.2541201710700989,
      "learning_rate": 7.353264769193973e-05,
      "loss": 0.0196,
      "step": 16287
    },
    {
      "epoch": 2.2828311142256483,
      "grad_norm": 0.2207149714231491,
      "learning_rate": 7.351829705812007e-05,
      "loss": 0.0294,
      "step": 16288
    },
    {
      "epoch": 2.282971268395235,
      "grad_norm": 0.24132980406284332,
      "learning_rate": 7.35039464243004e-05,
      "loss": 0.0539,
      "step": 16289
    },
    {
      "epoch": 2.2831114225648212,
      "grad_norm": 0.40868696570396423,
      "learning_rate": 7.348959579048074e-05,
      "loss": 0.033,
      "step": 16290
    },
    {
      "epoch": 2.2832515767344077,
      "grad_norm": 0.31463509798049927,
      "learning_rate": 7.347524515666108e-05,
      "loss": 0.0759,
      "step": 16291
    },
    {
      "epoch": 2.283391730903994,
      "grad_norm": 0.13443753123283386,
      "learning_rate": 7.346089452284142e-05,
      "loss": 0.0389,
      "step": 16292
    },
    {
      "epoch": 2.283531885073581,
      "grad_norm": 0.29175665974617004,
      "learning_rate": 7.344654388902175e-05,
      "loss": 0.0554,
      "step": 16293
    },
    {
      "epoch": 2.2836720392431675,
      "grad_norm": 0.25446051359176636,
      "learning_rate": 7.34321932552021e-05,
      "loss": 0.0358,
      "step": 16294
    },
    {
      "epoch": 2.283812193412754,
      "grad_norm": 0.47193390130996704,
      "learning_rate": 7.341784262138244e-05,
      "loss": 0.0573,
      "step": 16295
    },
    {
      "epoch": 2.2839523475823404,
      "grad_norm": 0.2153552621603012,
      "learning_rate": 7.340349198756278e-05,
      "loss": 0.0265,
      "step": 16296
    },
    {
      "epoch": 2.2840925017519274,
      "grad_norm": 0.1195971667766571,
      "learning_rate": 7.338914135374312e-05,
      "loss": 0.0187,
      "step": 16297
    },
    {
      "epoch": 2.284232655921514,
      "grad_norm": 0.1436283141374588,
      "learning_rate": 7.337479071992345e-05,
      "loss": 0.0197,
      "step": 16298
    },
    {
      "epoch": 2.2843728100911003,
      "grad_norm": 0.06564446538686752,
      "learning_rate": 7.33604400861038e-05,
      "loss": 0.003,
      "step": 16299
    },
    {
      "epoch": 2.2845129642606867,
      "grad_norm": 0.29224902391433716,
      "learning_rate": 7.334608945228413e-05,
      "loss": 0.0384,
      "step": 16300
    },
    {
      "epoch": 2.284653118430273,
      "grad_norm": 0.16283342242240906,
      "learning_rate": 7.333173881846448e-05,
      "loss": 0.0364,
      "step": 16301
    },
    {
      "epoch": 2.2847932725998596,
      "grad_norm": 0.2010299563407898,
      "learning_rate": 7.331738818464482e-05,
      "loss": 0.0303,
      "step": 16302
    },
    {
      "epoch": 2.2849334267694466,
      "grad_norm": 0.18469111621379852,
      "learning_rate": 7.330303755082516e-05,
      "loss": 0.0181,
      "step": 16303
    },
    {
      "epoch": 2.285073580939033,
      "grad_norm": 0.007796700578182936,
      "learning_rate": 7.32886869170055e-05,
      "loss": 0.0009,
      "step": 16304
    },
    {
      "epoch": 2.2852137351086195,
      "grad_norm": 0.08528345078229904,
      "learning_rate": 7.327433628318583e-05,
      "loss": 0.0079,
      "step": 16305
    },
    {
      "epoch": 2.285353889278206,
      "grad_norm": 0.5259838104248047,
      "learning_rate": 7.325998564936617e-05,
      "loss": 0.0759,
      "step": 16306
    },
    {
      "epoch": 2.2854940434477924,
      "grad_norm": 0.224366694688797,
      "learning_rate": 7.324563501554652e-05,
      "loss": 0.0124,
      "step": 16307
    },
    {
      "epoch": 2.2856341976173793,
      "grad_norm": 0.21579058468341827,
      "learning_rate": 7.323128438172686e-05,
      "loss": 0.0183,
      "step": 16308
    },
    {
      "epoch": 2.2857743517869658,
      "grad_norm": 0.19897300004959106,
      "learning_rate": 7.321693374790719e-05,
      "loss": 0.0107,
      "step": 16309
    },
    {
      "epoch": 2.285914505956552,
      "grad_norm": 0.20075054466724396,
      "learning_rate": 7.320258311408753e-05,
      "loss": 0.0172,
      "step": 16310
    },
    {
      "epoch": 2.2860546601261387,
      "grad_norm": 0.17577716708183289,
      "learning_rate": 7.318823248026787e-05,
      "loss": 0.0063,
      "step": 16311
    },
    {
      "epoch": 2.286194814295725,
      "grad_norm": 0.272735595703125,
      "learning_rate": 7.317388184644821e-05,
      "loss": 0.0192,
      "step": 16312
    },
    {
      "epoch": 2.286334968465312,
      "grad_norm": 0.24795500934123993,
      "learning_rate": 7.315953121262855e-05,
      "loss": 0.0545,
      "step": 16313
    },
    {
      "epoch": 2.2864751226348985,
      "grad_norm": 0.12107976526021957,
      "learning_rate": 7.314518057880888e-05,
      "loss": 0.0099,
      "step": 16314
    },
    {
      "epoch": 2.286615276804485,
      "grad_norm": 0.8941563963890076,
      "learning_rate": 7.313082994498924e-05,
      "loss": 0.1324,
      "step": 16315
    },
    {
      "epoch": 2.2867554309740714,
      "grad_norm": 0.04670083522796631,
      "learning_rate": 7.311647931116957e-05,
      "loss": 0.0012,
      "step": 16316
    },
    {
      "epoch": 2.286895585143658,
      "grad_norm": 0.36447498202323914,
      "learning_rate": 7.310212867734991e-05,
      "loss": 0.044,
      "step": 16317
    },
    {
      "epoch": 2.2870357393132448,
      "grad_norm": 2.088002920150757,
      "learning_rate": 7.308777804353025e-05,
      "loss": 0.0318,
      "step": 16318
    },
    {
      "epoch": 2.2871758934828312,
      "grad_norm": 0.15274091064929962,
      "learning_rate": 7.30734274097106e-05,
      "loss": 0.0032,
      "step": 16319
    },
    {
      "epoch": 2.2873160476524177,
      "grad_norm": 2.037416696548462,
      "learning_rate": 7.305907677589094e-05,
      "loss": 0.1003,
      "step": 16320
    },
    {
      "epoch": 2.287456201822004,
      "grad_norm": 0.3803187906742096,
      "learning_rate": 7.304472614207126e-05,
      "loss": 0.0916,
      "step": 16321
    },
    {
      "epoch": 2.2875963559915906,
      "grad_norm": 0.6628861427307129,
      "learning_rate": 7.30303755082516e-05,
      "loss": 0.0753,
      "step": 16322
    },
    {
      "epoch": 2.287736510161177,
      "grad_norm": 0.47854968905448914,
      "learning_rate": 7.301602487443195e-05,
      "loss": 0.0316,
      "step": 16323
    },
    {
      "epoch": 2.287876664330764,
      "grad_norm": 0.3248313367366791,
      "learning_rate": 7.300167424061229e-05,
      "loss": 0.0359,
      "step": 16324
    },
    {
      "epoch": 2.2880168185003504,
      "grad_norm": 0.2358720898628235,
      "learning_rate": 7.298732360679262e-05,
      "loss": 0.0341,
      "step": 16325
    },
    {
      "epoch": 2.288156972669937,
      "grad_norm": 0.08605606108903885,
      "learning_rate": 7.297297297297297e-05,
      "loss": 0.0053,
      "step": 16326
    },
    {
      "epoch": 2.2882971268395234,
      "grad_norm": 0.2811298370361328,
      "learning_rate": 7.29586223391533e-05,
      "loss": 0.0236,
      "step": 16327
    },
    {
      "epoch": 2.2884372810091103,
      "grad_norm": 0.33712103962898254,
      "learning_rate": 7.294427170533365e-05,
      "loss": 0.0477,
      "step": 16328
    },
    {
      "epoch": 2.2885774351786967,
      "grad_norm": 0.2777279317378998,
      "learning_rate": 7.292992107151399e-05,
      "loss": 0.0284,
      "step": 16329
    },
    {
      "epoch": 2.288717589348283,
      "grad_norm": 0.25908640027046204,
      "learning_rate": 7.291557043769433e-05,
      "loss": 0.0489,
      "step": 16330
    },
    {
      "epoch": 2.2888577435178696,
      "grad_norm": 0.1270764172077179,
      "learning_rate": 7.290121980387467e-05,
      "loss": 0.0164,
      "step": 16331
    },
    {
      "epoch": 2.288997897687456,
      "grad_norm": 0.21425771713256836,
      "learning_rate": 7.2886869170055e-05,
      "loss": 0.0117,
      "step": 16332
    },
    {
      "epoch": 2.2891380518570426,
      "grad_norm": 0.23401951789855957,
      "learning_rate": 7.287251853623534e-05,
      "loss": 0.0491,
      "step": 16333
    },
    {
      "epoch": 2.2892782060266295,
      "grad_norm": 0.3998829424381256,
      "learning_rate": 7.285816790241568e-05,
      "loss": 0.0284,
      "step": 16334
    },
    {
      "epoch": 2.289418360196216,
      "grad_norm": 0.08534751087427139,
      "learning_rate": 7.284381726859603e-05,
      "loss": 0.0142,
      "step": 16335
    },
    {
      "epoch": 2.2895585143658024,
      "grad_norm": 0.36604636907577515,
      "learning_rate": 7.282946663477637e-05,
      "loss": 0.0983,
      "step": 16336
    },
    {
      "epoch": 2.289698668535389,
      "grad_norm": 0.08008506149053574,
      "learning_rate": 7.28151160009567e-05,
      "loss": 0.0088,
      "step": 16337
    },
    {
      "epoch": 2.2898388227049753,
      "grad_norm": 0.2987242639064789,
      "learning_rate": 7.280076536713704e-05,
      "loss": 0.0384,
      "step": 16338
    },
    {
      "epoch": 2.289978976874562,
      "grad_norm": 0.2669370770454407,
      "learning_rate": 7.278641473331738e-05,
      "loss": 0.0319,
      "step": 16339
    },
    {
      "epoch": 2.2901191310441487,
      "grad_norm": 0.08875153213739395,
      "learning_rate": 7.277206409949772e-05,
      "loss": 0.0087,
      "step": 16340
    },
    {
      "epoch": 2.290259285213735,
      "grad_norm": 0.06805703043937683,
      "learning_rate": 7.275771346567805e-05,
      "loss": 0.0164,
      "step": 16341
    },
    {
      "epoch": 2.2903994393833216,
      "grad_norm": 0.2065509557723999,
      "learning_rate": 7.274336283185841e-05,
      "loss": 0.0142,
      "step": 16342
    },
    {
      "epoch": 2.290539593552908,
      "grad_norm": 2.216456413269043,
      "learning_rate": 7.272901219803874e-05,
      "loss": 0.0278,
      "step": 16343
    },
    {
      "epoch": 2.2906797477224945,
      "grad_norm": 0.38408058881759644,
      "learning_rate": 7.271466156421908e-05,
      "loss": 0.0238,
      "step": 16344
    },
    {
      "epoch": 2.2908199018920814,
      "grad_norm": 0.1543806493282318,
      "learning_rate": 7.270031093039942e-05,
      "loss": 0.0136,
      "step": 16345
    },
    {
      "epoch": 2.290960056061668,
      "grad_norm": 0.4030522406101227,
      "learning_rate": 7.268596029657976e-05,
      "loss": 0.0333,
      "step": 16346
    },
    {
      "epoch": 2.2911002102312543,
      "grad_norm": 0.1624830812215805,
      "learning_rate": 7.26716096627601e-05,
      "loss": 0.0099,
      "step": 16347
    },
    {
      "epoch": 2.291240364400841,
      "grad_norm": 0.27439552545547485,
      "learning_rate": 7.265725902894043e-05,
      "loss": 0.0219,
      "step": 16348
    },
    {
      "epoch": 2.2913805185704277,
      "grad_norm": 0.6404179334640503,
      "learning_rate": 7.264290839512078e-05,
      "loss": 0.0535,
      "step": 16349
    },
    {
      "epoch": 2.291520672740014,
      "grad_norm": 0.11418481171131134,
      "learning_rate": 7.262855776130112e-05,
      "loss": 0.019,
      "step": 16350
    },
    {
      "epoch": 2.2916608269096006,
      "grad_norm": 0.1833251416683197,
      "learning_rate": 7.261420712748146e-05,
      "loss": 0.0615,
      "step": 16351
    },
    {
      "epoch": 2.291800981079187,
      "grad_norm": 0.22833088040351868,
      "learning_rate": 7.25998564936618e-05,
      "loss": 0.0354,
      "step": 16352
    },
    {
      "epoch": 2.2919411352487735,
      "grad_norm": 0.33519577980041504,
      "learning_rate": 7.258550585984213e-05,
      "loss": 0.0465,
      "step": 16353
    },
    {
      "epoch": 2.29208128941836,
      "grad_norm": 0.040395405143499374,
      "learning_rate": 7.257115522602247e-05,
      "loss": 0.0025,
      "step": 16354
    },
    {
      "epoch": 2.292221443587947,
      "grad_norm": 0.13844487071037292,
      "learning_rate": 7.255680459220281e-05,
      "loss": 0.0079,
      "step": 16355
    },
    {
      "epoch": 2.2923615977575333,
      "grad_norm": 0.22500017285346985,
      "learning_rate": 7.254245395838316e-05,
      "loss": 0.0231,
      "step": 16356
    },
    {
      "epoch": 2.29250175192712,
      "grad_norm": 0.5496867895126343,
      "learning_rate": 7.252810332456349e-05,
      "loss": 0.1337,
      "step": 16357
    },
    {
      "epoch": 2.2926419060967063,
      "grad_norm": 0.4275802671909332,
      "learning_rate": 7.251375269074384e-05,
      "loss": 0.0364,
      "step": 16358
    },
    {
      "epoch": 2.292782060266293,
      "grad_norm": 0.505795419216156,
      "learning_rate": 7.249940205692417e-05,
      "loss": 0.0275,
      "step": 16359
    },
    {
      "epoch": 2.2929222144358796,
      "grad_norm": 0.12260551005601883,
      "learning_rate": 7.248505142310451e-05,
      "loss": 0.0248,
      "step": 16360
    },
    {
      "epoch": 2.293062368605466,
      "grad_norm": 0.14141011238098145,
      "learning_rate": 7.247070078928485e-05,
      "loss": 0.013,
      "step": 16361
    },
    {
      "epoch": 2.2932025227750525,
      "grad_norm": 0.7091691493988037,
      "learning_rate": 7.24563501554652e-05,
      "loss": 0.0433,
      "step": 16362
    },
    {
      "epoch": 2.293342676944639,
      "grad_norm": 0.9841392040252686,
      "learning_rate": 7.244199952164554e-05,
      "loss": 0.0728,
      "step": 16363
    },
    {
      "epoch": 2.2934828311142255,
      "grad_norm": 0.2956642508506775,
      "learning_rate": 7.242764888782587e-05,
      "loss": 0.045,
      "step": 16364
    },
    {
      "epoch": 2.2936229852838124,
      "grad_norm": 0.1715112179517746,
      "learning_rate": 7.241329825400622e-05,
      "loss": 0.0177,
      "step": 16365
    },
    {
      "epoch": 2.293763139453399,
      "grad_norm": 0.5652415156364441,
      "learning_rate": 7.239894762018655e-05,
      "loss": 0.0579,
      "step": 16366
    },
    {
      "epoch": 2.2939032936229853,
      "grad_norm": 0.11872896552085876,
      "learning_rate": 7.238459698636689e-05,
      "loss": 0.0048,
      "step": 16367
    },
    {
      "epoch": 2.2940434477925717,
      "grad_norm": 0.46811461448669434,
      "learning_rate": 7.237024635254723e-05,
      "loss": 0.131,
      "step": 16368
    },
    {
      "epoch": 2.294183601962158,
      "grad_norm": 1.2988077402114868,
      "learning_rate": 7.235589571872758e-05,
      "loss": 0.0951,
      "step": 16369
    },
    {
      "epoch": 2.294323756131745,
      "grad_norm": NaN,
      "learning_rate": 7.23415450849079e-05,
      "loss": 0.1121,
      "step": 16370
    },
    {
      "epoch": 2.2944639103013316,
      "grad_norm": 0.4405251443386078,
      "learning_rate": 7.23415450849079e-05,
      "loss": 0.0448,
      "step": 16371
    },
    {
      "epoch": 2.294604064470918,
      "grad_norm": 0.16474972665309906,
      "learning_rate": 7.232719445108825e-05,
      "loss": 0.0375,
      "step": 16372
    },
    {
      "epoch": 2.2947442186405045,
      "grad_norm": 0.173185795545578,
      "learning_rate": 7.231284381726859e-05,
      "loss": 0.0182,
      "step": 16373
    },
    {
      "epoch": 2.294884372810091,
      "grad_norm": 0.31825515627861023,
      "learning_rate": 7.229849318344893e-05,
      "loss": 0.0203,
      "step": 16374
    },
    {
      "epoch": 2.2950245269796774,
      "grad_norm": 0.2430480718612671,
      "learning_rate": 7.228414254962927e-05,
      "loss": 0.0107,
      "step": 16375
    },
    {
      "epoch": 2.2951646811492643,
      "grad_norm": 0.23854903876781464,
      "learning_rate": 7.22697919158096e-05,
      "loss": 0.0229,
      "step": 16376
    },
    {
      "epoch": 2.2953048353188508,
      "grad_norm": 0.28108254075050354,
      "learning_rate": 7.225544128198994e-05,
      "loss": 0.0238,
      "step": 16377
    },
    {
      "epoch": 2.2954449894884372,
      "grad_norm": 0.5692513585090637,
      "learning_rate": 7.224109064817029e-05,
      "loss": 0.0547,
      "step": 16378
    },
    {
      "epoch": 2.2955851436580237,
      "grad_norm": 0.27059462666511536,
      "learning_rate": 7.222674001435063e-05,
      "loss": 0.0507,
      "step": 16379
    },
    {
      "epoch": 2.2957252978276106,
      "grad_norm": 0.2959274351596832,
      "learning_rate": 7.221238938053097e-05,
      "loss": 0.0656,
      "step": 16380
    },
    {
      "epoch": 2.295865451997197,
      "grad_norm": 0.17645077407360077,
      "learning_rate": 7.21980387467113e-05,
      "loss": 0.023,
      "step": 16381
    },
    {
      "epoch": 2.2960056061667835,
      "grad_norm": 0.11586400866508484,
      "learning_rate": 7.218368811289166e-05,
      "loss": 0.0098,
      "step": 16382
    },
    {
      "epoch": 2.29614576033637,
      "grad_norm": 0.11652031540870667,
      "learning_rate": 7.216933747907198e-05,
      "loss": 0.0116,
      "step": 16383
    },
    {
      "epoch": 2.2962859145059564,
      "grad_norm": 0.13560156524181366,
      "learning_rate": 7.215498684525233e-05,
      "loss": 0.0134,
      "step": 16384
    },
    {
      "epoch": 2.296426068675543,
      "grad_norm": 0.07631959021091461,
      "learning_rate": 7.214063621143267e-05,
      "loss": 0.0091,
      "step": 16385
    },
    {
      "epoch": 2.29656622284513,
      "grad_norm": 0.24796761572360992,
      "learning_rate": 7.212628557761301e-05,
      "loss": 0.0394,
      "step": 16386
    },
    {
      "epoch": 2.2967063770147162,
      "grad_norm": 0.46616461873054504,
      "learning_rate": 7.211193494379334e-05,
      "loss": 0.0587,
      "step": 16387
    },
    {
      "epoch": 2.2968465311843027,
      "grad_norm": 0.3734676241874695,
      "learning_rate": 7.209758430997368e-05,
      "loss": 0.0272,
      "step": 16388
    },
    {
      "epoch": 2.296986685353889,
      "grad_norm": 0.336534708738327,
      "learning_rate": 7.208323367615402e-05,
      "loss": 0.0687,
      "step": 16389
    },
    {
      "epoch": 2.297126839523476,
      "grad_norm": 0.2475401759147644,
      "learning_rate": 7.206888304233437e-05,
      "loss": 0.0117,
      "step": 16390
    },
    {
      "epoch": 2.2972669936930625,
      "grad_norm": 0.6748811602592468,
      "learning_rate": 7.205453240851471e-05,
      "loss": 0.0101,
      "step": 16391
    },
    {
      "epoch": 2.297407147862649,
      "grad_norm": 0.2517526149749756,
      "learning_rate": 7.204018177469504e-05,
      "loss": 0.0379,
      "step": 16392
    },
    {
      "epoch": 2.2975473020322355,
      "grad_norm": 0.1673772782087326,
      "learning_rate": 7.202583114087538e-05,
      "loss": 0.0148,
      "step": 16393
    },
    {
      "epoch": 2.297687456201822,
      "grad_norm": 0.7198790311813354,
      "learning_rate": 7.201148050705572e-05,
      "loss": 0.03,
      "step": 16394
    },
    {
      "epoch": 2.2978276103714084,
      "grad_norm": 0.09669344872236252,
      "learning_rate": 7.199712987323606e-05,
      "loss": 0.0135,
      "step": 16395
    },
    {
      "epoch": 2.2979677645409953,
      "grad_norm": 0.15343192219734192,
      "learning_rate": 7.19827792394164e-05,
      "loss": 0.0114,
      "step": 16396
    },
    {
      "epoch": 2.2981079187105817,
      "grad_norm": 0.22420866787433624,
      "learning_rate": 7.196842860559673e-05,
      "loss": 0.0189,
      "step": 16397
    },
    {
      "epoch": 2.298248072880168,
      "grad_norm": 0.3045988380908966,
      "learning_rate": 7.195407797177709e-05,
      "loss": 0.069,
      "step": 16398
    },
    {
      "epoch": 2.2983882270497547,
      "grad_norm": 0.2505404055118561,
      "learning_rate": 7.193972733795742e-05,
      "loss": 0.0485,
      "step": 16399
    },
    {
      "epoch": 2.298528381219341,
      "grad_norm": 0.3031961917877197,
      "learning_rate": 7.192537670413776e-05,
      "loss": 0.0406,
      "step": 16400
    },
    {
      "epoch": 2.298668535388928,
      "grad_norm": 0.17877697944641113,
      "learning_rate": 7.19110260703181e-05,
      "loss": 0.0236,
      "step": 16401
    },
    {
      "epoch": 2.2988086895585145,
      "grad_norm": 0.10980509221553802,
      "learning_rate": 7.189667543649844e-05,
      "loss": 0.0262,
      "step": 16402
    },
    {
      "epoch": 2.298948843728101,
      "grad_norm": 0.047711316496133804,
      "learning_rate": 7.188232480267877e-05,
      "loss": 0.0041,
      "step": 16403
    },
    {
      "epoch": 2.2990889978976874,
      "grad_norm": 0.2273794263601303,
      "learning_rate": 7.186797416885911e-05,
      "loss": 0.0515,
      "step": 16404
    },
    {
      "epoch": 2.299229152067274,
      "grad_norm": 0.40179771184921265,
      "learning_rate": 7.185362353503946e-05,
      "loss": 0.0349,
      "step": 16405
    },
    {
      "epoch": 2.2993693062368603,
      "grad_norm": 0.45591670274734497,
      "learning_rate": 7.18392729012198e-05,
      "loss": 0.0544,
      "step": 16406
    },
    {
      "epoch": 2.299509460406447,
      "grad_norm": 0.04963712394237518,
      "learning_rate": 7.182492226740014e-05,
      "loss": 0.0045,
      "step": 16407
    },
    {
      "epoch": 2.2996496145760337,
      "grad_norm": 0.1084568053483963,
      "learning_rate": 7.181057163358047e-05,
      "loss": 0.008,
      "step": 16408
    },
    {
      "epoch": 2.29978976874562,
      "grad_norm": 0.43498218059539795,
      "learning_rate": 7.179622099976082e-05,
      "loss": 0.0438,
      "step": 16409
    },
    {
      "epoch": 2.2999299229152066,
      "grad_norm": 0.3103965222835541,
      "learning_rate": 7.178187036594115e-05,
      "loss": 0.0363,
      "step": 16410
    },
    {
      "epoch": 2.3000700770847935,
      "grad_norm": 0.18497788906097412,
      "learning_rate": 7.17675197321215e-05,
      "loss": 0.0108,
      "step": 16411
    },
    {
      "epoch": 2.30021023125438,
      "grad_norm": 0.17044216394424438,
      "learning_rate": 7.175316909830184e-05,
      "loss": 0.0301,
      "step": 16412
    },
    {
      "epoch": 2.3003503854239664,
      "grad_norm": 0.08862544596195221,
      "learning_rate": 7.173881846448218e-05,
      "loss": 0.0048,
      "step": 16413
    },
    {
      "epoch": 2.300490539593553,
      "grad_norm": 0.09470432996749878,
      "learning_rate": 7.172446783066252e-05,
      "loss": 0.0047,
      "step": 16414
    },
    {
      "epoch": 2.3006306937631393,
      "grad_norm": 0.9463232159614563,
      "learning_rate": 7.171011719684285e-05,
      "loss": 0.1228,
      "step": 16415
    },
    {
      "epoch": 2.300770847932726,
      "grad_norm": 0.13871662318706512,
      "learning_rate": 7.169576656302319e-05,
      "loss": 0.046,
      "step": 16416
    },
    {
      "epoch": 2.3009110021023127,
      "grad_norm": 0.7979040145874023,
      "learning_rate": 7.168141592920353e-05,
      "loss": 0.2621,
      "step": 16417
    },
    {
      "epoch": 2.301051156271899,
      "grad_norm": 1.1652113199234009,
      "learning_rate": 7.166706529538388e-05,
      "loss": 0.1517,
      "step": 16418
    },
    {
      "epoch": 2.3011913104414856,
      "grad_norm": 0.9334061741828918,
      "learning_rate": 7.16527146615642e-05,
      "loss": 0.1352,
      "step": 16419
    },
    {
      "epoch": 2.301331464611072,
      "grad_norm": 0.4707517623901367,
      "learning_rate": 7.163836402774455e-05,
      "loss": 0.0162,
      "step": 16420
    },
    {
      "epoch": 2.3014716187806585,
      "grad_norm": 0.13064664602279663,
      "learning_rate": 7.162401339392489e-05,
      "loss": 0.012,
      "step": 16421
    },
    {
      "epoch": 2.3016117729502454,
      "grad_norm": 0.22265811264514923,
      "learning_rate": 7.160966276010523e-05,
      "loss": 0.0158,
      "step": 16422
    },
    {
      "epoch": 2.301751927119832,
      "grad_norm": 0.3326801359653473,
      "learning_rate": 7.159531212628557e-05,
      "loss": 0.0821,
      "step": 16423
    },
    {
      "epoch": 2.3018920812894184,
      "grad_norm": 0.2284831553697586,
      "learning_rate": 7.15809614924659e-05,
      "loss": 0.011,
      "step": 16424
    },
    {
      "epoch": 2.302032235459005,
      "grad_norm": 0.26042962074279785,
      "learning_rate": 7.156661085864626e-05,
      "loss": 0.0196,
      "step": 16425
    },
    {
      "epoch": 2.3021723896285913,
      "grad_norm": 0.26026061177253723,
      "learning_rate": 7.155226022482659e-05,
      "loss": 0.0512,
      "step": 16426
    },
    {
      "epoch": 2.302312543798178,
      "grad_norm": 0.09287726134061813,
      "learning_rate": 7.153790959100693e-05,
      "loss": 0.0058,
      "step": 16427
    },
    {
      "epoch": 2.3024526979677646,
      "grad_norm": 0.13203004002571106,
      "learning_rate": 7.152355895718727e-05,
      "loss": 0.0183,
      "step": 16428
    },
    {
      "epoch": 2.302592852137351,
      "grad_norm": 0.09747445583343506,
      "learning_rate": 7.150920832336761e-05,
      "loss": 0.013,
      "step": 16429
    },
    {
      "epoch": 2.3027330063069376,
      "grad_norm": 0.21391847729682922,
      "learning_rate": 7.149485768954795e-05,
      "loss": 0.0099,
      "step": 16430
    },
    {
      "epoch": 2.302873160476524,
      "grad_norm": 0.1325349509716034,
      "learning_rate": 7.148050705572828e-05,
      "loss": 0.0115,
      "step": 16431
    },
    {
      "epoch": 2.303013314646111,
      "grad_norm": 0.28195393085479736,
      "learning_rate": 7.146615642190863e-05,
      "loss": 0.0423,
      "step": 16432
    },
    {
      "epoch": 2.3031534688156974,
      "grad_norm": 0.4755287766456604,
      "learning_rate": 7.145180578808897e-05,
      "loss": 0.0469,
      "step": 16433
    },
    {
      "epoch": 2.303293622985284,
      "grad_norm": 0.4658677875995636,
      "learning_rate": 7.143745515426931e-05,
      "loss": 0.0675,
      "step": 16434
    },
    {
      "epoch": 2.3034337771548703,
      "grad_norm": 0.09921735525131226,
      "learning_rate": 7.142310452044964e-05,
      "loss": 0.008,
      "step": 16435
    },
    {
      "epoch": 2.3035739313244568,
      "grad_norm": 0.17093031108379364,
      "learning_rate": 7.140875388662998e-05,
      "loss": 0.0396,
      "step": 16436
    },
    {
      "epoch": 2.303714085494043,
      "grad_norm": 0.15276773273944855,
      "learning_rate": 7.139440325281032e-05,
      "loss": 0.0099,
      "step": 16437
    },
    {
      "epoch": 2.30385423966363,
      "grad_norm": 0.16145962476730347,
      "learning_rate": 7.138005261899066e-05,
      "loss": 0.0207,
      "step": 16438
    },
    {
      "epoch": 2.3039943938332166,
      "grad_norm": 0.545348048210144,
      "learning_rate": 7.1365701985171e-05,
      "loss": 0.0933,
      "step": 16439
    },
    {
      "epoch": 2.304134548002803,
      "grad_norm": 0.23149898648262024,
      "learning_rate": 7.135135135135135e-05,
      "loss": 0.0337,
      "step": 16440
    },
    {
      "epoch": 2.3042747021723895,
      "grad_norm": 0.13768118619918823,
      "learning_rate": 7.133700071753169e-05,
      "loss": 0.0206,
      "step": 16441
    },
    {
      "epoch": 2.3044148563419764,
      "grad_norm": 0.3924516439437866,
      "learning_rate": 7.132265008371202e-05,
      "loss": 0.0431,
      "step": 16442
    },
    {
      "epoch": 2.304555010511563,
      "grad_norm": 0.274938702583313,
      "learning_rate": 7.130829944989236e-05,
      "loss": 0.0437,
      "step": 16443
    },
    {
      "epoch": 2.3046951646811493,
      "grad_norm": 0.17318908870220184,
      "learning_rate": 7.12939488160727e-05,
      "loss": 0.014,
      "step": 16444
    },
    {
      "epoch": 2.304835318850736,
      "grad_norm": 0.1927107274532318,
      "learning_rate": 7.127959818225305e-05,
      "loss": 0.0638,
      "step": 16445
    },
    {
      "epoch": 2.3049754730203222,
      "grad_norm": 0.32449328899383545,
      "learning_rate": 7.126524754843339e-05,
      "loss": 0.0459,
      "step": 16446
    },
    {
      "epoch": 2.3051156271899087,
      "grad_norm": 0.20378804206848145,
      "learning_rate": 7.125089691461372e-05,
      "loss": 0.0174,
      "step": 16447
    },
    {
      "epoch": 2.3052557813594956,
      "grad_norm": 0.08240165561437607,
      "learning_rate": 7.123654628079407e-05,
      "loss": 0.025,
      "step": 16448
    },
    {
      "epoch": 2.305395935529082,
      "grad_norm": 0.07225339114665985,
      "learning_rate": 7.12221956469744e-05,
      "loss": 0.004,
      "step": 16449
    },
    {
      "epoch": 2.3055360896986685,
      "grad_norm": 0.2991752624511719,
      "learning_rate": 7.120784501315474e-05,
      "loss": 0.0625,
      "step": 16450
    },
    {
      "epoch": 2.305676243868255,
      "grad_norm": 0.2670639157295227,
      "learning_rate": 7.119349437933508e-05,
      "loss": 0.0133,
      "step": 16451
    },
    {
      "epoch": 2.3058163980378414,
      "grad_norm": 0.24177685379981995,
      "learning_rate": 7.117914374551543e-05,
      "loss": 0.0228,
      "step": 16452
    },
    {
      "epoch": 2.3059565522074283,
      "grad_norm": 0.20552009344100952,
      "learning_rate": 7.116479311169576e-05,
      "loss": 0.0173,
      "step": 16453
    },
    {
      "epoch": 2.306096706377015,
      "grad_norm": 0.09374982118606567,
      "learning_rate": 7.11504424778761e-05,
      "loss": 0.0101,
      "step": 16454
    },
    {
      "epoch": 2.3062368605466013,
      "grad_norm": 0.21385590732097626,
      "learning_rate": 7.113609184405644e-05,
      "loss": 0.0132,
      "step": 16455
    },
    {
      "epoch": 2.3063770147161877,
      "grad_norm": 0.26455244421958923,
      "learning_rate": 7.112174121023678e-05,
      "loss": 0.0389,
      "step": 16456
    },
    {
      "epoch": 2.306517168885774,
      "grad_norm": 0.07907243818044662,
      "learning_rate": 7.110739057641712e-05,
      "loss": 0.0077,
      "step": 16457
    },
    {
      "epoch": 2.306657323055361,
      "grad_norm": 0.5661972165107727,
      "learning_rate": 7.109303994259745e-05,
      "loss": 0.0192,
      "step": 16458
    },
    {
      "epoch": 2.3067974772249475,
      "grad_norm": 0.2675645053386688,
      "learning_rate": 7.10786893087778e-05,
      "loss": 0.0492,
      "step": 16459
    },
    {
      "epoch": 2.306937631394534,
      "grad_norm": 0.1989121437072754,
      "learning_rate": 7.106433867495814e-05,
      "loss": 0.0319,
      "step": 16460
    },
    {
      "epoch": 2.3070777855641205,
      "grad_norm": 0.20769351720809937,
      "learning_rate": 7.104998804113848e-05,
      "loss": 0.016,
      "step": 16461
    },
    {
      "epoch": 2.307217939733707,
      "grad_norm": 0.15482890605926514,
      "learning_rate": 7.103563740731882e-05,
      "loss": 0.0151,
      "step": 16462
    },
    {
      "epoch": 2.307358093903294,
      "grad_norm": 0.05521903559565544,
      "learning_rate": 7.102128677349915e-05,
      "loss": 0.0034,
      "step": 16463
    },
    {
      "epoch": 2.3074982480728803,
      "grad_norm": 0.11155806481838226,
      "learning_rate": 7.10069361396795e-05,
      "loss": 0.008,
      "step": 16464
    },
    {
      "epoch": 2.3076384022424667,
      "grad_norm": 0.18680016696453094,
      "learning_rate": 7.099258550585983e-05,
      "loss": 0.0313,
      "step": 16465
    },
    {
      "epoch": 2.307778556412053,
      "grad_norm": 0.42223021388053894,
      "learning_rate": 7.097823487204018e-05,
      "loss": 0.0118,
      "step": 16466
    },
    {
      "epoch": 2.3079187105816397,
      "grad_norm": 1.411315679550171,
      "learning_rate": 7.096388423822052e-05,
      "loss": 0.1292,
      "step": 16467
    },
    {
      "epoch": 2.308058864751226,
      "grad_norm": 0.14049489796161652,
      "learning_rate": 7.094953360440086e-05,
      "loss": 0.0076,
      "step": 16468
    },
    {
      "epoch": 2.308199018920813,
      "grad_norm": 0.29055240750312805,
      "learning_rate": 7.093518297058119e-05,
      "loss": 0.0534,
      "step": 16469
    },
    {
      "epoch": 2.3083391730903995,
      "grad_norm": 2.0605289936065674,
      "learning_rate": 7.092083233676153e-05,
      "loss": 0.1397,
      "step": 16470
    },
    {
      "epoch": 2.308479327259986,
      "grad_norm": 0.11836738884449005,
      "learning_rate": 7.090648170294187e-05,
      "loss": 0.0119,
      "step": 16471
    },
    {
      "epoch": 2.3086194814295724,
      "grad_norm": 0.33166682720184326,
      "learning_rate": 7.089213106912222e-05,
      "loss": 0.0435,
      "step": 16472
    },
    {
      "epoch": 2.3087596355991593,
      "grad_norm": 0.16832590103149414,
      "learning_rate": 7.087778043530256e-05,
      "loss": 0.0142,
      "step": 16473
    },
    {
      "epoch": 2.3088997897687458,
      "grad_norm": 0.24544009566307068,
      "learning_rate": 7.086342980148289e-05,
      "loss": 0.0314,
      "step": 16474
    },
    {
      "epoch": 2.3090399439383322,
      "grad_norm": 0.13889656960964203,
      "learning_rate": 7.084907916766324e-05,
      "loss": 0.0216,
      "step": 16475
    },
    {
      "epoch": 2.3091800981079187,
      "grad_norm": 0.09510023891925812,
      "learning_rate": 7.083472853384357e-05,
      "loss": 0.0045,
      "step": 16476
    },
    {
      "epoch": 2.309320252277505,
      "grad_norm": 0.2245909720659256,
      "learning_rate": 7.082037790002391e-05,
      "loss": 0.0493,
      "step": 16477
    },
    {
      "epoch": 2.3094604064470916,
      "grad_norm": 2.259777784347534,
      "learning_rate": 7.080602726620425e-05,
      "loss": 0.0666,
      "step": 16478
    },
    {
      "epoch": 2.3096005606166785,
      "grad_norm": 0.4320547580718994,
      "learning_rate": 7.07916766323846e-05,
      "loss": 0.0143,
      "step": 16479
    },
    {
      "epoch": 2.309740714786265,
      "grad_norm": 0.45345693826675415,
      "learning_rate": 7.077732599856494e-05,
      "loss": 0.0528,
      "step": 16480
    },
    {
      "epoch": 2.3098808689558514,
      "grad_norm": 1.4806920289993286,
      "learning_rate": 7.076297536474527e-05,
      "loss": 0.0282,
      "step": 16481
    },
    {
      "epoch": 2.310021023125438,
      "grad_norm": 0.08577558398246765,
      "learning_rate": 7.074862473092561e-05,
      "loss": 0.0083,
      "step": 16482
    },
    {
      "epoch": 2.3101611772950243,
      "grad_norm": 0.10238148272037506,
      "learning_rate": 7.073427409710595e-05,
      "loss": 0.0057,
      "step": 16483
    },
    {
      "epoch": 2.3103013314646113,
      "grad_norm": 0.34073325991630554,
      "learning_rate": 7.07199234632863e-05,
      "loss": 0.0291,
      "step": 16484
    },
    {
      "epoch": 2.3104414856341977,
      "grad_norm": 0.16331274807453156,
      "learning_rate": 7.070557282946662e-05,
      "loss": 0.0141,
      "step": 16485
    },
    {
      "epoch": 2.310581639803784,
      "grad_norm": 0.19547033309936523,
      "learning_rate": 7.069122219564696e-05,
      "loss": 0.0253,
      "step": 16486
    },
    {
      "epoch": 2.3107217939733706,
      "grad_norm": 0.6395230889320374,
      "learning_rate": 7.06768715618273e-05,
      "loss": 0.0128,
      "step": 16487
    },
    {
      "epoch": 2.310861948142957,
      "grad_norm": 0.197127565741539,
      "learning_rate": 7.066252092800765e-05,
      "loss": 0.0275,
      "step": 16488
    },
    {
      "epoch": 2.3110021023125435,
      "grad_norm": 0.33264726400375366,
      "learning_rate": 7.064817029418799e-05,
      "loss": 0.0436,
      "step": 16489
    },
    {
      "epoch": 2.3111422564821305,
      "grad_norm": 0.2518979012966156,
      "learning_rate": 7.063381966036832e-05,
      "loss": 0.0292,
      "step": 16490
    },
    {
      "epoch": 2.311282410651717,
      "grad_norm": 0.41997990012168884,
      "learning_rate": 7.061946902654867e-05,
      "loss": 0.0196,
      "step": 16491
    },
    {
      "epoch": 2.3114225648213034,
      "grad_norm": 0.4283941984176636,
      "learning_rate": 7.0605118392729e-05,
      "loss": 0.0316,
      "step": 16492
    },
    {
      "epoch": 2.31156271899089,
      "grad_norm": 0.19397379457950592,
      "learning_rate": 7.059076775890935e-05,
      "loss": 0.0207,
      "step": 16493
    },
    {
      "epoch": 2.3117028731604767,
      "grad_norm": 0.05297606810927391,
      "learning_rate": 7.057641712508969e-05,
      "loss": 0.0036,
      "step": 16494
    },
    {
      "epoch": 2.311843027330063,
      "grad_norm": 0.45471590757369995,
      "learning_rate": 7.056206649127003e-05,
      "loss": 0.0363,
      "step": 16495
    },
    {
      "epoch": 2.3119831814996497,
      "grad_norm": 0.48851683735847473,
      "learning_rate": 7.054771585745037e-05,
      "loss": 0.0268,
      "step": 16496
    },
    {
      "epoch": 2.312123335669236,
      "grad_norm": 0.15683528780937195,
      "learning_rate": 7.05333652236307e-05,
      "loss": 0.0294,
      "step": 16497
    },
    {
      "epoch": 2.3122634898388226,
      "grad_norm": 0.48048844933509827,
      "learning_rate": 7.051901458981104e-05,
      "loss": 0.0434,
      "step": 16498
    },
    {
      "epoch": 2.312403644008409,
      "grad_norm": 0.44901102781295776,
      "learning_rate": 7.050466395599138e-05,
      "loss": 0.0883,
      "step": 16499
    },
    {
      "epoch": 2.312543798177996,
      "grad_norm": 0.22067351639270782,
      "learning_rate": 7.049031332217173e-05,
      "loss": 0.0188,
      "step": 16500
    },
    {
      "epoch": 2.3126839523475824,
      "grad_norm": 0.2735159695148468,
      "learning_rate": 7.047596268835206e-05,
      "loss": 0.0562,
      "step": 16501
    },
    {
      "epoch": 2.312824106517169,
      "grad_norm": 0.5904778242111206,
      "learning_rate": 7.04616120545324e-05,
      "loss": 0.0232,
      "step": 16502
    },
    {
      "epoch": 2.3129642606867553,
      "grad_norm": 0.3933092951774597,
      "learning_rate": 7.044726142071274e-05,
      "loss": 0.043,
      "step": 16503
    },
    {
      "epoch": 2.313104414856342,
      "grad_norm": 0.12955988943576813,
      "learning_rate": 7.043291078689308e-05,
      "loss": 0.0085,
      "step": 16504
    },
    {
      "epoch": 2.3132445690259287,
      "grad_norm": 0.10079353302717209,
      "learning_rate": 7.041856015307342e-05,
      "loss": 0.0137,
      "step": 16505
    },
    {
      "epoch": 2.313384723195515,
      "grad_norm": 0.20393581688404083,
      "learning_rate": 7.040420951925375e-05,
      "loss": 0.0246,
      "step": 16506
    },
    {
      "epoch": 2.3135248773651016,
      "grad_norm": 0.1674930900335312,
      "learning_rate": 7.038985888543411e-05,
      "loss": 0.0541,
      "step": 16507
    },
    {
      "epoch": 2.313665031534688,
      "grad_norm": 0.34796109795570374,
      "learning_rate": 7.037550825161444e-05,
      "loss": 0.0149,
      "step": 16508
    },
    {
      "epoch": 2.3138051857042745,
      "grad_norm": 0.27727648615837097,
      "learning_rate": 7.036115761779478e-05,
      "loss": 0.0274,
      "step": 16509
    },
    {
      "epoch": 2.3139453398738614,
      "grad_norm": 0.8513749837875366,
      "learning_rate": 7.034680698397512e-05,
      "loss": 0.0098,
      "step": 16510
    },
    {
      "epoch": 2.314085494043448,
      "grad_norm": 0.42540132999420166,
      "learning_rate": 7.033245635015546e-05,
      "loss": 0.0627,
      "step": 16511
    },
    {
      "epoch": 2.3142256482130343,
      "grad_norm": 0.18111921846866608,
      "learning_rate": 7.03181057163358e-05,
      "loss": 0.0758,
      "step": 16512
    },
    {
      "epoch": 2.314365802382621,
      "grad_norm": 0.2865699827671051,
      "learning_rate": 7.030375508251613e-05,
      "loss": 0.0328,
      "step": 16513
    },
    {
      "epoch": 2.3145059565522073,
      "grad_norm": 0.21320120990276337,
      "learning_rate": 7.028940444869648e-05,
      "loss": 0.0504,
      "step": 16514
    },
    {
      "epoch": 2.314646110721794,
      "grad_norm": 1.5606576204299927,
      "learning_rate": 7.027505381487682e-05,
      "loss": 0.1099,
      "step": 16515
    },
    {
      "epoch": 2.3147862648913806,
      "grad_norm": 0.4962499737739563,
      "learning_rate": 7.026070318105716e-05,
      "loss": 0.0322,
      "step": 16516
    },
    {
      "epoch": 2.314926419060967,
      "grad_norm": 0.5625011920928955,
      "learning_rate": 7.024635254723749e-05,
      "loss": 0.0445,
      "step": 16517
    },
    {
      "epoch": 2.3150665732305535,
      "grad_norm": 1.182071566581726,
      "learning_rate": 7.023200191341784e-05,
      "loss": 0.1061,
      "step": 16518
    },
    {
      "epoch": 2.31520672740014,
      "grad_norm": 1.0522364377975464,
      "learning_rate": 7.021765127959817e-05,
      "loss": 0.1033,
      "step": 16519
    },
    {
      "epoch": 2.3153468815697265,
      "grad_norm": 2.5701210498809814,
      "learning_rate": 7.020330064577851e-05,
      "loss": 0.2004,
      "step": 16520
    },
    {
      "epoch": 2.3154870357393134,
      "grad_norm": 0.0727614015340805,
      "learning_rate": 7.018895001195886e-05,
      "loss": 0.0098,
      "step": 16521
    },
    {
      "epoch": 2.3156271899089,
      "grad_norm": 0.03042016364634037,
      "learning_rate": 7.01745993781392e-05,
      "loss": 0.0025,
      "step": 16522
    },
    {
      "epoch": 2.3157673440784863,
      "grad_norm": 0.16228154301643372,
      "learning_rate": 7.016024874431954e-05,
      "loss": 0.01,
      "step": 16523
    },
    {
      "epoch": 2.3159074982480727,
      "grad_norm": 0.16910363733768463,
      "learning_rate": 7.014589811049987e-05,
      "loss": 0.0338,
      "step": 16524
    },
    {
      "epoch": 2.3160476524176596,
      "grad_norm": 0.3106030821800232,
      "learning_rate": 7.013154747668021e-05,
      "loss": 0.0345,
      "step": 16525
    },
    {
      "epoch": 2.316187806587246,
      "grad_norm": 0.3267758786678314,
      "learning_rate": 7.011719684286055e-05,
      "loss": 0.0551,
      "step": 16526
    },
    {
      "epoch": 2.3163279607568326,
      "grad_norm": 0.26353809237480164,
      "learning_rate": 7.01028462090409e-05,
      "loss": 0.0176,
      "step": 16527
    },
    {
      "epoch": 2.316468114926419,
      "grad_norm": 0.26869645714759827,
      "learning_rate": 7.008849557522124e-05,
      "loss": 0.0139,
      "step": 16528
    },
    {
      "epoch": 2.3166082690960055,
      "grad_norm": 0.18929380178451538,
      "learning_rate": 7.007414494140157e-05,
      "loss": 0.0237,
      "step": 16529
    },
    {
      "epoch": 2.316748423265592,
      "grad_norm": 0.19186298549175262,
      "learning_rate": 7.005979430758191e-05,
      "loss": 0.0155,
      "step": 16530
    },
    {
      "epoch": 2.316888577435179,
      "grad_norm": 0.3253980278968811,
      "learning_rate": 7.004544367376225e-05,
      "loss": 0.0572,
      "step": 16531
    },
    {
      "epoch": 2.3170287316047653,
      "grad_norm": 0.05737369880080223,
      "learning_rate": 7.003109303994259e-05,
      "loss": 0.0045,
      "step": 16532
    },
    {
      "epoch": 2.3171688857743518,
      "grad_norm": 0.27716708183288574,
      "learning_rate": 7.001674240612292e-05,
      "loss": 0.0634,
      "step": 16533
    },
    {
      "epoch": 2.317309039943938,
      "grad_norm": 0.07707148045301437,
      "learning_rate": 7.000239177230328e-05,
      "loss": 0.0096,
      "step": 16534
    },
    {
      "epoch": 2.317449194113525,
      "grad_norm": 0.4619060158729553,
      "learning_rate": 6.99880411384836e-05,
      "loss": 0.0681,
      "step": 16535
    },
    {
      "epoch": 2.3175893482831116,
      "grad_norm": 0.12493345886468887,
      "learning_rate": 6.997369050466395e-05,
      "loss": 0.0299,
      "step": 16536
    },
    {
      "epoch": 2.317729502452698,
      "grad_norm": 0.1722351759672165,
      "learning_rate": 6.995933987084429e-05,
      "loss": 0.0179,
      "step": 16537
    },
    {
      "epoch": 2.3178696566222845,
      "grad_norm": 0.061598047614097595,
      "learning_rate": 6.994498923702463e-05,
      "loss": 0.0168,
      "step": 16538
    },
    {
      "epoch": 2.318009810791871,
      "grad_norm": 0.05584261566400528,
      "learning_rate": 6.993063860320497e-05,
      "loss": 0.0095,
      "step": 16539
    },
    {
      "epoch": 2.3181499649614574,
      "grad_norm": 0.2388603240251541,
      "learning_rate": 6.99162879693853e-05,
      "loss": 0.0504,
      "step": 16540
    },
    {
      "epoch": 2.3182901191310443,
      "grad_norm": 0.19127263128757477,
      "learning_rate": 6.990193733556564e-05,
      "loss": 0.027,
      "step": 16541
    },
    {
      "epoch": 2.318430273300631,
      "grad_norm": 0.16364222764968872,
      "learning_rate": 6.988758670174599e-05,
      "loss": 0.0404,
      "step": 16542
    },
    {
      "epoch": 2.3185704274702172,
      "grad_norm": 0.24606789648532867,
      "learning_rate": 6.987323606792633e-05,
      "loss": 0.0081,
      "step": 16543
    },
    {
      "epoch": 2.3187105816398037,
      "grad_norm": 0.1390107274055481,
      "learning_rate": 6.985888543410667e-05,
      "loss": 0.0408,
      "step": 16544
    },
    {
      "epoch": 2.31885073580939,
      "grad_norm": 0.08847220987081528,
      "learning_rate": 6.9844534800287e-05,
      "loss": 0.009,
      "step": 16545
    },
    {
      "epoch": 2.318990889978977,
      "grad_norm": 0.1411711573600769,
      "learning_rate": 6.983018416646734e-05,
      "loss": 0.014,
      "step": 16546
    },
    {
      "epoch": 2.3191310441485635,
      "grad_norm": 0.2588651180267334,
      "learning_rate": 6.981583353264768e-05,
      "loss": 0.0444,
      "step": 16547
    },
    {
      "epoch": 2.31927119831815,
      "grad_norm": 0.25281545519828796,
      "learning_rate": 6.980148289882803e-05,
      "loss": 0.0132,
      "step": 16548
    },
    {
      "epoch": 2.3194113524877364,
      "grad_norm": 0.2586451768875122,
      "learning_rate": 6.978713226500837e-05,
      "loss": 0.0109,
      "step": 16549
    },
    {
      "epoch": 2.319551506657323,
      "grad_norm": 0.25907331705093384,
      "learning_rate": 6.977278163118871e-05,
      "loss": 0.047,
      "step": 16550
    },
    {
      "epoch": 2.3196916608269094,
      "grad_norm": 0.2584594488143921,
      "learning_rate": 6.975843099736904e-05,
      "loss": 0.0527,
      "step": 16551
    },
    {
      "epoch": 2.3198318149964963,
      "grad_norm": 0.24839650094509125,
      "learning_rate": 6.974408036354938e-05,
      "loss": 0.0759,
      "step": 16552
    },
    {
      "epoch": 2.3199719691660827,
      "grad_norm": 0.10946990549564362,
      "learning_rate": 6.972972972972972e-05,
      "loss": 0.012,
      "step": 16553
    },
    {
      "epoch": 2.320112123335669,
      "grad_norm": 0.18611249327659607,
      "learning_rate": 6.971537909591007e-05,
      "loss": 0.0539,
      "step": 16554
    },
    {
      "epoch": 2.3202522775052556,
      "grad_norm": 0.11179360002279282,
      "learning_rate": 6.970102846209041e-05,
      "loss": 0.0106,
      "step": 16555
    },
    {
      "epoch": 2.3203924316748425,
      "grad_norm": 0.1899394989013672,
      "learning_rate": 6.968667782827074e-05,
      "loss": 0.0348,
      "step": 16556
    },
    {
      "epoch": 2.320532585844429,
      "grad_norm": 0.11880426853895187,
      "learning_rate": 6.967232719445109e-05,
      "loss": 0.0075,
      "step": 16557
    },
    {
      "epoch": 2.3206727400140155,
      "grad_norm": 0.2622591555118561,
      "learning_rate": 6.965797656063142e-05,
      "loss": 0.0092,
      "step": 16558
    },
    {
      "epoch": 2.320812894183602,
      "grad_norm": 0.2903350591659546,
      "learning_rate": 6.964362592681176e-05,
      "loss": 0.0953,
      "step": 16559
    },
    {
      "epoch": 2.3209530483531884,
      "grad_norm": 0.3928678631782532,
      "learning_rate": 6.96292752929921e-05,
      "loss": 0.036,
      "step": 16560
    },
    {
      "epoch": 2.321093202522775,
      "grad_norm": 0.08735954016447067,
      "learning_rate": 6.961492465917245e-05,
      "loss": 0.009,
      "step": 16561
    },
    {
      "epoch": 2.3212333566923617,
      "grad_norm": 0.163041889667511,
      "learning_rate": 6.960057402535279e-05,
      "loss": 0.0136,
      "step": 16562
    },
    {
      "epoch": 2.321373510861948,
      "grad_norm": 0.37632039189338684,
      "learning_rate": 6.958622339153312e-05,
      "loss": 0.0776,
      "step": 16563
    },
    {
      "epoch": 2.3215136650315347,
      "grad_norm": 0.13588887453079224,
      "learning_rate": 6.957187275771346e-05,
      "loss": 0.0127,
      "step": 16564
    },
    {
      "epoch": 2.321653819201121,
      "grad_norm": 0.24518993496894836,
      "learning_rate": 6.95575221238938e-05,
      "loss": 0.0178,
      "step": 16565
    },
    {
      "epoch": 2.3217939733707076,
      "grad_norm": 0.15324679017066956,
      "learning_rate": 6.954317149007414e-05,
      "loss": 0.0157,
      "step": 16566
    },
    {
      "epoch": 2.3219341275402945,
      "grad_norm": 1.1067732572555542,
      "learning_rate": 6.952882085625447e-05,
      "loss": 0.0873,
      "step": 16567
    },
    {
      "epoch": 2.322074281709881,
      "grad_norm": 1.524893045425415,
      "learning_rate": 6.951447022243481e-05,
      "loss": 0.2479,
      "step": 16568
    },
    {
      "epoch": 2.3222144358794674,
      "grad_norm": 0.33871299028396606,
      "learning_rate": 6.950011958861516e-05,
      "loss": 0.0112,
      "step": 16569
    },
    {
      "epoch": 2.322354590049054,
      "grad_norm": 0.41682717204093933,
      "learning_rate": 6.94857689547955e-05,
      "loss": 0.011,
      "step": 16570
    },
    {
      "epoch": 2.3224947442186403,
      "grad_norm": 0.1875259429216385,
      "learning_rate": 6.947141832097584e-05,
      "loss": 0.023,
      "step": 16571
    },
    {
      "epoch": 2.3226348983882272,
      "grad_norm": 0.153966024518013,
      "learning_rate": 6.945706768715617e-05,
      "loss": 0.0126,
      "step": 16572
    },
    {
      "epoch": 2.3227750525578137,
      "grad_norm": 0.37229520082473755,
      "learning_rate": 6.944271705333652e-05,
      "loss": 0.0285,
      "step": 16573
    },
    {
      "epoch": 2.3229152067274,
      "grad_norm": 0.264862984418869,
      "learning_rate": 6.942836641951685e-05,
      "loss": 0.0537,
      "step": 16574
    },
    {
      "epoch": 2.3230553608969866,
      "grad_norm": 0.10851848870515823,
      "learning_rate": 6.94140157856972e-05,
      "loss": 0.0107,
      "step": 16575
    },
    {
      "epoch": 2.323195515066573,
      "grad_norm": 0.44682520627975464,
      "learning_rate": 6.939966515187754e-05,
      "loss": 0.052,
      "step": 16576
    },
    {
      "epoch": 2.32333566923616,
      "grad_norm": 0.08117534965276718,
      "learning_rate": 6.938531451805788e-05,
      "loss": 0.0085,
      "step": 16577
    },
    {
      "epoch": 2.3234758234057464,
      "grad_norm": 0.5897807478904724,
      "learning_rate": 6.937096388423822e-05,
      "loss": 0.0725,
      "step": 16578
    },
    {
      "epoch": 2.323615977575333,
      "grad_norm": 0.36941468715667725,
      "learning_rate": 6.935661325041855e-05,
      "loss": 0.0315,
      "step": 16579
    },
    {
      "epoch": 2.3237561317449194,
      "grad_norm": 0.22469641268253326,
      "learning_rate": 6.934226261659889e-05,
      "loss": 0.0203,
      "step": 16580
    },
    {
      "epoch": 2.323896285914506,
      "grad_norm": 0.16602832078933716,
      "learning_rate": 6.932791198277923e-05,
      "loss": 0.0251,
      "step": 16581
    },
    {
      "epoch": 2.3240364400840923,
      "grad_norm": 0.17652295529842377,
      "learning_rate": 6.931356134895958e-05,
      "loss": 0.015,
      "step": 16582
    },
    {
      "epoch": 2.324176594253679,
      "grad_norm": 0.08235567063093185,
      "learning_rate": 6.92992107151399e-05,
      "loss": 0.0184,
      "step": 16583
    },
    {
      "epoch": 2.3243167484232656,
      "grad_norm": 0.13511085510253906,
      "learning_rate": 6.928486008132025e-05,
      "loss": 0.0075,
      "step": 16584
    },
    {
      "epoch": 2.324456902592852,
      "grad_norm": 0.3036658465862274,
      "learning_rate": 6.927050944750059e-05,
      "loss": 0.0202,
      "step": 16585
    },
    {
      "epoch": 2.3245970567624386,
      "grad_norm": 0.25037023425102234,
      "learning_rate": 6.925615881368093e-05,
      "loss": 0.0203,
      "step": 16586
    },
    {
      "epoch": 2.3247372109320255,
      "grad_norm": 0.24066416919231415,
      "learning_rate": 6.924180817986127e-05,
      "loss": 0.0532,
      "step": 16587
    },
    {
      "epoch": 2.324877365101612,
      "grad_norm": 0.22306126356124878,
      "learning_rate": 6.922745754604162e-05,
      "loss": 0.0322,
      "step": 16588
    },
    {
      "epoch": 2.3250175192711984,
      "grad_norm": 0.10941849648952484,
      "learning_rate": 6.921310691222196e-05,
      "loss": 0.0044,
      "step": 16589
    },
    {
      "epoch": 2.325157673440785,
      "grad_norm": 0.04568088799715042,
      "learning_rate": 6.919875627840229e-05,
      "loss": 0.0037,
      "step": 16590
    },
    {
      "epoch": 2.3252978276103713,
      "grad_norm": 0.08783835917711258,
      "learning_rate": 6.918440564458263e-05,
      "loss": 0.0028,
      "step": 16591
    },
    {
      "epoch": 2.3254379817799578,
      "grad_norm": 0.16091756522655487,
      "learning_rate": 6.917005501076297e-05,
      "loss": 0.0065,
      "step": 16592
    },
    {
      "epoch": 2.3255781359495447,
      "grad_norm": 0.07261763513088226,
      "learning_rate": 6.915570437694331e-05,
      "loss": 0.0055,
      "step": 16593
    },
    {
      "epoch": 2.325718290119131,
      "grad_norm": 0.2247370183467865,
      "learning_rate": 6.914135374312365e-05,
      "loss": 0.0258,
      "step": 16594
    },
    {
      "epoch": 2.3258584442887176,
      "grad_norm": 0.28299105167388916,
      "learning_rate": 6.912700310930398e-05,
      "loss": 0.0244,
      "step": 16595
    },
    {
      "epoch": 2.325998598458304,
      "grad_norm": 0.21893605589866638,
      "learning_rate": 6.911265247548433e-05,
      "loss": 0.0374,
      "step": 16596
    },
    {
      "epoch": 2.3261387526278905,
      "grad_norm": 0.22427575290203094,
      "learning_rate": 6.909830184166467e-05,
      "loss": 0.0141,
      "step": 16597
    },
    {
      "epoch": 2.3262789067974774,
      "grad_norm": 0.2727227210998535,
      "learning_rate": 6.908395120784501e-05,
      "loss": 0.0345,
      "step": 16598
    },
    {
      "epoch": 2.326419060967064,
      "grad_norm": 0.2515237331390381,
      "learning_rate": 6.906960057402534e-05,
      "loss": 0.0283,
      "step": 16599
    },
    {
      "epoch": 2.3265592151366503,
      "grad_norm": 0.25063416361808777,
      "learning_rate": 6.90552499402057e-05,
      "loss": 0.0216,
      "step": 16600
    },
    {
      "epoch": 2.3266993693062368,
      "grad_norm": 0.5554426312446594,
      "learning_rate": 6.904089930638602e-05,
      "loss": 0.0457,
      "step": 16601
    },
    {
      "epoch": 2.3268395234758232,
      "grad_norm": 0.1692066490650177,
      "learning_rate": 6.902654867256636e-05,
      "loss": 0.0195,
      "step": 16602
    },
    {
      "epoch": 2.32697967764541,
      "grad_norm": 0.251560240983963,
      "learning_rate": 6.90121980387467e-05,
      "loss": 0.0163,
      "step": 16603
    },
    {
      "epoch": 2.3271198318149966,
      "grad_norm": 0.14772725105285645,
      "learning_rate": 6.899784740492705e-05,
      "loss": 0.0194,
      "step": 16604
    },
    {
      "epoch": 2.327259985984583,
      "grad_norm": 0.22039751708507538,
      "learning_rate": 6.898349677110739e-05,
      "loss": 0.0158,
      "step": 16605
    },
    {
      "epoch": 2.3274001401541695,
      "grad_norm": 0.2137940227985382,
      "learning_rate": 6.896914613728772e-05,
      "loss": 0.0368,
      "step": 16606
    },
    {
      "epoch": 2.327540294323756,
      "grad_norm": 0.11543194204568863,
      "learning_rate": 6.895479550346806e-05,
      "loss": 0.0105,
      "step": 16607
    },
    {
      "epoch": 2.327680448493343,
      "grad_norm": 0.48380768299102783,
      "learning_rate": 6.89404448696484e-05,
      "loss": 0.042,
      "step": 16608
    },
    {
      "epoch": 2.3278206026629293,
      "grad_norm": 0.3692674934864044,
      "learning_rate": 6.892609423582875e-05,
      "loss": 0.0426,
      "step": 16609
    },
    {
      "epoch": 2.327960756832516,
      "grad_norm": 0.16432121396064758,
      "learning_rate": 6.891174360200909e-05,
      "loss": 0.0759,
      "step": 16610
    },
    {
      "epoch": 2.3281009110021023,
      "grad_norm": 0.05551853030920029,
      "learning_rate": 6.889739296818942e-05,
      "loss": 0.007,
      "step": 16611
    },
    {
      "epoch": 2.3282410651716887,
      "grad_norm": 0.15788644552230835,
      "learning_rate": 6.888304233436976e-05,
      "loss": 0.0082,
      "step": 16612
    },
    {
      "epoch": 2.328381219341275,
      "grad_norm": 0.351199209690094,
      "learning_rate": 6.88686917005501e-05,
      "loss": 0.0183,
      "step": 16613
    },
    {
      "epoch": 2.328521373510862,
      "grad_norm": 0.5869712829589844,
      "learning_rate": 6.885434106673044e-05,
      "loss": 0.0302,
      "step": 16614
    },
    {
      "epoch": 2.3286615276804485,
      "grad_norm": 1.0614866018295288,
      "learning_rate": 6.883999043291077e-05,
      "loss": 0.0993,
      "step": 16615
    },
    {
      "epoch": 2.328801681850035,
      "grad_norm": 0.5097002387046814,
      "learning_rate": 6.882563979909113e-05,
      "loss": 0.0321,
      "step": 16616
    },
    {
      "epoch": 2.3289418360196215,
      "grad_norm": 1.2023595571517944,
      "learning_rate": 6.881128916527146e-05,
      "loss": 0.0556,
      "step": 16617
    },
    {
      "epoch": 2.3290819901892084,
      "grad_norm": 1.4112474918365479,
      "learning_rate": 6.87969385314518e-05,
      "loss": 0.1111,
      "step": 16618
    },
    {
      "epoch": 2.329222144358795,
      "grad_norm": 0.38839244842529297,
      "learning_rate": 6.878258789763214e-05,
      "loss": 0.0271,
      "step": 16619
    },
    {
      "epoch": 2.3293622985283813,
      "grad_norm": 2.6200151443481445,
      "learning_rate": 6.876823726381248e-05,
      "loss": 0.0648,
      "step": 16620
    },
    {
      "epoch": 2.3295024526979677,
      "grad_norm": 0.07727036625146866,
      "learning_rate": 6.875388662999282e-05,
      "loss": 0.0226,
      "step": 16621
    },
    {
      "epoch": 2.329642606867554,
      "grad_norm": 0.5386391878128052,
      "learning_rate": 6.873953599617315e-05,
      "loss": 0.0478,
      "step": 16622
    },
    {
      "epoch": 2.3297827610371407,
      "grad_norm": 0.23182600736618042,
      "learning_rate": 6.872518536235351e-05,
      "loss": 0.0566,
      "step": 16623
    },
    {
      "epoch": 2.3299229152067276,
      "grad_norm": 0.31552937626838684,
      "learning_rate": 6.871083472853384e-05,
      "loss": 0.0298,
      "step": 16624
    },
    {
      "epoch": 2.330063069376314,
      "grad_norm": 0.32090187072753906,
      "learning_rate": 6.869648409471418e-05,
      "loss": 0.0298,
      "step": 16625
    },
    {
      "epoch": 2.3302032235459005,
      "grad_norm": 0.07858520746231079,
      "learning_rate": 6.868213346089452e-05,
      "loss": 0.0051,
      "step": 16626
    },
    {
      "epoch": 2.330343377715487,
      "grad_norm": 0.20682689547538757,
      "learning_rate": 6.866778282707486e-05,
      "loss": 0.0102,
      "step": 16627
    },
    {
      "epoch": 2.3304835318850734,
      "grad_norm": 0.23187360167503357,
      "learning_rate": 6.865343219325519e-05,
      "loss": 0.0336,
      "step": 16628
    },
    {
      "epoch": 2.3306236860546603,
      "grad_norm": 0.10584904253482819,
      "learning_rate": 6.863908155943553e-05,
      "loss": 0.0204,
      "step": 16629
    },
    {
      "epoch": 2.3307638402242468,
      "grad_norm": 0.225706547498703,
      "learning_rate": 6.862473092561588e-05,
      "loss": 0.0156,
      "step": 16630
    },
    {
      "epoch": 2.3309039943938332,
      "grad_norm": 0.4158373475074768,
      "learning_rate": 6.861038029179622e-05,
      "loss": 0.0235,
      "step": 16631
    },
    {
      "epoch": 2.3310441485634197,
      "grad_norm": 0.1999945193529129,
      "learning_rate": 6.859602965797656e-05,
      "loss": 0.0293,
      "step": 16632
    },
    {
      "epoch": 2.331184302733006,
      "grad_norm": 0.2068648487329483,
      "learning_rate": 6.858167902415689e-05,
      "loss": 0.0451,
      "step": 16633
    },
    {
      "epoch": 2.331324456902593,
      "grad_norm": 0.5172854661941528,
      "learning_rate": 6.856732839033723e-05,
      "loss": 0.0407,
      "step": 16634
    },
    {
      "epoch": 2.3314646110721795,
      "grad_norm": 0.08726966381072998,
      "learning_rate": 6.855297775651757e-05,
      "loss": 0.0148,
      "step": 16635
    },
    {
      "epoch": 2.331604765241766,
      "grad_norm": 0.06812809407711029,
      "learning_rate": 6.853862712269791e-05,
      "loss": 0.0053,
      "step": 16636
    },
    {
      "epoch": 2.3317449194113524,
      "grad_norm": 0.0888265073299408,
      "learning_rate": 6.852427648887826e-05,
      "loss": 0.0069,
      "step": 16637
    },
    {
      "epoch": 2.331885073580939,
      "grad_norm": 0.27313244342803955,
      "learning_rate": 6.850992585505859e-05,
      "loss": 0.016,
      "step": 16638
    },
    {
      "epoch": 2.332025227750526,
      "grad_norm": 0.28939148783683777,
      "learning_rate": 6.849557522123894e-05,
      "loss": 0.0227,
      "step": 16639
    },
    {
      "epoch": 2.3321653819201122,
      "grad_norm": 0.12822197377681732,
      "learning_rate": 6.848122458741927e-05,
      "loss": 0.013,
      "step": 16640
    },
    {
      "epoch": 2.3323055360896987,
      "grad_norm": 0.08915102481842041,
      "learning_rate": 6.846687395359961e-05,
      "loss": 0.0059,
      "step": 16641
    },
    {
      "epoch": 2.332445690259285,
      "grad_norm": 0.1277698278427124,
      "learning_rate": 6.845252331977995e-05,
      "loss": 0.0072,
      "step": 16642
    },
    {
      "epoch": 2.3325858444288716,
      "grad_norm": 0.4313455820083618,
      "learning_rate": 6.84381726859603e-05,
      "loss": 0.0311,
      "step": 16643
    },
    {
      "epoch": 2.332725998598458,
      "grad_norm": 0.332746297121048,
      "learning_rate": 6.842382205214062e-05,
      "loss": 0.0338,
      "step": 16644
    },
    {
      "epoch": 2.332866152768045,
      "grad_norm": 0.29175901412963867,
      "learning_rate": 6.840947141832097e-05,
      "loss": 0.0207,
      "step": 16645
    },
    {
      "epoch": 2.3330063069376314,
      "grad_norm": 0.17617519199848175,
      "learning_rate": 6.839512078450131e-05,
      "loss": 0.0188,
      "step": 16646
    },
    {
      "epoch": 2.333146461107218,
      "grad_norm": 0.321696400642395,
      "learning_rate": 6.838077015068165e-05,
      "loss": 0.0365,
      "step": 16647
    },
    {
      "epoch": 2.3332866152768044,
      "grad_norm": 0.1628393679857254,
      "learning_rate": 6.836641951686199e-05,
      "loss": 0.0272,
      "step": 16648
    },
    {
      "epoch": 2.3334267694463913,
      "grad_norm": 0.10594715923070908,
      "learning_rate": 6.835206888304232e-05,
      "loss": 0.0066,
      "step": 16649
    },
    {
      "epoch": 2.3335669236159777,
      "grad_norm": 0.7272409796714783,
      "learning_rate": 6.833771824922266e-05,
      "loss": 0.0346,
      "step": 16650
    },
    {
      "epoch": 2.333707077785564,
      "grad_norm": 0.6533247232437134,
      "learning_rate": 6.8323367615403e-05,
      "loss": 0.042,
      "step": 16651
    },
    {
      "epoch": 2.3338472319551506,
      "grad_norm": 0.10759581625461578,
      "learning_rate": 6.830901698158335e-05,
      "loss": 0.0171,
      "step": 16652
    },
    {
      "epoch": 2.333987386124737,
      "grad_norm": 0.5260326862335205,
      "learning_rate": 6.829466634776369e-05,
      "loss": 0.0673,
      "step": 16653
    },
    {
      "epoch": 2.3341275402943236,
      "grad_norm": 0.2641003429889679,
      "learning_rate": 6.828031571394402e-05,
      "loss": 0.0217,
      "step": 16654
    },
    {
      "epoch": 2.3342676944639105,
      "grad_norm": 0.18134334683418274,
      "learning_rate": 6.826596508012437e-05,
      "loss": 0.0119,
      "step": 16655
    },
    {
      "epoch": 2.334407848633497,
      "grad_norm": 0.748595654964447,
      "learning_rate": 6.82516144463047e-05,
      "loss": 0.0551,
      "step": 16656
    },
    {
      "epoch": 2.3345480028030834,
      "grad_norm": 0.5826172828674316,
      "learning_rate": 6.823726381248505e-05,
      "loss": 0.0509,
      "step": 16657
    },
    {
      "epoch": 2.33468815697267,
      "grad_norm": 0.22067295014858246,
      "learning_rate": 6.822291317866539e-05,
      "loss": 0.028,
      "step": 16658
    },
    {
      "epoch": 2.3348283111422563,
      "grad_norm": 0.44468802213668823,
      "learning_rate": 6.820856254484573e-05,
      "loss": 0.0376,
      "step": 16659
    },
    {
      "epoch": 2.334968465311843,
      "grad_norm": 0.5872768759727478,
      "learning_rate": 6.819421191102606e-05,
      "loss": 0.0672,
      "step": 16660
    },
    {
      "epoch": 2.3351086194814297,
      "grad_norm": 0.05444268882274628,
      "learning_rate": 6.81798612772064e-05,
      "loss": 0.0051,
      "step": 16661
    },
    {
      "epoch": 2.335248773651016,
      "grad_norm": 0.23349133133888245,
      "learning_rate": 6.816551064338674e-05,
      "loss": 0.0094,
      "step": 16662
    },
    {
      "epoch": 2.3353889278206026,
      "grad_norm": 0.21105697751045227,
      "learning_rate": 6.815116000956708e-05,
      "loss": 0.003,
      "step": 16663
    },
    {
      "epoch": 2.335529081990189,
      "grad_norm": 0.48818713426589966,
      "learning_rate": 6.813680937574743e-05,
      "loss": 0.0213,
      "step": 16664
    },
    {
      "epoch": 2.3356692361597755,
      "grad_norm": 0.726647138595581,
      "learning_rate": 6.812245874192775e-05,
      "loss": 0.0158,
      "step": 16665
    },
    {
      "epoch": 2.3358093903293624,
      "grad_norm": 0.3100503981113434,
      "learning_rate": 6.810810810810811e-05,
      "loss": 0.0227,
      "step": 16666
    },
    {
      "epoch": 2.335949544498949,
      "grad_norm": 1.2361353635787964,
      "learning_rate": 6.809375747428844e-05,
      "loss": 0.0699,
      "step": 16667
    },
    {
      "epoch": 2.3360896986685353,
      "grad_norm": 2.3035330772399902,
      "learning_rate": 6.807940684046878e-05,
      "loss": 0.17,
      "step": 16668
    },
    {
      "epoch": 2.336229852838122,
      "grad_norm": 1.7631738185882568,
      "learning_rate": 6.806505620664912e-05,
      "loss": 0.0744,
      "step": 16669
    },
    {
      "epoch": 2.3363700070077087,
      "grad_norm": 2.252312421798706,
      "learning_rate": 6.805070557282947e-05,
      "loss": 0.4268,
      "step": 16670
    },
    {
      "epoch": 2.336510161177295,
      "grad_norm": 0.48670411109924316,
      "learning_rate": 6.803635493900981e-05,
      "loss": 0.0744,
      "step": 16671
    },
    {
      "epoch": 2.3366503153468816,
      "grad_norm": 0.2907535135746002,
      "learning_rate": 6.802200430519014e-05,
      "loss": 0.0518,
      "step": 16672
    },
    {
      "epoch": 2.336790469516468,
      "grad_norm": 0.04928178712725639,
      "learning_rate": 6.800765367137048e-05,
      "loss": 0.0074,
      "step": 16673
    },
    {
      "epoch": 2.3369306236860545,
      "grad_norm": 0.09691784530878067,
      "learning_rate": 6.799330303755082e-05,
      "loss": 0.0054,
      "step": 16674
    },
    {
      "epoch": 2.337070777855641,
      "grad_norm": 0.08753497153520584,
      "learning_rate": 6.797895240373116e-05,
      "loss": 0.0209,
      "step": 16675
    },
    {
      "epoch": 2.337210932025228,
      "grad_norm": 0.23312413692474365,
      "learning_rate": 6.796460176991149e-05,
      "loss": 0.0241,
      "step": 16676
    },
    {
      "epoch": 2.3373510861948144,
      "grad_norm": 0.7392073273658752,
      "learning_rate": 6.795025113609183e-05,
      "loss": 0.0297,
      "step": 16677
    },
    {
      "epoch": 2.337491240364401,
      "grad_norm": 0.46310803294181824,
      "learning_rate": 6.793590050227218e-05,
      "loss": 0.0458,
      "step": 16678
    },
    {
      "epoch": 2.3376313945339873,
      "grad_norm": 0.08045361191034317,
      "learning_rate": 6.792154986845252e-05,
      "loss": 0.011,
      "step": 16679
    },
    {
      "epoch": 2.337771548703574,
      "grad_norm": 0.34028133749961853,
      "learning_rate": 6.790719923463286e-05,
      "loss": 0.0525,
      "step": 16680
    },
    {
      "epoch": 2.3379117028731606,
      "grad_norm": 0.08520636707544327,
      "learning_rate": 6.789284860081319e-05,
      "loss": 0.0301,
      "step": 16681
    },
    {
      "epoch": 2.338051857042747,
      "grad_norm": 0.811071515083313,
      "learning_rate": 6.787849796699354e-05,
      "loss": 0.0564,
      "step": 16682
    },
    {
      "epoch": 2.3381920112123336,
      "grad_norm": 0.18612436950206757,
      "learning_rate": 6.786414733317387e-05,
      "loss": 0.016,
      "step": 16683
    },
    {
      "epoch": 2.33833216538192,
      "grad_norm": 0.12141390889883041,
      "learning_rate": 6.784979669935421e-05,
      "loss": 0.0138,
      "step": 16684
    },
    {
      "epoch": 2.3384723195515065,
      "grad_norm": 0.5466933250427246,
      "learning_rate": 6.783544606553456e-05,
      "loss": 0.05,
      "step": 16685
    },
    {
      "epoch": 2.3386124737210934,
      "grad_norm": 0.4337654709815979,
      "learning_rate": 6.78210954317149e-05,
      "loss": 0.0381,
      "step": 16686
    },
    {
      "epoch": 2.33875262789068,
      "grad_norm": 0.5993040204048157,
      "learning_rate": 6.780674479789524e-05,
      "loss": 0.105,
      "step": 16687
    },
    {
      "epoch": 2.3388927820602663,
      "grad_norm": 0.10743822902441025,
      "learning_rate": 6.779239416407557e-05,
      "loss": 0.0171,
      "step": 16688
    },
    {
      "epoch": 2.3390329362298528,
      "grad_norm": 0.3625826835632324,
      "learning_rate": 6.777804353025591e-05,
      "loss": 0.075,
      "step": 16689
    },
    {
      "epoch": 2.339173090399439,
      "grad_norm": 0.14484982192516327,
      "learning_rate": 6.776369289643625e-05,
      "loss": 0.013,
      "step": 16690
    },
    {
      "epoch": 2.339313244569026,
      "grad_norm": 0.14316824078559875,
      "learning_rate": 6.77493422626166e-05,
      "loss": 0.011,
      "step": 16691
    },
    {
      "epoch": 2.3394533987386126,
      "grad_norm": 0.30650871992111206,
      "learning_rate": 6.773499162879692e-05,
      "loss": 0.0766,
      "step": 16692
    },
    {
      "epoch": 2.339593552908199,
      "grad_norm": 0.4431057870388031,
      "learning_rate": 6.772064099497727e-05,
      "loss": 0.0579,
      "step": 16693
    },
    {
      "epoch": 2.3397337070777855,
      "grad_norm": 0.1763039082288742,
      "learning_rate": 6.770629036115761e-05,
      "loss": 0.0178,
      "step": 16694
    },
    {
      "epoch": 2.339873861247372,
      "grad_norm": 0.08771751075983047,
      "learning_rate": 6.769193972733795e-05,
      "loss": 0.0097,
      "step": 16695
    },
    {
      "epoch": 2.3400140154169584,
      "grad_norm": 0.2820626497268677,
      "learning_rate": 6.767758909351829e-05,
      "loss": 0.029,
      "step": 16696
    },
    {
      "epoch": 2.3401541695865453,
      "grad_norm": 0.1556430459022522,
      "learning_rate": 6.766323845969863e-05,
      "loss": 0.0152,
      "step": 16697
    },
    {
      "epoch": 2.340294323756132,
      "grad_norm": 0.20324398577213287,
      "learning_rate": 6.764888782587898e-05,
      "loss": 0.0319,
      "step": 16698
    },
    {
      "epoch": 2.3404344779257182,
      "grad_norm": 0.1523885428905487,
      "learning_rate": 6.76345371920593e-05,
      "loss": 0.0238,
      "step": 16699
    },
    {
      "epoch": 2.3405746320953047,
      "grad_norm": 0.1927792876958847,
      "learning_rate": 6.762018655823965e-05,
      "loss": 0.0238,
      "step": 16700
    },
    {
      "epoch": 2.3407147862648916,
      "grad_norm": 1.7138257026672363,
      "learning_rate": 6.760583592441999e-05,
      "loss": 0.0583,
      "step": 16701
    },
    {
      "epoch": 2.340854940434478,
      "grad_norm": 0.130364328622818,
      "learning_rate": 6.759148529060033e-05,
      "loss": 0.0125,
      "step": 16702
    },
    {
      "epoch": 2.3409950946040645,
      "grad_norm": 0.31931135058403015,
      "learning_rate": 6.757713465678067e-05,
      "loss": 0.0141,
      "step": 16703
    },
    {
      "epoch": 2.341135248773651,
      "grad_norm": 0.1332167088985443,
      "learning_rate": 6.7562784022961e-05,
      "loss": 0.0141,
      "step": 16704
    },
    {
      "epoch": 2.3412754029432374,
      "grad_norm": 0.16314949095249176,
      "learning_rate": 6.754843338914136e-05,
      "loss": 0.0557,
      "step": 16705
    },
    {
      "epoch": 2.341415557112824,
      "grad_norm": 0.2002786546945572,
      "learning_rate": 6.753408275532169e-05,
      "loss": 0.0344,
      "step": 16706
    },
    {
      "epoch": 2.341555711282411,
      "grad_norm": 0.42262527346611023,
      "learning_rate": 6.751973212150203e-05,
      "loss": 0.009,
      "step": 16707
    },
    {
      "epoch": 2.3416958654519973,
      "grad_norm": 0.6499032378196716,
      "learning_rate": 6.750538148768236e-05,
      "loss": 0.0362,
      "step": 16708
    },
    {
      "epoch": 2.3418360196215837,
      "grad_norm": 0.23660887777805328,
      "learning_rate": 6.749103085386271e-05,
      "loss": 0.0119,
      "step": 16709
    },
    {
      "epoch": 2.34197617379117,
      "grad_norm": 0.5324763059616089,
      "learning_rate": 6.747668022004304e-05,
      "loss": 0.0345,
      "step": 16710
    },
    {
      "epoch": 2.3421163279607566,
      "grad_norm": 1.1433472633361816,
      "learning_rate": 6.746232958622338e-05,
      "loss": 0.1076,
      "step": 16711
    },
    {
      "epoch": 2.3422564821303435,
      "grad_norm": 0.4222467541694641,
      "learning_rate": 6.744797895240373e-05,
      "loss": 0.0739,
      "step": 16712
    },
    {
      "epoch": 2.34239663629993,
      "grad_norm": 0.2540854811668396,
      "learning_rate": 6.743362831858407e-05,
      "loss": 0.0084,
      "step": 16713
    },
    {
      "epoch": 2.3425367904695165,
      "grad_norm": 0.015154390595853329,
      "learning_rate": 6.741927768476441e-05,
      "loss": 0.0014,
      "step": 16714
    },
    {
      "epoch": 2.342676944639103,
      "grad_norm": 0.059180036187171936,
      "learning_rate": 6.740492705094474e-05,
      "loss": 0.0081,
      "step": 16715
    },
    {
      "epoch": 2.3428170988086894,
      "grad_norm": 0.4236818253993988,
      "learning_rate": 6.739057641712508e-05,
      "loss": 0.0134,
      "step": 16716
    },
    {
      "epoch": 2.3429572529782763,
      "grad_norm": 0.03651474788784981,
      "learning_rate": 6.737622578330542e-05,
      "loss": 0.0029,
      "step": 16717
    },
    {
      "epoch": 2.3430974071478627,
      "grad_norm": 0.2646081745624542,
      "learning_rate": 6.736187514948576e-05,
      "loss": 0.0084,
      "step": 16718
    },
    {
      "epoch": 2.343237561317449,
      "grad_norm": 0.1457076519727707,
      "learning_rate": 6.734752451566611e-05,
      "loss": 0.0208,
      "step": 16719
    },
    {
      "epoch": 2.3433777154870357,
      "grad_norm": 3.895206928253174,
      "learning_rate": 6.733317388184644e-05,
      "loss": 0.3217,
      "step": 16720
    },
    {
      "epoch": 2.343517869656622,
      "grad_norm": 0.08148682862520218,
      "learning_rate": 6.731882324802679e-05,
      "loss": 0.009,
      "step": 16721
    },
    {
      "epoch": 2.343658023826209,
      "grad_norm": 0.38824546337127686,
      "learning_rate": 6.730447261420712e-05,
      "loss": 0.0375,
      "step": 16722
    },
    {
      "epoch": 2.3437981779957955,
      "grad_norm": 0.48277169466018677,
      "learning_rate": 6.729012198038746e-05,
      "loss": 0.0223,
      "step": 16723
    },
    {
      "epoch": 2.343938332165382,
      "grad_norm": 0.07657844573259354,
      "learning_rate": 6.727577134656779e-05,
      "loss": 0.0026,
      "step": 16724
    },
    {
      "epoch": 2.3440784863349684,
      "grad_norm": 0.08366616815328598,
      "learning_rate": 6.726142071274815e-05,
      "loss": 0.0049,
      "step": 16725
    },
    {
      "epoch": 2.344218640504555,
      "grad_norm": 0.1319780796766281,
      "learning_rate": 6.724707007892847e-05,
      "loss": 0.0258,
      "step": 16726
    },
    {
      "epoch": 2.3443587946741413,
      "grad_norm": 0.03102542832493782,
      "learning_rate": 6.723271944510882e-05,
      "loss": 0.0076,
      "step": 16727
    },
    {
      "epoch": 2.3444989488437282,
      "grad_norm": 0.32032671570777893,
      "learning_rate": 6.721836881128916e-05,
      "loss": 0.0235,
      "step": 16728
    },
    {
      "epoch": 2.3446391030133147,
      "grad_norm": 0.23940148949623108,
      "learning_rate": 6.72040181774695e-05,
      "loss": 0.0126,
      "step": 16729
    },
    {
      "epoch": 2.344779257182901,
      "grad_norm": 0.26261255145072937,
      "learning_rate": 6.718966754364984e-05,
      "loss": 0.034,
      "step": 16730
    },
    {
      "epoch": 2.3449194113524876,
      "grad_norm": 0.20490874350070953,
      "learning_rate": 6.717531690983017e-05,
      "loss": 0.0265,
      "step": 16731
    },
    {
      "epoch": 2.3450595655220745,
      "grad_norm": 0.20463722944259644,
      "learning_rate": 6.716096627601051e-05,
      "loss": 0.0186,
      "step": 16732
    },
    {
      "epoch": 2.345199719691661,
      "grad_norm": 0.15001113712787628,
      "learning_rate": 6.714661564219086e-05,
      "loss": 0.0223,
      "step": 16733
    },
    {
      "epoch": 2.3453398738612474,
      "grad_norm": 0.056073326617479324,
      "learning_rate": 6.71322650083712e-05,
      "loss": 0.0043,
      "step": 16734
    },
    {
      "epoch": 2.345480028030834,
      "grad_norm": 0.18617713451385498,
      "learning_rate": 6.711791437455154e-05,
      "loss": 0.026,
      "step": 16735
    },
    {
      "epoch": 2.3456201822004203,
      "grad_norm": 0.18807999789714813,
      "learning_rate": 6.710356374073188e-05,
      "loss": 0.0369,
      "step": 16736
    },
    {
      "epoch": 2.345760336370007,
      "grad_norm": 0.16750648617744446,
      "learning_rate": 6.708921310691222e-05,
      "loss": 0.0091,
      "step": 16737
    },
    {
      "epoch": 2.3459004905395937,
      "grad_norm": 0.3186100423336029,
      "learning_rate": 6.707486247309255e-05,
      "loss": 0.0231,
      "step": 16738
    },
    {
      "epoch": 2.34604064470918,
      "grad_norm": 0.3226334750652313,
      "learning_rate": 6.70605118392729e-05,
      "loss": 0.0141,
      "step": 16739
    },
    {
      "epoch": 2.3461807988787666,
      "grad_norm": 0.27671194076538086,
      "learning_rate": 6.704616120545324e-05,
      "loss": 0.0178,
      "step": 16740
    },
    {
      "epoch": 2.346320953048353,
      "grad_norm": 0.35524260997772217,
      "learning_rate": 6.703181057163358e-05,
      "loss": 0.0237,
      "step": 16741
    },
    {
      "epoch": 2.3464611072179395,
      "grad_norm": 0.4764234721660614,
      "learning_rate": 6.701745993781391e-05,
      "loss": 0.0558,
      "step": 16742
    },
    {
      "epoch": 2.3466012613875264,
      "grad_norm": 0.16321344673633575,
      "learning_rate": 6.700310930399425e-05,
      "loss": 0.0106,
      "step": 16743
    },
    {
      "epoch": 2.346741415557113,
      "grad_norm": 0.4195289611816406,
      "learning_rate": 6.698875867017459e-05,
      "loss": 0.0147,
      "step": 16744
    },
    {
      "epoch": 2.3468815697266994,
      "grad_norm": 0.0997488722205162,
      "learning_rate": 6.697440803635493e-05,
      "loss": 0.011,
      "step": 16745
    },
    {
      "epoch": 2.347021723896286,
      "grad_norm": 0.530870258808136,
      "learning_rate": 6.696005740253528e-05,
      "loss": 0.0268,
      "step": 16746
    },
    {
      "epoch": 2.3471618780658723,
      "grad_norm": 0.08970702439546585,
      "learning_rate": 6.69457067687156e-05,
      "loss": 0.0058,
      "step": 16747
    },
    {
      "epoch": 2.347302032235459,
      "grad_norm": 0.16014617681503296,
      "learning_rate": 6.693135613489596e-05,
      "loss": 0.0359,
      "step": 16748
    },
    {
      "epoch": 2.3474421864050456,
      "grad_norm": 0.21170884370803833,
      "learning_rate": 6.691700550107629e-05,
      "loss": 0.0098,
      "step": 16749
    },
    {
      "epoch": 2.347582340574632,
      "grad_norm": 0.20161627233028412,
      "learning_rate": 6.690265486725663e-05,
      "loss": 0.0165,
      "step": 16750
    },
    {
      "epoch": 2.3477224947442186,
      "grad_norm": 0.08059614896774292,
      "learning_rate": 6.688830423343697e-05,
      "loss": 0.0395,
      "step": 16751
    },
    {
      "epoch": 2.347862648913805,
      "grad_norm": 0.3088594377040863,
      "learning_rate": 6.687395359961732e-05,
      "loss": 0.0165,
      "step": 16752
    },
    {
      "epoch": 2.348002803083392,
      "grad_norm": 0.1841048002243042,
      "learning_rate": 6.685960296579766e-05,
      "loss": 0.0107,
      "step": 16753
    },
    {
      "epoch": 2.3481429572529784,
      "grad_norm": 0.7148263454437256,
      "learning_rate": 6.684525233197799e-05,
      "loss": 0.0472,
      "step": 16754
    },
    {
      "epoch": 2.348283111422565,
      "grad_norm": 0.0530632808804512,
      "learning_rate": 6.683090169815833e-05,
      "loss": 0.0036,
      "step": 16755
    },
    {
      "epoch": 2.3484232655921513,
      "grad_norm": 0.16300493478775024,
      "learning_rate": 6.681655106433867e-05,
      "loss": 0.0114,
      "step": 16756
    },
    {
      "epoch": 2.3485634197617378,
      "grad_norm": 0.0721244141459465,
      "learning_rate": 6.680220043051901e-05,
      "loss": 0.0035,
      "step": 16757
    },
    {
      "epoch": 2.3487035739313242,
      "grad_norm": 0.13145971298217773,
      "learning_rate": 6.678784979669934e-05,
      "loss": 0.0508,
      "step": 16758
    },
    {
      "epoch": 2.348843728100911,
      "grad_norm": 0.07286734879016876,
      "learning_rate": 6.677349916287968e-05,
      "loss": 0.0031,
      "step": 16759
    },
    {
      "epoch": 2.3489838822704976,
      "grad_norm": 0.17474566400051117,
      "learning_rate": 6.675914852906003e-05,
      "loss": 0.0191,
      "step": 16760
    },
    {
      "epoch": 2.349124036440084,
      "grad_norm": 0.2069219946861267,
      "learning_rate": 6.674479789524037e-05,
      "loss": 0.0066,
      "step": 16761
    },
    {
      "epoch": 2.3492641906096705,
      "grad_norm": 0.22322790324687958,
      "learning_rate": 6.673044726142071e-05,
      "loss": 0.0192,
      "step": 16762
    },
    {
      "epoch": 2.3494043447792574,
      "grad_norm": 0.15062567591667175,
      "learning_rate": 6.671609662760104e-05,
      "loss": 0.0241,
      "step": 16763
    },
    {
      "epoch": 2.349544498948844,
      "grad_norm": 0.467182993888855,
      "learning_rate": 6.67017459937814e-05,
      "loss": 0.0133,
      "step": 16764
    },
    {
      "epoch": 2.3496846531184303,
      "grad_norm": 0.08561839163303375,
      "learning_rate": 6.668739535996172e-05,
      "loss": 0.0056,
      "step": 16765
    },
    {
      "epoch": 2.349824807288017,
      "grad_norm": 0.26732197403907776,
      "learning_rate": 6.667304472614206e-05,
      "loss": 0.0224,
      "step": 16766
    },
    {
      "epoch": 2.3499649614576033,
      "grad_norm": 0.8566556572914124,
      "learning_rate": 6.66586940923224e-05,
      "loss": 0.1034,
      "step": 16767
    },
    {
      "epoch": 2.3501051156271897,
      "grad_norm": 1.312525987625122,
      "learning_rate": 6.664434345850275e-05,
      "loss": 0.0357,
      "step": 16768
    },
    {
      "epoch": 2.3502452697967766,
      "grad_norm": 0.05718226358294487,
      "learning_rate": 6.662999282468309e-05,
      "loss": 0.0034,
      "step": 16769
    },
    {
      "epoch": 2.350385423966363,
      "grad_norm": 1.637704610824585,
      "learning_rate": 6.661564219086342e-05,
      "loss": 0.0364,
      "step": 16770
    },
    {
      "epoch": 2.3505255781359495,
      "grad_norm": 0.2637943625450134,
      "learning_rate": 6.660129155704376e-05,
      "loss": 0.046,
      "step": 16771
    },
    {
      "epoch": 2.350665732305536,
      "grad_norm": 0.11931415647268295,
      "learning_rate": 6.65869409232241e-05,
      "loss": 0.0058,
      "step": 16772
    },
    {
      "epoch": 2.3508058864751225,
      "grad_norm": 0.2528848946094513,
      "learning_rate": 6.657259028940445e-05,
      "loss": 0.0317,
      "step": 16773
    },
    {
      "epoch": 2.3509460406447094,
      "grad_norm": 0.5890038013458252,
      "learning_rate": 6.655823965558477e-05,
      "loss": 0.0386,
      "step": 16774
    },
    {
      "epoch": 2.351086194814296,
      "grad_norm": 0.020809343084692955,
      "learning_rate": 6.654388902176513e-05,
      "loss": 0.0012,
      "step": 16775
    },
    {
      "epoch": 2.3512263489838823,
      "grad_norm": 0.3048439919948578,
      "learning_rate": 6.652953838794546e-05,
      "loss": 0.0096,
      "step": 16776
    },
    {
      "epoch": 2.3513665031534687,
      "grad_norm": 0.29299676418304443,
      "learning_rate": 6.65151877541258e-05,
      "loss": 0.042,
      "step": 16777
    },
    {
      "epoch": 2.351506657323055,
      "grad_norm": 0.18414941430091858,
      "learning_rate": 6.650083712030614e-05,
      "loss": 0.0148,
      "step": 16778
    },
    {
      "epoch": 2.351646811492642,
      "grad_norm": 0.07064701616764069,
      "learning_rate": 6.648648648648648e-05,
      "loss": 0.0077,
      "step": 16779
    },
    {
      "epoch": 2.3517869656622286,
      "grad_norm": 0.2584095299243927,
      "learning_rate": 6.647213585266683e-05,
      "loss": 0.0431,
      "step": 16780
    },
    {
      "epoch": 2.351927119831815,
      "grad_norm": 0.36756014823913574,
      "learning_rate": 6.645778521884716e-05,
      "loss": 0.0328,
      "step": 16781
    },
    {
      "epoch": 2.3520672740014015,
      "grad_norm": 0.2210976928472519,
      "learning_rate": 6.64434345850275e-05,
      "loss": 0.0256,
      "step": 16782
    },
    {
      "epoch": 2.352207428170988,
      "grad_norm": 0.2337346076965332,
      "learning_rate": 6.642908395120784e-05,
      "loss": 0.009,
      "step": 16783
    },
    {
      "epoch": 2.352347582340575,
      "grad_norm": 0.2046017199754715,
      "learning_rate": 6.641473331738818e-05,
      "loss": 0.0282,
      "step": 16784
    },
    {
      "epoch": 2.3524877365101613,
      "grad_norm": 0.3241329491138458,
      "learning_rate": 6.640038268356852e-05,
      "loss": 0.0384,
      "step": 16785
    },
    {
      "epoch": 2.3526278906797478,
      "grad_norm": 0.11610342562198639,
      "learning_rate": 6.638603204974885e-05,
      "loss": 0.0065,
      "step": 16786
    },
    {
      "epoch": 2.352768044849334,
      "grad_norm": 0.15356920659542084,
      "learning_rate": 6.63716814159292e-05,
      "loss": 0.0214,
      "step": 16787
    },
    {
      "epoch": 2.3529081990189207,
      "grad_norm": 0.391592800617218,
      "learning_rate": 6.635733078210954e-05,
      "loss": 0.0359,
      "step": 16788
    },
    {
      "epoch": 2.353048353188507,
      "grad_norm": 0.25107723474502563,
      "learning_rate": 6.634298014828988e-05,
      "loss": 0.034,
      "step": 16789
    },
    {
      "epoch": 2.353188507358094,
      "grad_norm": 0.09637567400932312,
      "learning_rate": 6.632862951447021e-05,
      "loss": 0.0138,
      "step": 16790
    },
    {
      "epoch": 2.3533286615276805,
      "grad_norm": 0.8719984889030457,
      "learning_rate": 6.631427888065056e-05,
      "loss": 0.07,
      "step": 16791
    },
    {
      "epoch": 2.353468815697267,
      "grad_norm": 0.027512352913618088,
      "learning_rate": 6.629992824683089e-05,
      "loss": 0.0028,
      "step": 16792
    },
    {
      "epoch": 2.3536089698668534,
      "grad_norm": 0.12451263517141342,
      "learning_rate": 6.628557761301123e-05,
      "loss": 0.018,
      "step": 16793
    },
    {
      "epoch": 2.3537491240364403,
      "grad_norm": 0.1623818278312683,
      "learning_rate": 6.627122697919158e-05,
      "loss": 0.0174,
      "step": 16794
    },
    {
      "epoch": 2.353889278206027,
      "grad_norm": 0.326121062040329,
      "learning_rate": 6.625687634537192e-05,
      "loss": 0.0224,
      "step": 16795
    },
    {
      "epoch": 2.3540294323756132,
      "grad_norm": 0.5130059123039246,
      "learning_rate": 6.624252571155226e-05,
      "loss": 0.0731,
      "step": 16796
    },
    {
      "epoch": 2.3541695865451997,
      "grad_norm": 0.166997030377388,
      "learning_rate": 6.622817507773259e-05,
      "loss": 0.0168,
      "step": 16797
    },
    {
      "epoch": 2.354309740714786,
      "grad_norm": 0.03297094255685806,
      "learning_rate": 6.621382444391293e-05,
      "loss": 0.004,
      "step": 16798
    },
    {
      "epoch": 2.3544498948843726,
      "grad_norm": 0.11285286396741867,
      "learning_rate": 6.619947381009327e-05,
      "loss": 0.0515,
      "step": 16799
    },
    {
      "epoch": 2.3545900490539595,
      "grad_norm": 0.1945682018995285,
      "learning_rate": 6.618512317627361e-05,
      "loss": 0.0165,
      "step": 16800
    },
    {
      "epoch": 2.354730203223546,
      "grad_norm": 0.27851366996765137,
      "learning_rate": 6.617077254245396e-05,
      "loss": 0.0182,
      "step": 16801
    },
    {
      "epoch": 2.3548703573931324,
      "grad_norm": 0.2644082009792328,
      "learning_rate": 6.615642190863429e-05,
      "loss": 0.0518,
      "step": 16802
    },
    {
      "epoch": 2.355010511562719,
      "grad_norm": 0.4891209304332733,
      "learning_rate": 6.614207127481463e-05,
      "loss": 0.0563,
      "step": 16803
    },
    {
      "epoch": 2.3551506657323054,
      "grad_norm": 0.18028511106967926,
      "learning_rate": 6.612772064099497e-05,
      "loss": 0.0075,
      "step": 16804
    },
    {
      "epoch": 2.3552908199018923,
      "grad_norm": 0.07767307758331299,
      "learning_rate": 6.611337000717531e-05,
      "loss": 0.0074,
      "step": 16805
    },
    {
      "epoch": 2.3554309740714787,
      "grad_norm": 0.05823894217610359,
      "learning_rate": 6.609901937335564e-05,
      "loss": 0.0031,
      "step": 16806
    },
    {
      "epoch": 2.355571128241065,
      "grad_norm": 0.3098495602607727,
      "learning_rate": 6.6084668739536e-05,
      "loss": 0.0513,
      "step": 16807
    },
    {
      "epoch": 2.3557112824106516,
      "grad_norm": 0.051881611347198486,
      "learning_rate": 6.607031810571632e-05,
      "loss": 0.0044,
      "step": 16808
    },
    {
      "epoch": 2.355851436580238,
      "grad_norm": 0.2774588167667389,
      "learning_rate": 6.605596747189667e-05,
      "loss": 0.0577,
      "step": 16809
    },
    {
      "epoch": 2.3559915907498246,
      "grad_norm": 0.86620032787323,
      "learning_rate": 6.604161683807701e-05,
      "loss": 0.1558,
      "step": 16810
    },
    {
      "epoch": 2.3561317449194115,
      "grad_norm": 0.6263657212257385,
      "learning_rate": 6.602726620425735e-05,
      "loss": 0.0308,
      "step": 16811
    },
    {
      "epoch": 2.356271899088998,
      "grad_norm": 0.3518466353416443,
      "learning_rate": 6.601291557043769e-05,
      "loss": 0.0751,
      "step": 16812
    },
    {
      "epoch": 2.3564120532585844,
      "grad_norm": 0.5598931312561035,
      "learning_rate": 6.599856493661802e-05,
      "loss": 0.0562,
      "step": 16813
    },
    {
      "epoch": 2.356552207428171,
      "grad_norm": 0.41986972093582153,
      "learning_rate": 6.598421430279838e-05,
      "loss": 0.032,
      "step": 16814
    },
    {
      "epoch": 2.3566923615977577,
      "grad_norm": 0.3903028964996338,
      "learning_rate": 6.59698636689787e-05,
      "loss": 0.0715,
      "step": 16815
    },
    {
      "epoch": 2.356832515767344,
      "grad_norm": 1.2094265222549438,
      "learning_rate": 6.595551303515905e-05,
      "loss": 0.0303,
      "step": 16816
    },
    {
      "epoch": 2.3569726699369307,
      "grad_norm": 1.2694437503814697,
      "learning_rate": 6.594116240133939e-05,
      "loss": 0.0623,
      "step": 16817
    },
    {
      "epoch": 2.357112824106517,
      "grad_norm": 1.1664642095565796,
      "learning_rate": 6.592681176751973e-05,
      "loss": 0.0516,
      "step": 16818
    },
    {
      "epoch": 2.3572529782761036,
      "grad_norm": 1.7885206937789917,
      "learning_rate": 6.591246113370006e-05,
      "loss": 0.0481,
      "step": 16819
    },
    {
      "epoch": 2.35739313244569,
      "grad_norm": 0.8192026019096375,
      "learning_rate": 6.58981104998804e-05,
      "loss": 0.0995,
      "step": 16820
    },
    {
      "epoch": 2.357533286615277,
      "grad_norm": 0.7249272465705872,
      "learning_rate": 6.588375986606075e-05,
      "loss": 0.014,
      "step": 16821
    },
    {
      "epoch": 2.3576734407848634,
      "grad_norm": 0.3018769323825836,
      "learning_rate": 6.586940923224109e-05,
      "loss": 0.0325,
      "step": 16822
    },
    {
      "epoch": 2.35781359495445,
      "grad_norm": 0.2909446656703949,
      "learning_rate": 6.585505859842143e-05,
      "loss": 0.0352,
      "step": 16823
    },
    {
      "epoch": 2.3579537491240363,
      "grad_norm": 0.2735503911972046,
      "learning_rate": 6.584070796460176e-05,
      "loss": 0.0152,
      "step": 16824
    },
    {
      "epoch": 2.3580939032936232,
      "grad_norm": 0.3283838927745819,
      "learning_rate": 6.58263573307821e-05,
      "loss": 0.0351,
      "step": 16825
    },
    {
      "epoch": 2.3582340574632097,
      "grad_norm": 0.2914266884326935,
      "learning_rate": 6.581200669696244e-05,
      "loss": 0.0706,
      "step": 16826
    },
    {
      "epoch": 2.358374211632796,
      "grad_norm": 0.2754274904727936,
      "learning_rate": 6.579765606314278e-05,
      "loss": 0.0355,
      "step": 16827
    },
    {
      "epoch": 2.3585143658023826,
      "grad_norm": 0.2437049150466919,
      "learning_rate": 6.578330542932313e-05,
      "loss": 0.0664,
      "step": 16828
    },
    {
      "epoch": 2.358654519971969,
      "grad_norm": 0.3542616665363312,
      "learning_rate": 6.576895479550345e-05,
      "loss": 0.0498,
      "step": 16829
    },
    {
      "epoch": 2.3587946741415555,
      "grad_norm": 0.3559741973876953,
      "learning_rate": 6.575460416168381e-05,
      "loss": 0.019,
      "step": 16830
    },
    {
      "epoch": 2.3589348283111424,
      "grad_norm": 0.08941761404275894,
      "learning_rate": 6.574025352786414e-05,
      "loss": 0.0056,
      "step": 16831
    },
    {
      "epoch": 2.359074982480729,
      "grad_norm": 0.10273107886314392,
      "learning_rate": 6.572590289404448e-05,
      "loss": 0.0084,
      "step": 16832
    },
    {
      "epoch": 2.3592151366503153,
      "grad_norm": 0.27030569314956665,
      "learning_rate": 6.571155226022482e-05,
      "loss": 0.0252,
      "step": 16833
    },
    {
      "epoch": 2.359355290819902,
      "grad_norm": 0.17541047930717468,
      "learning_rate": 6.569720162640517e-05,
      "loss": 0.0208,
      "step": 16834
    },
    {
      "epoch": 2.3594954449894883,
      "grad_norm": 0.171622171998024,
      "learning_rate": 6.56828509925855e-05,
      "loss": 0.0261,
      "step": 16835
    },
    {
      "epoch": 2.359635599159075,
      "grad_norm": 0.1687568575143814,
      "learning_rate": 6.566850035876584e-05,
      "loss": 0.0221,
      "step": 16836
    },
    {
      "epoch": 2.3597757533286616,
      "grad_norm": 0.2137746810913086,
      "learning_rate": 6.565414972494618e-05,
      "loss": 0.0412,
      "step": 16837
    },
    {
      "epoch": 2.359915907498248,
      "grad_norm": 0.1906731128692627,
      "learning_rate": 6.563979909112652e-05,
      "loss": 0.015,
      "step": 16838
    },
    {
      "epoch": 2.3600560616678345,
      "grad_norm": 0.12018901854753494,
      "learning_rate": 6.562544845730686e-05,
      "loss": 0.0068,
      "step": 16839
    },
    {
      "epoch": 2.360196215837421,
      "grad_norm": 0.25816285610198975,
      "learning_rate": 6.561109782348719e-05,
      "loss": 0.0246,
      "step": 16840
    },
    {
      "epoch": 2.3603363700070075,
      "grad_norm": 0.22464999556541443,
      "learning_rate": 6.559674718966753e-05,
      "loss": 0.031,
      "step": 16841
    },
    {
      "epoch": 2.3604765241765944,
      "grad_norm": 0.2570783495903015,
      "learning_rate": 6.558239655584788e-05,
      "loss": 0.0389,
      "step": 16842
    },
    {
      "epoch": 2.360616678346181,
      "grad_norm": 0.08131103217601776,
      "learning_rate": 6.556804592202822e-05,
      "loss": 0.005,
      "step": 16843
    },
    {
      "epoch": 2.3607568325157673,
      "grad_norm": 0.4477812647819519,
      "learning_rate": 6.555369528820856e-05,
      "loss": 0.0459,
      "step": 16844
    },
    {
      "epoch": 2.3608969866853537,
      "grad_norm": 0.36922189593315125,
      "learning_rate": 6.55393446543889e-05,
      "loss": 0.0109,
      "step": 16845
    },
    {
      "epoch": 2.3610371408549407,
      "grad_norm": 0.19609679281711578,
      "learning_rate": 6.552499402056924e-05,
      "loss": 0.0048,
      "step": 16846
    },
    {
      "epoch": 2.361177295024527,
      "grad_norm": 0.03386911377310753,
      "learning_rate": 6.551064338674957e-05,
      "loss": 0.0032,
      "step": 16847
    },
    {
      "epoch": 2.3613174491941136,
      "grad_norm": 0.6084646582603455,
      "learning_rate": 6.549629275292991e-05,
      "loss": 0.0682,
      "step": 16848
    },
    {
      "epoch": 2.3614576033637,
      "grad_norm": 0.17862890660762787,
      "learning_rate": 6.548194211911026e-05,
      "loss": 0.0171,
      "step": 16849
    },
    {
      "epoch": 2.3615977575332865,
      "grad_norm": 0.38720476627349854,
      "learning_rate": 6.54675914852906e-05,
      "loss": 0.0075,
      "step": 16850
    },
    {
      "epoch": 2.361737911702873,
      "grad_norm": 0.11374688148498535,
      "learning_rate": 6.545324085147094e-05,
      "loss": 0.0113,
      "step": 16851
    },
    {
      "epoch": 2.36187806587246,
      "grad_norm": 0.3034529983997345,
      "learning_rate": 6.543889021765127e-05,
      "loss": 0.0185,
      "step": 16852
    },
    {
      "epoch": 2.3620182200420463,
      "grad_norm": 0.37107303738594055,
      "learning_rate": 6.542453958383161e-05,
      "loss": 0.0446,
      "step": 16853
    },
    {
      "epoch": 2.3621583742116328,
      "grad_norm": 0.2727849781513214,
      "learning_rate": 6.541018895001195e-05,
      "loss": 0.0659,
      "step": 16854
    },
    {
      "epoch": 2.3622985283812192,
      "grad_norm": 0.15175366401672363,
      "learning_rate": 6.53958383161923e-05,
      "loss": 0.0052,
      "step": 16855
    },
    {
      "epoch": 2.362438682550806,
      "grad_norm": 0.2139585167169571,
      "learning_rate": 6.538148768237262e-05,
      "loss": 0.0286,
      "step": 16856
    },
    {
      "epoch": 2.3625788367203926,
      "grad_norm": 0.06553781032562256,
      "learning_rate": 6.536713704855298e-05,
      "loss": 0.0034,
      "step": 16857
    },
    {
      "epoch": 2.362718990889979,
      "grad_norm": 0.25779879093170166,
      "learning_rate": 6.535278641473331e-05,
      "loss": 0.0318,
      "step": 16858
    },
    {
      "epoch": 2.3628591450595655,
      "grad_norm": 0.28648388385772705,
      "learning_rate": 6.533843578091365e-05,
      "loss": 0.0087,
      "step": 16859
    },
    {
      "epoch": 2.362999299229152,
      "grad_norm": 0.3281601667404175,
      "learning_rate": 6.532408514709399e-05,
      "loss": 0.0429,
      "step": 16860
    },
    {
      "epoch": 2.3631394533987384,
      "grad_norm": 0.2057129293680191,
      "learning_rate": 6.530973451327433e-05,
      "loss": 0.0171,
      "step": 16861
    },
    {
      "epoch": 2.3632796075683253,
      "grad_norm": 0.4731660485267639,
      "learning_rate": 6.529538387945468e-05,
      "loss": 0.0912,
      "step": 16862
    },
    {
      "epoch": 2.363419761737912,
      "grad_norm": 0.7325529456138611,
      "learning_rate": 6.5281033245635e-05,
      "loss": 0.0377,
      "step": 16863
    },
    {
      "epoch": 2.3635599159074983,
      "grad_norm": 0.16006779670715332,
      "learning_rate": 6.526668261181535e-05,
      "loss": 0.0186,
      "step": 16864
    },
    {
      "epoch": 2.3637000700770847,
      "grad_norm": 0.43588986992836,
      "learning_rate": 6.525233197799569e-05,
      "loss": 0.0601,
      "step": 16865
    },
    {
      "epoch": 2.363840224246671,
      "grad_norm": 0.025887934491038322,
      "learning_rate": 6.523798134417603e-05,
      "loss": 0.0026,
      "step": 16866
    },
    {
      "epoch": 2.363980378416258,
      "grad_norm": 0.4535498321056366,
      "learning_rate": 6.522363071035637e-05,
      "loss": 0.0832,
      "step": 16867
    },
    {
      "epoch": 2.3641205325858445,
      "grad_norm": 2.6256422996520996,
      "learning_rate": 6.52092800765367e-05,
      "loss": 0.0216,
      "step": 16868
    },
    {
      "epoch": 2.364260686755431,
      "grad_norm": 2.908099412918091,
      "learning_rate": 6.519492944271704e-05,
      "loss": 0.0863,
      "step": 16869
    },
    {
      "epoch": 2.3644008409250175,
      "grad_norm": 0.8879013061523438,
      "learning_rate": 6.518057880889739e-05,
      "loss": 0.2711,
      "step": 16870
    },
    {
      "epoch": 2.364540995094604,
      "grad_norm": 0.25158190727233887,
      "learning_rate": 6.516622817507773e-05,
      "loss": 0.009,
      "step": 16871
    },
    {
      "epoch": 2.3646811492641904,
      "grad_norm": 0.13140380382537842,
      "learning_rate": 6.515187754125806e-05,
      "loss": 0.02,
      "step": 16872
    },
    {
      "epoch": 2.3648213034337773,
      "grad_norm": 0.1571803241968155,
      "learning_rate": 6.513752690743841e-05,
      "loss": 0.0107,
      "step": 16873
    },
    {
      "epoch": 2.3649614576033637,
      "grad_norm": 0.08212386816740036,
      "learning_rate": 6.512317627361874e-05,
      "loss": 0.0089,
      "step": 16874
    },
    {
      "epoch": 2.36510161177295,
      "grad_norm": 0.5521555542945862,
      "learning_rate": 6.510882563979908e-05,
      "loss": 0.0331,
      "step": 16875
    },
    {
      "epoch": 2.3652417659425367,
      "grad_norm": 0.0706278383731842,
      "learning_rate": 6.509447500597943e-05,
      "loss": 0.0048,
      "step": 16876
    },
    {
      "epoch": 2.3653819201121236,
      "grad_norm": 0.12548606097698212,
      "learning_rate": 6.508012437215977e-05,
      "loss": 0.0057,
      "step": 16877
    },
    {
      "epoch": 2.36552207428171,
      "grad_norm": 0.2095845490694046,
      "learning_rate": 6.506577373834011e-05,
      "loss": 0.0121,
      "step": 16878
    },
    {
      "epoch": 2.3656622284512965,
      "grad_norm": 0.23944036662578583,
      "learning_rate": 6.505142310452044e-05,
      "loss": 0.0224,
      "step": 16879
    },
    {
      "epoch": 2.365802382620883,
      "grad_norm": 0.7114379405975342,
      "learning_rate": 6.503707247070078e-05,
      "loss": 0.0444,
      "step": 16880
    },
    {
      "epoch": 2.3659425367904694,
      "grad_norm": 0.24444064497947693,
      "learning_rate": 6.502272183688112e-05,
      "loss": 0.0221,
      "step": 16881
    },
    {
      "epoch": 2.366082690960056,
      "grad_norm": 0.22728760540485382,
      "learning_rate": 6.500837120306146e-05,
      "loss": 0.0072,
      "step": 16882
    },
    {
      "epoch": 2.3662228451296428,
      "grad_norm": 0.7124162912368774,
      "learning_rate": 6.499402056924181e-05,
      "loss": 0.0407,
      "step": 16883
    },
    {
      "epoch": 2.366362999299229,
      "grad_norm": 0.05480144917964935,
      "learning_rate": 6.497966993542215e-05,
      "loss": 0.004,
      "step": 16884
    },
    {
      "epoch": 2.3665031534688157,
      "grad_norm": 0.20836380124092102,
      "learning_rate": 6.496531930160248e-05,
      "loss": 0.0376,
      "step": 16885
    },
    {
      "epoch": 2.366643307638402,
      "grad_norm": 0.25173670053482056,
      "learning_rate": 6.495096866778282e-05,
      "loss": 0.0205,
      "step": 16886
    },
    {
      "epoch": 2.3667834618079886,
      "grad_norm": 0.23888419568538666,
      "learning_rate": 6.493661803396316e-05,
      "loss": 0.0152,
      "step": 16887
    },
    {
      "epoch": 2.3669236159775755,
      "grad_norm": 0.11593123525381088,
      "learning_rate": 6.49222674001435e-05,
      "loss": 0.0155,
      "step": 16888
    },
    {
      "epoch": 2.367063770147162,
      "grad_norm": 0.3511277437210083,
      "learning_rate": 6.490791676632385e-05,
      "loss": 0.0882,
      "step": 16889
    },
    {
      "epoch": 2.3672039243167484,
      "grad_norm": 0.3233483135700226,
      "learning_rate": 6.489356613250417e-05,
      "loss": 0.0222,
      "step": 16890
    },
    {
      "epoch": 2.367344078486335,
      "grad_norm": 0.4679713845252991,
      "learning_rate": 6.487921549868452e-05,
      "loss": 0.0248,
      "step": 16891
    },
    {
      "epoch": 2.3674842326559213,
      "grad_norm": 0.18504492938518524,
      "learning_rate": 6.486486486486486e-05,
      "loss": 0.0027,
      "step": 16892
    },
    {
      "epoch": 2.3676243868255082,
      "grad_norm": 0.7173232436180115,
      "learning_rate": 6.48505142310452e-05,
      "loss": 0.0448,
      "step": 16893
    },
    {
      "epoch": 2.3677645409950947,
      "grad_norm": 0.06801613420248032,
      "learning_rate": 6.483616359722554e-05,
      "loss": 0.0055,
      "step": 16894
    },
    {
      "epoch": 2.367904695164681,
      "grad_norm": 0.16512177884578705,
      "learning_rate": 6.482181296340587e-05,
      "loss": 0.0344,
      "step": 16895
    },
    {
      "epoch": 2.3680448493342676,
      "grad_norm": 0.10589629411697388,
      "learning_rate": 6.480746232958623e-05,
      "loss": 0.0082,
      "step": 16896
    },
    {
      "epoch": 2.368185003503854,
      "grad_norm": 0.14586491882801056,
      "learning_rate": 6.479311169576656e-05,
      "loss": 0.0237,
      "step": 16897
    },
    {
      "epoch": 2.368325157673441,
      "grad_norm": 0.32704222202301025,
      "learning_rate": 6.47787610619469e-05,
      "loss": 0.0199,
      "step": 16898
    },
    {
      "epoch": 2.3684653118430274,
      "grad_norm": 0.5472651720046997,
      "learning_rate": 6.476441042812724e-05,
      "loss": 0.0393,
      "step": 16899
    },
    {
      "epoch": 2.368605466012614,
      "grad_norm": 0.47295185923576355,
      "learning_rate": 6.475005979430758e-05,
      "loss": 0.0436,
      "step": 16900
    },
    {
      "epoch": 2.3687456201822004,
      "grad_norm": 0.12014143913984299,
      "learning_rate": 6.473570916048791e-05,
      "loss": 0.0093,
      "step": 16901
    },
    {
      "epoch": 2.368885774351787,
      "grad_norm": 0.21537567675113678,
      "learning_rate": 6.472135852666825e-05,
      "loss": 0.0399,
      "step": 16902
    },
    {
      "epoch": 2.3690259285213733,
      "grad_norm": 0.19364473223686218,
      "learning_rate": 6.47070078928486e-05,
      "loss": 0.0223,
      "step": 16903
    },
    {
      "epoch": 2.36916608269096,
      "grad_norm": 0.11488724499940872,
      "learning_rate": 6.469265725902894e-05,
      "loss": 0.0117,
      "step": 16904
    },
    {
      "epoch": 2.3693062368605466,
      "grad_norm": 0.580854594707489,
      "learning_rate": 6.467830662520928e-05,
      "loss": 0.0442,
      "step": 16905
    },
    {
      "epoch": 2.369446391030133,
      "grad_norm": 0.08569135516881943,
      "learning_rate": 6.466395599138961e-05,
      "loss": 0.0061,
      "step": 16906
    },
    {
      "epoch": 2.3695865451997196,
      "grad_norm": 0.30035510659217834,
      "learning_rate": 6.464960535756995e-05,
      "loss": 0.0371,
      "step": 16907
    },
    {
      "epoch": 2.3697266993693065,
      "grad_norm": 0.14449165761470795,
      "learning_rate": 6.463525472375029e-05,
      "loss": 0.0671,
      "step": 16908
    },
    {
      "epoch": 2.369866853538893,
      "grad_norm": 0.2254241555929184,
      "learning_rate": 6.462090408993063e-05,
      "loss": 0.0507,
      "step": 16909
    },
    {
      "epoch": 2.3700070077084794,
      "grad_norm": 0.3636128306388855,
      "learning_rate": 6.460655345611098e-05,
      "loss": 0.0525,
      "step": 16910
    },
    {
      "epoch": 2.370147161878066,
      "grad_norm": 0.032167065888643265,
      "learning_rate": 6.45922028222913e-05,
      "loss": 0.0024,
      "step": 16911
    },
    {
      "epoch": 2.3702873160476523,
      "grad_norm": 0.24461814761161804,
      "learning_rate": 6.457785218847166e-05,
      "loss": 0.0101,
      "step": 16912
    },
    {
      "epoch": 2.3704274702172388,
      "grad_norm": 0.3655865788459778,
      "learning_rate": 6.456350155465199e-05,
      "loss": 0.0853,
      "step": 16913
    },
    {
      "epoch": 2.3705676243868257,
      "grad_norm": 0.13492873311042786,
      "learning_rate": 6.454915092083233e-05,
      "loss": 0.015,
      "step": 16914
    },
    {
      "epoch": 2.370707778556412,
      "grad_norm": 0.04546618461608887,
      "learning_rate": 6.453480028701267e-05,
      "loss": 0.0035,
      "step": 16915
    },
    {
      "epoch": 2.3708479327259986,
      "grad_norm": 1.2768014669418335,
      "learning_rate": 6.452044965319302e-05,
      "loss": 0.0374,
      "step": 16916
    },
    {
      "epoch": 2.370988086895585,
      "grad_norm": 0.06340478360652924,
      "learning_rate": 6.450609901937334e-05,
      "loss": 0.0041,
      "step": 16917
    },
    {
      "epoch": 2.3711282410651715,
      "grad_norm": 0.0500447079539299,
      "learning_rate": 6.449174838555369e-05,
      "loss": 0.0044,
      "step": 16918
    },
    {
      "epoch": 2.3712683952347584,
      "grad_norm": 0.08931964635848999,
      "learning_rate": 6.447739775173403e-05,
      "loss": 0.0043,
      "step": 16919
    },
    {
      "epoch": 2.371408549404345,
      "grad_norm": 0.6721488833427429,
      "learning_rate": 6.446304711791437e-05,
      "loss": 0.0757,
      "step": 16920
    },
    {
      "epoch": 2.3715487035739313,
      "grad_norm": 1.0046268701553345,
      "learning_rate": 6.444869648409471e-05,
      "loss": 0.0556,
      "step": 16921
    },
    {
      "epoch": 2.371688857743518,
      "grad_norm": 0.34156569838523865,
      "learning_rate": 6.443434585027504e-05,
      "loss": 0.0433,
      "step": 16922
    },
    {
      "epoch": 2.3718290119131042,
      "grad_norm": 0.37228503823280334,
      "learning_rate": 6.44199952164554e-05,
      "loss": 0.0209,
      "step": 16923
    },
    {
      "epoch": 2.371969166082691,
      "grad_norm": 0.11845259368419647,
      "learning_rate": 6.440564458263573e-05,
      "loss": 0.0109,
      "step": 16924
    },
    {
      "epoch": 2.3721093202522776,
      "grad_norm": 0.22649669647216797,
      "learning_rate": 6.439129394881607e-05,
      "loss": 0.0462,
      "step": 16925
    },
    {
      "epoch": 2.372249474421864,
      "grad_norm": 0.13704326748847961,
      "learning_rate": 6.437694331499641e-05,
      "loss": 0.0219,
      "step": 16926
    },
    {
      "epoch": 2.3723896285914505,
      "grad_norm": 0.22121179103851318,
      "learning_rate": 6.436259268117675e-05,
      "loss": 0.0283,
      "step": 16927
    },
    {
      "epoch": 2.372529782761037,
      "grad_norm": 0.034840602427721024,
      "learning_rate": 6.43482420473571e-05,
      "loss": 0.004,
      "step": 16928
    },
    {
      "epoch": 2.372669936930624,
      "grad_norm": 0.22619760036468506,
      "learning_rate": 6.433389141353742e-05,
      "loss": 0.0215,
      "step": 16929
    },
    {
      "epoch": 2.3728100911002104,
      "grad_norm": 0.05948106572031975,
      "learning_rate": 6.431954077971776e-05,
      "loss": 0.0031,
      "step": 16930
    },
    {
      "epoch": 2.372950245269797,
      "grad_norm": 0.21795515716075897,
      "learning_rate": 6.43051901458981e-05,
      "loss": 0.0292,
      "step": 16931
    },
    {
      "epoch": 2.3730903994393833,
      "grad_norm": 0.21289026737213135,
      "learning_rate": 6.429083951207845e-05,
      "loss": 0.0184,
      "step": 16932
    },
    {
      "epoch": 2.3732305536089697,
      "grad_norm": 0.5294535756111145,
      "learning_rate": 6.427648887825878e-05,
      "loss": 0.0706,
      "step": 16933
    },
    {
      "epoch": 2.373370707778556,
      "grad_norm": 0.1806413233280182,
      "learning_rate": 6.426213824443912e-05,
      "loss": 0.0233,
      "step": 16934
    },
    {
      "epoch": 2.373510861948143,
      "grad_norm": 0.522103488445282,
      "learning_rate": 6.424778761061946e-05,
      "loss": 0.0895,
      "step": 16935
    },
    {
      "epoch": 2.3736510161177296,
      "grad_norm": 0.21771401166915894,
      "learning_rate": 6.42334369767998e-05,
      "loss": 0.0243,
      "step": 16936
    },
    {
      "epoch": 2.373791170287316,
      "grad_norm": 0.4710409343242645,
      "learning_rate": 6.421908634298015e-05,
      "loss": 0.0294,
      "step": 16937
    },
    {
      "epoch": 2.3739313244569025,
      "grad_norm": 0.23137691617012024,
      "learning_rate": 6.420473570916047e-05,
      "loss": 0.0054,
      "step": 16938
    },
    {
      "epoch": 2.3740714786264894,
      "grad_norm": 0.07563072443008423,
      "learning_rate": 6.419038507534083e-05,
      "loss": 0.0073,
      "step": 16939
    },
    {
      "epoch": 2.374211632796076,
      "grad_norm": 0.3973282277584076,
      "learning_rate": 6.417603444152116e-05,
      "loss": 0.0855,
      "step": 16940
    },
    {
      "epoch": 2.3743517869656623,
      "grad_norm": 0.1551436185836792,
      "learning_rate": 6.41616838077015e-05,
      "loss": 0.0244,
      "step": 16941
    },
    {
      "epoch": 2.3744919411352488,
      "grad_norm": 0.05691460147500038,
      "learning_rate": 6.414733317388184e-05,
      "loss": 0.0038,
      "step": 16942
    },
    {
      "epoch": 2.374632095304835,
      "grad_norm": 0.09008825570344925,
      "learning_rate": 6.413298254006218e-05,
      "loss": 0.0043,
      "step": 16943
    },
    {
      "epoch": 2.3747722494744217,
      "grad_norm": 0.8951126337051392,
      "learning_rate": 6.411863190624253e-05,
      "loss": 0.0254,
      "step": 16944
    },
    {
      "epoch": 2.3749124036440086,
      "grad_norm": 0.2815404236316681,
      "learning_rate": 6.410428127242286e-05,
      "loss": 0.0304,
      "step": 16945
    },
    {
      "epoch": 2.375052557813595,
      "grad_norm": 0.06636140495538712,
      "learning_rate": 6.40899306386032e-05,
      "loss": 0.004,
      "step": 16946
    },
    {
      "epoch": 2.3751927119831815,
      "grad_norm": 0.09949080646038055,
      "learning_rate": 6.407558000478354e-05,
      "loss": 0.0138,
      "step": 16947
    },
    {
      "epoch": 2.375332866152768,
      "grad_norm": 0.12217723578214645,
      "learning_rate": 6.406122937096388e-05,
      "loss": 0.0222,
      "step": 16948
    },
    {
      "epoch": 2.3754730203223544,
      "grad_norm": 0.3615403175354004,
      "learning_rate": 6.404687873714421e-05,
      "loss": 0.0345,
      "step": 16949
    },
    {
      "epoch": 2.3756131744919413,
      "grad_norm": 0.9572797417640686,
      "learning_rate": 6.403252810332455e-05,
      "loss": 0.0984,
      "step": 16950
    },
    {
      "epoch": 2.3757533286615278,
      "grad_norm": 0.18464066088199615,
      "learning_rate": 6.40181774695049e-05,
      "loss": 0.0169,
      "step": 16951
    },
    {
      "epoch": 2.3758934828311142,
      "grad_norm": 0.45925700664520264,
      "learning_rate": 6.400382683568524e-05,
      "loss": 0.0685,
      "step": 16952
    },
    {
      "epoch": 2.3760336370007007,
      "grad_norm": 0.2794213891029358,
      "learning_rate": 6.398947620186558e-05,
      "loss": 0.0405,
      "step": 16953
    },
    {
      "epoch": 2.376173791170287,
      "grad_norm": 0.3778564929962158,
      "learning_rate": 6.397512556804591e-05,
      "loss": 0.0765,
      "step": 16954
    },
    {
      "epoch": 2.3763139453398736,
      "grad_norm": 0.20111976563930511,
      "learning_rate": 6.396077493422626e-05,
      "loss": 0.0313,
      "step": 16955
    },
    {
      "epoch": 2.3764540995094605,
      "grad_norm": 0.09110995382070541,
      "learning_rate": 6.394642430040659e-05,
      "loss": 0.0086,
      "step": 16956
    },
    {
      "epoch": 2.376594253679047,
      "grad_norm": 0.6317217350006104,
      "learning_rate": 6.393207366658693e-05,
      "loss": 0.0821,
      "step": 16957
    },
    {
      "epoch": 2.3767344078486334,
      "grad_norm": 0.19801975786685944,
      "learning_rate": 6.391772303276728e-05,
      "loss": 0.0261,
      "step": 16958
    },
    {
      "epoch": 2.37687456201822,
      "grad_norm": 0.2943643629550934,
      "learning_rate": 6.390337239894762e-05,
      "loss": 0.0488,
      "step": 16959
    },
    {
      "epoch": 2.377014716187807,
      "grad_norm": 0.7233700156211853,
      "learning_rate": 6.388902176512796e-05,
      "loss": 0.0126,
      "step": 16960
    },
    {
      "epoch": 2.3771548703573933,
      "grad_norm": 0.24063602089881897,
      "learning_rate": 6.387467113130829e-05,
      "loss": 0.0259,
      "step": 16961
    },
    {
      "epoch": 2.3772950245269797,
      "grad_norm": 0.4068490266799927,
      "learning_rate": 6.386032049748864e-05,
      "loss": 0.0168,
      "step": 16962
    },
    {
      "epoch": 2.377435178696566,
      "grad_norm": 0.6211187839508057,
      "learning_rate": 6.384596986366897e-05,
      "loss": 0.0334,
      "step": 16963
    },
    {
      "epoch": 2.3775753328661526,
      "grad_norm": 0.5214171409606934,
      "learning_rate": 6.383161922984931e-05,
      "loss": 0.0496,
      "step": 16964
    },
    {
      "epoch": 2.377715487035739,
      "grad_norm": 0.5902215838432312,
      "learning_rate": 6.381726859602964e-05,
      "loss": 0.082,
      "step": 16965
    },
    {
      "epoch": 2.377855641205326,
      "grad_norm": 0.17820778489112854,
      "learning_rate": 6.380291796221e-05,
      "loss": 0.0077,
      "step": 16966
    },
    {
      "epoch": 2.3779957953749125,
      "grad_norm": 0.2765914499759674,
      "learning_rate": 6.378856732839033e-05,
      "loss": 0.065,
      "step": 16967
    },
    {
      "epoch": 2.378135949544499,
      "grad_norm": 0.18553400039672852,
      "learning_rate": 6.377421669457067e-05,
      "loss": 0.019,
      "step": 16968
    },
    {
      "epoch": 2.3782761037140854,
      "grad_norm": 0.09079540520906448,
      "learning_rate": 6.375986606075101e-05,
      "loss": 0.0031,
      "step": 16969
    },
    {
      "epoch": 2.3784162578836723,
      "grad_norm": 1.549363613128662,
      "learning_rate": 6.374551542693135e-05,
      "loss": 0.0565,
      "step": 16970
    },
    {
      "epoch": 2.3785564120532587,
      "grad_norm": 0.2311146855354309,
      "learning_rate": 6.37311647931117e-05,
      "loss": 0.0333,
      "step": 16971
    },
    {
      "epoch": 2.378696566222845,
      "grad_norm": 0.4168628454208374,
      "learning_rate": 6.371681415929202e-05,
      "loss": 0.0536,
      "step": 16972
    },
    {
      "epoch": 2.3788367203924317,
      "grad_norm": 0.18378330767154694,
      "learning_rate": 6.370246352547237e-05,
      "loss": 0.0139,
      "step": 16973
    },
    {
      "epoch": 2.378976874562018,
      "grad_norm": 0.2749338448047638,
      "learning_rate": 6.368811289165271e-05,
      "loss": 0.0304,
      "step": 16974
    },
    {
      "epoch": 2.3791170287316046,
      "grad_norm": 0.36670419573783875,
      "learning_rate": 6.367376225783305e-05,
      "loss": 0.025,
      "step": 16975
    },
    {
      "epoch": 2.3792571829011915,
      "grad_norm": 0.2734379172325134,
      "learning_rate": 6.365941162401339e-05,
      "loss": 0.0168,
      "step": 16976
    },
    {
      "epoch": 2.379397337070778,
      "grad_norm": 0.16084024310112,
      "learning_rate": 6.364506099019372e-05,
      "loss": 0.0136,
      "step": 16977
    },
    {
      "epoch": 2.3795374912403644,
      "grad_norm": 0.11295800656080246,
      "learning_rate": 6.363071035637408e-05,
      "loss": 0.0157,
      "step": 16978
    },
    {
      "epoch": 2.379677645409951,
      "grad_norm": 0.13842664659023285,
      "learning_rate": 6.36163597225544e-05,
      "loss": 0.0221,
      "step": 16979
    },
    {
      "epoch": 2.3798177995795373,
      "grad_norm": 0.22612878680229187,
      "learning_rate": 6.360200908873475e-05,
      "loss": 0.0274,
      "step": 16980
    },
    {
      "epoch": 2.379957953749124,
      "grad_norm": 0.19449256360530853,
      "learning_rate": 6.358765845491508e-05,
      "loss": 0.0131,
      "step": 16981
    },
    {
      "epoch": 2.3800981079187107,
      "grad_norm": 0.5510298609733582,
      "learning_rate": 6.357330782109543e-05,
      "loss": 0.0995,
      "step": 16982
    },
    {
      "epoch": 2.380238262088297,
      "grad_norm": 0.11415030062198639,
      "learning_rate": 6.355895718727576e-05,
      "loss": 0.0173,
      "step": 16983
    },
    {
      "epoch": 2.3803784162578836,
      "grad_norm": 0.18728147447109222,
      "learning_rate": 6.35446065534561e-05,
      "loss": 0.0205,
      "step": 16984
    },
    {
      "epoch": 2.38051857042747,
      "grad_norm": 0.26597216725349426,
      "learning_rate": 6.353025591963644e-05,
      "loss": 0.0665,
      "step": 16985
    },
    {
      "epoch": 2.3806587245970565,
      "grad_norm": 0.30623021721839905,
      "learning_rate": 6.351590528581679e-05,
      "loss": 0.0136,
      "step": 16986
    },
    {
      "epoch": 2.3807988787666434,
      "grad_norm": 0.10489777475595474,
      "learning_rate": 6.350155465199713e-05,
      "loss": 0.0269,
      "step": 16987
    },
    {
      "epoch": 2.38093903293623,
      "grad_norm": 0.08576521277427673,
      "learning_rate": 6.348720401817746e-05,
      "loss": 0.0135,
      "step": 16988
    },
    {
      "epoch": 2.3810791871058163,
      "grad_norm": 0.13533248007297516,
      "learning_rate": 6.34728533843578e-05,
      "loss": 0.0064,
      "step": 16989
    },
    {
      "epoch": 2.381219341275403,
      "grad_norm": 0.4142076075077057,
      "learning_rate": 6.345850275053814e-05,
      "loss": 0.0404,
      "step": 16990
    },
    {
      "epoch": 2.3813594954449897,
      "grad_norm": 0.19384825229644775,
      "learning_rate": 6.344415211671848e-05,
      "loss": 0.0123,
      "step": 16991
    },
    {
      "epoch": 2.381499649614576,
      "grad_norm": 0.19663523137569427,
      "learning_rate": 6.342980148289883e-05,
      "loss": 0.0216,
      "step": 16992
    },
    {
      "epoch": 2.3816398037841626,
      "grad_norm": 0.09719817340373993,
      "learning_rate": 6.341545084907915e-05,
      "loss": 0.0086,
      "step": 16993
    },
    {
      "epoch": 2.381779957953749,
      "grad_norm": 0.12020457535982132,
      "learning_rate": 6.340110021525951e-05,
      "loss": 0.0275,
      "step": 16994
    },
    {
      "epoch": 2.3819201121233355,
      "grad_norm": 0.06197202578186989,
      "learning_rate": 6.338674958143984e-05,
      "loss": 0.0093,
      "step": 16995
    },
    {
      "epoch": 2.382060266292922,
      "grad_norm": 0.34650909900665283,
      "learning_rate": 6.337239894762018e-05,
      "loss": 0.0213,
      "step": 16996
    },
    {
      "epoch": 2.382200420462509,
      "grad_norm": 0.09400416910648346,
      "learning_rate": 6.335804831380052e-05,
      "loss": 0.0527,
      "step": 16997
    },
    {
      "epoch": 2.3823405746320954,
      "grad_norm": 0.31884637475013733,
      "learning_rate": 6.334369767998087e-05,
      "loss": 0.0204,
      "step": 16998
    },
    {
      "epoch": 2.382480728801682,
      "grad_norm": 0.6473954319953918,
      "learning_rate": 6.33293470461612e-05,
      "loss": 0.0566,
      "step": 16999
    },
    {
      "epoch": 2.3826208829712683,
      "grad_norm": 0.12085790187120438,
      "learning_rate": 6.331499641234154e-05,
      "loss": 0.0044,
      "step": 17000
    },
    {
      "epoch": 2.382761037140855,
      "grad_norm": 0.3617757260799408,
      "learning_rate": 6.330064577852188e-05,
      "loss": 0.0176,
      "step": 17001
    },
    {
      "epoch": 2.3829011913104416,
      "grad_norm": 0.2565879225730896,
      "learning_rate": 6.328629514470222e-05,
      "loss": 0.0696,
      "step": 17002
    },
    {
      "epoch": 2.383041345480028,
      "grad_norm": 0.631227970123291,
      "learning_rate": 6.327194451088256e-05,
      "loss": 0.0283,
      "step": 17003
    },
    {
      "epoch": 2.3831814996496146,
      "grad_norm": 0.08526162058115005,
      "learning_rate": 6.325759387706289e-05,
      "loss": 0.0048,
      "step": 17004
    },
    {
      "epoch": 2.383321653819201,
      "grad_norm": 0.17534476518630981,
      "learning_rate": 6.324324324324325e-05,
      "loss": 0.0226,
      "step": 17005
    },
    {
      "epoch": 2.3834618079887875,
      "grad_norm": 0.22487549483776093,
      "learning_rate": 6.322889260942358e-05,
      "loss": 0.0076,
      "step": 17006
    },
    {
      "epoch": 2.3836019621583744,
      "grad_norm": 0.3976882994174957,
      "learning_rate": 6.321454197560392e-05,
      "loss": 0.016,
      "step": 17007
    },
    {
      "epoch": 2.383742116327961,
      "grad_norm": 0.2527986764907837,
      "learning_rate": 6.320019134178426e-05,
      "loss": 0.0214,
      "step": 17008
    },
    {
      "epoch": 2.3838822704975473,
      "grad_norm": 0.5134919285774231,
      "learning_rate": 6.31858407079646e-05,
      "loss": 0.0468,
      "step": 17009
    },
    {
      "epoch": 2.3840224246671338,
      "grad_norm": 0.12431655824184418,
      "learning_rate": 6.317149007414494e-05,
      "loss": 0.0076,
      "step": 17010
    },
    {
      "epoch": 2.3841625788367202,
      "grad_norm": 0.6028428077697754,
      "learning_rate": 6.315713944032527e-05,
      "loss": 0.0682,
      "step": 17011
    },
    {
      "epoch": 2.384302733006307,
      "grad_norm": 0.18822795152664185,
      "learning_rate": 6.314278880650561e-05,
      "loss": 0.0298,
      "step": 17012
    },
    {
      "epoch": 2.3844428871758936,
      "grad_norm": 0.029102494940161705,
      "learning_rate": 6.312843817268596e-05,
      "loss": 0.0022,
      "step": 17013
    },
    {
      "epoch": 2.38458304134548,
      "grad_norm": 0.260664165019989,
      "learning_rate": 6.31140875388663e-05,
      "loss": 0.0095,
      "step": 17014
    },
    {
      "epoch": 2.3847231955150665,
      "grad_norm": 0.4358069598674774,
      "learning_rate": 6.309973690504663e-05,
      "loss": 0.0242,
      "step": 17015
    },
    {
      "epoch": 2.384863349684653,
      "grad_norm": 0.20669299364089966,
      "learning_rate": 6.308538627122697e-05,
      "loss": 0.0241,
      "step": 17016
    },
    {
      "epoch": 2.3850035038542394,
      "grad_norm": 0.2118389755487442,
      "learning_rate": 6.307103563740731e-05,
      "loss": 0.024,
      "step": 17017
    },
    {
      "epoch": 2.3851436580238263,
      "grad_norm": 0.528901994228363,
      "learning_rate": 6.305668500358765e-05,
      "loss": 0.0783,
      "step": 17018
    },
    {
      "epoch": 2.385283812193413,
      "grad_norm": 0.2793574333190918,
      "learning_rate": 6.3042334369768e-05,
      "loss": 0.0168,
      "step": 17019
    },
    {
      "epoch": 2.3854239663629992,
      "grad_norm": 0.46677568554878235,
      "learning_rate": 6.302798373594832e-05,
      "loss": 0.1444,
      "step": 17020
    },
    {
      "epoch": 2.3855641205325857,
      "grad_norm": 0.04272417724132538,
      "learning_rate": 6.301363310212868e-05,
      "loss": 0.0047,
      "step": 17021
    },
    {
      "epoch": 2.3857042747021726,
      "grad_norm": 0.28152692317962646,
      "learning_rate": 6.299928246830901e-05,
      "loss": 0.0315,
      "step": 17022
    },
    {
      "epoch": 2.385844428871759,
      "grad_norm": 0.12600915133953094,
      "learning_rate": 6.298493183448935e-05,
      "loss": 0.0116,
      "step": 17023
    },
    {
      "epoch": 2.3859845830413455,
      "grad_norm": 0.2111283391714096,
      "learning_rate": 6.297058120066969e-05,
      "loss": 0.0219,
      "step": 17024
    },
    {
      "epoch": 2.386124737210932,
      "grad_norm": 0.5127525329589844,
      "learning_rate": 6.295623056685003e-05,
      "loss": 0.0508,
      "step": 17025
    },
    {
      "epoch": 2.3862648913805184,
      "grad_norm": 0.4175175428390503,
      "learning_rate": 6.294187993303038e-05,
      "loss": 0.0161,
      "step": 17026
    },
    {
      "epoch": 2.386405045550105,
      "grad_norm": 0.27285531163215637,
      "learning_rate": 6.29275292992107e-05,
      "loss": 0.0204,
      "step": 17027
    },
    {
      "epoch": 2.386545199719692,
      "grad_norm": 0.11451488733291626,
      "learning_rate": 6.291317866539105e-05,
      "loss": 0.0375,
      "step": 17028
    },
    {
      "epoch": 2.3866853538892783,
      "grad_norm": 0.044639017432928085,
      "learning_rate": 6.289882803157139e-05,
      "loss": 0.0038,
      "step": 17029
    },
    {
      "epoch": 2.3868255080588647,
      "grad_norm": 0.06948194652795792,
      "learning_rate": 6.288447739775173e-05,
      "loss": 0.0084,
      "step": 17030
    },
    {
      "epoch": 2.386965662228451,
      "grad_norm": 0.08171617984771729,
      "learning_rate": 6.287012676393206e-05,
      "loss": 0.0066,
      "step": 17031
    },
    {
      "epoch": 2.3871058163980376,
      "grad_norm": 0.37806883454322815,
      "learning_rate": 6.285577613011242e-05,
      "loss": 0.0142,
      "step": 17032
    },
    {
      "epoch": 2.3872459705676246,
      "grad_norm": 0.224162757396698,
      "learning_rate": 6.284142549629274e-05,
      "loss": 0.0672,
      "step": 17033
    },
    {
      "epoch": 2.387386124737211,
      "grad_norm": 0.05461925268173218,
      "learning_rate": 6.282707486247309e-05,
      "loss": 0.004,
      "step": 17034
    },
    {
      "epoch": 2.3875262789067975,
      "grad_norm": 0.07444392889738083,
      "learning_rate": 6.281272422865343e-05,
      "loss": 0.0077,
      "step": 17035
    },
    {
      "epoch": 2.387666433076384,
      "grad_norm": 0.08721990138292313,
      "learning_rate": 6.279837359483377e-05,
      "loss": 0.0104,
      "step": 17036
    },
    {
      "epoch": 2.3878065872459704,
      "grad_norm": 0.18916510045528412,
      "learning_rate": 6.278402296101411e-05,
      "loss": 0.0195,
      "step": 17037
    },
    {
      "epoch": 2.3879467414155573,
      "grad_norm": 0.2810911536216736,
      "learning_rate": 6.276967232719444e-05,
      "loss": 0.0436,
      "step": 17038
    },
    {
      "epoch": 2.3880868955851438,
      "grad_norm": 0.20790895819664001,
      "learning_rate": 6.275532169337478e-05,
      "loss": 0.0267,
      "step": 17039
    },
    {
      "epoch": 2.38822704975473,
      "grad_norm": 0.1411064714193344,
      "learning_rate": 6.274097105955513e-05,
      "loss": 0.0339,
      "step": 17040
    },
    {
      "epoch": 2.3883672039243167,
      "grad_norm": 0.36535367369651794,
      "learning_rate": 6.272662042573547e-05,
      "loss": 0.0737,
      "step": 17041
    },
    {
      "epoch": 2.388507358093903,
      "grad_norm": 0.5073526501655579,
      "learning_rate": 6.271226979191581e-05,
      "loss": 0.0403,
      "step": 17042
    },
    {
      "epoch": 2.38864751226349,
      "grad_norm": 0.14355416595935822,
      "learning_rate": 6.269791915809614e-05,
      "loss": 0.015,
      "step": 17043
    },
    {
      "epoch": 2.3887876664330765,
      "grad_norm": 0.2587110102176666,
      "learning_rate": 6.268356852427648e-05,
      "loss": 0.0043,
      "step": 17044
    },
    {
      "epoch": 2.388927820602663,
      "grad_norm": 0.1335587054491043,
      "learning_rate": 6.266921789045682e-05,
      "loss": 0.0178,
      "step": 17045
    },
    {
      "epoch": 2.3890679747722494,
      "grad_norm": 0.18810881674289703,
      "learning_rate": 6.265486725663716e-05,
      "loss": 0.0176,
      "step": 17046
    },
    {
      "epoch": 2.389208128941836,
      "grad_norm": 0.09448571503162384,
      "learning_rate": 6.26405166228175e-05,
      "loss": 0.0102,
      "step": 17047
    },
    {
      "epoch": 2.3893482831114223,
      "grad_norm": 0.2839357554912567,
      "learning_rate": 6.262616598899785e-05,
      "loss": 0.0606,
      "step": 17048
    },
    {
      "epoch": 2.3894884372810092,
      "grad_norm": 0.2945462465286255,
      "learning_rate": 6.261181535517818e-05,
      "loss": 0.0486,
      "step": 17049
    },
    {
      "epoch": 2.3896285914505957,
      "grad_norm": 0.7806060910224915,
      "learning_rate": 6.259746472135852e-05,
      "loss": 0.0193,
      "step": 17050
    },
    {
      "epoch": 2.389768745620182,
      "grad_norm": 0.15069469809532166,
      "learning_rate": 6.258311408753886e-05,
      "loss": 0.0589,
      "step": 17051
    },
    {
      "epoch": 2.3899088997897686,
      "grad_norm": 0.37087157368659973,
      "learning_rate": 6.25687634537192e-05,
      "loss": 0.0124,
      "step": 17052
    },
    {
      "epoch": 2.3900490539593555,
      "grad_norm": 0.5365495681762695,
      "learning_rate": 6.255441281989955e-05,
      "loss": 0.0604,
      "step": 17053
    },
    {
      "epoch": 2.390189208128942,
      "grad_norm": 0.5000490546226501,
      "learning_rate": 6.254006218607987e-05,
      "loss": 0.0914,
      "step": 17054
    },
    {
      "epoch": 2.3903293622985284,
      "grad_norm": 0.43623706698417664,
      "learning_rate": 6.252571155226022e-05,
      "loss": 0.0302,
      "step": 17055
    },
    {
      "epoch": 2.390469516468115,
      "grad_norm": 0.13209204375743866,
      "learning_rate": 6.251136091844056e-05,
      "loss": 0.0212,
      "step": 17056
    },
    {
      "epoch": 2.3906096706377014,
      "grad_norm": 0.6652680039405823,
      "learning_rate": 6.24970102846209e-05,
      "loss": 0.1109,
      "step": 17057
    },
    {
      "epoch": 2.390749824807288,
      "grad_norm": 0.5584889650344849,
      "learning_rate": 6.248265965080124e-05,
      "loss": 0.0574,
      "step": 17058
    },
    {
      "epoch": 2.3908899789768747,
      "grad_norm": 0.271617591381073,
      "learning_rate": 6.246830901698157e-05,
      "loss": 0.0114,
      "step": 17059
    },
    {
      "epoch": 2.391030133146461,
      "grad_norm": 0.459026962518692,
      "learning_rate": 6.245395838316191e-05,
      "loss": 0.0261,
      "step": 17060
    },
    {
      "epoch": 2.3911702873160476,
      "grad_norm": 0.3595461845397949,
      "learning_rate": 6.243960774934226e-05,
      "loss": 0.0291,
      "step": 17061
    },
    {
      "epoch": 2.391310441485634,
      "grad_norm": 0.1368018090724945,
      "learning_rate": 6.24252571155226e-05,
      "loss": 0.0233,
      "step": 17062
    },
    {
      "epoch": 2.3914505956552206,
      "grad_norm": 0.17785823345184326,
      "learning_rate": 6.241090648170293e-05,
      "loss": 0.0122,
      "step": 17063
    },
    {
      "epoch": 2.3915907498248075,
      "grad_norm": 0.34287145733833313,
      "learning_rate": 6.239655584788328e-05,
      "loss": 0.0073,
      "step": 17064
    },
    {
      "epoch": 2.391730903994394,
      "grad_norm": 0.23111079633235931,
      "learning_rate": 6.238220521406361e-05,
      "loss": 0.0647,
      "step": 17065
    },
    {
      "epoch": 2.3918710581639804,
      "grad_norm": 0.14815792441368103,
      "learning_rate": 6.236785458024395e-05,
      "loss": 0.0059,
      "step": 17066
    },
    {
      "epoch": 2.392011212333567,
      "grad_norm": 0.7067782282829285,
      "learning_rate": 6.23535039464243e-05,
      "loss": 0.0763,
      "step": 17067
    },
    {
      "epoch": 2.3921513665031533,
      "grad_norm": 0.8463408350944519,
      "learning_rate": 6.233915331260464e-05,
      "loss": 0.2086,
      "step": 17068
    },
    {
      "epoch": 2.39229152067274,
      "grad_norm": 0.7380649447441101,
      "learning_rate": 6.232480267878498e-05,
      "loss": 0.1277,
      "step": 17069
    },
    {
      "epoch": 2.3924316748423267,
      "grad_norm": 0.692986786365509,
      "learning_rate": 6.231045204496531e-05,
      "loss": 0.0661,
      "step": 17070
    },
    {
      "epoch": 2.392571829011913,
      "grad_norm": 0.20089690387248993,
      "learning_rate": 6.229610141114566e-05,
      "loss": 0.0286,
      "step": 17071
    },
    {
      "epoch": 2.3927119831814996,
      "grad_norm": 0.1232534870505333,
      "learning_rate": 6.228175077732599e-05,
      "loss": 0.0205,
      "step": 17072
    },
    {
      "epoch": 2.392852137351086,
      "grad_norm": 0.2921312153339386,
      "learning_rate": 6.226740014350633e-05,
      "loss": 0.0275,
      "step": 17073
    },
    {
      "epoch": 2.392992291520673,
      "grad_norm": 0.2044769823551178,
      "learning_rate": 6.225304950968668e-05,
      "loss": 0.0397,
      "step": 17074
    },
    {
      "epoch": 2.3931324456902594,
      "grad_norm": 0.10084021091461182,
      "learning_rate": 6.223869887586702e-05,
      "loss": 0.0198,
      "step": 17075
    },
    {
      "epoch": 2.393272599859846,
      "grad_norm": 0.13797718286514282,
      "learning_rate": 6.222434824204735e-05,
      "loss": 0.0097,
      "step": 17076
    },
    {
      "epoch": 2.3934127540294323,
      "grad_norm": 0.49426141381263733,
      "learning_rate": 6.220999760822769e-05,
      "loss": 0.0251,
      "step": 17077
    },
    {
      "epoch": 2.393552908199019,
      "grad_norm": 0.14347510039806366,
      "learning_rate": 6.219564697440803e-05,
      "loss": 0.026,
      "step": 17078
    },
    {
      "epoch": 2.3936930623686052,
      "grad_norm": 0.24426138401031494,
      "learning_rate": 6.218129634058837e-05,
      "loss": 0.0398,
      "step": 17079
    },
    {
      "epoch": 2.393833216538192,
      "grad_norm": 0.09398007392883301,
      "learning_rate": 6.216694570676872e-05,
      "loss": 0.0113,
      "step": 17080
    },
    {
      "epoch": 2.3939733707077786,
      "grad_norm": 0.15549708902835846,
      "learning_rate": 6.215259507294904e-05,
      "loss": 0.0227,
      "step": 17081
    },
    {
      "epoch": 2.394113524877365,
      "grad_norm": 0.0769689753651619,
      "learning_rate": 6.213824443912939e-05,
      "loss": 0.006,
      "step": 17082
    },
    {
      "epoch": 2.3942536790469515,
      "grad_norm": 0.1584383100271225,
      "learning_rate": 6.212389380530973e-05,
      "loss": 0.0071,
      "step": 17083
    },
    {
      "epoch": 2.3943938332165384,
      "grad_norm": 0.049454204738140106,
      "learning_rate": 6.210954317149007e-05,
      "loss": 0.0029,
      "step": 17084
    },
    {
      "epoch": 2.394533987386125,
      "grad_norm": 0.21866178512573242,
      "learning_rate": 6.209519253767041e-05,
      "loss": 0.0178,
      "step": 17085
    },
    {
      "epoch": 2.3946741415557113,
      "grad_norm": 0.3654237985610962,
      "learning_rate": 6.208084190385074e-05,
      "loss": 0.0177,
      "step": 17086
    },
    {
      "epoch": 2.394814295725298,
      "grad_norm": 0.10090696811676025,
      "learning_rate": 6.20664912700311e-05,
      "loss": 0.0057,
      "step": 17087
    },
    {
      "epoch": 2.3949544498948843,
      "grad_norm": 0.1193164810538292,
      "learning_rate": 6.205214063621143e-05,
      "loss": 0.0117,
      "step": 17088
    },
    {
      "epoch": 2.3950946040644707,
      "grad_norm": 0.17001782357692719,
      "learning_rate": 6.203779000239177e-05,
      "loss": 0.0477,
      "step": 17089
    },
    {
      "epoch": 2.3952347582340576,
      "grad_norm": 0.25766634941101074,
      "learning_rate": 6.202343936857211e-05,
      "loss": 0.036,
      "step": 17090
    },
    {
      "epoch": 2.395374912403644,
      "grad_norm": 0.3608410954475403,
      "learning_rate": 6.200908873475245e-05,
      "loss": 0.042,
      "step": 17091
    },
    {
      "epoch": 2.3955150665732305,
      "grad_norm": 0.2884196639060974,
      "learning_rate": 6.199473810093278e-05,
      "loss": 0.0171,
      "step": 17092
    },
    {
      "epoch": 2.395655220742817,
      "grad_norm": 0.11337452381849289,
      "learning_rate": 6.198038746711312e-05,
      "loss": 0.0069,
      "step": 17093
    },
    {
      "epoch": 2.3957953749124035,
      "grad_norm": 0.06342663615942001,
      "learning_rate": 6.196603683329346e-05,
      "loss": 0.0106,
      "step": 17094
    },
    {
      "epoch": 2.3959355290819904,
      "grad_norm": 0.5680875778198242,
      "learning_rate": 6.19516861994738e-05,
      "loss": 0.0159,
      "step": 17095
    },
    {
      "epoch": 2.396075683251577,
      "grad_norm": 0.2034035176038742,
      "learning_rate": 6.193733556565415e-05,
      "loss": 0.0563,
      "step": 17096
    },
    {
      "epoch": 2.3962158374211633,
      "grad_norm": 0.18437014520168304,
      "learning_rate": 6.192298493183448e-05,
      "loss": 0.0262,
      "step": 17097
    },
    {
      "epoch": 2.3963559915907497,
      "grad_norm": 0.22780919075012207,
      "learning_rate": 6.190863429801482e-05,
      "loss": 0.0471,
      "step": 17098
    },
    {
      "epoch": 2.396496145760336,
      "grad_norm": 0.04214702546596527,
      "learning_rate": 6.189428366419516e-05,
      "loss": 0.0049,
      "step": 17099
    },
    {
      "epoch": 2.396636299929923,
      "grad_norm": 0.21062171459197998,
      "learning_rate": 6.18799330303755e-05,
      "loss": 0.0278,
      "step": 17100
    },
    {
      "epoch": 2.3967764540995096,
      "grad_norm": 0.2688031494617462,
      "learning_rate": 6.186558239655585e-05,
      "loss": 0.0177,
      "step": 17101
    },
    {
      "epoch": 2.396916608269096,
      "grad_norm": 0.3999437093734741,
      "learning_rate": 6.185123176273617e-05,
      "loss": 0.0302,
      "step": 17102
    },
    {
      "epoch": 2.3970567624386825,
      "grad_norm": 0.4475868046283722,
      "learning_rate": 6.183688112891653e-05,
      "loss": 0.0964,
      "step": 17103
    },
    {
      "epoch": 2.397196916608269,
      "grad_norm": 0.11797259747982025,
      "learning_rate": 6.182253049509686e-05,
      "loss": 0.0162,
      "step": 17104
    },
    {
      "epoch": 2.397337070777856,
      "grad_norm": 0.3219282031059265,
      "learning_rate": 6.18081798612772e-05,
      "loss": 0.0204,
      "step": 17105
    },
    {
      "epoch": 2.3974772249474423,
      "grad_norm": 0.22857385873794556,
      "learning_rate": 6.179382922745754e-05,
      "loss": 0.0279,
      "step": 17106
    },
    {
      "epoch": 2.3976173791170288,
      "grad_norm": 0.30190396308898926,
      "learning_rate": 6.177947859363788e-05,
      "loss": 0.0264,
      "step": 17107
    },
    {
      "epoch": 2.3977575332866152,
      "grad_norm": 0.02618985064327717,
      "learning_rate": 6.176512795981821e-05,
      "loss": 0.0026,
      "step": 17108
    },
    {
      "epoch": 2.3978976874562017,
      "grad_norm": 0.17172715067863464,
      "learning_rate": 6.175077732599856e-05,
      "loss": 0.0201,
      "step": 17109
    },
    {
      "epoch": 2.398037841625788,
      "grad_norm": 0.09613469243049622,
      "learning_rate": 6.17364266921789e-05,
      "loss": 0.0074,
      "step": 17110
    },
    {
      "epoch": 2.398177995795375,
      "grad_norm": 0.32952970266342163,
      "learning_rate": 6.172207605835924e-05,
      "loss": 0.1115,
      "step": 17111
    },
    {
      "epoch": 2.3983181499649615,
      "grad_norm": 0.013354174792766571,
      "learning_rate": 6.170772542453958e-05,
      "loss": 0.0011,
      "step": 17112
    },
    {
      "epoch": 2.398458304134548,
      "grad_norm": 0.5007480382919312,
      "learning_rate": 6.169337479071991e-05,
      "loss": 0.099,
      "step": 17113
    },
    {
      "epoch": 2.3985984583041344,
      "grad_norm": 0.1851712465286255,
      "learning_rate": 6.167902415690027e-05,
      "loss": 0.0179,
      "step": 17114
    },
    {
      "epoch": 2.3987386124737213,
      "grad_norm": 0.47097906470298767,
      "learning_rate": 6.16646735230806e-05,
      "loss": 0.0175,
      "step": 17115
    },
    {
      "epoch": 2.398878766643308,
      "grad_norm": 0.1594603806734085,
      "learning_rate": 6.165032288926094e-05,
      "loss": 0.0088,
      "step": 17116
    },
    {
      "epoch": 2.3990189208128943,
      "grad_norm": 0.8486140370368958,
      "learning_rate": 6.163597225544128e-05,
      "loss": 0.073,
      "step": 17117
    },
    {
      "epoch": 2.3991590749824807,
      "grad_norm": 0.13665708899497986,
      "learning_rate": 6.162162162162162e-05,
      "loss": 0.0079,
      "step": 17118
    },
    {
      "epoch": 2.399299229152067,
      "grad_norm": 0.17007233202457428,
      "learning_rate": 6.160727098780196e-05,
      "loss": 0.0122,
      "step": 17119
    },
    {
      "epoch": 2.3994393833216536,
      "grad_norm": 2.2276971340179443,
      "learning_rate": 6.159292035398229e-05,
      "loss": 0.1992,
      "step": 17120
    },
    {
      "epoch": 2.3995795374912405,
      "grad_norm": 0.050837937742471695,
      "learning_rate": 6.157856972016263e-05,
      "loss": 0.0053,
      "step": 17121
    },
    {
      "epoch": 2.399719691660827,
      "grad_norm": 0.096001535654068,
      "learning_rate": 6.156421908634298e-05,
      "loss": 0.0138,
      "step": 17122
    },
    {
      "epoch": 2.3998598458304135,
      "grad_norm": 0.14186835289001465,
      "learning_rate": 6.154986845252332e-05,
      "loss": 0.0163,
      "step": 17123
    },
    {
      "epoch": 2.4,
      "grad_norm": 0.10023430734872818,
      "learning_rate": 6.153551781870365e-05,
      "loss": 0.0274,
      "step": 17124
    },
    {
      "epoch": 2.4001401541695864,
      "grad_norm": 0.10408713668584824,
      "learning_rate": 6.152116718488399e-05,
      "loss": 0.013,
      "step": 17125
    },
    {
      "epoch": 2.4002803083391733,
      "grad_norm": 0.24739912152290344,
      "learning_rate": 6.150681655106433e-05,
      "loss": 0.043,
      "step": 17126
    },
    {
      "epoch": 2.4004204625087597,
      "grad_norm": 0.07923022657632828,
      "learning_rate": 6.149246591724467e-05,
      "loss": 0.0049,
      "step": 17127
    },
    {
      "epoch": 2.400560616678346,
      "grad_norm": 0.17069591581821442,
      "learning_rate": 6.147811528342501e-05,
      "loss": 0.0119,
      "step": 17128
    },
    {
      "epoch": 2.4007007708479327,
      "grad_norm": 0.25277915596961975,
      "learning_rate": 6.146376464960534e-05,
      "loss": 0.0388,
      "step": 17129
    },
    {
      "epoch": 2.400840925017519,
      "grad_norm": 0.35724565386772156,
      "learning_rate": 6.14494140157857e-05,
      "loss": 0.0458,
      "step": 17130
    },
    {
      "epoch": 2.4009810791871056,
      "grad_norm": 0.16841921210289001,
      "learning_rate": 6.143506338196603e-05,
      "loss": 0.0178,
      "step": 17131
    },
    {
      "epoch": 2.4011212333566925,
      "grad_norm": 0.16430306434631348,
      "learning_rate": 6.142071274814637e-05,
      "loss": 0.0101,
      "step": 17132
    },
    {
      "epoch": 2.401261387526279,
      "grad_norm": 0.18938927352428436,
      "learning_rate": 6.140636211432671e-05,
      "loss": 0.0185,
      "step": 17133
    },
    {
      "epoch": 2.4014015416958654,
      "grad_norm": 0.11738921701908112,
      "learning_rate": 6.139201148050705e-05,
      "loss": 0.009,
      "step": 17134
    },
    {
      "epoch": 2.401541695865452,
      "grad_norm": 0.740653932094574,
      "learning_rate": 6.13776608466874e-05,
      "loss": 0.0246,
      "step": 17135
    },
    {
      "epoch": 2.4016818500350388,
      "grad_norm": 0.157445028424263,
      "learning_rate": 6.136331021286772e-05,
      "loss": 0.0305,
      "step": 17136
    },
    {
      "epoch": 2.401822004204625,
      "grad_norm": 0.12229357659816742,
      "learning_rate": 6.134895957904807e-05,
      "loss": 0.0263,
      "step": 17137
    },
    {
      "epoch": 2.4019621583742117,
      "grad_norm": 0.18288549780845642,
      "learning_rate": 6.133460894522841e-05,
      "loss": 0.0161,
      "step": 17138
    },
    {
      "epoch": 2.402102312543798,
      "grad_norm": 0.20269018411636353,
      "learning_rate": 6.132025831140875e-05,
      "loss": 0.0162,
      "step": 17139
    },
    {
      "epoch": 2.4022424667133846,
      "grad_norm": 0.4243325889110565,
      "learning_rate": 6.130590767758909e-05,
      "loss": 0.0396,
      "step": 17140
    },
    {
      "epoch": 2.402382620882971,
      "grad_norm": 0.398271381855011,
      "learning_rate": 6.129155704376942e-05,
      "loss": 0.0585,
      "step": 17141
    },
    {
      "epoch": 2.402522775052558,
      "grad_norm": 0.22428776323795319,
      "learning_rate": 6.127720640994976e-05,
      "loss": 0.0098,
      "step": 17142
    },
    {
      "epoch": 2.4026629292221444,
      "grad_norm": 0.3369782865047455,
      "learning_rate": 6.12628557761301e-05,
      "loss": 0.0252,
      "step": 17143
    },
    {
      "epoch": 2.402803083391731,
      "grad_norm": 0.1796584576368332,
      "learning_rate": 6.124850514231045e-05,
      "loss": 0.0107,
      "step": 17144
    },
    {
      "epoch": 2.4029432375613173,
      "grad_norm": 0.08359088748693466,
      "learning_rate": 6.123415450849079e-05,
      "loss": 0.007,
      "step": 17145
    },
    {
      "epoch": 2.4030833917309042,
      "grad_norm": 0.16979293525218964,
      "learning_rate": 6.121980387467113e-05,
      "loss": 0.0312,
      "step": 17146
    },
    {
      "epoch": 2.4032235459004907,
      "grad_norm": 0.3132004737854004,
      "learning_rate": 6.120545324085146e-05,
      "loss": 0.027,
      "step": 17147
    },
    {
      "epoch": 2.403363700070077,
      "grad_norm": 0.12117159366607666,
      "learning_rate": 6.11911026070318e-05,
      "loss": 0.0291,
      "step": 17148
    },
    {
      "epoch": 2.4035038542396636,
      "grad_norm": 0.24035337567329407,
      "learning_rate": 6.117675197321214e-05,
      "loss": 0.0236,
      "step": 17149
    },
    {
      "epoch": 2.40364400840925,
      "grad_norm": 0.13443993031978607,
      "learning_rate": 6.116240133939249e-05,
      "loss": 0.0125,
      "step": 17150
    },
    {
      "epoch": 2.4037841625788365,
      "grad_norm": 0.19733117520809174,
      "learning_rate": 6.114805070557283e-05,
      "loss": 0.0283,
      "step": 17151
    },
    {
      "epoch": 2.4039243167484234,
      "grad_norm": 0.1524735391139984,
      "learning_rate": 6.113370007175316e-05,
      "loss": 0.0095,
      "step": 17152
    },
    {
      "epoch": 2.40406447091801,
      "grad_norm": 0.05169086903333664,
      "learning_rate": 6.111934943793351e-05,
      "loss": 0.0028,
      "step": 17153
    },
    {
      "epoch": 2.4042046250875964,
      "grad_norm": 0.28921663761138916,
      "learning_rate": 6.110499880411384e-05,
      "loss": 0.0354,
      "step": 17154
    },
    {
      "epoch": 2.404344779257183,
      "grad_norm": 0.08542416989803314,
      "learning_rate": 6.109064817029418e-05,
      "loss": 0.0074,
      "step": 17155
    },
    {
      "epoch": 2.4044849334267693,
      "grad_norm": 0.2869613766670227,
      "learning_rate": 6.107629753647453e-05,
      "loss": 0.0511,
      "step": 17156
    },
    {
      "epoch": 2.404625087596356,
      "grad_norm": 0.23059694468975067,
      "learning_rate": 6.106194690265487e-05,
      "loss": 0.0321,
      "step": 17157
    },
    {
      "epoch": 2.4047652417659426,
      "grad_norm": 0.15354962646961212,
      "learning_rate": 6.10475962688352e-05,
      "loss": 0.0233,
      "step": 17158
    },
    {
      "epoch": 2.404905395935529,
      "grad_norm": 0.1864965856075287,
      "learning_rate": 6.103324563501554e-05,
      "loss": 0.0346,
      "step": 17159
    },
    {
      "epoch": 2.4050455501051156,
      "grad_norm": 0.3018342852592468,
      "learning_rate": 6.101889500119588e-05,
      "loss": 0.012,
      "step": 17160
    },
    {
      "epoch": 2.405185704274702,
      "grad_norm": 0.1295814961194992,
      "learning_rate": 6.100454436737622e-05,
      "loss": 0.0081,
      "step": 17161
    },
    {
      "epoch": 2.4053258584442885,
      "grad_norm": 0.947309672832489,
      "learning_rate": 6.099019373355656e-05,
      "loss": 0.1207,
      "step": 17162
    },
    {
      "epoch": 2.4054660126138754,
      "grad_norm": 0.6175588369369507,
      "learning_rate": 6.09758430997369e-05,
      "loss": 0.0639,
      "step": 17163
    },
    {
      "epoch": 2.405606166783462,
      "grad_norm": 0.21148596704006195,
      "learning_rate": 6.0961492465917236e-05,
      "loss": 0.0142,
      "step": 17164
    },
    {
      "epoch": 2.4057463209530483,
      "grad_norm": 0.03760712593793869,
      "learning_rate": 6.0947141832097585e-05,
      "loss": 0.0019,
      "step": 17165
    },
    {
      "epoch": 2.4058864751226348,
      "grad_norm": 0.044599004089832306,
      "learning_rate": 6.093279119827792e-05,
      "loss": 0.0018,
      "step": 17166
    },
    {
      "epoch": 2.4060266292922217,
      "grad_norm": 0.15554621815681458,
      "learning_rate": 6.0918440564458255e-05,
      "loss": 0.0057,
      "step": 17167
    },
    {
      "epoch": 2.406166783461808,
      "grad_norm": 0.46567302942276,
      "learning_rate": 6.09040899306386e-05,
      "loss": 0.0275,
      "step": 17168
    },
    {
      "epoch": 2.4063069376313946,
      "grad_norm": 0.03271007910370827,
      "learning_rate": 6.088973929681894e-05,
      "loss": 0.0028,
      "step": 17169
    },
    {
      "epoch": 2.406447091800981,
      "grad_norm": 0.197012260556221,
      "learning_rate": 6.0875388662999275e-05,
      "loss": 0.0103,
      "step": 17170
    },
    {
      "epoch": 2.4065872459705675,
      "grad_norm": 0.25466659665107727,
      "learning_rate": 6.086103802917962e-05,
      "loss": 0.0852,
      "step": 17171
    },
    {
      "epoch": 2.406727400140154,
      "grad_norm": 0.13236021995544434,
      "learning_rate": 6.084668739535995e-05,
      "loss": 0.0108,
      "step": 17172
    },
    {
      "epoch": 2.406867554309741,
      "grad_norm": 0.19801558554172516,
      "learning_rate": 6.08323367615403e-05,
      "loss": 0.0964,
      "step": 17173
    },
    {
      "epoch": 2.4070077084793273,
      "grad_norm": 0.3901863098144531,
      "learning_rate": 6.081798612772064e-05,
      "loss": 0.0081,
      "step": 17174
    },
    {
      "epoch": 2.407147862648914,
      "grad_norm": 0.2413594275712967,
      "learning_rate": 6.080363549390097e-05,
      "loss": 0.0493,
      "step": 17175
    },
    {
      "epoch": 2.4072880168185002,
      "grad_norm": 0.1466943770647049,
      "learning_rate": 6.0789284860081314e-05,
      "loss": 0.0201,
      "step": 17176
    },
    {
      "epoch": 2.4074281709880867,
      "grad_norm": 0.16011151671409607,
      "learning_rate": 6.0774934226261656e-05,
      "loss": 0.0365,
      "step": 17177
    },
    {
      "epoch": 2.4075683251576736,
      "grad_norm": 0.18179325759410858,
      "learning_rate": 6.076058359244199e-05,
      "loss": 0.0489,
      "step": 17178
    },
    {
      "epoch": 2.40770847932726,
      "grad_norm": 0.2514638900756836,
      "learning_rate": 6.0746232958622334e-05,
      "loss": 0.0337,
      "step": 17179
    },
    {
      "epoch": 2.4078486334968465,
      "grad_norm": 0.2746260166168213,
      "learning_rate": 6.0731882324802676e-05,
      "loss": 0.0207,
      "step": 17180
    },
    {
      "epoch": 2.407988787666433,
      "grad_norm": 0.18683600425720215,
      "learning_rate": 6.071753169098302e-05,
      "loss": 0.0135,
      "step": 17181
    },
    {
      "epoch": 2.4081289418360194,
      "grad_norm": 0.3731100857257843,
      "learning_rate": 6.070318105716335e-05,
      "loss": 0.0747,
      "step": 17182
    },
    {
      "epoch": 2.4082690960056063,
      "grad_norm": 0.07254581153392792,
      "learning_rate": 6.068883042334369e-05,
      "loss": 0.004,
      "step": 17183
    },
    {
      "epoch": 2.408409250175193,
      "grad_norm": 0.1650509238243103,
      "learning_rate": 6.067447978952404e-05,
      "loss": 0.0296,
      "step": 17184
    },
    {
      "epoch": 2.4085494043447793,
      "grad_norm": 0.20255090296268463,
      "learning_rate": 6.066012915570437e-05,
      "loss": 0.0188,
      "step": 17185
    },
    {
      "epoch": 2.4086895585143657,
      "grad_norm": 0.3189687430858612,
      "learning_rate": 6.064577852188471e-05,
      "loss": 0.0855,
      "step": 17186
    },
    {
      "epoch": 2.408829712683952,
      "grad_norm": 0.12048765271902084,
      "learning_rate": 6.063142788806505e-05,
      "loss": 0.0155,
      "step": 17187
    },
    {
      "epoch": 2.408969866853539,
      "grad_norm": 0.3445298671722412,
      "learning_rate": 6.061707725424539e-05,
      "loss": 0.0524,
      "step": 17188
    },
    {
      "epoch": 2.4091100210231255,
      "grad_norm": 0.22000162303447723,
      "learning_rate": 6.0602726620425734e-05,
      "loss": 0.0196,
      "step": 17189
    },
    {
      "epoch": 2.409250175192712,
      "grad_norm": 0.22827807068824768,
      "learning_rate": 6.058837598660607e-05,
      "loss": 0.0283,
      "step": 17190
    },
    {
      "epoch": 2.4093903293622985,
      "grad_norm": 0.277342289686203,
      "learning_rate": 6.0574025352786405e-05,
      "loss": 0.0076,
      "step": 17191
    },
    {
      "epoch": 2.409530483531885,
      "grad_norm": 0.2629663348197937,
      "learning_rate": 6.0559674718966754e-05,
      "loss": 0.0227,
      "step": 17192
    },
    {
      "epoch": 2.4096706377014714,
      "grad_norm": 0.47268515825271606,
      "learning_rate": 6.054532408514709e-05,
      "loss": 0.0381,
      "step": 17193
    },
    {
      "epoch": 2.4098107918710583,
      "grad_norm": 0.14151810109615326,
      "learning_rate": 6.0530973451327425e-05,
      "loss": 0.0242,
      "step": 17194
    },
    {
      "epoch": 2.4099509460406447,
      "grad_norm": 0.12931035459041595,
      "learning_rate": 6.051662281750777e-05,
      "loss": 0.017,
      "step": 17195
    },
    {
      "epoch": 2.410091100210231,
      "grad_norm": 0.18927617371082306,
      "learning_rate": 6.050227218368811e-05,
      "loss": 0.024,
      "step": 17196
    },
    {
      "epoch": 2.4102312543798177,
      "grad_norm": 0.3010213077068329,
      "learning_rate": 6.048792154986845e-05,
      "loss": 0.0333,
      "step": 17197
    },
    {
      "epoch": 2.4103714085494046,
      "grad_norm": 0.17706774175167084,
      "learning_rate": 6.0473570916048786e-05,
      "loss": 0.061,
      "step": 17198
    },
    {
      "epoch": 2.410511562718991,
      "grad_norm": 0.17203454673290253,
      "learning_rate": 6.045922028222912e-05,
      "loss": 0.0108,
      "step": 17199
    },
    {
      "epoch": 2.4106517168885775,
      "grad_norm": 0.19356462359428406,
      "learning_rate": 6.044486964840947e-05,
      "loss": 0.0446,
      "step": 17200
    },
    {
      "epoch": 2.410791871058164,
      "grad_norm": 0.8297691345214844,
      "learning_rate": 6.0430519014589806e-05,
      "loss": 0.0288,
      "step": 17201
    },
    {
      "epoch": 2.4109320252277504,
      "grad_norm": 0.3473604917526245,
      "learning_rate": 6.041616838077014e-05,
      "loss": 0.0261,
      "step": 17202
    },
    {
      "epoch": 2.411072179397337,
      "grad_norm": 0.14792002737522125,
      "learning_rate": 6.040181774695048e-05,
      "loss": 0.0149,
      "step": 17203
    },
    {
      "epoch": 2.4112123335669238,
      "grad_norm": 0.31362733244895935,
      "learning_rate": 6.0387467113130825e-05,
      "loss": 0.0364,
      "step": 17204
    },
    {
      "epoch": 2.4113524877365102,
      "grad_norm": 0.07948626577854156,
      "learning_rate": 6.037311647931117e-05,
      "loss": 0.0072,
      "step": 17205
    },
    {
      "epoch": 2.4114926419060967,
      "grad_norm": 0.21225471794605255,
      "learning_rate": 6.03587658454915e-05,
      "loss": 0.0658,
      "step": 17206
    },
    {
      "epoch": 2.411632796075683,
      "grad_norm": 0.14901341497898102,
      "learning_rate": 6.034441521167184e-05,
      "loss": 0.03,
      "step": 17207
    },
    {
      "epoch": 2.4117729502452696,
      "grad_norm": 0.20639602839946747,
      "learning_rate": 6.033006457785219e-05,
      "loss": 0.0253,
      "step": 17208
    },
    {
      "epoch": 2.4119131044148565,
      "grad_norm": 0.17323389649391174,
      "learning_rate": 6.031571394403252e-05,
      "loss": 0.0405,
      "step": 17209
    },
    {
      "epoch": 2.412053258584443,
      "grad_norm": 0.16269272565841675,
      "learning_rate": 6.030136331021286e-05,
      "loss": 0.0109,
      "step": 17210
    },
    {
      "epoch": 2.4121934127540294,
      "grad_norm": 0.3181898295879364,
      "learning_rate": 6.02870126763932e-05,
      "loss": 0.023,
      "step": 17211
    },
    {
      "epoch": 2.412333566923616,
      "grad_norm": 0.4775742292404175,
      "learning_rate": 6.027266204257354e-05,
      "loss": 0.1531,
      "step": 17212
    },
    {
      "epoch": 2.4124737210932024,
      "grad_norm": 0.21781279146671295,
      "learning_rate": 6.0258311408753884e-05,
      "loss": 0.0239,
      "step": 17213
    },
    {
      "epoch": 2.4126138752627893,
      "grad_norm": 0.09848164767026901,
      "learning_rate": 6.024396077493422e-05,
      "loss": 0.0075,
      "step": 17214
    },
    {
      "epoch": 2.4127540294323757,
      "grad_norm": 0.6712706685066223,
      "learning_rate": 6.0229610141114555e-05,
      "loss": 0.1075,
      "step": 17215
    },
    {
      "epoch": 2.412894183601962,
      "grad_norm": 1.1856974363327026,
      "learning_rate": 6.0215259507294904e-05,
      "loss": 0.0955,
      "step": 17216
    },
    {
      "epoch": 2.4130343377715486,
      "grad_norm": 0.43746069073677063,
      "learning_rate": 6.020090887347524e-05,
      "loss": 0.0472,
      "step": 17217
    },
    {
      "epoch": 2.413174491941135,
      "grad_norm": 0.17862793803215027,
      "learning_rate": 6.0186558239655574e-05,
      "loss": 0.0096,
      "step": 17218
    },
    {
      "epoch": 2.413314646110722,
      "grad_norm": 2.1139705181121826,
      "learning_rate": 6.017220760583592e-05,
      "loss": 0.2104,
      "step": 17219
    },
    {
      "epoch": 2.4134548002803085,
      "grad_norm": 1.1476143598556519,
      "learning_rate": 6.015785697201626e-05,
      "loss": 0.1617,
      "step": 17220
    },
    {
      "epoch": 2.413594954449895,
      "grad_norm": 0.20820845663547516,
      "learning_rate": 6.01435063381966e-05,
      "loss": 0.0289,
      "step": 17221
    },
    {
      "epoch": 2.4137351086194814,
      "grad_norm": 0.27490726113319397,
      "learning_rate": 6.0129155704376936e-05,
      "loss": 0.0213,
      "step": 17222
    },
    {
      "epoch": 2.413875262789068,
      "grad_norm": 0.13856208324432373,
      "learning_rate": 6.011480507055728e-05,
      "loss": 0.0087,
      "step": 17223
    },
    {
      "epoch": 2.4140154169586543,
      "grad_norm": 0.24667789041996002,
      "learning_rate": 6.010045443673762e-05,
      "loss": 0.03,
      "step": 17224
    },
    {
      "epoch": 2.414155571128241,
      "grad_norm": 0.09652167558670044,
      "learning_rate": 6.0086103802917956e-05,
      "loss": 0.0119,
      "step": 17225
    },
    {
      "epoch": 2.4142957252978277,
      "grad_norm": 0.2773926556110382,
      "learning_rate": 6.007175316909829e-05,
      "loss": 0.0278,
      "step": 17226
    },
    {
      "epoch": 2.414435879467414,
      "grad_norm": 0.25053325295448303,
      "learning_rate": 6.005740253527864e-05,
      "loss": 0.0232,
      "step": 17227
    },
    {
      "epoch": 2.4145760336370006,
      "grad_norm": 0.09703636169433594,
      "learning_rate": 6.0043051901458975e-05,
      "loss": 0.0085,
      "step": 17228
    },
    {
      "epoch": 2.4147161878065875,
      "grad_norm": 0.12566767632961273,
      "learning_rate": 6.002870126763932e-05,
      "loss": 0.0367,
      "step": 17229
    },
    {
      "epoch": 2.414856341976174,
      "grad_norm": 0.35741427540779114,
      "learning_rate": 6.001435063381965e-05,
      "loss": 0.0277,
      "step": 17230
    },
    {
      "epoch": 2.4149964961457604,
      "grad_norm": 0.29923439025878906,
      "learning_rate": 5.9999999999999995e-05,
      "loss": 0.0533,
      "step": 17231
    },
    {
      "epoch": 2.415136650315347,
      "grad_norm": 0.3132534325122833,
      "learning_rate": 5.998564936618034e-05,
      "loss": 0.0444,
      "step": 17232
    },
    {
      "epoch": 2.4152768044849333,
      "grad_norm": 0.17771215736865997,
      "learning_rate": 5.997129873236067e-05,
      "loss": 0.0247,
      "step": 17233
    },
    {
      "epoch": 2.4154169586545198,
      "grad_norm": 0.2619757056236267,
      "learning_rate": 5.995694809854101e-05,
      "loss": 0.049,
      "step": 17234
    },
    {
      "epoch": 2.4155571128241067,
      "grad_norm": 0.3328060209751129,
      "learning_rate": 5.9942597464721356e-05,
      "loss": 0.0341,
      "step": 17235
    },
    {
      "epoch": 2.415697266993693,
      "grad_norm": 0.21630752086639404,
      "learning_rate": 5.992824683090169e-05,
      "loss": 0.0182,
      "step": 17236
    },
    {
      "epoch": 2.4158374211632796,
      "grad_norm": 0.219991534948349,
      "learning_rate": 5.9913896197082034e-05,
      "loss": 0.0286,
      "step": 17237
    },
    {
      "epoch": 2.415977575332866,
      "grad_norm": 0.647718608379364,
      "learning_rate": 5.989954556326237e-05,
      "loss": 0.0338,
      "step": 17238
    },
    {
      "epoch": 2.4161177295024525,
      "grad_norm": 0.11938625574111938,
      "learning_rate": 5.988519492944271e-05,
      "loss": 0.0125,
      "step": 17239
    },
    {
      "epoch": 2.4162578836720394,
      "grad_norm": 0.048991985619068146,
      "learning_rate": 5.9870844295623053e-05,
      "loss": 0.004,
      "step": 17240
    },
    {
      "epoch": 2.416398037841626,
      "grad_norm": 0.24102351069450378,
      "learning_rate": 5.985649366180339e-05,
      "loss": 0.0184,
      "step": 17241
    },
    {
      "epoch": 2.4165381920112123,
      "grad_norm": 0.06479820609092712,
      "learning_rate": 5.9842143027983724e-05,
      "loss": 0.0092,
      "step": 17242
    },
    {
      "epoch": 2.416678346180799,
      "grad_norm": 0.18250589072704315,
      "learning_rate": 5.982779239416407e-05,
      "loss": 0.0329,
      "step": 17243
    },
    {
      "epoch": 2.4168185003503853,
      "grad_norm": 0.042573247104883194,
      "learning_rate": 5.981344176034441e-05,
      "loss": 0.0019,
      "step": 17244
    },
    {
      "epoch": 2.416958654519972,
      "grad_norm": 0.14258728921413422,
      "learning_rate": 5.979909112652475e-05,
      "loss": 0.0151,
      "step": 17245
    },
    {
      "epoch": 2.4170988086895586,
      "grad_norm": 0.08728601038455963,
      "learning_rate": 5.9784740492705086e-05,
      "loss": 0.0036,
      "step": 17246
    },
    {
      "epoch": 2.417238962859145,
      "grad_norm": 0.496239572763443,
      "learning_rate": 5.9770389858885435e-05,
      "loss": 0.0282,
      "step": 17247
    },
    {
      "epoch": 2.4173791170287315,
      "grad_norm": 0.19870057702064514,
      "learning_rate": 5.975603922506577e-05,
      "loss": 0.025,
      "step": 17248
    },
    {
      "epoch": 2.417519271198318,
      "grad_norm": 0.16261160373687744,
      "learning_rate": 5.9741688591246105e-05,
      "loss": 0.0277,
      "step": 17249
    },
    {
      "epoch": 2.417659425367905,
      "grad_norm": 0.3400171101093292,
      "learning_rate": 5.972733795742644e-05,
      "loss": 0.0126,
      "step": 17250
    },
    {
      "epoch": 2.4177995795374914,
      "grad_norm": 0.06124268099665642,
      "learning_rate": 5.971298732360679e-05,
      "loss": 0.0075,
      "step": 17251
    },
    {
      "epoch": 2.417939733707078,
      "grad_norm": 0.09326623380184174,
      "learning_rate": 5.9698636689787125e-05,
      "loss": 0.0115,
      "step": 17252
    },
    {
      "epoch": 2.4180798878766643,
      "grad_norm": 0.1605793833732605,
      "learning_rate": 5.968428605596747e-05,
      "loss": 0.0358,
      "step": 17253
    },
    {
      "epoch": 2.4182200420462507,
      "grad_norm": 0.15432000160217285,
      "learning_rate": 5.966993542214781e-05,
      "loss": 0.007,
      "step": 17254
    },
    {
      "epoch": 2.418360196215837,
      "grad_norm": 0.1933581382036209,
      "learning_rate": 5.965558478832815e-05,
      "loss": 0.016,
      "step": 17255
    },
    {
      "epoch": 2.418500350385424,
      "grad_norm": 0.3474451005458832,
      "learning_rate": 5.9641234154508487e-05,
      "loss": 0.0534,
      "step": 17256
    },
    {
      "epoch": 2.4186405045550106,
      "grad_norm": 0.2471638321876526,
      "learning_rate": 5.962688352068882e-05,
      "loss": 0.0809,
      "step": 17257
    },
    {
      "epoch": 2.418780658724597,
      "grad_norm": 0.18646566569805145,
      "learning_rate": 5.961253288686917e-05,
      "loss": 0.0078,
      "step": 17258
    },
    {
      "epoch": 2.4189208128941835,
      "grad_norm": 0.09521641582250595,
      "learning_rate": 5.9598182253049506e-05,
      "loss": 0.0076,
      "step": 17259
    },
    {
      "epoch": 2.4190609670637704,
      "grad_norm": 0.19135162234306335,
      "learning_rate": 5.958383161922984e-05,
      "loss": 0.0565,
      "step": 17260
    },
    {
      "epoch": 2.419201121233357,
      "grad_norm": 0.15246811509132385,
      "learning_rate": 5.9569480985410184e-05,
      "loss": 0.0129,
      "step": 17261
    },
    {
      "epoch": 2.4193412754029433,
      "grad_norm": 0.038744255900382996,
      "learning_rate": 5.9555130351590526e-05,
      "loss": 0.0035,
      "step": 17262
    },
    {
      "epoch": 2.4194814295725298,
      "grad_norm": 0.09676238894462585,
      "learning_rate": 5.954077971777087e-05,
      "loss": 0.0068,
      "step": 17263
    },
    {
      "epoch": 2.419621583742116,
      "grad_norm": 0.3693442940711975,
      "learning_rate": 5.95264290839512e-05,
      "loss": 0.0159,
      "step": 17264
    },
    {
      "epoch": 2.4197617379117027,
      "grad_norm": 0.20712009072303772,
      "learning_rate": 5.951207845013154e-05,
      "loss": 0.0171,
      "step": 17265
    },
    {
      "epoch": 2.4199018920812896,
      "grad_norm": 0.1251690536737442,
      "learning_rate": 5.949772781631189e-05,
      "loss": 0.0039,
      "step": 17266
    },
    {
      "epoch": 2.420042046250876,
      "grad_norm": 0.5674152970314026,
      "learning_rate": 5.948337718249222e-05,
      "loss": 0.0474,
      "step": 17267
    },
    {
      "epoch": 2.4201822004204625,
      "grad_norm": 0.5214142203330994,
      "learning_rate": 5.946902654867256e-05,
      "loss": 0.0746,
      "step": 17268
    },
    {
      "epoch": 2.420322354590049,
      "grad_norm": 0.6095026135444641,
      "learning_rate": 5.94546759148529e-05,
      "loss": 0.0226,
      "step": 17269
    },
    {
      "epoch": 2.4204625087596354,
      "grad_norm": 2.0127460956573486,
      "learning_rate": 5.944032528103324e-05,
      "loss": 0.1201,
      "step": 17270
    },
    {
      "epoch": 2.4206026629292223,
      "grad_norm": 0.1310524046421051,
      "learning_rate": 5.9425974647213584e-05,
      "loss": 0.0238,
      "step": 17271
    },
    {
      "epoch": 2.420742817098809,
      "grad_norm": 0.0937386304140091,
      "learning_rate": 5.941162401339392e-05,
      "loss": 0.0119,
      "step": 17272
    },
    {
      "epoch": 2.4208829712683952,
      "grad_norm": 0.21025891602039337,
      "learning_rate": 5.9397273379574255e-05,
      "loss": 0.0178,
      "step": 17273
    },
    {
      "epoch": 2.4210231254379817,
      "grad_norm": 0.5389658808708191,
      "learning_rate": 5.9382922745754604e-05,
      "loss": 0.0327,
      "step": 17274
    },
    {
      "epoch": 2.421163279607568,
      "grad_norm": 0.14517338573932648,
      "learning_rate": 5.936857211193494e-05,
      "loss": 0.0289,
      "step": 17275
    },
    {
      "epoch": 2.4213034337771546,
      "grad_norm": 0.1003274917602539,
      "learning_rate": 5.9354221478115275e-05,
      "loss": 0.0076,
      "step": 17276
    },
    {
      "epoch": 2.4214435879467415,
      "grad_norm": 0.267656147480011,
      "learning_rate": 5.933987084429562e-05,
      "loss": 0.0478,
      "step": 17277
    },
    {
      "epoch": 2.421583742116328,
      "grad_norm": 0.13005879521369934,
      "learning_rate": 5.932552021047596e-05,
      "loss": 0.0119,
      "step": 17278
    },
    {
      "epoch": 2.4217238962859144,
      "grad_norm": 0.1691494882106781,
      "learning_rate": 5.93111695766563e-05,
      "loss": 0.0215,
      "step": 17279
    },
    {
      "epoch": 2.421864050455501,
      "grad_norm": 0.2108124941587448,
      "learning_rate": 5.9296818942836636e-05,
      "loss": 0.0204,
      "step": 17280
    },
    {
      "epoch": 2.422004204625088,
      "grad_norm": 0.24599264562129974,
      "learning_rate": 5.928246830901697e-05,
      "loss": 0.0239,
      "step": 17281
    },
    {
      "epoch": 2.4221443587946743,
      "grad_norm": 0.1283554881811142,
      "learning_rate": 5.926811767519732e-05,
      "loss": 0.0059,
      "step": 17282
    },
    {
      "epoch": 2.4222845129642607,
      "grad_norm": 0.1940176784992218,
      "learning_rate": 5.9253767041377656e-05,
      "loss": 0.0213,
      "step": 17283
    },
    {
      "epoch": 2.422424667133847,
      "grad_norm": 0.07534559816122055,
      "learning_rate": 5.923941640755799e-05,
      "loss": 0.0093,
      "step": 17284
    },
    {
      "epoch": 2.4225648213034336,
      "grad_norm": 0.35346484184265137,
      "learning_rate": 5.922506577373833e-05,
      "loss": 0.0114,
      "step": 17285
    },
    {
      "epoch": 2.42270497547302,
      "grad_norm": 0.2003057897090912,
      "learning_rate": 5.9210715139918675e-05,
      "loss": 0.0156,
      "step": 17286
    },
    {
      "epoch": 2.422845129642607,
      "grad_norm": 0.21672599017620087,
      "learning_rate": 5.919636450609902e-05,
      "loss": 0.0144,
      "step": 17287
    },
    {
      "epoch": 2.4229852838121935,
      "grad_norm": 0.14388997852802277,
      "learning_rate": 5.918201387227935e-05,
      "loss": 0.0403,
      "step": 17288
    },
    {
      "epoch": 2.42312543798178,
      "grad_norm": 0.2987062335014343,
      "learning_rate": 5.916766323845969e-05,
      "loss": 0.0106,
      "step": 17289
    },
    {
      "epoch": 2.4232655921513664,
      "grad_norm": 0.37020429968833923,
      "learning_rate": 5.915331260464004e-05,
      "loss": 0.0394,
      "step": 17290
    },
    {
      "epoch": 2.4234057463209533,
      "grad_norm": 0.27939000725746155,
      "learning_rate": 5.913896197082037e-05,
      "loss": 0.0354,
      "step": 17291
    },
    {
      "epoch": 2.4235459004905398,
      "grad_norm": 0.3336494565010071,
      "learning_rate": 5.912461133700071e-05,
      "loss": 0.0293,
      "step": 17292
    },
    {
      "epoch": 2.423686054660126,
      "grad_norm": 0.3211798071861267,
      "learning_rate": 5.911026070318106e-05,
      "loss": 0.0222,
      "step": 17293
    },
    {
      "epoch": 2.4238262088297127,
      "grad_norm": 0.16890409588813782,
      "learning_rate": 5.909591006936139e-05,
      "loss": 0.0112,
      "step": 17294
    },
    {
      "epoch": 2.423966362999299,
      "grad_norm": 0.09941990673542023,
      "learning_rate": 5.9081559435541734e-05,
      "loss": 0.018,
      "step": 17295
    },
    {
      "epoch": 2.4241065171688856,
      "grad_norm": 0.26585209369659424,
      "learning_rate": 5.906720880172207e-05,
      "loss": 0.01,
      "step": 17296
    },
    {
      "epoch": 2.4242466713384725,
      "grad_norm": 0.28566795587539673,
      "learning_rate": 5.905285816790241e-05,
      "loss": 0.0231,
      "step": 17297
    },
    {
      "epoch": 2.424386825508059,
      "grad_norm": 0.16020673513412476,
      "learning_rate": 5.9038507534082754e-05,
      "loss": 0.0088,
      "step": 17298
    },
    {
      "epoch": 2.4245269796776454,
      "grad_norm": 0.20671851933002472,
      "learning_rate": 5.902415690026309e-05,
      "loss": 0.0462,
      "step": 17299
    },
    {
      "epoch": 2.424667133847232,
      "grad_norm": 0.24831578135490417,
      "learning_rate": 5.9009806266443424e-05,
      "loss": 0.0132,
      "step": 17300
    },
    {
      "epoch": 2.4248072880168183,
      "grad_norm": 0.11230577528476715,
      "learning_rate": 5.899545563262377e-05,
      "loss": 0.0159,
      "step": 17301
    },
    {
      "epoch": 2.4249474421864052,
      "grad_norm": 0.40549013018608093,
      "learning_rate": 5.898110499880411e-05,
      "loss": 0.0288,
      "step": 17302
    },
    {
      "epoch": 2.4250875963559917,
      "grad_norm": 0.21482624113559723,
      "learning_rate": 5.896675436498445e-05,
      "loss": 0.0411,
      "step": 17303
    },
    {
      "epoch": 2.425227750525578,
      "grad_norm": 0.7297971248626709,
      "learning_rate": 5.8952403731164786e-05,
      "loss": 0.041,
      "step": 17304
    },
    {
      "epoch": 2.4253679046951646,
      "grad_norm": 0.20108667016029358,
      "learning_rate": 5.893805309734513e-05,
      "loss": 0.0224,
      "step": 17305
    },
    {
      "epoch": 2.425508058864751,
      "grad_norm": 0.18769244849681854,
      "learning_rate": 5.892370246352547e-05,
      "loss": 0.0152,
      "step": 17306
    },
    {
      "epoch": 2.4256482130343375,
      "grad_norm": 0.3743934631347656,
      "learning_rate": 5.8909351829705806e-05,
      "loss": 0.0271,
      "step": 17307
    },
    {
      "epoch": 2.4257883672039244,
      "grad_norm": 0.06821400672197342,
      "learning_rate": 5.889500119588614e-05,
      "loss": 0.0068,
      "step": 17308
    },
    {
      "epoch": 2.425928521373511,
      "grad_norm": 0.1807606965303421,
      "learning_rate": 5.888065056206649e-05,
      "loss": 0.0239,
      "step": 17309
    },
    {
      "epoch": 2.4260686755430974,
      "grad_norm": 0.2810732126235962,
      "learning_rate": 5.8866299928246825e-05,
      "loss": 0.0242,
      "step": 17310
    },
    {
      "epoch": 2.426208829712684,
      "grad_norm": 0.326339989900589,
      "learning_rate": 5.885194929442717e-05,
      "loss": 0.0131,
      "step": 17311
    },
    {
      "epoch": 2.4263489838822707,
      "grad_norm": 0.3636963367462158,
      "learning_rate": 5.88375986606075e-05,
      "loss": 0.0584,
      "step": 17312
    },
    {
      "epoch": 2.426489138051857,
      "grad_norm": 0.2233114242553711,
      "learning_rate": 5.8823248026787845e-05,
      "loss": 0.0705,
      "step": 17313
    },
    {
      "epoch": 2.4266292922214436,
      "grad_norm": 0.05255798250436783,
      "learning_rate": 5.880889739296819e-05,
      "loss": 0.0047,
      "step": 17314
    },
    {
      "epoch": 2.42676944639103,
      "grad_norm": 0.307024210691452,
      "learning_rate": 5.879454675914852e-05,
      "loss": 0.0427,
      "step": 17315
    },
    {
      "epoch": 2.4269096005606166,
      "grad_norm": 0.539598286151886,
      "learning_rate": 5.878019612532886e-05,
      "loss": 0.0325,
      "step": 17316
    },
    {
      "epoch": 2.427049754730203,
      "grad_norm": 0.13242220878601074,
      "learning_rate": 5.8765845491509206e-05,
      "loss": 0.0048,
      "step": 17317
    },
    {
      "epoch": 2.42718990889979,
      "grad_norm": 0.012009680271148682,
      "learning_rate": 5.875149485768954e-05,
      "loss": 0.0013,
      "step": 17318
    },
    {
      "epoch": 2.4273300630693764,
      "grad_norm": 0.25054246187210083,
      "learning_rate": 5.8737144223869884e-05,
      "loss": 0.0126,
      "step": 17319
    },
    {
      "epoch": 2.427470217238963,
      "grad_norm": 0.5420270562171936,
      "learning_rate": 5.872279359005022e-05,
      "loss": 0.0058,
      "step": 17320
    },
    {
      "epoch": 2.4276103714085493,
      "grad_norm": 0.31170839071273804,
      "learning_rate": 5.870844295623056e-05,
      "loss": 0.0161,
      "step": 17321
    },
    {
      "epoch": 2.427750525578136,
      "grad_norm": 0.280725359916687,
      "learning_rate": 5.86940923224109e-05,
      "loss": 0.0295,
      "step": 17322
    },
    {
      "epoch": 2.4278906797477227,
      "grad_norm": 0.22767262160778046,
      "learning_rate": 5.867974168859124e-05,
      "loss": 0.0279,
      "step": 17323
    },
    {
      "epoch": 2.428030833917309,
      "grad_norm": 0.08386383205652237,
      "learning_rate": 5.8665391054771574e-05,
      "loss": 0.0053,
      "step": 17324
    },
    {
      "epoch": 2.4281709880868956,
      "grad_norm": 0.27017468214035034,
      "learning_rate": 5.865104042095192e-05,
      "loss": 0.0277,
      "step": 17325
    },
    {
      "epoch": 2.428311142256482,
      "grad_norm": 0.16565978527069092,
      "learning_rate": 5.863668978713226e-05,
      "loss": 0.0219,
      "step": 17326
    },
    {
      "epoch": 2.4284512964260685,
      "grad_norm": 0.319987028837204,
      "learning_rate": 5.86223391533126e-05,
      "loss": 0.0342,
      "step": 17327
    },
    {
      "epoch": 2.4285914505956554,
      "grad_norm": 0.11172568798065186,
      "learning_rate": 5.860798851949294e-05,
      "loss": 0.0219,
      "step": 17328
    },
    {
      "epoch": 2.428731604765242,
      "grad_norm": 0.1598750203847885,
      "learning_rate": 5.859363788567328e-05,
      "loss": 0.0101,
      "step": 17329
    },
    {
      "epoch": 2.4288717589348283,
      "grad_norm": 0.28091269731521606,
      "learning_rate": 5.857928725185362e-05,
      "loss": 0.0572,
      "step": 17330
    },
    {
      "epoch": 2.4290119131044148,
      "grad_norm": 0.3136027753353119,
      "learning_rate": 5.8564936618033955e-05,
      "loss": 0.0133,
      "step": 17331
    },
    {
      "epoch": 2.4291520672740012,
      "grad_norm": 0.4446505308151245,
      "learning_rate": 5.8550585984214304e-05,
      "loss": 0.0511,
      "step": 17332
    },
    {
      "epoch": 2.429292221443588,
      "grad_norm": 0.09604140371084213,
      "learning_rate": 5.853623535039464e-05,
      "loss": 0.0047,
      "step": 17333
    },
    {
      "epoch": 2.4294323756131746,
      "grad_norm": 0.22032104432582855,
      "learning_rate": 5.8521884716574975e-05,
      "loss": 0.0268,
      "step": 17334
    },
    {
      "epoch": 2.429572529782761,
      "grad_norm": 0.2517751455307007,
      "learning_rate": 5.850753408275532e-05,
      "loss": 0.0415,
      "step": 17335
    },
    {
      "epoch": 2.4297126839523475,
      "grad_norm": 0.58295738697052,
      "learning_rate": 5.849318344893566e-05,
      "loss": 0.0272,
      "step": 17336
    },
    {
      "epoch": 2.429852838121934,
      "grad_norm": 0.1690925508737564,
      "learning_rate": 5.8478832815115994e-05,
      "loss": 0.0376,
      "step": 17337
    },
    {
      "epoch": 2.4299929922915204,
      "grad_norm": 0.230002298951149,
      "learning_rate": 5.8464482181296337e-05,
      "loss": 0.0109,
      "step": 17338
    },
    {
      "epoch": 2.4301331464611073,
      "grad_norm": 0.20594878494739532,
      "learning_rate": 5.845013154747667e-05,
      "loss": 0.0234,
      "step": 17339
    },
    {
      "epoch": 2.430273300630694,
      "grad_norm": 0.13403496146202087,
      "learning_rate": 5.843578091365702e-05,
      "loss": 0.0467,
      "step": 17340
    },
    {
      "epoch": 2.4304134548002803,
      "grad_norm": 0.27012327313423157,
      "learning_rate": 5.8421430279837356e-05,
      "loss": 0.0699,
      "step": 17341
    },
    {
      "epoch": 2.4305536089698667,
      "grad_norm": 0.32172754406929016,
      "learning_rate": 5.840707964601769e-05,
      "loss": 0.0428,
      "step": 17342
    },
    {
      "epoch": 2.4306937631394536,
      "grad_norm": 0.2663648724555969,
      "learning_rate": 5.8392729012198034e-05,
      "loss": 0.0393,
      "step": 17343
    },
    {
      "epoch": 2.43083391730904,
      "grad_norm": 0.3988592028617859,
      "learning_rate": 5.8378378378378376e-05,
      "loss": 0.0315,
      "step": 17344
    },
    {
      "epoch": 2.4309740714786265,
      "grad_norm": 0.34296631813049316,
      "learning_rate": 5.836402774455871e-05,
      "loss": 0.0445,
      "step": 17345
    },
    {
      "epoch": 2.431114225648213,
      "grad_norm": 0.21390260756015778,
      "learning_rate": 5.834967711073905e-05,
      "loss": 0.031,
      "step": 17346
    },
    {
      "epoch": 2.4312543798177995,
      "grad_norm": 0.09705667197704315,
      "learning_rate": 5.833532647691939e-05,
      "loss": 0.0035,
      "step": 17347
    },
    {
      "epoch": 2.431394533987386,
      "grad_norm": 0.2565990388393402,
      "learning_rate": 5.832097584309974e-05,
      "loss": 0.066,
      "step": 17348
    },
    {
      "epoch": 2.431534688156973,
      "grad_norm": 0.4618794918060303,
      "learning_rate": 5.830662520928007e-05,
      "loss": 0.0331,
      "step": 17349
    },
    {
      "epoch": 2.4316748423265593,
      "grad_norm": 0.4538531005382538,
      "learning_rate": 5.829227457546041e-05,
      "loss": 0.0417,
      "step": 17350
    },
    {
      "epoch": 2.4318149964961457,
      "grad_norm": 0.22675305604934692,
      "learning_rate": 5.827792394164075e-05,
      "loss": 0.0175,
      "step": 17351
    },
    {
      "epoch": 2.431955150665732,
      "grad_norm": 0.23802068829536438,
      "learning_rate": 5.826357330782109e-05,
      "loss": 0.0353,
      "step": 17352
    },
    {
      "epoch": 2.4320953048353187,
      "grad_norm": 0.2600937783718109,
      "learning_rate": 5.824922267400143e-05,
      "loss": 0.0213,
      "step": 17353
    },
    {
      "epoch": 2.4322354590049056,
      "grad_norm": 0.11470136046409607,
      "learning_rate": 5.823487204018177e-05,
      "loss": 0.0062,
      "step": 17354
    },
    {
      "epoch": 2.432375613174492,
      "grad_norm": 0.25565436482429504,
      "learning_rate": 5.8220521406362105e-05,
      "loss": 0.0394,
      "step": 17355
    },
    {
      "epoch": 2.4325157673440785,
      "grad_norm": 0.11170201748609543,
      "learning_rate": 5.8206170772542454e-05,
      "loss": 0.0094,
      "step": 17356
    },
    {
      "epoch": 2.432655921513665,
      "grad_norm": 0.4334985315799713,
      "learning_rate": 5.819182013872279e-05,
      "loss": 0.0428,
      "step": 17357
    },
    {
      "epoch": 2.4327960756832514,
      "grad_norm": 0.13811059296131134,
      "learning_rate": 5.8177469504903125e-05,
      "loss": 0.0345,
      "step": 17358
    },
    {
      "epoch": 2.4329362298528383,
      "grad_norm": 0.4813917577266693,
      "learning_rate": 5.816311887108347e-05,
      "loss": 0.0228,
      "step": 17359
    },
    {
      "epoch": 2.4330763840224248,
      "grad_norm": 0.3816911280155182,
      "learning_rate": 5.814876823726381e-05,
      "loss": 0.0362,
      "step": 17360
    },
    {
      "epoch": 2.4332165381920112,
      "grad_norm": 0.399700403213501,
      "learning_rate": 5.8134417603444144e-05,
      "loss": 0.0483,
      "step": 17361
    },
    {
      "epoch": 2.4333566923615977,
      "grad_norm": 0.0950688049197197,
      "learning_rate": 5.8120066969624486e-05,
      "loss": 0.0156,
      "step": 17362
    },
    {
      "epoch": 2.433496846531184,
      "grad_norm": 0.7060579657554626,
      "learning_rate": 5.810571633580482e-05,
      "loss": 0.0337,
      "step": 17363
    },
    {
      "epoch": 2.433637000700771,
      "grad_norm": 0.18039365112781525,
      "learning_rate": 5.809136570198517e-05,
      "loss": 0.0162,
      "step": 17364
    },
    {
      "epoch": 2.4337771548703575,
      "grad_norm": 0.18923288583755493,
      "learning_rate": 5.8077015068165506e-05,
      "loss": 0.0043,
      "step": 17365
    },
    {
      "epoch": 2.433917309039944,
      "grad_norm": 0.23057551681995392,
      "learning_rate": 5.806266443434584e-05,
      "loss": 0.0093,
      "step": 17366
    },
    {
      "epoch": 2.4340574632095304,
      "grad_norm": 0.0744875818490982,
      "learning_rate": 5.804831380052619e-05,
      "loss": 0.0024,
      "step": 17367
    },
    {
      "epoch": 2.434197617379117,
      "grad_norm": 0.4068446159362793,
      "learning_rate": 5.8033963166706525e-05,
      "loss": 0.0457,
      "step": 17368
    },
    {
      "epoch": 2.4343377715487033,
      "grad_norm": 2.500253915786743,
      "learning_rate": 5.801961253288686e-05,
      "loss": 0.036,
      "step": 17369
    },
    {
      "epoch": 2.4344779257182902,
      "grad_norm": 0.30666816234588623,
      "learning_rate": 5.80052618990672e-05,
      "loss": 0.0257,
      "step": 17370
    },
    {
      "epoch": 2.4346180798878767,
      "grad_norm": 0.4612908959388733,
      "learning_rate": 5.7990911265247545e-05,
      "loss": 0.0832,
      "step": 17371
    },
    {
      "epoch": 2.434758234057463,
      "grad_norm": 0.2421182096004486,
      "learning_rate": 5.797656063142789e-05,
      "loss": 0.0618,
      "step": 17372
    },
    {
      "epoch": 2.4348983882270496,
      "grad_norm": 0.2944526672363281,
      "learning_rate": 5.796220999760822e-05,
      "loss": 0.0074,
      "step": 17373
    },
    {
      "epoch": 2.4350385423966365,
      "grad_norm": 0.16928522288799286,
      "learning_rate": 5.794785936378856e-05,
      "loss": 0.0171,
      "step": 17374
    },
    {
      "epoch": 2.435178696566223,
      "grad_norm": 0.4732314944267273,
      "learning_rate": 5.7933508729968907e-05,
      "loss": 0.051,
      "step": 17375
    },
    {
      "epoch": 2.4353188507358094,
      "grad_norm": 0.07006821781396866,
      "learning_rate": 5.791915809614924e-05,
      "loss": 0.0042,
      "step": 17376
    },
    {
      "epoch": 2.435459004905396,
      "grad_norm": 0.11372796446084976,
      "learning_rate": 5.790480746232958e-05,
      "loss": 0.0113,
      "step": 17377
    },
    {
      "epoch": 2.4355991590749824,
      "grad_norm": 0.13307808339595795,
      "learning_rate": 5.789045682850992e-05,
      "loss": 0.036,
      "step": 17378
    },
    {
      "epoch": 2.435739313244569,
      "grad_norm": 0.17327964305877686,
      "learning_rate": 5.787610619469026e-05,
      "loss": 0.0437,
      "step": 17379
    },
    {
      "epoch": 2.4358794674141557,
      "grad_norm": 0.1326773762702942,
      "learning_rate": 5.7861755560870604e-05,
      "loss": 0.0149,
      "step": 17380
    },
    {
      "epoch": 2.436019621583742,
      "grad_norm": 0.1691352277994156,
      "learning_rate": 5.784740492705094e-05,
      "loss": 0.0106,
      "step": 17381
    },
    {
      "epoch": 2.4361597757533286,
      "grad_norm": 0.09650550782680511,
      "learning_rate": 5.7833054293231274e-05,
      "loss": 0.0189,
      "step": 17382
    },
    {
      "epoch": 2.436299929922915,
      "grad_norm": 0.33244314789772034,
      "learning_rate": 5.781870365941162e-05,
      "loss": 0.0315,
      "step": 17383
    },
    {
      "epoch": 2.4364400840925016,
      "grad_norm": 0.4329344630241394,
      "learning_rate": 5.780435302559196e-05,
      "loss": 0.0661,
      "step": 17384
    },
    {
      "epoch": 2.4365802382620885,
      "grad_norm": 0.2881946861743927,
      "learning_rate": 5.7790002391772294e-05,
      "loss": 0.0722,
      "step": 17385
    },
    {
      "epoch": 2.436720392431675,
      "grad_norm": 0.14996686577796936,
      "learning_rate": 5.7775651757952636e-05,
      "loss": 0.0505,
      "step": 17386
    },
    {
      "epoch": 2.4368605466012614,
      "grad_norm": 0.24359512329101562,
      "learning_rate": 5.776130112413298e-05,
      "loss": 0.0178,
      "step": 17387
    },
    {
      "epoch": 2.437000700770848,
      "grad_norm": 0.017672166228294373,
      "learning_rate": 5.774695049031332e-05,
      "loss": 0.0014,
      "step": 17388
    },
    {
      "epoch": 2.4371408549404343,
      "grad_norm": 0.05534207820892334,
      "learning_rate": 5.7732599856493656e-05,
      "loss": 0.0035,
      "step": 17389
    },
    {
      "epoch": 2.437281009110021,
      "grad_norm": 0.18137279152870178,
      "learning_rate": 5.771824922267399e-05,
      "loss": 0.0245,
      "step": 17390
    },
    {
      "epoch": 2.4374211632796077,
      "grad_norm": 0.19302916526794434,
      "learning_rate": 5.770389858885434e-05,
      "loss": 0.0181,
      "step": 17391
    },
    {
      "epoch": 2.437561317449194,
      "grad_norm": 0.3653002977371216,
      "learning_rate": 5.7689547955034675e-05,
      "loss": 0.0407,
      "step": 17392
    },
    {
      "epoch": 2.4377014716187806,
      "grad_norm": 0.15861906111240387,
      "learning_rate": 5.767519732121501e-05,
      "loss": 0.0138,
      "step": 17393
    },
    {
      "epoch": 2.437841625788367,
      "grad_norm": 0.16883985698223114,
      "learning_rate": 5.766084668739535e-05,
      "loss": 0.0417,
      "step": 17394
    },
    {
      "epoch": 2.437981779957954,
      "grad_norm": 0.16244134306907654,
      "learning_rate": 5.7646496053575695e-05,
      "loss": 0.0162,
      "step": 17395
    },
    {
      "epoch": 2.4381219341275404,
      "grad_norm": 0.1126076802611351,
      "learning_rate": 5.763214541975604e-05,
      "loss": 0.0048,
      "step": 17396
    },
    {
      "epoch": 2.438262088297127,
      "grad_norm": 0.1689811497926712,
      "learning_rate": 5.761779478593637e-05,
      "loss": 0.0408,
      "step": 17397
    },
    {
      "epoch": 2.4384022424667133,
      "grad_norm": 0.13655903935432434,
      "learning_rate": 5.760344415211671e-05,
      "loss": 0.0364,
      "step": 17398
    },
    {
      "epoch": 2.4385423966363,
      "grad_norm": 0.10317081958055496,
      "learning_rate": 5.7589093518297056e-05,
      "loss": 0.0214,
      "step": 17399
    },
    {
      "epoch": 2.4386825508058863,
      "grad_norm": 0.3053319752216339,
      "learning_rate": 5.757474288447739e-05,
      "loss": 0.0277,
      "step": 17400
    },
    {
      "epoch": 2.438822704975473,
      "grad_norm": 0.12531666457653046,
      "learning_rate": 5.756039225065773e-05,
      "loss": 0.0126,
      "step": 17401
    },
    {
      "epoch": 2.4389628591450596,
      "grad_norm": 0.3033042252063751,
      "learning_rate": 5.7546041616838076e-05,
      "loss": 0.0106,
      "step": 17402
    },
    {
      "epoch": 2.439103013314646,
      "grad_norm": 0.42690351605415344,
      "learning_rate": 5.753169098301841e-05,
      "loss": 0.0338,
      "step": 17403
    },
    {
      "epoch": 2.4392431674842325,
      "grad_norm": 0.2444438338279724,
      "learning_rate": 5.751734034919875e-05,
      "loss": 0.0169,
      "step": 17404
    },
    {
      "epoch": 2.4393833216538194,
      "grad_norm": 0.16978374123573303,
      "learning_rate": 5.750298971537909e-05,
      "loss": 0.0177,
      "step": 17405
    },
    {
      "epoch": 2.439523475823406,
      "grad_norm": 0.29568228125572205,
      "learning_rate": 5.748863908155944e-05,
      "loss": 0.0532,
      "step": 17406
    },
    {
      "epoch": 2.4396636299929924,
      "grad_norm": 0.24909085035324097,
      "learning_rate": 5.747428844773977e-05,
      "loss": 0.0243,
      "step": 17407
    },
    {
      "epoch": 2.439803784162579,
      "grad_norm": 0.12027157843112946,
      "learning_rate": 5.745993781392011e-05,
      "loss": 0.0176,
      "step": 17408
    },
    {
      "epoch": 2.4399439383321653,
      "grad_norm": 0.17350029945373535,
      "learning_rate": 5.7445587180100444e-05,
      "loss": 0.0298,
      "step": 17409
    },
    {
      "epoch": 2.4400840925017517,
      "grad_norm": 0.5006131529808044,
      "learning_rate": 5.743123654628079e-05,
      "loss": 0.0316,
      "step": 17410
    },
    {
      "epoch": 2.4402242466713386,
      "grad_norm": 0.3299851417541504,
      "learning_rate": 5.741688591246113e-05,
      "loss": 0.0667,
      "step": 17411
    },
    {
      "epoch": 2.440364400840925,
      "grad_norm": 0.17065083980560303,
      "learning_rate": 5.740253527864147e-05,
      "loss": 0.0261,
      "step": 17412
    },
    {
      "epoch": 2.4405045550105116,
      "grad_norm": 0.5163525342941284,
      "learning_rate": 5.7388184644821805e-05,
      "loss": 0.0069,
      "step": 17413
    },
    {
      "epoch": 2.440644709180098,
      "grad_norm": 0.18288275599479675,
      "learning_rate": 5.7373834011002154e-05,
      "loss": 0.0126,
      "step": 17414
    },
    {
      "epoch": 2.4407848633496845,
      "grad_norm": 0.3170284032821655,
      "learning_rate": 5.735948337718249e-05,
      "loss": 0.019,
      "step": 17415
    },
    {
      "epoch": 2.4409250175192714,
      "grad_norm": 0.024866115301847458,
      "learning_rate": 5.7345132743362825e-05,
      "loss": 0.002,
      "step": 17416
    },
    {
      "epoch": 2.441065171688858,
      "grad_norm": 0.8360458612442017,
      "learning_rate": 5.733078210954317e-05,
      "loss": 0.0731,
      "step": 17417
    },
    {
      "epoch": 2.4412053258584443,
      "grad_norm": 2.647637367248535,
      "learning_rate": 5.731643147572351e-05,
      "loss": 0.1484,
      "step": 17418
    },
    {
      "epoch": 2.4413454800280308,
      "grad_norm": 0.12292732298374176,
      "learning_rate": 5.7302080841903844e-05,
      "loss": 0.0027,
      "step": 17419
    },
    {
      "epoch": 2.441485634197617,
      "grad_norm": 1.0630799531936646,
      "learning_rate": 5.7287730208084186e-05,
      "loss": 0.0933,
      "step": 17420
    },
    {
      "epoch": 2.4416257883672037,
      "grad_norm": 0.20364363491535187,
      "learning_rate": 5.727337957426452e-05,
      "loss": 0.0352,
      "step": 17421
    },
    {
      "epoch": 2.4417659425367906,
      "grad_norm": 0.1532624065876007,
      "learning_rate": 5.725902894044487e-05,
      "loss": 0.0471,
      "step": 17422
    },
    {
      "epoch": 2.441906096706377,
      "grad_norm": 0.17431098222732544,
      "learning_rate": 5.7244678306625206e-05,
      "loss": 0.026,
      "step": 17423
    },
    {
      "epoch": 2.4420462508759635,
      "grad_norm": 0.20087915658950806,
      "learning_rate": 5.723032767280554e-05,
      "loss": 0.0163,
      "step": 17424
    },
    {
      "epoch": 2.44218640504555,
      "grad_norm": 0.3538360893726349,
      "learning_rate": 5.7215977038985883e-05,
      "loss": 0.0563,
      "step": 17425
    },
    {
      "epoch": 2.442326559215137,
      "grad_norm": 0.11413371562957764,
      "learning_rate": 5.7201626405166226e-05,
      "loss": 0.0148,
      "step": 17426
    },
    {
      "epoch": 2.4424667133847233,
      "grad_norm": 0.06987227499485016,
      "learning_rate": 5.718727577134656e-05,
      "loss": 0.007,
      "step": 17427
    },
    {
      "epoch": 2.44260686755431,
      "grad_norm": 0.07358165085315704,
      "learning_rate": 5.71729251375269e-05,
      "loss": 0.0067,
      "step": 17428
    },
    {
      "epoch": 2.4427470217238962,
      "grad_norm": 0.15052703022956848,
      "learning_rate": 5.715857450370724e-05,
      "loss": 0.0143,
      "step": 17429
    },
    {
      "epoch": 2.4428871758934827,
      "grad_norm": 0.33657753467559814,
      "learning_rate": 5.714422386988759e-05,
      "loss": 0.0784,
      "step": 17430
    },
    {
      "epoch": 2.443027330063069,
      "grad_norm": 0.2944040894508362,
      "learning_rate": 5.712987323606792e-05,
      "loss": 0.0328,
      "step": 17431
    },
    {
      "epoch": 2.443167484232656,
      "grad_norm": 0.07326116412878036,
      "learning_rate": 5.711552260224826e-05,
      "loss": 0.0082,
      "step": 17432
    },
    {
      "epoch": 2.4433076384022425,
      "grad_norm": 0.20281057059764862,
      "learning_rate": 5.71011719684286e-05,
      "loss": 0.0101,
      "step": 17433
    },
    {
      "epoch": 2.443447792571829,
      "grad_norm": 0.8643790483474731,
      "learning_rate": 5.708682133460894e-05,
      "loss": 0.0396,
      "step": 17434
    },
    {
      "epoch": 2.4435879467414154,
      "grad_norm": 1.076904296875,
      "learning_rate": 5.707247070078928e-05,
      "loss": 0.0353,
      "step": 17435
    },
    {
      "epoch": 2.4437281009110023,
      "grad_norm": 0.2580803632736206,
      "learning_rate": 5.705812006696962e-05,
      "loss": 0.0192,
      "step": 17436
    },
    {
      "epoch": 2.443868255080589,
      "grad_norm": 0.2819705307483673,
      "learning_rate": 5.7043769433149955e-05,
      "loss": 0.0255,
      "step": 17437
    },
    {
      "epoch": 2.4440084092501753,
      "grad_norm": 0.20648692548274994,
      "learning_rate": 5.7029418799330304e-05,
      "loss": 0.0304,
      "step": 17438
    },
    {
      "epoch": 2.4441485634197617,
      "grad_norm": 0.06767088174819946,
      "learning_rate": 5.701506816551064e-05,
      "loss": 0.0038,
      "step": 17439
    },
    {
      "epoch": 2.444288717589348,
      "grad_norm": 0.6174846291542053,
      "learning_rate": 5.7000717531690975e-05,
      "loss": 0.065,
      "step": 17440
    },
    {
      "epoch": 2.4444288717589346,
      "grad_norm": 0.2815219461917877,
      "learning_rate": 5.6986366897871323e-05,
      "loss": 0.0319,
      "step": 17441
    },
    {
      "epoch": 2.4445690259285215,
      "grad_norm": 0.1457332819700241,
      "learning_rate": 5.697201626405166e-05,
      "loss": 0.0252,
      "step": 17442
    },
    {
      "epoch": 2.444709180098108,
      "grad_norm": 0.15533718466758728,
      "learning_rate": 5.6957665630231994e-05,
      "loss": 0.0166,
      "step": 17443
    },
    {
      "epoch": 2.4448493342676945,
      "grad_norm": 0.12442232668399811,
      "learning_rate": 5.6943314996412336e-05,
      "loss": 0.0225,
      "step": 17444
    },
    {
      "epoch": 2.444989488437281,
      "grad_norm": 0.30915507674217224,
      "learning_rate": 5.692896436259268e-05,
      "loss": 0.0482,
      "step": 17445
    },
    {
      "epoch": 2.4451296426068674,
      "grad_norm": 0.30458346009254456,
      "learning_rate": 5.691461372877302e-05,
      "loss": 0.0841,
      "step": 17446
    },
    {
      "epoch": 2.4452697967764543,
      "grad_norm": 0.2273937165737152,
      "learning_rate": 5.6900263094953356e-05,
      "loss": 0.054,
      "step": 17447
    },
    {
      "epoch": 2.4454099509460407,
      "grad_norm": 0.4468878209590912,
      "learning_rate": 5.688591246113369e-05,
      "loss": 0.0224,
      "step": 17448
    },
    {
      "epoch": 2.445550105115627,
      "grad_norm": 0.16137811541557312,
      "learning_rate": 5.687156182731404e-05,
      "loss": 0.004,
      "step": 17449
    },
    {
      "epoch": 2.4456902592852137,
      "grad_norm": 0.13575440645217896,
      "learning_rate": 5.6857211193494375e-05,
      "loss": 0.006,
      "step": 17450
    },
    {
      "epoch": 2.4458304134548,
      "grad_norm": 0.18281446397304535,
      "learning_rate": 5.684286055967471e-05,
      "loss": 0.0161,
      "step": 17451
    },
    {
      "epoch": 2.4459705676243866,
      "grad_norm": 0.5374318957328796,
      "learning_rate": 5.682850992585505e-05,
      "loss": 0.0254,
      "step": 17452
    },
    {
      "epoch": 2.4461107217939735,
      "grad_norm": 0.19403211772441864,
      "learning_rate": 5.6814159292035395e-05,
      "loss": 0.0133,
      "step": 17453
    },
    {
      "epoch": 2.44625087596356,
      "grad_norm": 0.4154665768146515,
      "learning_rate": 5.679980865821574e-05,
      "loss": 0.0546,
      "step": 17454
    },
    {
      "epoch": 2.4463910301331464,
      "grad_norm": 0.028070323169231415,
      "learning_rate": 5.678545802439607e-05,
      "loss": 0.0032,
      "step": 17455
    },
    {
      "epoch": 2.446531184302733,
      "grad_norm": 0.08635620772838593,
      "learning_rate": 5.677110739057641e-05,
      "loss": 0.0131,
      "step": 17456
    },
    {
      "epoch": 2.4466713384723198,
      "grad_norm": 0.25716128945350647,
      "learning_rate": 5.6756756756756757e-05,
      "loss": 0.056,
      "step": 17457
    },
    {
      "epoch": 2.4468114926419062,
      "grad_norm": 0.40723538398742676,
      "learning_rate": 5.674240612293709e-05,
      "loss": 0.0142,
      "step": 17458
    },
    {
      "epoch": 2.4469516468114927,
      "grad_norm": 0.4177103042602539,
      "learning_rate": 5.672805548911743e-05,
      "loss": 0.0807,
      "step": 17459
    },
    {
      "epoch": 2.447091800981079,
      "grad_norm": 0.30821430683135986,
      "learning_rate": 5.671370485529777e-05,
      "loss": 0.0191,
      "step": 17460
    },
    {
      "epoch": 2.4472319551506656,
      "grad_norm": 1.9172149896621704,
      "learning_rate": 5.669935422147811e-05,
      "loss": 0.0251,
      "step": 17461
    },
    {
      "epoch": 2.447372109320252,
      "grad_norm": 0.11550029367208481,
      "learning_rate": 5.6685003587658454e-05,
      "loss": 0.0294,
      "step": 17462
    },
    {
      "epoch": 2.447512263489839,
      "grad_norm": 0.34596455097198486,
      "learning_rate": 5.667065295383879e-05,
      "loss": 0.0537,
      "step": 17463
    },
    {
      "epoch": 2.4476524176594254,
      "grad_norm": 0.04013461992144585,
      "learning_rate": 5.6656302320019124e-05,
      "loss": 0.0016,
      "step": 17464
    },
    {
      "epoch": 2.447792571829012,
      "grad_norm": 0.657112181186676,
      "learning_rate": 5.664195168619947e-05,
      "loss": 0.0693,
      "step": 17465
    },
    {
      "epoch": 2.4479327259985983,
      "grad_norm": 0.0632171779870987,
      "learning_rate": 5.662760105237981e-05,
      "loss": 0.0034,
      "step": 17466
    },
    {
      "epoch": 2.4480728801681853,
      "grad_norm": 0.027072517201304436,
      "learning_rate": 5.6613250418560144e-05,
      "loss": 0.0023,
      "step": 17467
    },
    {
      "epoch": 2.4482130343377717,
      "grad_norm": 0.5208009481430054,
      "learning_rate": 5.6598899784740486e-05,
      "loss": 0.0369,
      "step": 17468
    },
    {
      "epoch": 2.448353188507358,
      "grad_norm": 0.1418853998184204,
      "learning_rate": 5.658454915092083e-05,
      "loss": 0.0089,
      "step": 17469
    },
    {
      "epoch": 2.4484933426769446,
      "grad_norm": 2.169551134109497,
      "learning_rate": 5.657019851710117e-05,
      "loss": 0.0949,
      "step": 17470
    },
    {
      "epoch": 2.448633496846531,
      "grad_norm": 0.23169922828674316,
      "learning_rate": 5.6555847883281505e-05,
      "loss": 0.0216,
      "step": 17471
    },
    {
      "epoch": 2.4487736510161175,
      "grad_norm": 0.2709794044494629,
      "learning_rate": 5.654149724946184e-05,
      "loss": 0.0089,
      "step": 17472
    },
    {
      "epoch": 2.4489138051857045,
      "grad_norm": 0.1287136971950531,
      "learning_rate": 5.652714661564219e-05,
      "loss": 0.0119,
      "step": 17473
    },
    {
      "epoch": 2.449053959355291,
      "grad_norm": 0.1448056697845459,
      "learning_rate": 5.6512795981822525e-05,
      "loss": 0.0095,
      "step": 17474
    },
    {
      "epoch": 2.4491941135248774,
      "grad_norm": 0.4334147572517395,
      "learning_rate": 5.649844534800286e-05,
      "loss": 0.0405,
      "step": 17475
    },
    {
      "epoch": 2.449334267694464,
      "grad_norm": 0.12910759449005127,
      "learning_rate": 5.64840947141832e-05,
      "loss": 0.0163,
      "step": 17476
    },
    {
      "epoch": 2.4494744218640503,
      "grad_norm": 0.2877131998538971,
      "learning_rate": 5.6469744080363545e-05,
      "loss": 0.0403,
      "step": 17477
    },
    {
      "epoch": 2.449614576033637,
      "grad_norm": 0.12174680829048157,
      "learning_rate": 5.645539344654389e-05,
      "loss": 0.0103,
      "step": 17478
    },
    {
      "epoch": 2.4497547302032237,
      "grad_norm": 0.2639316916465759,
      "learning_rate": 5.644104281272422e-05,
      "loss": 0.021,
      "step": 17479
    },
    {
      "epoch": 2.44989488437281,
      "grad_norm": 0.12304729968309402,
      "learning_rate": 5.6426692178904564e-05,
      "loss": 0.0077,
      "step": 17480
    },
    {
      "epoch": 2.4500350385423966,
      "grad_norm": 0.39205509424209595,
      "learning_rate": 5.6412341545084906e-05,
      "loss": 0.0241,
      "step": 17481
    },
    {
      "epoch": 2.450175192711983,
      "grad_norm": 0.2815830707550049,
      "learning_rate": 5.639799091126524e-05,
      "loss": 0.0266,
      "step": 17482
    },
    {
      "epoch": 2.4503153468815695,
      "grad_norm": 0.5846973061561584,
      "learning_rate": 5.638364027744558e-05,
      "loss": 0.0653,
      "step": 17483
    },
    {
      "epoch": 2.4504555010511564,
      "grad_norm": 0.3415982723236084,
      "learning_rate": 5.6369289643625926e-05,
      "loss": 0.0608,
      "step": 17484
    },
    {
      "epoch": 2.450595655220743,
      "grad_norm": 0.2679758667945862,
      "learning_rate": 5.635493900980626e-05,
      "loss": 0.0476,
      "step": 17485
    },
    {
      "epoch": 2.4507358093903293,
      "grad_norm": 0.3460039496421814,
      "learning_rate": 5.63405883759866e-05,
      "loss": 0.0226,
      "step": 17486
    },
    {
      "epoch": 2.4508759635599158,
      "grad_norm": 0.1832917332649231,
      "learning_rate": 5.632623774216694e-05,
      "loss": 0.0155,
      "step": 17487
    },
    {
      "epoch": 2.4510161177295027,
      "grad_norm": 0.25096654891967773,
      "learning_rate": 5.631188710834728e-05,
      "loss": 0.0315,
      "step": 17488
    },
    {
      "epoch": 2.451156271899089,
      "grad_norm": 0.05631822720170021,
      "learning_rate": 5.629753647452762e-05,
      "loss": 0.0062,
      "step": 17489
    },
    {
      "epoch": 2.4512964260686756,
      "grad_norm": 0.1662900298833847,
      "learning_rate": 5.628318584070796e-05,
      "loss": 0.0425,
      "step": 17490
    },
    {
      "epoch": 2.451436580238262,
      "grad_norm": 0.2209007441997528,
      "learning_rate": 5.6268835206888294e-05,
      "loss": 0.0341,
      "step": 17491
    },
    {
      "epoch": 2.4515767344078485,
      "grad_norm": 0.0636403039097786,
      "learning_rate": 5.625448457306864e-05,
      "loss": 0.0069,
      "step": 17492
    },
    {
      "epoch": 2.451716888577435,
      "grad_norm": 0.33765727281570435,
      "learning_rate": 5.624013393924898e-05,
      "loss": 0.0168,
      "step": 17493
    },
    {
      "epoch": 2.451857042747022,
      "grad_norm": 0.23613616824150085,
      "learning_rate": 5.622578330542932e-05,
      "loss": 0.0193,
      "step": 17494
    },
    {
      "epoch": 2.4519971969166083,
      "grad_norm": 0.1330777257680893,
      "learning_rate": 5.6211432671609655e-05,
      "loss": 0.007,
      "step": 17495
    },
    {
      "epoch": 2.452137351086195,
      "grad_norm": 0.13238130509853363,
      "learning_rate": 5.619708203779e-05,
      "loss": 0.0388,
      "step": 17496
    },
    {
      "epoch": 2.4522775052557813,
      "grad_norm": 0.31312137842178345,
      "learning_rate": 5.618273140397034e-05,
      "loss": 0.013,
      "step": 17497
    },
    {
      "epoch": 2.4524176594253677,
      "grad_norm": 0.1287730187177658,
      "learning_rate": 5.6168380770150675e-05,
      "loss": 0.0074,
      "step": 17498
    },
    {
      "epoch": 2.4525578135949546,
      "grad_norm": 0.31204330921173096,
      "learning_rate": 5.615403013633101e-05,
      "loss": 0.0354,
      "step": 17499
    },
    {
      "epoch": 2.452697967764541,
      "grad_norm": 0.17967405915260315,
      "learning_rate": 5.613967950251136e-05,
      "loss": 0.0257,
      "step": 17500
    },
    {
      "epoch": 2.4528381219341275,
      "grad_norm": 0.07374734431505203,
      "learning_rate": 5.6125328868691694e-05,
      "loss": 0.0067,
      "step": 17501
    },
    {
      "epoch": 2.452978276103714,
      "grad_norm": 0.5038384199142456,
      "learning_rate": 5.6110978234872036e-05,
      "loss": 0.0577,
      "step": 17502
    },
    {
      "epoch": 2.4531184302733005,
      "grad_norm": 0.05403308570384979,
      "learning_rate": 5.609662760105237e-05,
      "loss": 0.0034,
      "step": 17503
    },
    {
      "epoch": 2.4532585844428874,
      "grad_norm": 0.1436837762594223,
      "learning_rate": 5.6082276967232714e-05,
      "loss": 0.0133,
      "step": 17504
    },
    {
      "epoch": 2.453398738612474,
      "grad_norm": 0.16623951494693756,
      "learning_rate": 5.6067926333413056e-05,
      "loss": 0.0316,
      "step": 17505
    },
    {
      "epoch": 2.4535388927820603,
      "grad_norm": 0.14871586859226227,
      "learning_rate": 5.605357569959339e-05,
      "loss": 0.0439,
      "step": 17506
    },
    {
      "epoch": 2.4536790469516467,
      "grad_norm": 0.09176642447710037,
      "learning_rate": 5.603922506577373e-05,
      "loss": 0.0106,
      "step": 17507
    },
    {
      "epoch": 2.453819201121233,
      "grad_norm": 0.13082154095172882,
      "learning_rate": 5.6024874431954076e-05,
      "loss": 0.0126,
      "step": 17508
    },
    {
      "epoch": 2.45395935529082,
      "grad_norm": 0.09946896135807037,
      "learning_rate": 5.601052379813441e-05,
      "loss": 0.0034,
      "step": 17509
    },
    {
      "epoch": 2.4540995094604066,
      "grad_norm": 0.396707683801651,
      "learning_rate": 5.599617316431475e-05,
      "loss": 0.068,
      "step": 17510
    },
    {
      "epoch": 2.454239663629993,
      "grad_norm": 0.3288958966732025,
      "learning_rate": 5.598182253049509e-05,
      "loss": 0.0312,
      "step": 17511
    },
    {
      "epoch": 2.4543798177995795,
      "grad_norm": 0.20087780058383942,
      "learning_rate": 5.596747189667543e-05,
      "loss": 0.0107,
      "step": 17512
    },
    {
      "epoch": 2.454519971969166,
      "grad_norm": 0.5589919090270996,
      "learning_rate": 5.595312126285577e-05,
      "loss": 0.0252,
      "step": 17513
    },
    {
      "epoch": 2.4546601261387524,
      "grad_norm": 0.13789360225200653,
      "learning_rate": 5.593877062903611e-05,
      "loss": 0.0206,
      "step": 17514
    },
    {
      "epoch": 2.4548002803083393,
      "grad_norm": 0.6786129474639893,
      "learning_rate": 5.592441999521646e-05,
      "loss": 0.13,
      "step": 17515
    },
    {
      "epoch": 2.4549404344779258,
      "grad_norm": 0.46774470806121826,
      "learning_rate": 5.591006936139679e-05,
      "loss": 0.0776,
      "step": 17516
    },
    {
      "epoch": 2.455080588647512,
      "grad_norm": 0.47916993498802185,
      "learning_rate": 5.589571872757713e-05,
      "loss": 0.0791,
      "step": 17517
    },
    {
      "epoch": 2.4552207428170987,
      "grad_norm": 0.1951342672109604,
      "learning_rate": 5.588136809375747e-05,
      "loss": 0.0523,
      "step": 17518
    },
    {
      "epoch": 2.4553608969866856,
      "grad_norm": 2.039923667907715,
      "learning_rate": 5.586701745993781e-05,
      "loss": 0.0448,
      "step": 17519
    },
    {
      "epoch": 2.455501051156272,
      "grad_norm": 0.44729429483413696,
      "learning_rate": 5.585266682611815e-05,
      "loss": 0.0136,
      "step": 17520
    },
    {
      "epoch": 2.4556412053258585,
      "grad_norm": 0.17004776000976562,
      "learning_rate": 5.583831619229849e-05,
      "loss": 0.0418,
      "step": 17521
    },
    {
      "epoch": 2.455781359495445,
      "grad_norm": 0.45309966802597046,
      "learning_rate": 5.5823965558478824e-05,
      "loss": 0.0583,
      "step": 17522
    },
    {
      "epoch": 2.4559215136650314,
      "grad_norm": 0.3523557186126709,
      "learning_rate": 5.580961492465917e-05,
      "loss": 0.0547,
      "step": 17523
    },
    {
      "epoch": 2.456061667834618,
      "grad_norm": 0.1562371402978897,
      "learning_rate": 5.579526429083951e-05,
      "loss": 0.0291,
      "step": 17524
    },
    {
      "epoch": 2.456201822004205,
      "grad_norm": 0.46670010685920715,
      "learning_rate": 5.5780913657019844e-05,
      "loss": 0.0736,
      "step": 17525
    },
    {
      "epoch": 2.4563419761737912,
      "grad_norm": 0.29604804515838623,
      "learning_rate": 5.5766563023200186e-05,
      "loss": 0.0314,
      "step": 17526
    },
    {
      "epoch": 2.4564821303433777,
      "grad_norm": 0.048980217427015305,
      "learning_rate": 5.575221238938053e-05,
      "loss": 0.0048,
      "step": 17527
    },
    {
      "epoch": 2.456622284512964,
      "grad_norm": 0.3164384961128235,
      "learning_rate": 5.573786175556087e-05,
      "loss": 0.0215,
      "step": 17528
    },
    {
      "epoch": 2.4567624386825506,
      "grad_norm": 0.14032945036888123,
      "learning_rate": 5.5723511121741206e-05,
      "loss": 0.0135,
      "step": 17529
    },
    {
      "epoch": 2.4569025928521375,
      "grad_norm": 0.16116999089717865,
      "learning_rate": 5.570916048792154e-05,
      "loss": 0.0191,
      "step": 17530
    },
    {
      "epoch": 2.457042747021724,
      "grad_norm": 0.5274431109428406,
      "learning_rate": 5.569480985410189e-05,
      "loss": 0.0459,
      "step": 17531
    },
    {
      "epoch": 2.4571829011913104,
      "grad_norm": 0.07985707372426987,
      "learning_rate": 5.5680459220282225e-05,
      "loss": 0.0149,
      "step": 17532
    },
    {
      "epoch": 2.457323055360897,
      "grad_norm": 0.3965173661708832,
      "learning_rate": 5.566610858646256e-05,
      "loss": 0.0689,
      "step": 17533
    },
    {
      "epoch": 2.4574632095304834,
      "grad_norm": 0.10106916725635529,
      "learning_rate": 5.56517579526429e-05,
      "loss": 0.007,
      "step": 17534
    },
    {
      "epoch": 2.4576033637000703,
      "grad_norm": 0.07537499815225601,
      "learning_rate": 5.5637407318823245e-05,
      "loss": 0.0121,
      "step": 17535
    },
    {
      "epoch": 2.4577435178696567,
      "grad_norm": 0.08306626230478287,
      "learning_rate": 5.562305668500359e-05,
      "loss": 0.0073,
      "step": 17536
    },
    {
      "epoch": 2.457883672039243,
      "grad_norm": 0.32017800211906433,
      "learning_rate": 5.560870605118392e-05,
      "loss": 0.0142,
      "step": 17537
    },
    {
      "epoch": 2.4580238262088296,
      "grad_norm": 0.06587784737348557,
      "learning_rate": 5.559435541736426e-05,
      "loss": 0.0091,
      "step": 17538
    },
    {
      "epoch": 2.458163980378416,
      "grad_norm": 0.1399422138929367,
      "learning_rate": 5.5580004783544606e-05,
      "loss": 0.0247,
      "step": 17539
    },
    {
      "epoch": 2.458304134548003,
      "grad_norm": 0.08159590512514114,
      "learning_rate": 5.556565414972494e-05,
      "loss": 0.0087,
      "step": 17540
    },
    {
      "epoch": 2.4584442887175895,
      "grad_norm": 0.5555731058120728,
      "learning_rate": 5.555130351590528e-05,
      "loss": 0.0305,
      "step": 17541
    },
    {
      "epoch": 2.458584442887176,
      "grad_norm": 0.1639791578054428,
      "learning_rate": 5.553695288208562e-05,
      "loss": 0.0296,
      "step": 17542
    },
    {
      "epoch": 2.4587245970567624,
      "grad_norm": 0.13340994715690613,
      "learning_rate": 5.552260224826596e-05,
      "loss": 0.0116,
      "step": 17543
    },
    {
      "epoch": 2.458864751226349,
      "grad_norm": 0.5196988582611084,
      "learning_rate": 5.5508251614446304e-05,
      "loss": 0.0336,
      "step": 17544
    },
    {
      "epoch": 2.4590049053959353,
      "grad_norm": 0.3339941203594208,
      "learning_rate": 5.549390098062664e-05,
      "loss": 0.042,
      "step": 17545
    },
    {
      "epoch": 2.459145059565522,
      "grad_norm": 0.15877309441566467,
      "learning_rate": 5.5479550346806974e-05,
      "loss": 0.0296,
      "step": 17546
    },
    {
      "epoch": 2.4592852137351087,
      "grad_norm": 0.10402316600084305,
      "learning_rate": 5.546519971298732e-05,
      "loss": 0.0129,
      "step": 17547
    },
    {
      "epoch": 2.459425367904695,
      "grad_norm": 0.21463969349861145,
      "learning_rate": 5.545084907916766e-05,
      "loss": 0.0391,
      "step": 17548
    },
    {
      "epoch": 2.4595655220742816,
      "grad_norm": 0.2541576623916626,
      "learning_rate": 5.5436498445347994e-05,
      "loss": 0.0364,
      "step": 17549
    },
    {
      "epoch": 2.4597056762438685,
      "grad_norm": 0.19378821551799774,
      "learning_rate": 5.5422147811528336e-05,
      "loss": 0.0175,
      "step": 17550
    },
    {
      "epoch": 2.459845830413455,
      "grad_norm": 0.27218207716941833,
      "learning_rate": 5.540779717770868e-05,
      "loss": 0.0074,
      "step": 17551
    },
    {
      "epoch": 2.4599859845830414,
      "grad_norm": 0.25482678413391113,
      "learning_rate": 5.539344654388902e-05,
      "loss": 0.0747,
      "step": 17552
    },
    {
      "epoch": 2.460126138752628,
      "grad_norm": 0.2299598753452301,
      "learning_rate": 5.5379095910069355e-05,
      "loss": 0.0277,
      "step": 17553
    },
    {
      "epoch": 2.4602662929222143,
      "grad_norm": 0.4402224123477936,
      "learning_rate": 5.53647452762497e-05,
      "loss": 0.0313,
      "step": 17554
    },
    {
      "epoch": 2.460406447091801,
      "grad_norm": 0.05747057870030403,
      "learning_rate": 5.535039464243004e-05,
      "loss": 0.005,
      "step": 17555
    },
    {
      "epoch": 2.4605466012613877,
      "grad_norm": 0.39516496658325195,
      "learning_rate": 5.5336044008610375e-05,
      "loss": 0.0731,
      "step": 17556
    },
    {
      "epoch": 2.460686755430974,
      "grad_norm": 0.16670043766498566,
      "learning_rate": 5.532169337479071e-05,
      "loss": 0.0072,
      "step": 17557
    },
    {
      "epoch": 2.4608269096005606,
      "grad_norm": 0.23560599982738495,
      "learning_rate": 5.530734274097106e-05,
      "loss": 0.0062,
      "step": 17558
    },
    {
      "epoch": 2.460967063770147,
      "grad_norm": 0.23532281816005707,
      "learning_rate": 5.5292992107151395e-05,
      "loss": 0.0309,
      "step": 17559
    },
    {
      "epoch": 2.4611072179397335,
      "grad_norm": 0.13903413712978363,
      "learning_rate": 5.527864147333174e-05,
      "loss": 0.0121,
      "step": 17560
    },
    {
      "epoch": 2.4612473721093204,
      "grad_norm": 0.14731024205684662,
      "learning_rate": 5.526429083951207e-05,
      "loss": 0.0058,
      "step": 17561
    },
    {
      "epoch": 2.461387526278907,
      "grad_norm": 0.22105124592781067,
      "learning_rate": 5.5249940205692414e-05,
      "loss": 0.0334,
      "step": 17562
    },
    {
      "epoch": 2.4615276804484933,
      "grad_norm": 0.18250513076782227,
      "learning_rate": 5.5235589571872756e-05,
      "loss": 0.028,
      "step": 17563
    },
    {
      "epoch": 2.46166783461808,
      "grad_norm": 0.2927420437335968,
      "learning_rate": 5.522123893805309e-05,
      "loss": 0.0202,
      "step": 17564
    },
    {
      "epoch": 2.4618079887876663,
      "grad_norm": 0.17645567655563354,
      "learning_rate": 5.520688830423343e-05,
      "loss": 0.0055,
      "step": 17565
    },
    {
      "epoch": 2.461948142957253,
      "grad_norm": 0.1739467978477478,
      "learning_rate": 5.5192537670413776e-05,
      "loss": 0.0071,
      "step": 17566
    },
    {
      "epoch": 2.4620882971268396,
      "grad_norm": 0.12015873938798904,
      "learning_rate": 5.517818703659411e-05,
      "loss": 0.0292,
      "step": 17567
    },
    {
      "epoch": 2.462228451296426,
      "grad_norm": 0.15247364342212677,
      "learning_rate": 5.516383640277445e-05,
      "loss": 0.0073,
      "step": 17568
    },
    {
      "epoch": 2.4623686054660125,
      "grad_norm": 1.001036524772644,
      "learning_rate": 5.514948576895479e-05,
      "loss": 0.0779,
      "step": 17569
    },
    {
      "epoch": 2.462508759635599,
      "grad_norm": 1.0480105876922607,
      "learning_rate": 5.513513513513513e-05,
      "loss": 0.0321,
      "step": 17570
    },
    {
      "epoch": 2.462648913805186,
      "grad_norm": 0.24105487763881683,
      "learning_rate": 5.512078450131547e-05,
      "loss": 0.0369,
      "step": 17571
    },
    {
      "epoch": 2.4627890679747724,
      "grad_norm": 0.09204325824975967,
      "learning_rate": 5.510643386749581e-05,
      "loss": 0.0125,
      "step": 17572
    },
    {
      "epoch": 2.462929222144359,
      "grad_norm": 0.32665038108825684,
      "learning_rate": 5.5092083233676143e-05,
      "loss": 0.0316,
      "step": 17573
    },
    {
      "epoch": 2.4630693763139453,
      "grad_norm": 1.4087618589401245,
      "learning_rate": 5.507773259985649e-05,
      "loss": 0.0309,
      "step": 17574
    },
    {
      "epoch": 2.4632095304835318,
      "grad_norm": 0.37037238478660583,
      "learning_rate": 5.506338196603683e-05,
      "loss": 0.0332,
      "step": 17575
    },
    {
      "epoch": 2.463349684653118,
      "grad_norm": 0.26116031408309937,
      "learning_rate": 5.504903133221717e-05,
      "loss": 0.0419,
      "step": 17576
    },
    {
      "epoch": 2.463489838822705,
      "grad_norm": 0.46755537390708923,
      "learning_rate": 5.5034680698397505e-05,
      "loss": 0.0267,
      "step": 17577
    },
    {
      "epoch": 2.4636299929922916,
      "grad_norm": 0.26234257221221924,
      "learning_rate": 5.502033006457785e-05,
      "loss": 0.0459,
      "step": 17578
    },
    {
      "epoch": 2.463770147161878,
      "grad_norm": 0.5037165880203247,
      "learning_rate": 5.500597943075819e-05,
      "loss": 0.0349,
      "step": 17579
    },
    {
      "epoch": 2.4639103013314645,
      "grad_norm": 0.15681907534599304,
      "learning_rate": 5.4991628796938525e-05,
      "loss": 0.0134,
      "step": 17580
    },
    {
      "epoch": 2.4640504555010514,
      "grad_norm": 0.6365145444869995,
      "learning_rate": 5.497727816311886e-05,
      "loss": 0.071,
      "step": 17581
    },
    {
      "epoch": 2.464190609670638,
      "grad_norm": 0.26379770040512085,
      "learning_rate": 5.496292752929921e-05,
      "loss": 0.027,
      "step": 17582
    },
    {
      "epoch": 2.4643307638402243,
      "grad_norm": 0.3846650719642639,
      "learning_rate": 5.4948576895479544e-05,
      "loss": 0.0255,
      "step": 17583
    },
    {
      "epoch": 2.4644709180098108,
      "grad_norm": 0.5092746615409851,
      "learning_rate": 5.4934226261659886e-05,
      "loss": 0.0303,
      "step": 17584
    },
    {
      "epoch": 2.4646110721793972,
      "grad_norm": 0.4900977909564972,
      "learning_rate": 5.491987562784022e-05,
      "loss": 0.0111,
      "step": 17585
    },
    {
      "epoch": 2.4647512263489837,
      "grad_norm": 0.0670226439833641,
      "learning_rate": 5.4905524994020564e-05,
      "loss": 0.0043,
      "step": 17586
    },
    {
      "epoch": 2.4648913805185706,
      "grad_norm": 0.06610870361328125,
      "learning_rate": 5.4891174360200906e-05,
      "loss": 0.0079,
      "step": 17587
    },
    {
      "epoch": 2.465031534688157,
      "grad_norm": 0.180598184466362,
      "learning_rate": 5.487682372638124e-05,
      "loss": 0.0086,
      "step": 17588
    },
    {
      "epoch": 2.4651716888577435,
      "grad_norm": 0.18171022832393646,
      "learning_rate": 5.486247309256159e-05,
      "loss": 0.0269,
      "step": 17589
    },
    {
      "epoch": 2.46531184302733,
      "grad_norm": 0.2641808092594147,
      "learning_rate": 5.4848122458741925e-05,
      "loss": 0.0335,
      "step": 17590
    },
    {
      "epoch": 2.4654519971969164,
      "grad_norm": 0.29543328285217285,
      "learning_rate": 5.483377182492226e-05,
      "loss": 0.0268,
      "step": 17591
    },
    {
      "epoch": 2.4655921513665033,
      "grad_norm": 0.20150195062160492,
      "learning_rate": 5.48194211911026e-05,
      "loss": 0.0955,
      "step": 17592
    },
    {
      "epoch": 2.46573230553609,
      "grad_norm": 0.3302060067653656,
      "learning_rate": 5.4805070557282945e-05,
      "loss": 0.0383,
      "step": 17593
    },
    {
      "epoch": 2.4658724597056763,
      "grad_norm": 0.22340722382068634,
      "learning_rate": 5.479071992346328e-05,
      "loss": 0.0537,
      "step": 17594
    },
    {
      "epoch": 2.4660126138752627,
      "grad_norm": 0.20863080024719238,
      "learning_rate": 5.477636928964362e-05,
      "loss": 0.0174,
      "step": 17595
    },
    {
      "epoch": 2.466152768044849,
      "grad_norm": 0.280705988407135,
      "learning_rate": 5.476201865582396e-05,
      "loss": 0.0259,
      "step": 17596
    },
    {
      "epoch": 2.4662929222144356,
      "grad_norm": 0.23006755113601685,
      "learning_rate": 5.474766802200431e-05,
      "loss": 0.0542,
      "step": 17597
    },
    {
      "epoch": 2.4664330763840225,
      "grad_norm": 0.44066715240478516,
      "learning_rate": 5.473331738818464e-05,
      "loss": 0.0239,
      "step": 17598
    },
    {
      "epoch": 2.466573230553609,
      "grad_norm": 0.1422925889492035,
      "learning_rate": 5.471896675436498e-05,
      "loss": 0.0156,
      "step": 17599
    },
    {
      "epoch": 2.4667133847231955,
      "grad_norm": 0.5817549824714661,
      "learning_rate": 5.470461612054532e-05,
      "loss": 0.073,
      "step": 17600
    },
    {
      "epoch": 2.466853538892782,
      "grad_norm": 0.14191415905952454,
      "learning_rate": 5.469026548672566e-05,
      "loss": 0.0057,
      "step": 17601
    },
    {
      "epoch": 2.466993693062369,
      "grad_norm": 0.1116810217499733,
      "learning_rate": 5.4675914852906e-05,
      "loss": 0.0091,
      "step": 17602
    },
    {
      "epoch": 2.4671338472319553,
      "grad_norm": 0.2270602583885193,
      "learning_rate": 5.466156421908634e-05,
      "loss": 0.0264,
      "step": 17603
    },
    {
      "epoch": 2.4672740014015417,
      "grad_norm": 0.38715335726737976,
      "learning_rate": 5.4647213585266674e-05,
      "loss": 0.0669,
      "step": 17604
    },
    {
      "epoch": 2.467414155571128,
      "grad_norm": 0.12069467455148697,
      "learning_rate": 5.463286295144702e-05,
      "loss": 0.0071,
      "step": 17605
    },
    {
      "epoch": 2.4675543097407147,
      "grad_norm": 0.2475578635931015,
      "learning_rate": 5.461851231762736e-05,
      "loss": 0.0549,
      "step": 17606
    },
    {
      "epoch": 2.467694463910301,
      "grad_norm": 0.4554625451564789,
      "learning_rate": 5.4604161683807694e-05,
      "loss": 0.0532,
      "step": 17607
    },
    {
      "epoch": 2.467834618079888,
      "grad_norm": 0.09981324523687363,
      "learning_rate": 5.4589811049988036e-05,
      "loss": 0.0165,
      "step": 17608
    },
    {
      "epoch": 2.4679747722494745,
      "grad_norm": 0.5379135012626648,
      "learning_rate": 5.457546041616838e-05,
      "loss": 0.0228,
      "step": 17609
    },
    {
      "epoch": 2.468114926419061,
      "grad_norm": 0.11988509446382523,
      "learning_rate": 5.4561109782348714e-05,
      "loss": 0.0386,
      "step": 17610
    },
    {
      "epoch": 2.4682550805886474,
      "grad_norm": 0.15549993515014648,
      "learning_rate": 5.4546759148529056e-05,
      "loss": 0.0082,
      "step": 17611
    },
    {
      "epoch": 2.4683952347582343,
      "grad_norm": 0.26520514488220215,
      "learning_rate": 5.453240851470939e-05,
      "loss": 0.0377,
      "step": 17612
    },
    {
      "epoch": 2.4685353889278208,
      "grad_norm": 0.12857557833194733,
      "learning_rate": 5.451805788088974e-05,
      "loss": 0.0082,
      "step": 17613
    },
    {
      "epoch": 2.468675543097407,
      "grad_norm": 0.21515578031539917,
      "learning_rate": 5.4503707247070075e-05,
      "loss": 0.0166,
      "step": 17614
    },
    {
      "epoch": 2.4688156972669937,
      "grad_norm": 0.1563521772623062,
      "learning_rate": 5.448935661325041e-05,
      "loss": 0.0186,
      "step": 17615
    },
    {
      "epoch": 2.46895585143658,
      "grad_norm": 0.4665060341358185,
      "learning_rate": 5.447500597943075e-05,
      "loss": 0.0302,
      "step": 17616
    },
    {
      "epoch": 2.4690960056061666,
      "grad_norm": 0.22005833685398102,
      "learning_rate": 5.4460655345611095e-05,
      "loss": 0.018,
      "step": 17617
    },
    {
      "epoch": 2.4692361597757535,
      "grad_norm": 0.328733891248703,
      "learning_rate": 5.444630471179143e-05,
      "loss": 0.0121,
      "step": 17618
    },
    {
      "epoch": 2.46937631394534,
      "grad_norm": 0.10675746947526932,
      "learning_rate": 5.443195407797177e-05,
      "loss": 0.0039,
      "step": 17619
    },
    {
      "epoch": 2.4695164681149264,
      "grad_norm": 0.25655025243759155,
      "learning_rate": 5.441760344415211e-05,
      "loss": 0.0289,
      "step": 17620
    },
    {
      "epoch": 2.469656622284513,
      "grad_norm": 0.28004828095436096,
      "learning_rate": 5.4403252810332456e-05,
      "loss": 0.0334,
      "step": 17621
    },
    {
      "epoch": 2.4697967764540993,
      "grad_norm": 0.6077659726142883,
      "learning_rate": 5.438890217651279e-05,
      "loss": 0.0644,
      "step": 17622
    },
    {
      "epoch": 2.4699369306236862,
      "grad_norm": 0.19305074214935303,
      "learning_rate": 5.437455154269313e-05,
      "loss": 0.0216,
      "step": 17623
    },
    {
      "epoch": 2.4700770847932727,
      "grad_norm": 0.4131547808647156,
      "learning_rate": 5.436020090887347e-05,
      "loss": 0.0347,
      "step": 17624
    },
    {
      "epoch": 2.470217238962859,
      "grad_norm": 0.14775022864341736,
      "learning_rate": 5.434585027505381e-05,
      "loss": 0.0172,
      "step": 17625
    },
    {
      "epoch": 2.4703573931324456,
      "grad_norm": 0.17525510489940643,
      "learning_rate": 5.433149964123415e-05,
      "loss": 0.0164,
      "step": 17626
    },
    {
      "epoch": 2.470497547302032,
      "grad_norm": 0.15028899908065796,
      "learning_rate": 5.431714900741449e-05,
      "loss": 0.0291,
      "step": 17627
    },
    {
      "epoch": 2.4706377014716185,
      "grad_norm": 0.42396828532218933,
      "learning_rate": 5.430279837359483e-05,
      "loss": 0.0373,
      "step": 17628
    },
    {
      "epoch": 2.4707778556412054,
      "grad_norm": 0.26741477847099304,
      "learning_rate": 5.428844773977517e-05,
      "loss": 0.0815,
      "step": 17629
    },
    {
      "epoch": 2.470918009810792,
      "grad_norm": 0.23611553013324738,
      "learning_rate": 5.427409710595551e-05,
      "loss": 0.0191,
      "step": 17630
    },
    {
      "epoch": 2.4710581639803784,
      "grad_norm": 0.18713533878326416,
      "learning_rate": 5.4259746472135844e-05,
      "loss": 0.0185,
      "step": 17631
    },
    {
      "epoch": 2.471198318149965,
      "grad_norm": 0.11704926937818527,
      "learning_rate": 5.424539583831619e-05,
      "loss": 0.0178,
      "step": 17632
    },
    {
      "epoch": 2.4713384723195517,
      "grad_norm": 0.16930925846099854,
      "learning_rate": 5.423104520449653e-05,
      "loss": 0.0136,
      "step": 17633
    },
    {
      "epoch": 2.471478626489138,
      "grad_norm": 0.339443564414978,
      "learning_rate": 5.421669457067686e-05,
      "loss": 0.008,
      "step": 17634
    },
    {
      "epoch": 2.4716187806587246,
      "grad_norm": 0.1321626454591751,
      "learning_rate": 5.4202343936857205e-05,
      "loss": 0.017,
      "step": 17635
    },
    {
      "epoch": 2.471758934828311,
      "grad_norm": 0.36218515038490295,
      "learning_rate": 5.418799330303755e-05,
      "loss": 0.0414,
      "step": 17636
    },
    {
      "epoch": 2.4718990889978976,
      "grad_norm": 0.28529658913612366,
      "learning_rate": 5.417364266921789e-05,
      "loss": 0.0315,
      "step": 17637
    },
    {
      "epoch": 2.472039243167484,
      "grad_norm": 0.1235649436712265,
      "learning_rate": 5.4159292035398225e-05,
      "loss": 0.0083,
      "step": 17638
    },
    {
      "epoch": 2.472179397337071,
      "grad_norm": 0.03561408445239067,
      "learning_rate": 5.414494140157856e-05,
      "loss": 0.002,
      "step": 17639
    },
    {
      "epoch": 2.4723195515066574,
      "grad_norm": 0.22890983521938324,
      "learning_rate": 5.413059076775891e-05,
      "loss": 0.0323,
      "step": 17640
    },
    {
      "epoch": 2.472459705676244,
      "grad_norm": 0.1588759422302246,
      "learning_rate": 5.4116240133939244e-05,
      "loss": 0.0307,
      "step": 17641
    },
    {
      "epoch": 2.4725998598458303,
      "grad_norm": 0.1341257095336914,
      "learning_rate": 5.410188950011958e-05,
      "loss": 0.0085,
      "step": 17642
    },
    {
      "epoch": 2.4727400140154168,
      "grad_norm": 0.7717081904411316,
      "learning_rate": 5.408753886629992e-05,
      "loss": 0.0716,
      "step": 17643
    },
    {
      "epoch": 2.4728801681850037,
      "grad_norm": 0.20445075631141663,
      "learning_rate": 5.4073188232480264e-05,
      "loss": 0.0345,
      "step": 17644
    },
    {
      "epoch": 2.47302032235459,
      "grad_norm": 0.12482832372188568,
      "learning_rate": 5.4058837598660606e-05,
      "loss": 0.0418,
      "step": 17645
    },
    {
      "epoch": 2.4731604765241766,
      "grad_norm": 0.09583435952663422,
      "learning_rate": 5.404448696484094e-05,
      "loss": 0.0172,
      "step": 17646
    },
    {
      "epoch": 2.473300630693763,
      "grad_norm": 0.8656417727470398,
      "learning_rate": 5.403013633102128e-05,
      "loss": 0.1083,
      "step": 17647
    },
    {
      "epoch": 2.4734407848633495,
      "grad_norm": 0.4089510142803192,
      "learning_rate": 5.4015785697201626e-05,
      "loss": 0.0417,
      "step": 17648
    },
    {
      "epoch": 2.4735809390329364,
      "grad_norm": 0.29118308424949646,
      "learning_rate": 5.400143506338196e-05,
      "loss": 0.065,
      "step": 17649
    },
    {
      "epoch": 2.473721093202523,
      "grad_norm": 0.3107341527938843,
      "learning_rate": 5.3987084429562296e-05,
      "loss": 0.0104,
      "step": 17650
    },
    {
      "epoch": 2.4738612473721093,
      "grad_norm": 0.5279232263565063,
      "learning_rate": 5.397273379574264e-05,
      "loss": 0.0311,
      "step": 17651
    },
    {
      "epoch": 2.474001401541696,
      "grad_norm": 0.13390517234802246,
      "learning_rate": 5.395838316192298e-05,
      "loss": 0.0137,
      "step": 17652
    },
    {
      "epoch": 2.4741415557112822,
      "grad_norm": 0.5480524301528931,
      "learning_rate": 5.394403252810332e-05,
      "loss": 0.1103,
      "step": 17653
    },
    {
      "epoch": 2.474281709880869,
      "grad_norm": 0.47438445687294006,
      "learning_rate": 5.392968189428366e-05,
      "loss": 0.0625,
      "step": 17654
    },
    {
      "epoch": 2.4744218640504556,
      "grad_norm": 0.2500709593296051,
      "learning_rate": 5.3915331260463993e-05,
      "loss": 0.0353,
      "step": 17655
    },
    {
      "epoch": 2.474562018220042,
      "grad_norm": 0.1397605985403061,
      "learning_rate": 5.390098062664434e-05,
      "loss": 0.0055,
      "step": 17656
    },
    {
      "epoch": 2.4747021723896285,
      "grad_norm": 0.27476930618286133,
      "learning_rate": 5.388662999282468e-05,
      "loss": 0.0647,
      "step": 17657
    },
    {
      "epoch": 2.474842326559215,
      "grad_norm": 0.3947570025920868,
      "learning_rate": 5.387227935900501e-05,
      "loss": 0.0598,
      "step": 17658
    },
    {
      "epoch": 2.4749824807288014,
      "grad_norm": 0.08438500761985779,
      "learning_rate": 5.3857928725185355e-05,
      "loss": 0.0097,
      "step": 17659
    },
    {
      "epoch": 2.4751226348983884,
      "grad_norm": 0.05265098065137863,
      "learning_rate": 5.38435780913657e-05,
      "loss": 0.0062,
      "step": 17660
    },
    {
      "epoch": 2.475262789067975,
      "grad_norm": 0.5026751756668091,
      "learning_rate": 5.382922745754604e-05,
      "loss": 0.0414,
      "step": 17661
    },
    {
      "epoch": 2.4754029432375613,
      "grad_norm": 0.4059378206729889,
      "learning_rate": 5.3814876823726375e-05,
      "loss": 0.0441,
      "step": 17662
    },
    {
      "epoch": 2.4755430974071477,
      "grad_norm": 0.09328906983137131,
      "learning_rate": 5.3800526189906724e-05,
      "loss": 0.004,
      "step": 17663
    },
    {
      "epoch": 2.4756832515767346,
      "grad_norm": 0.09670199453830719,
      "learning_rate": 5.378617555608706e-05,
      "loss": 0.0052,
      "step": 17664
    },
    {
      "epoch": 2.475823405746321,
      "grad_norm": 0.5110570192337036,
      "learning_rate": 5.3771824922267394e-05,
      "loss": 0.0458,
      "step": 17665
    },
    {
      "epoch": 2.4759635599159076,
      "grad_norm": 0.22674758732318878,
      "learning_rate": 5.375747428844773e-05,
      "loss": 0.0129,
      "step": 17666
    },
    {
      "epoch": 2.476103714085494,
      "grad_norm": 0.4692128896713257,
      "learning_rate": 5.374312365462808e-05,
      "loss": 0.0542,
      "step": 17667
    },
    {
      "epoch": 2.4762438682550805,
      "grad_norm": 0.6973477602005005,
      "learning_rate": 5.3728773020808414e-05,
      "loss": 0.0369,
      "step": 17668
    },
    {
      "epoch": 2.476384022424667,
      "grad_norm": 0.5574290752410889,
      "learning_rate": 5.3714422386988756e-05,
      "loss": 0.0109,
      "step": 17669
    },
    {
      "epoch": 2.476524176594254,
      "grad_norm": 1.2437050342559814,
      "learning_rate": 5.370007175316909e-05,
      "loss": 0.0368,
      "step": 17670
    },
    {
      "epoch": 2.4766643307638403,
      "grad_norm": 0.08141283690929413,
      "learning_rate": 5.368572111934944e-05,
      "loss": 0.0088,
      "step": 17671
    },
    {
      "epoch": 2.4768044849334268,
      "grad_norm": 0.17542196810245514,
      "learning_rate": 5.3671370485529775e-05,
      "loss": 0.0457,
      "step": 17672
    },
    {
      "epoch": 2.476944639103013,
      "grad_norm": 0.1895805150270462,
      "learning_rate": 5.365701985171011e-05,
      "loss": 0.0133,
      "step": 17673
    },
    {
      "epoch": 2.4770847932725997,
      "grad_norm": 0.13209833204746246,
      "learning_rate": 5.3642669217890446e-05,
      "loss": 0.0093,
      "step": 17674
    },
    {
      "epoch": 2.4772249474421866,
      "grad_norm": 0.3393106758594513,
      "learning_rate": 5.3628318584070795e-05,
      "loss": 0.0854,
      "step": 17675
    },
    {
      "epoch": 2.477365101611773,
      "grad_norm": 0.27093663811683655,
      "learning_rate": 5.361396795025113e-05,
      "loss": 0.0307,
      "step": 17676
    },
    {
      "epoch": 2.4775052557813595,
      "grad_norm": 0.15531690418720245,
      "learning_rate": 5.359961731643147e-05,
      "loss": 0.0171,
      "step": 17677
    },
    {
      "epoch": 2.477645409950946,
      "grad_norm": 0.531705379486084,
      "learning_rate": 5.358526668261181e-05,
      "loss": 0.0695,
      "step": 17678
    },
    {
      "epoch": 2.4777855641205324,
      "grad_norm": 0.34920862317085266,
      "learning_rate": 5.357091604879216e-05,
      "loss": 0.0275,
      "step": 17679
    },
    {
      "epoch": 2.4779257182901193,
      "grad_norm": 0.2613199055194855,
      "learning_rate": 5.355656541497249e-05,
      "loss": 0.0248,
      "step": 17680
    },
    {
      "epoch": 2.4780658724597058,
      "grad_norm": 0.15108931064605713,
      "learning_rate": 5.354221478115283e-05,
      "loss": 0.0274,
      "step": 17681
    },
    {
      "epoch": 2.4782060266292922,
      "grad_norm": 0.1552342027425766,
      "learning_rate": 5.352786414733316e-05,
      "loss": 0.0259,
      "step": 17682
    },
    {
      "epoch": 2.4783461807988787,
      "grad_norm": 0.320279985666275,
      "learning_rate": 5.351351351351351e-05,
      "loss": 0.0606,
      "step": 17683
    },
    {
      "epoch": 2.478486334968465,
      "grad_norm": 0.5443435311317444,
      "learning_rate": 5.349916287969385e-05,
      "loss": 0.0536,
      "step": 17684
    },
    {
      "epoch": 2.478626489138052,
      "grad_norm": 0.31841275095939636,
      "learning_rate": 5.348481224587419e-05,
      "loss": 0.049,
      "step": 17685
    },
    {
      "epoch": 2.4787666433076385,
      "grad_norm": 0.12419672310352325,
      "learning_rate": 5.3470461612054524e-05,
      "loss": 0.0108,
      "step": 17686
    },
    {
      "epoch": 2.478906797477225,
      "grad_norm": 0.3704558312892914,
      "learning_rate": 5.345611097823487e-05,
      "loss": 0.0406,
      "step": 17687
    },
    {
      "epoch": 2.4790469516468114,
      "grad_norm": 0.18540392816066742,
      "learning_rate": 5.344176034441521e-05,
      "loss": 0.0224,
      "step": 17688
    },
    {
      "epoch": 2.479187105816398,
      "grad_norm": 0.49803242087364197,
      "learning_rate": 5.3427409710595544e-05,
      "loss": 0.0808,
      "step": 17689
    },
    {
      "epoch": 2.4793272599859844,
      "grad_norm": 0.23828963935375214,
      "learning_rate": 5.341305907677588e-05,
      "loss": 0.0576,
      "step": 17690
    },
    {
      "epoch": 2.4794674141555713,
      "grad_norm": 0.24287208914756775,
      "learning_rate": 5.339870844295623e-05,
      "loss": 0.0731,
      "step": 17691
    },
    {
      "epoch": 2.4796075683251577,
      "grad_norm": 0.43838050961494446,
      "learning_rate": 5.3384357809136563e-05,
      "loss": 0.0264,
      "step": 17692
    },
    {
      "epoch": 2.479747722494744,
      "grad_norm": 0.13406634330749512,
      "learning_rate": 5.3370007175316906e-05,
      "loss": 0.01,
      "step": 17693
    },
    {
      "epoch": 2.4798878766643306,
      "grad_norm": 0.28545108437538147,
      "learning_rate": 5.335565654149724e-05,
      "loss": 0.0296,
      "step": 17694
    },
    {
      "epoch": 2.4800280308339175,
      "grad_norm": 0.38217201828956604,
      "learning_rate": 5.334130590767759e-05,
      "loss": 0.0609,
      "step": 17695
    },
    {
      "epoch": 2.480168185003504,
      "grad_norm": 0.2534046471118927,
      "learning_rate": 5.3326955273857925e-05,
      "loss": 0.0286,
      "step": 17696
    },
    {
      "epoch": 2.4803083391730905,
      "grad_norm": 0.2855016589164734,
      "learning_rate": 5.331260464003826e-05,
      "loss": 0.0329,
      "step": 17697
    },
    {
      "epoch": 2.480448493342677,
      "grad_norm": 0.42521926760673523,
      "learning_rate": 5.3298254006218596e-05,
      "loss": 0.0241,
      "step": 17698
    },
    {
      "epoch": 2.4805886475122634,
      "grad_norm": 0.10707966983318329,
      "learning_rate": 5.3283903372398945e-05,
      "loss": 0.0061,
      "step": 17699
    },
    {
      "epoch": 2.48072880168185,
      "grad_norm": 0.18590058386325836,
      "learning_rate": 5.326955273857928e-05,
      "loss": 0.0431,
      "step": 17700
    },
    {
      "epoch": 2.4808689558514367,
      "grad_norm": 0.3210992217063904,
      "learning_rate": 5.325520210475962e-05,
      "loss": 0.0851,
      "step": 17701
    },
    {
      "epoch": 2.481009110021023,
      "grad_norm": 0.20693227648735046,
      "learning_rate": 5.3240851470939964e-05,
      "loss": 0.0533,
      "step": 17702
    },
    {
      "epoch": 2.4811492641906097,
      "grad_norm": 0.2266017347574234,
      "learning_rate": 5.3226500837120306e-05,
      "loss": 0.0502,
      "step": 17703
    },
    {
      "epoch": 2.481289418360196,
      "grad_norm": 0.15091189742088318,
      "learning_rate": 5.321215020330064e-05,
      "loss": 0.0174,
      "step": 17704
    },
    {
      "epoch": 2.4814295725297826,
      "grad_norm": 0.3703337609767914,
      "learning_rate": 5.319779956948098e-05,
      "loss": 0.0424,
      "step": 17705
    },
    {
      "epoch": 2.4815697266993695,
      "grad_norm": 0.11492913216352463,
      "learning_rate": 5.3183448935661326e-05,
      "loss": 0.0119,
      "step": 17706
    },
    {
      "epoch": 2.481709880868956,
      "grad_norm": 0.14657877385616302,
      "learning_rate": 5.316909830184166e-05,
      "loss": 0.0093,
      "step": 17707
    },
    {
      "epoch": 2.4818500350385424,
      "grad_norm": 0.2525840699672699,
      "learning_rate": 5.3154747668022e-05,
      "loss": 0.035,
      "step": 17708
    },
    {
      "epoch": 2.481990189208129,
      "grad_norm": 0.18262861669063568,
      "learning_rate": 5.314039703420234e-05,
      "loss": 0.0216,
      "step": 17709
    },
    {
      "epoch": 2.4821303433777153,
      "grad_norm": 0.1263977289199829,
      "learning_rate": 5.312604640038268e-05,
      "loss": 0.0093,
      "step": 17710
    },
    {
      "epoch": 2.4822704975473022,
      "grad_norm": 0.3245948553085327,
      "learning_rate": 5.311169576656302e-05,
      "loss": 0.1534,
      "step": 17711
    },
    {
      "epoch": 2.4824106517168887,
      "grad_norm": 0.1624259203672409,
      "learning_rate": 5.309734513274336e-05,
      "loss": 0.0159,
      "step": 17712
    },
    {
      "epoch": 2.482550805886475,
      "grad_norm": 0.8008989095687866,
      "learning_rate": 5.3082994498923694e-05,
      "loss": 0.0603,
      "step": 17713
    },
    {
      "epoch": 2.4826909600560616,
      "grad_norm": 0.06288202852010727,
      "learning_rate": 5.306864386510404e-05,
      "loss": 0.003,
      "step": 17714
    },
    {
      "epoch": 2.482831114225648,
      "grad_norm": 1.5275949239730835,
      "learning_rate": 5.305429323128438e-05,
      "loss": 0.0289,
      "step": 17715
    },
    {
      "epoch": 2.482971268395235,
      "grad_norm": 0.06765114516019821,
      "learning_rate": 5.303994259746471e-05,
      "loss": 0.0037,
      "step": 17716
    },
    {
      "epoch": 2.4831114225648214,
      "grad_norm": 1.492302417755127,
      "learning_rate": 5.3025591963645055e-05,
      "loss": 0.0832,
      "step": 17717
    },
    {
      "epoch": 2.483251576734408,
      "grad_norm": 0.49175533652305603,
      "learning_rate": 5.30112413298254e-05,
      "loss": 0.0294,
      "step": 17718
    },
    {
      "epoch": 2.4833917309039943,
      "grad_norm": 0.13394728302955627,
      "learning_rate": 5.299689069600574e-05,
      "loss": 0.0057,
      "step": 17719
    },
    {
      "epoch": 2.483531885073581,
      "grad_norm": 0.11188614368438721,
      "learning_rate": 5.2982540062186075e-05,
      "loss": 0.0076,
      "step": 17720
    },
    {
      "epoch": 2.4836720392431673,
      "grad_norm": 0.12141694128513336,
      "learning_rate": 5.296818942836641e-05,
      "loss": 0.0126,
      "step": 17721
    },
    {
      "epoch": 2.483812193412754,
      "grad_norm": 0.2904088795185089,
      "learning_rate": 5.295383879454676e-05,
      "loss": 0.0125,
      "step": 17722
    },
    {
      "epoch": 2.4839523475823406,
      "grad_norm": 0.16423878073692322,
      "learning_rate": 5.2939488160727094e-05,
      "loss": 0.0288,
      "step": 17723
    },
    {
      "epoch": 2.484092501751927,
      "grad_norm": 0.20900902152061462,
      "learning_rate": 5.292513752690743e-05,
      "loss": 0.0248,
      "step": 17724
    },
    {
      "epoch": 2.4842326559215135,
      "grad_norm": 0.08876301348209381,
      "learning_rate": 5.291078689308777e-05,
      "loss": 0.0119,
      "step": 17725
    },
    {
      "epoch": 2.4843728100911004,
      "grad_norm": 0.2671598196029663,
      "learning_rate": 5.2896436259268114e-05,
      "loss": 0.0315,
      "step": 17726
    },
    {
      "epoch": 2.484512964260687,
      "grad_norm": 0.18206124007701874,
      "learning_rate": 5.2882085625448456e-05,
      "loss": 0.0104,
      "step": 17727
    },
    {
      "epoch": 2.4846531184302734,
      "grad_norm": 0.06854202598333359,
      "learning_rate": 5.286773499162879e-05,
      "loss": 0.0057,
      "step": 17728
    },
    {
      "epoch": 2.48479327259986,
      "grad_norm": 0.4192564785480499,
      "learning_rate": 5.285338435780913e-05,
      "loss": 0.0278,
      "step": 17729
    },
    {
      "epoch": 2.4849334267694463,
      "grad_norm": 0.18883728981018066,
      "learning_rate": 5.2839033723989476e-05,
      "loss": 0.0073,
      "step": 17730
    },
    {
      "epoch": 2.4850735809390327,
      "grad_norm": 0.2082323282957077,
      "learning_rate": 5.282468309016981e-05,
      "loss": 0.0676,
      "step": 17731
    },
    {
      "epoch": 2.4852137351086196,
      "grad_norm": 0.48897141218185425,
      "learning_rate": 5.2810332456350146e-05,
      "loss": 0.0222,
      "step": 17732
    },
    {
      "epoch": 2.485353889278206,
      "grad_norm": 0.2984616458415985,
      "learning_rate": 5.279598182253049e-05,
      "loss": 0.0338,
      "step": 17733
    },
    {
      "epoch": 2.4854940434477926,
      "grad_norm": 0.1615888625383377,
      "learning_rate": 5.278163118871083e-05,
      "loss": 0.0184,
      "step": 17734
    },
    {
      "epoch": 2.485634197617379,
      "grad_norm": 0.5336517095565796,
      "learning_rate": 5.276728055489117e-05,
      "loss": 0.0573,
      "step": 17735
    },
    {
      "epoch": 2.4857743517869655,
      "grad_norm": 0.13963228464126587,
      "learning_rate": 5.275292992107151e-05,
      "loss": 0.0261,
      "step": 17736
    },
    {
      "epoch": 2.4859145059565524,
      "grad_norm": 0.3494892120361328,
      "learning_rate": 5.273857928725185e-05,
      "loss": 0.0749,
      "step": 17737
    },
    {
      "epoch": 2.486054660126139,
      "grad_norm": 0.13461467623710632,
      "learning_rate": 5.272422865343219e-05,
      "loss": 0.0203,
      "step": 17738
    },
    {
      "epoch": 2.4861948142957253,
      "grad_norm": 0.47826001048088074,
      "learning_rate": 5.270987801961253e-05,
      "loss": 0.0671,
      "step": 17739
    },
    {
      "epoch": 2.4863349684653118,
      "grad_norm": 0.34306755661964417,
      "learning_rate": 5.269552738579286e-05,
      "loss": 0.0623,
      "step": 17740
    },
    {
      "epoch": 2.4864751226348982,
      "grad_norm": 0.5603805184364319,
      "learning_rate": 5.268117675197321e-05,
      "loss": 0.0388,
      "step": 17741
    },
    {
      "epoch": 2.4866152768044847,
      "grad_norm": 0.4433344900608063,
      "learning_rate": 5.266682611815355e-05,
      "loss": 0.0183,
      "step": 17742
    },
    {
      "epoch": 2.4867554309740716,
      "grad_norm": 0.486017644405365,
      "learning_rate": 5.265247548433389e-05,
      "loss": 0.0395,
      "step": 17743
    },
    {
      "epoch": 2.486895585143658,
      "grad_norm": 0.16959120333194733,
      "learning_rate": 5.2638124850514225e-05,
      "loss": 0.0187,
      "step": 17744
    },
    {
      "epoch": 2.4870357393132445,
      "grad_norm": 0.2675526738166809,
      "learning_rate": 5.262377421669457e-05,
      "loss": 0.0483,
      "step": 17745
    },
    {
      "epoch": 2.487175893482831,
      "grad_norm": 0.2669757306575775,
      "learning_rate": 5.260942358287491e-05,
      "loss": 0.0316,
      "step": 17746
    },
    {
      "epoch": 2.487316047652418,
      "grad_norm": 0.12552404403686523,
      "learning_rate": 5.2595072949055244e-05,
      "loss": 0.0118,
      "step": 17747
    },
    {
      "epoch": 2.4874562018220043,
      "grad_norm": 0.2678147256374359,
      "learning_rate": 5.258072231523558e-05,
      "loss": 0.0158,
      "step": 17748
    },
    {
      "epoch": 2.487596355991591,
      "grad_norm": 0.11293377727270126,
      "learning_rate": 5.256637168141593e-05,
      "loss": 0.0195,
      "step": 17749
    },
    {
      "epoch": 2.4877365101611773,
      "grad_norm": 0.2692854106426239,
      "learning_rate": 5.2552021047596264e-05,
      "loss": 0.0182,
      "step": 17750
    },
    {
      "epoch": 2.4878766643307637,
      "grad_norm": 0.23372231423854828,
      "learning_rate": 5.2537670413776606e-05,
      "loss": 0.0176,
      "step": 17751
    },
    {
      "epoch": 2.48801681850035,
      "grad_norm": 0.8065145611763,
      "learning_rate": 5.252331977995694e-05,
      "loss": 0.0388,
      "step": 17752
    },
    {
      "epoch": 2.488156972669937,
      "grad_norm": 0.4757090210914612,
      "learning_rate": 5.250896914613728e-05,
      "loss": 0.022,
      "step": 17753
    },
    {
      "epoch": 2.4882971268395235,
      "grad_norm": 0.09404882788658142,
      "learning_rate": 5.2494618512317625e-05,
      "loss": 0.0174,
      "step": 17754
    },
    {
      "epoch": 2.48843728100911,
      "grad_norm": 0.21451681852340698,
      "learning_rate": 5.248026787849796e-05,
      "loss": 0.0332,
      "step": 17755
    },
    {
      "epoch": 2.4885774351786965,
      "grad_norm": 0.23228569328784943,
      "learning_rate": 5.2465917244678296e-05,
      "loss": 0.0358,
      "step": 17756
    },
    {
      "epoch": 2.4887175893482834,
      "grad_norm": 0.11712727695703506,
      "learning_rate": 5.2451566610858645e-05,
      "loss": 0.0068,
      "step": 17757
    },
    {
      "epoch": 2.48885774351787,
      "grad_norm": 0.360446572303772,
      "learning_rate": 5.243721597703898e-05,
      "loss": 0.0313,
      "step": 17758
    },
    {
      "epoch": 2.4889978976874563,
      "grad_norm": 0.20387250185012817,
      "learning_rate": 5.242286534321932e-05,
      "loss": 0.0262,
      "step": 17759
    },
    {
      "epoch": 2.4891380518570427,
      "grad_norm": 0.3580976724624634,
      "learning_rate": 5.240851470939966e-05,
      "loss": 0.0592,
      "step": 17760
    },
    {
      "epoch": 2.489278206026629,
      "grad_norm": 0.13282306492328644,
      "learning_rate": 5.239416407558e-05,
      "loss": 0.0361,
      "step": 17761
    },
    {
      "epoch": 2.4894183601962157,
      "grad_norm": 0.6758518815040588,
      "learning_rate": 5.237981344176034e-05,
      "loss": 0.1175,
      "step": 17762
    },
    {
      "epoch": 2.4895585143658026,
      "grad_norm": 0.40552985668182373,
      "learning_rate": 5.236546280794068e-05,
      "loss": 0.0431,
      "step": 17763
    },
    {
      "epoch": 2.489698668535389,
      "grad_norm": 0.19911421835422516,
      "learning_rate": 5.235111217412101e-05,
      "loss": 0.0135,
      "step": 17764
    },
    {
      "epoch": 2.4898388227049755,
      "grad_norm": 0.6081478595733643,
      "learning_rate": 5.233676154030136e-05,
      "loss": 0.0935,
      "step": 17765
    },
    {
      "epoch": 2.489978976874562,
      "grad_norm": 1.7193678617477417,
      "learning_rate": 5.23224109064817e-05,
      "loss": 0.0933,
      "step": 17766
    },
    {
      "epoch": 2.4901191310441484,
      "grad_norm": 0.49033698439598083,
      "learning_rate": 5.230806027266204e-05,
      "loss": 0.103,
      "step": 17767
    },
    {
      "epoch": 2.4902592852137353,
      "grad_norm": 0.5711290836334229,
      "learning_rate": 5.2293709638842374e-05,
      "loss": 0.0755,
      "step": 17768
    },
    {
      "epoch": 2.4903994393833218,
      "grad_norm": 1.0722718238830566,
      "learning_rate": 5.2279359005022716e-05,
      "loss": 0.037,
      "step": 17769
    },
    {
      "epoch": 2.490539593552908,
      "grad_norm": 0.3700416088104248,
      "learning_rate": 5.226500837120306e-05,
      "loss": 0.0461,
      "step": 17770
    },
    {
      "epoch": 2.4906797477224947,
      "grad_norm": 0.15610124170780182,
      "learning_rate": 5.2250657737383394e-05,
      "loss": 0.0167,
      "step": 17771
    },
    {
      "epoch": 2.490819901892081,
      "grad_norm": 0.229808047413826,
      "learning_rate": 5.223630710356373e-05,
      "loss": 0.0161,
      "step": 17772
    },
    {
      "epoch": 2.4909600560616676,
      "grad_norm": 0.12942081689834595,
      "learning_rate": 5.222195646974408e-05,
      "loss": 0.0204,
      "step": 17773
    },
    {
      "epoch": 2.4911002102312545,
      "grad_norm": 0.43906357884407043,
      "learning_rate": 5.2207605835924413e-05,
      "loss": 0.0271,
      "step": 17774
    },
    {
      "epoch": 2.491240364400841,
      "grad_norm": 0.4648025631904602,
      "learning_rate": 5.2193255202104756e-05,
      "loss": 0.0575,
      "step": 17775
    },
    {
      "epoch": 2.4913805185704274,
      "grad_norm": 0.1541581153869629,
      "learning_rate": 5.21789045682851e-05,
      "loss": 0.0276,
      "step": 17776
    },
    {
      "epoch": 2.491520672740014,
      "grad_norm": 0.5011047720909119,
      "learning_rate": 5.216455393446543e-05,
      "loss": 0.0301,
      "step": 17777
    },
    {
      "epoch": 2.491660826909601,
      "grad_norm": 0.18152040243148804,
      "learning_rate": 5.2150203300645775e-05,
      "loss": 0.0147,
      "step": 17778
    },
    {
      "epoch": 2.4918009810791872,
      "grad_norm": 0.07357105612754822,
      "learning_rate": 5.213585266682611e-05,
      "loss": 0.0076,
      "step": 17779
    },
    {
      "epoch": 2.4919411352487737,
      "grad_norm": 0.33727243542671204,
      "learning_rate": 5.212150203300646e-05,
      "loss": 0.0526,
      "step": 17780
    },
    {
      "epoch": 2.49208128941836,
      "grad_norm": 0.35229700803756714,
      "learning_rate": 5.2107151399186795e-05,
      "loss": 0.0462,
      "step": 17781
    },
    {
      "epoch": 2.4922214435879466,
      "grad_norm": 0.18100976943969727,
      "learning_rate": 5.209280076536713e-05,
      "loss": 0.0182,
      "step": 17782
    },
    {
      "epoch": 2.492361597757533,
      "grad_norm": 0.03994106501340866,
      "learning_rate": 5.207845013154747e-05,
      "loss": 0.0124,
      "step": 17783
    },
    {
      "epoch": 2.49250175192712,
      "grad_norm": 0.05880053713917732,
      "learning_rate": 5.2064099497727814e-05,
      "loss": 0.0049,
      "step": 17784
    },
    {
      "epoch": 2.4926419060967064,
      "grad_norm": 0.28882768750190735,
      "learning_rate": 5.204974886390815e-05,
      "loss": 0.0232,
      "step": 17785
    },
    {
      "epoch": 2.492782060266293,
      "grad_norm": 0.705376148223877,
      "learning_rate": 5.203539823008849e-05,
      "loss": 0.0876,
      "step": 17786
    },
    {
      "epoch": 2.4929222144358794,
      "grad_norm": 0.21390913426876068,
      "learning_rate": 5.202104759626883e-05,
      "loss": 0.0288,
      "step": 17787
    },
    {
      "epoch": 2.4930623686054663,
      "grad_norm": 0.29915696382522583,
      "learning_rate": 5.2006696962449176e-05,
      "loss": 0.0275,
      "step": 17788
    },
    {
      "epoch": 2.4932025227750527,
      "grad_norm": 0.30241236090660095,
      "learning_rate": 5.199234632862951e-05,
      "loss": 0.0409,
      "step": 17789
    },
    {
      "epoch": 2.493342676944639,
      "grad_norm": 0.09223033487796783,
      "learning_rate": 5.1977995694809847e-05,
      "loss": 0.0081,
      "step": 17790
    },
    {
      "epoch": 2.4934828311142256,
      "grad_norm": 0.24814589321613312,
      "learning_rate": 5.196364506099019e-05,
      "loss": 0.0375,
      "step": 17791
    },
    {
      "epoch": 2.493622985283812,
      "grad_norm": 0.16973090171813965,
      "learning_rate": 5.194929442717053e-05,
      "loss": 0.0295,
      "step": 17792
    },
    {
      "epoch": 2.4937631394533986,
      "grad_norm": 0.5886319875717163,
      "learning_rate": 5.1934943793350866e-05,
      "loss": 0.0561,
      "step": 17793
    },
    {
      "epoch": 2.4939032936229855,
      "grad_norm": 0.08989480882883072,
      "learning_rate": 5.192059315953121e-05,
      "loss": 0.0144,
      "step": 17794
    },
    {
      "epoch": 2.494043447792572,
      "grad_norm": 0.345582515001297,
      "learning_rate": 5.1906242525711544e-05,
      "loss": 0.0267,
      "step": 17795
    },
    {
      "epoch": 2.4941836019621584,
      "grad_norm": 0.1257106363773346,
      "learning_rate": 5.189189189189189e-05,
      "loss": 0.0133,
      "step": 17796
    },
    {
      "epoch": 2.494323756131745,
      "grad_norm": 0.583721935749054,
      "learning_rate": 5.187754125807223e-05,
      "loss": 0.0905,
      "step": 17797
    },
    {
      "epoch": 2.4944639103013313,
      "grad_norm": 0.09211540967226028,
      "learning_rate": 5.186319062425256e-05,
      "loss": 0.0064,
      "step": 17798
    },
    {
      "epoch": 2.494604064470918,
      "grad_norm": 0.1583651602268219,
      "learning_rate": 5.1848839990432905e-05,
      "loss": 0.0267,
      "step": 17799
    },
    {
      "epoch": 2.4947442186405047,
      "grad_norm": 0.1800818145275116,
      "learning_rate": 5.183448935661325e-05,
      "loss": 0.0237,
      "step": 17800
    },
    {
      "epoch": 2.494884372810091,
      "grad_norm": 0.19386602938175201,
      "learning_rate": 5.182013872279358e-05,
      "loss": 0.0147,
      "step": 17801
    },
    {
      "epoch": 2.4950245269796776,
      "grad_norm": 0.5707101821899414,
      "learning_rate": 5.1805788088973925e-05,
      "loss": 0.0198,
      "step": 17802
    },
    {
      "epoch": 2.495164681149264,
      "grad_norm": 0.2839837670326233,
      "learning_rate": 5.179143745515426e-05,
      "loss": 0.0081,
      "step": 17803
    },
    {
      "epoch": 2.4953048353188505,
      "grad_norm": 0.27072033286094666,
      "learning_rate": 5.177708682133461e-05,
      "loss": 0.0387,
      "step": 17804
    },
    {
      "epoch": 2.4954449894884374,
      "grad_norm": 0.1146271824836731,
      "learning_rate": 5.1762736187514944e-05,
      "loss": 0.0185,
      "step": 17805
    },
    {
      "epoch": 2.495585143658024,
      "grad_norm": 0.09938567131757736,
      "learning_rate": 5.174838555369528e-05,
      "loss": 0.0191,
      "step": 17806
    },
    {
      "epoch": 2.4957252978276103,
      "grad_norm": 0.302967369556427,
      "learning_rate": 5.173403491987562e-05,
      "loss": 0.0154,
      "step": 17807
    },
    {
      "epoch": 2.495865451997197,
      "grad_norm": 0.38114574551582336,
      "learning_rate": 5.1719684286055964e-05,
      "loss": 0.0298,
      "step": 17808
    },
    {
      "epoch": 2.4960056061667837,
      "grad_norm": 0.2538285553455353,
      "learning_rate": 5.17053336522363e-05,
      "loss": 0.0375,
      "step": 17809
    },
    {
      "epoch": 2.49614576033637,
      "grad_norm": 0.48061642050743103,
      "learning_rate": 5.169098301841664e-05,
      "loss": 0.0762,
      "step": 17810
    },
    {
      "epoch": 2.4962859145059566,
      "grad_norm": 0.08205193281173706,
      "learning_rate": 5.1676632384596984e-05,
      "loss": 0.0053,
      "step": 17811
    },
    {
      "epoch": 2.496426068675543,
      "grad_norm": 0.23486006259918213,
      "learning_rate": 5.1662281750777326e-05,
      "loss": 0.0045,
      "step": 17812
    },
    {
      "epoch": 2.4965662228451295,
      "grad_norm": 0.20154549181461334,
      "learning_rate": 5.164793111695766e-05,
      "loss": 0.0252,
      "step": 17813
    },
    {
      "epoch": 2.496706377014716,
      "grad_norm": 0.5714675188064575,
      "learning_rate": 5.1633580483137996e-05,
      "loss": 0.0272,
      "step": 17814
    },
    {
      "epoch": 2.496846531184303,
      "grad_norm": 0.399156779050827,
      "learning_rate": 5.1619229849318345e-05,
      "loss": 0.0265,
      "step": 17815
    },
    {
      "epoch": 2.4969866853538893,
      "grad_norm": 0.4697190821170807,
      "learning_rate": 5.160487921549868e-05,
      "loss": 0.0449,
      "step": 17816
    },
    {
      "epoch": 2.497126839523476,
      "grad_norm": 0.6566028594970703,
      "learning_rate": 5.159052858167902e-05,
      "loss": 0.053,
      "step": 17817
    },
    {
      "epoch": 2.4972669936930623,
      "grad_norm": 3.4883246421813965,
      "learning_rate": 5.157617794785936e-05,
      "loss": 0.1284,
      "step": 17818
    },
    {
      "epoch": 2.4974071478626487,
      "grad_norm": 0.5223955512046814,
      "learning_rate": 5.15618273140397e-05,
      "loss": 0.0453,
      "step": 17819
    },
    {
      "epoch": 2.4975473020322356,
      "grad_norm": 0.3521726727485657,
      "learning_rate": 5.154747668022004e-05,
      "loss": 0.0173,
      "step": 17820
    },
    {
      "epoch": 2.497687456201822,
      "grad_norm": 0.1914532631635666,
      "learning_rate": 5.153312604640038e-05,
      "loss": 0.0119,
      "step": 17821
    },
    {
      "epoch": 2.4978276103714085,
      "grad_norm": 0.17097818851470947,
      "learning_rate": 5.151877541258071e-05,
      "loss": 0.0392,
      "step": 17822
    },
    {
      "epoch": 2.497967764540995,
      "grad_norm": 0.0675303190946579,
      "learning_rate": 5.150442477876106e-05,
      "loss": 0.0053,
      "step": 17823
    },
    {
      "epoch": 2.4981079187105815,
      "grad_norm": 0.231094628572464,
      "learning_rate": 5.14900741449414e-05,
      "loss": 0.0227,
      "step": 17824
    },
    {
      "epoch": 2.4982480728801684,
      "grad_norm": 0.07156160473823547,
      "learning_rate": 5.147572351112174e-05,
      "loss": 0.0112,
      "step": 17825
    },
    {
      "epoch": 2.498388227049755,
      "grad_norm": 0.13417445123195648,
      "learning_rate": 5.1461372877302075e-05,
      "loss": 0.013,
      "step": 17826
    },
    {
      "epoch": 2.4985283812193413,
      "grad_norm": 0.22243309020996094,
      "learning_rate": 5.144702224348242e-05,
      "loss": 0.0225,
      "step": 17827
    },
    {
      "epoch": 2.4986685353889277,
      "grad_norm": 0.10465124249458313,
      "learning_rate": 5.143267160966276e-05,
      "loss": 0.0153,
      "step": 17828
    },
    {
      "epoch": 2.498808689558514,
      "grad_norm": 0.32272014021873474,
      "learning_rate": 5.1418320975843094e-05,
      "loss": 0.0404,
      "step": 17829
    },
    {
      "epoch": 2.498948843728101,
      "grad_norm": 0.08015695959329605,
      "learning_rate": 5.140397034202343e-05,
      "loss": 0.0088,
      "step": 17830
    },
    {
      "epoch": 2.4990889978976876,
      "grad_norm": 0.2525484561920166,
      "learning_rate": 5.138961970820378e-05,
      "loss": 0.0129,
      "step": 17831
    },
    {
      "epoch": 2.499229152067274,
      "grad_norm": 0.10982628166675568,
      "learning_rate": 5.1375269074384114e-05,
      "loss": 0.0237,
      "step": 17832
    },
    {
      "epoch": 2.4993693062368605,
      "grad_norm": 0.10212475061416626,
      "learning_rate": 5.1360918440564456e-05,
      "loss": 0.0216,
      "step": 17833
    },
    {
      "epoch": 2.499509460406447,
      "grad_norm": 0.04232310131192207,
      "learning_rate": 5.134656780674479e-05,
      "loss": 0.0035,
      "step": 17834
    },
    {
      "epoch": 2.4996496145760334,
      "grad_norm": 0.15495185554027557,
      "learning_rate": 5.133221717292513e-05,
      "loss": 0.0132,
      "step": 17835
    },
    {
      "epoch": 2.4997897687456203,
      "grad_norm": 0.09340164065361023,
      "learning_rate": 5.1317866539105475e-05,
      "loss": 0.0087,
      "step": 17836
    },
    {
      "epoch": 2.4999299229152068,
      "grad_norm": 0.23512424528598785,
      "learning_rate": 5.130351590528581e-05,
      "loss": 0.0423,
      "step": 17837
    },
    {
      "epoch": 2.5000700770847932,
      "grad_norm": 0.06448642164468765,
      "learning_rate": 5.1289165271466146e-05,
      "loss": 0.0094,
      "step": 17838
    },
    {
      "epoch": 2.5002102312543797,
      "grad_norm": 0.5939103960990906,
      "learning_rate": 5.1274814637646495e-05,
      "loss": 0.0179,
      "step": 17839
    },
    {
      "epoch": 2.5003503854239666,
      "grad_norm": 0.358553409576416,
      "learning_rate": 5.126046400382683e-05,
      "loss": 0.0102,
      "step": 17840
    },
    {
      "epoch": 2.500490539593553,
      "grad_norm": 0.13601501286029816,
      "learning_rate": 5.124611337000717e-05,
      "loss": 0.0067,
      "step": 17841
    },
    {
      "epoch": 2.5006306937631395,
      "grad_norm": 0.2634809613227844,
      "learning_rate": 5.123176273618751e-05,
      "loss": 0.0724,
      "step": 17842
    },
    {
      "epoch": 2.500770847932726,
      "grad_norm": 0.1829523891210556,
      "learning_rate": 5.121741210236785e-05,
      "loss": 0.0123,
      "step": 17843
    },
    {
      "epoch": 2.5009110021023124,
      "grad_norm": 0.3545670509338379,
      "learning_rate": 5.120306146854819e-05,
      "loss": 0.0115,
      "step": 17844
    },
    {
      "epoch": 2.501051156271899,
      "grad_norm": 0.0709671676158905,
      "learning_rate": 5.118871083472853e-05,
      "loss": 0.0068,
      "step": 17845
    },
    {
      "epoch": 2.501191310441486,
      "grad_norm": 0.4570164084434509,
      "learning_rate": 5.117436020090886e-05,
      "loss": 0.0406,
      "step": 17846
    },
    {
      "epoch": 2.5013314646110723,
      "grad_norm": 0.08937258273363113,
      "learning_rate": 5.116000956708921e-05,
      "loss": 0.0067,
      "step": 17847
    },
    {
      "epoch": 2.5014716187806587,
      "grad_norm": 0.0228976309299469,
      "learning_rate": 5.114565893326955e-05,
      "loss": 0.0024,
      "step": 17848
    },
    {
      "epoch": 2.501611772950245,
      "grad_norm": 0.4653420150279999,
      "learning_rate": 5.113130829944989e-05,
      "loss": 0.0248,
      "step": 17849
    },
    {
      "epoch": 2.501751927119832,
      "grad_norm": 0.36147162318229675,
      "learning_rate": 5.111695766563023e-05,
      "loss": 0.0326,
      "step": 17850
    },
    {
      "epoch": 2.5018920812894185,
      "grad_norm": 0.17428427934646606,
      "learning_rate": 5.1102607031810566e-05,
      "loss": 0.0192,
      "step": 17851
    },
    {
      "epoch": 2.502032235459005,
      "grad_norm": 0.3164307177066803,
      "learning_rate": 5.108825639799091e-05,
      "loss": 0.0234,
      "step": 17852
    },
    {
      "epoch": 2.5021723896285915,
      "grad_norm": 0.094609834253788,
      "learning_rate": 5.1073905764171244e-05,
      "loss": 0.0161,
      "step": 17853
    },
    {
      "epoch": 2.502312543798178,
      "grad_norm": 0.13780805468559265,
      "learning_rate": 5.105955513035159e-05,
      "loss": 0.0204,
      "step": 17854
    },
    {
      "epoch": 2.5024526979677644,
      "grad_norm": 0.14991626143455505,
      "learning_rate": 5.104520449653193e-05,
      "loss": 0.0168,
      "step": 17855
    },
    {
      "epoch": 2.502592852137351,
      "grad_norm": 0.33201947808265686,
      "learning_rate": 5.103085386271226e-05,
      "loss": 0.0066,
      "step": 17856
    },
    {
      "epoch": 2.5027330063069377,
      "grad_norm": 0.22748306393623352,
      "learning_rate": 5.1016503228892605e-05,
      "loss": 0.0656,
      "step": 17857
    },
    {
      "epoch": 2.502873160476524,
      "grad_norm": 0.15124912559986115,
      "learning_rate": 5.100215259507295e-05,
      "loss": 0.0152,
      "step": 17858
    },
    {
      "epoch": 2.5030133146461107,
      "grad_norm": 0.10268676280975342,
      "learning_rate": 5.098780196125328e-05,
      "loss": 0.0114,
      "step": 17859
    },
    {
      "epoch": 2.503153468815697,
      "grad_norm": 0.1801644265651703,
      "learning_rate": 5.0973451327433625e-05,
      "loss": 0.0282,
      "step": 17860
    },
    {
      "epoch": 2.503293622985284,
      "grad_norm": 0.17832498252391815,
      "learning_rate": 5.095910069361396e-05,
      "loss": 0.0061,
      "step": 17861
    },
    {
      "epoch": 2.5034337771548705,
      "grad_norm": 0.1257418543100357,
      "learning_rate": 5.094475005979431e-05,
      "loss": 0.0054,
      "step": 17862
    },
    {
      "epoch": 2.503573931324457,
      "grad_norm": 0.10041609406471252,
      "learning_rate": 5.0930399425974645e-05,
      "loss": 0.0028,
      "step": 17863
    },
    {
      "epoch": 2.5037140854940434,
      "grad_norm": 0.13054272532463074,
      "learning_rate": 5.091604879215498e-05,
      "loss": 0.0083,
      "step": 17864
    },
    {
      "epoch": 2.50385423966363,
      "grad_norm": 0.4925798177719116,
      "learning_rate": 5.090169815833532e-05,
      "loss": 0.1121,
      "step": 17865
    },
    {
      "epoch": 2.5039943938332163,
      "grad_norm": 0.5421183109283447,
      "learning_rate": 5.0887347524515664e-05,
      "loss": 0.0267,
      "step": 17866
    },
    {
      "epoch": 2.504134548002803,
      "grad_norm": 0.1838768720626831,
      "learning_rate": 5.0872996890696e-05,
      "loss": 0.0243,
      "step": 17867
    },
    {
      "epoch": 2.5042747021723897,
      "grad_norm": 0.565723180770874,
      "learning_rate": 5.085864625687634e-05,
      "loss": 0.1334,
      "step": 17868
    },
    {
      "epoch": 2.504414856341976,
      "grad_norm": 0.41988706588745117,
      "learning_rate": 5.084429562305668e-05,
      "loss": 0.018,
      "step": 17869
    },
    {
      "epoch": 2.5045550105115626,
      "grad_norm": 0.4841556251049042,
      "learning_rate": 5.0829944989237026e-05,
      "loss": 0.0718,
      "step": 17870
    },
    {
      "epoch": 2.5046951646811495,
      "grad_norm": 0.2065269649028778,
      "learning_rate": 5.081559435541736e-05,
      "loss": 0.0505,
      "step": 17871
    },
    {
      "epoch": 2.504835318850736,
      "grad_norm": 0.2573491632938385,
      "learning_rate": 5.0801243721597697e-05,
      "loss": 0.0093,
      "step": 17872
    },
    {
      "epoch": 2.5049754730203224,
      "grad_norm": 0.5838304162025452,
      "learning_rate": 5.078689308777804e-05,
      "loss": 0.033,
      "step": 17873
    },
    {
      "epoch": 2.505115627189909,
      "grad_norm": 0.19565308094024658,
      "learning_rate": 5.077254245395838e-05,
      "loss": 0.0289,
      "step": 17874
    },
    {
      "epoch": 2.5052557813594953,
      "grad_norm": 0.18960800766944885,
      "learning_rate": 5.0758191820138716e-05,
      "loss": 0.0407,
      "step": 17875
    },
    {
      "epoch": 2.505395935529082,
      "grad_norm": 0.43577390909194946,
      "learning_rate": 5.074384118631906e-05,
      "loss": 0.0325,
      "step": 17876
    },
    {
      "epoch": 2.5055360896986687,
      "grad_norm": 0.2558959424495697,
      "learning_rate": 5.0729490552499394e-05,
      "loss": 0.0124,
      "step": 17877
    },
    {
      "epoch": 2.505676243868255,
      "grad_norm": 0.2798151969909668,
      "learning_rate": 5.071513991867974e-05,
      "loss": 0.0419,
      "step": 17878
    },
    {
      "epoch": 2.5058163980378416,
      "grad_norm": 0.17561772465705872,
      "learning_rate": 5.070078928486008e-05,
      "loss": 0.0362,
      "step": 17879
    },
    {
      "epoch": 2.505956552207428,
      "grad_norm": 0.27153992652893066,
      "learning_rate": 5.068643865104041e-05,
      "loss": 0.0448,
      "step": 17880
    },
    {
      "epoch": 2.506096706377015,
      "grad_norm": 0.5290229320526123,
      "learning_rate": 5.0672088017220755e-05,
      "loss": 0.0669,
      "step": 17881
    },
    {
      "epoch": 2.5062368605466014,
      "grad_norm": 0.13243520259857178,
      "learning_rate": 5.06577373834011e-05,
      "loss": 0.0109,
      "step": 17882
    },
    {
      "epoch": 2.506377014716188,
      "grad_norm": 0.15251211822032928,
      "learning_rate": 5.064338674958143e-05,
      "loss": 0.0376,
      "step": 17883
    },
    {
      "epoch": 2.5065171688857744,
      "grad_norm": 0.11372412741184235,
      "learning_rate": 5.0629036115761775e-05,
      "loss": 0.0141,
      "step": 17884
    },
    {
      "epoch": 2.506657323055361,
      "grad_norm": 0.12586988508701324,
      "learning_rate": 5.061468548194212e-05,
      "loss": 0.034,
      "step": 17885
    },
    {
      "epoch": 2.5067974772249473,
      "grad_norm": 0.1825370043516159,
      "learning_rate": 5.060033484812246e-05,
      "loss": 0.0083,
      "step": 17886
    },
    {
      "epoch": 2.5069376313945337,
      "grad_norm": 0.25620436668395996,
      "learning_rate": 5.0585984214302794e-05,
      "loss": 0.0312,
      "step": 17887
    },
    {
      "epoch": 2.5070777855641206,
      "grad_norm": 0.1578015834093094,
      "learning_rate": 5.057163358048313e-05,
      "loss": 0.0129,
      "step": 17888
    },
    {
      "epoch": 2.507217939733707,
      "grad_norm": 0.12077666074037552,
      "learning_rate": 5.055728294666348e-05,
      "loss": 0.0148,
      "step": 17889
    },
    {
      "epoch": 2.5073580939032936,
      "grad_norm": 0.22039912641048431,
      "learning_rate": 5.0542932312843814e-05,
      "loss": 0.0269,
      "step": 17890
    },
    {
      "epoch": 2.50749824807288,
      "grad_norm": 0.05985488370060921,
      "learning_rate": 5.052858167902415e-05,
      "loss": 0.0039,
      "step": 17891
    },
    {
      "epoch": 2.507638402242467,
      "grad_norm": 0.26816192269325256,
      "learning_rate": 5.051423104520449e-05,
      "loss": 0.0115,
      "step": 17892
    },
    {
      "epoch": 2.5077785564120534,
      "grad_norm": 0.0780048593878746,
      "learning_rate": 5.0499880411384833e-05,
      "loss": 0.0064,
      "step": 17893
    },
    {
      "epoch": 2.50791871058164,
      "grad_norm": 0.14516665041446686,
      "learning_rate": 5.0485529777565176e-05,
      "loss": 0.0298,
      "step": 17894
    },
    {
      "epoch": 2.5080588647512263,
      "grad_norm": 0.05282115563750267,
      "learning_rate": 5.047117914374551e-05,
      "loss": 0.0042,
      "step": 17895
    },
    {
      "epoch": 2.5081990189208128,
      "grad_norm": 0.2411554604768753,
      "learning_rate": 5.0456828509925846e-05,
      "loss": 0.0554,
      "step": 17896
    },
    {
      "epoch": 2.508339173090399,
      "grad_norm": 0.05972570553421974,
      "learning_rate": 5.0442477876106195e-05,
      "loss": 0.0068,
      "step": 17897
    },
    {
      "epoch": 2.508479327259986,
      "grad_norm": 0.10814305394887924,
      "learning_rate": 5.042812724228653e-05,
      "loss": 0.0086,
      "step": 17898
    },
    {
      "epoch": 2.5086194814295726,
      "grad_norm": 0.38663139939308167,
      "learning_rate": 5.0413776608466866e-05,
      "loss": 0.0209,
      "step": 17899
    },
    {
      "epoch": 2.508759635599159,
      "grad_norm": 0.05171884223818779,
      "learning_rate": 5.039942597464721e-05,
      "loss": 0.0022,
      "step": 17900
    },
    {
      "epoch": 2.5088997897687455,
      "grad_norm": 0.2223225086927414,
      "learning_rate": 5.038507534082755e-05,
      "loss": 0.0345,
      "step": 17901
    },
    {
      "epoch": 2.5090399439383324,
      "grad_norm": 0.12274236232042313,
      "learning_rate": 5.037072470700789e-05,
      "loss": 0.0055,
      "step": 17902
    },
    {
      "epoch": 2.509180098107919,
      "grad_norm": 0.466958224773407,
      "learning_rate": 5.035637407318823e-05,
      "loss": 0.0068,
      "step": 17903
    },
    {
      "epoch": 2.5093202522775053,
      "grad_norm": 0.8914198279380798,
      "learning_rate": 5.034202343936856e-05,
      "loss": 0.0396,
      "step": 17904
    },
    {
      "epoch": 2.509460406447092,
      "grad_norm": 0.19931572675704956,
      "learning_rate": 5.032767280554891e-05,
      "loss": 0.0163,
      "step": 17905
    },
    {
      "epoch": 2.5096005606166782,
      "grad_norm": 0.18438127636909485,
      "learning_rate": 5.031332217172925e-05,
      "loss": 0.0248,
      "step": 17906
    },
    {
      "epoch": 2.5097407147862647,
      "grad_norm": 0.2518075406551361,
      "learning_rate": 5.029897153790958e-05,
      "loss": 0.0196,
      "step": 17907
    },
    {
      "epoch": 2.509880868955851,
      "grad_norm": 0.13407835364341736,
      "learning_rate": 5.0284620904089924e-05,
      "loss": 0.0188,
      "step": 17908
    },
    {
      "epoch": 2.510021023125438,
      "grad_norm": 0.33575475215911865,
      "learning_rate": 5.0270270270270267e-05,
      "loss": 0.0311,
      "step": 17909
    },
    {
      "epoch": 2.5101611772950245,
      "grad_norm": 0.16032351553440094,
      "learning_rate": 5.025591963645061e-05,
      "loss": 0.0097,
      "step": 17910
    },
    {
      "epoch": 2.510301331464611,
      "grad_norm": 0.022804399952292442,
      "learning_rate": 5.0241569002630944e-05,
      "loss": 0.0014,
      "step": 17911
    },
    {
      "epoch": 2.510441485634198,
      "grad_norm": 0.0834144577383995,
      "learning_rate": 5.022721836881128e-05,
      "loss": 0.0039,
      "step": 17912
    },
    {
      "epoch": 2.5105816398037843,
      "grad_norm": 0.2841784656047821,
      "learning_rate": 5.021286773499163e-05,
      "loss": 0.0255,
      "step": 17913
    },
    {
      "epoch": 2.510721793973371,
      "grad_norm": 0.21736538410186768,
      "learning_rate": 5.0198517101171964e-05,
      "loss": 0.0333,
      "step": 17914
    },
    {
      "epoch": 2.5108619481429573,
      "grad_norm": 0.07583963871002197,
      "learning_rate": 5.01841664673523e-05,
      "loss": 0.0437,
      "step": 17915
    },
    {
      "epoch": 2.5110021023125437,
      "grad_norm": 0.19257482886314392,
      "learning_rate": 5.016981583353264e-05,
      "loss": 0.0218,
      "step": 17916
    },
    {
      "epoch": 2.51114225648213,
      "grad_norm": 0.210791677236557,
      "learning_rate": 5.015546519971298e-05,
      "loss": 0.0242,
      "step": 17917
    },
    {
      "epoch": 2.5112824106517166,
      "grad_norm": 0.07649431377649307,
      "learning_rate": 5.0141114565893325e-05,
      "loss": 0.0045,
      "step": 17918
    },
    {
      "epoch": 2.5114225648213035,
      "grad_norm": 1.675681471824646,
      "learning_rate": 5.012676393207366e-05,
      "loss": 0.1565,
      "step": 17919
    },
    {
      "epoch": 2.51156271899089,
      "grad_norm": 0.8766339421272278,
      "learning_rate": 5.0112413298253996e-05,
      "loss": 0.2809,
      "step": 17920
    },
    {
      "epoch": 2.5117028731604765,
      "grad_norm": 0.15034359693527222,
      "learning_rate": 5.0098062664434345e-05,
      "loss": 0.0117,
      "step": 17921
    },
    {
      "epoch": 2.511843027330063,
      "grad_norm": 0.2890153229236603,
      "learning_rate": 5.008371203061468e-05,
      "loss": 0.0356,
      "step": 17922
    },
    {
      "epoch": 2.51198318149965,
      "grad_norm": 0.18044079840183258,
      "learning_rate": 5.0069361396795016e-05,
      "loss": 0.0427,
      "step": 17923
    },
    {
      "epoch": 2.5121233356692363,
      "grad_norm": 0.21608898043632507,
      "learning_rate": 5.0055010762975364e-05,
      "loss": 0.0431,
      "step": 17924
    },
    {
      "epoch": 2.5122634898388227,
      "grad_norm": 0.12421417981386185,
      "learning_rate": 5.00406601291557e-05,
      "loss": 0.0429,
      "step": 17925
    },
    {
      "epoch": 2.512403644008409,
      "grad_norm": 0.19879715144634247,
      "learning_rate": 5.002630949533604e-05,
      "loss": 0.0072,
      "step": 17926
    },
    {
      "epoch": 2.5125437981779957,
      "grad_norm": 0.24948570132255554,
      "learning_rate": 5.001195886151638e-05,
      "loss": 0.0392,
      "step": 17927
    },
    {
      "epoch": 2.512683952347582,
      "grad_norm": 0.09584183990955353,
      "learning_rate": 4.9997608227696726e-05,
      "loss": 0.0053,
      "step": 17928
    },
    {
      "epoch": 2.512824106517169,
      "grad_norm": 0.13370759785175323,
      "learning_rate": 4.998325759387706e-05,
      "loss": 0.0098,
      "step": 17929
    },
    {
      "epoch": 2.5129642606867555,
      "grad_norm": 0.12032349407672882,
      "learning_rate": 4.99689069600574e-05,
      "loss": 0.0065,
      "step": 17930
    },
    {
      "epoch": 2.513104414856342,
      "grad_norm": 0.09858205169439316,
      "learning_rate": 4.995455632623773e-05,
      "loss": 0.008,
      "step": 17931
    },
    {
      "epoch": 2.5132445690259284,
      "grad_norm": 0.18433667719364166,
      "learning_rate": 4.994020569241808e-05,
      "loss": 0.0193,
      "step": 17932
    },
    {
      "epoch": 2.5133847231955153,
      "grad_norm": 0.48852431774139404,
      "learning_rate": 4.9925855058598416e-05,
      "loss": 0.0302,
      "step": 17933
    },
    {
      "epoch": 2.5135248773651018,
      "grad_norm": 0.062215156853199005,
      "learning_rate": 4.991150442477876e-05,
      "loss": 0.0134,
      "step": 17934
    },
    {
      "epoch": 2.5136650315346882,
      "grad_norm": 0.131892591714859,
      "learning_rate": 4.9897153790959094e-05,
      "loss": 0.0384,
      "step": 17935
    },
    {
      "epoch": 2.5138051857042747,
      "grad_norm": 0.6051559448242188,
      "learning_rate": 4.988280315713944e-05,
      "loss": 0.049,
      "step": 17936
    },
    {
      "epoch": 2.513945339873861,
      "grad_norm": 0.29307442903518677,
      "learning_rate": 4.986845252331978e-05,
      "loss": 0.027,
      "step": 17937
    },
    {
      "epoch": 2.5140854940434476,
      "grad_norm": 0.25392740964889526,
      "learning_rate": 4.985410188950011e-05,
      "loss": 0.0241,
      "step": 17938
    },
    {
      "epoch": 2.514225648213034,
      "grad_norm": 0.1617138385772705,
      "learning_rate": 4.983975125568045e-05,
      "loss": 0.0208,
      "step": 17939
    },
    {
      "epoch": 2.514365802382621,
      "grad_norm": 0.42534786462783813,
      "learning_rate": 4.98254006218608e-05,
      "loss": 0.117,
      "step": 17940
    },
    {
      "epoch": 2.5145059565522074,
      "grad_norm": 0.0377785824239254,
      "learning_rate": 4.981104998804113e-05,
      "loss": 0.0031,
      "step": 17941
    },
    {
      "epoch": 2.514646110721794,
      "grad_norm": 0.2530292272567749,
      "learning_rate": 4.9796699354221475e-05,
      "loss": 0.0441,
      "step": 17942
    },
    {
      "epoch": 2.514786264891381,
      "grad_norm": 0.44305163621902466,
      "learning_rate": 4.978234872040181e-05,
      "loss": 0.0792,
      "step": 17943
    },
    {
      "epoch": 2.5149264190609673,
      "grad_norm": 0.2983666658401489,
      "learning_rate": 4.976799808658216e-05,
      "loss": 0.0439,
      "step": 17944
    },
    {
      "epoch": 2.5150665732305537,
      "grad_norm": 0.18985411524772644,
      "learning_rate": 4.9753647452762495e-05,
      "loss": 0.015,
      "step": 17945
    },
    {
      "epoch": 2.51520672740014,
      "grad_norm": 0.6229122281074524,
      "learning_rate": 4.973929681894283e-05,
      "loss": 0.0306,
      "step": 17946
    },
    {
      "epoch": 2.5153468815697266,
      "grad_norm": 0.11859645694494247,
      "learning_rate": 4.9724946185123165e-05,
      "loss": 0.0147,
      "step": 17947
    },
    {
      "epoch": 2.515487035739313,
      "grad_norm": 0.2446531057357788,
      "learning_rate": 4.9710595551303514e-05,
      "loss": 0.0491,
      "step": 17948
    },
    {
      "epoch": 2.5156271899088996,
      "grad_norm": 0.16358372569084167,
      "learning_rate": 4.969624491748385e-05,
      "loss": 0.006,
      "step": 17949
    },
    {
      "epoch": 2.5157673440784865,
      "grad_norm": 0.18392562866210938,
      "learning_rate": 4.968189428366419e-05,
      "loss": 0.0088,
      "step": 17950
    },
    {
      "epoch": 2.515907498248073,
      "grad_norm": 0.04929543659090996,
      "learning_rate": 4.966754364984453e-05,
      "loss": 0.005,
      "step": 17951
    },
    {
      "epoch": 2.5160476524176594,
      "grad_norm": 0.9538974761962891,
      "learning_rate": 4.9653193016024876e-05,
      "loss": 0.0479,
      "step": 17952
    },
    {
      "epoch": 2.516187806587246,
      "grad_norm": 0.12925632297992706,
      "learning_rate": 4.963884238220521e-05,
      "loss": 0.0051,
      "step": 17953
    },
    {
      "epoch": 2.5163279607568327,
      "grad_norm": 0.13811703026294708,
      "learning_rate": 4.9624491748385546e-05,
      "loss": 0.0122,
      "step": 17954
    },
    {
      "epoch": 2.516468114926419,
      "grad_norm": 0.37643539905548096,
      "learning_rate": 4.961014111456588e-05,
      "loss": 0.0191,
      "step": 17955
    },
    {
      "epoch": 2.5166082690960057,
      "grad_norm": 0.19138064980506897,
      "learning_rate": 4.959579048074623e-05,
      "loss": 0.0254,
      "step": 17956
    },
    {
      "epoch": 2.516748423265592,
      "grad_norm": 0.09747377783060074,
      "learning_rate": 4.9581439846926566e-05,
      "loss": 0.0075,
      "step": 17957
    },
    {
      "epoch": 2.5168885774351786,
      "grad_norm": 1.5282933712005615,
      "learning_rate": 4.956708921310691e-05,
      "loss": 0.0129,
      "step": 17958
    },
    {
      "epoch": 2.517028731604765,
      "grad_norm": 0.10173609107732773,
      "learning_rate": 4.955273857928725e-05,
      "loss": 0.0327,
      "step": 17959
    },
    {
      "epoch": 2.517168885774352,
      "grad_norm": 0.2733389139175415,
      "learning_rate": 4.953838794546759e-05,
      "loss": 0.0299,
      "step": 17960
    },
    {
      "epoch": 2.5173090399439384,
      "grad_norm": 0.1406165063381195,
      "learning_rate": 4.952403731164793e-05,
      "loss": 0.0147,
      "step": 17961
    },
    {
      "epoch": 2.517449194113525,
      "grad_norm": 0.24290359020233154,
      "learning_rate": 4.950968667782826e-05,
      "loss": 0.0946,
      "step": 17962
    },
    {
      "epoch": 2.5175893482831113,
      "grad_norm": 0.031261276453733444,
      "learning_rate": 4.949533604400861e-05,
      "loss": 0.0024,
      "step": 17963
    },
    {
      "epoch": 2.517729502452698,
      "grad_norm": 0.6725165247917175,
      "learning_rate": 4.948098541018895e-05,
      "loss": 0.0741,
      "step": 17964
    },
    {
      "epoch": 2.5178696566222847,
      "grad_norm": 0.22323952615261078,
      "learning_rate": 4.946663477636928e-05,
      "loss": 0.0956,
      "step": 17965
    },
    {
      "epoch": 2.518009810791871,
      "grad_norm": 0.058933075517416,
      "learning_rate": 4.9452284142549625e-05,
      "loss": 0.0045,
      "step": 17966
    },
    {
      "epoch": 2.5181499649614576,
      "grad_norm": 0.21088677644729614,
      "learning_rate": 4.943793350872997e-05,
      "loss": 0.0145,
      "step": 17967
    },
    {
      "epoch": 2.518290119131044,
      "grad_norm": 0.9093017578125,
      "learning_rate": 4.942358287491031e-05,
      "loss": 0.1194,
      "step": 17968
    },
    {
      "epoch": 2.5184302733006305,
      "grad_norm": 0.46114274859428406,
      "learning_rate": 4.9409232241090644e-05,
      "loss": 0.029,
      "step": 17969
    },
    {
      "epoch": 2.518570427470217,
      "grad_norm": 2.3653149604797363,
      "learning_rate": 4.939488160727098e-05,
      "loss": 0.1459,
      "step": 17970
    },
    {
      "epoch": 2.518710581639804,
      "grad_norm": 0.2956080138683319,
      "learning_rate": 4.938053097345133e-05,
      "loss": 0.0288,
      "step": 17971
    },
    {
      "epoch": 2.5188507358093903,
      "grad_norm": 0.23610498011112213,
      "learning_rate": 4.9366180339631664e-05,
      "loss": 0.0205,
      "step": 17972
    },
    {
      "epoch": 2.518990889978977,
      "grad_norm": 0.08853091299533844,
      "learning_rate": 4.9351829705812e-05,
      "loss": 0.0101,
      "step": 17973
    },
    {
      "epoch": 2.5191310441485633,
      "grad_norm": 0.11466262489557266,
      "learning_rate": 4.933747907199234e-05,
      "loss": 0.0077,
      "step": 17974
    },
    {
      "epoch": 2.51927119831815,
      "grad_norm": 0.03590298816561699,
      "learning_rate": 4.9323128438172683e-05,
      "loss": 0.0034,
      "step": 17975
    },
    {
      "epoch": 2.5194113524877366,
      "grad_norm": 0.3006124794483185,
      "learning_rate": 4.9308777804353026e-05,
      "loss": 0.0106,
      "step": 17976
    },
    {
      "epoch": 2.519551506657323,
      "grad_norm": 0.2731272578239441,
      "learning_rate": 4.929442717053336e-05,
      "loss": 0.0313,
      "step": 17977
    },
    {
      "epoch": 2.5196916608269095,
      "grad_norm": 0.16604268550872803,
      "learning_rate": 4.9280076536713696e-05,
      "loss": 0.0187,
      "step": 17978
    },
    {
      "epoch": 2.519831814996496,
      "grad_norm": 0.3185887932777405,
      "learning_rate": 4.9265725902894045e-05,
      "loss": 0.0409,
      "step": 17979
    },
    {
      "epoch": 2.5199719691660825,
      "grad_norm": 0.19657133519649506,
      "learning_rate": 4.925137526907438e-05,
      "loss": 0.0224,
      "step": 17980
    },
    {
      "epoch": 2.5201121233356694,
      "grad_norm": 0.2634916603565216,
      "learning_rate": 4.9237024635254716e-05,
      "loss": 0.0172,
      "step": 17981
    },
    {
      "epoch": 2.520252277505256,
      "grad_norm": 0.04718831926584244,
      "learning_rate": 4.922267400143506e-05,
      "loss": 0.0039,
      "step": 17982
    },
    {
      "epoch": 2.5203924316748423,
      "grad_norm": 0.15247592329978943,
      "learning_rate": 4.92083233676154e-05,
      "loss": 0.0193,
      "step": 17983
    },
    {
      "epoch": 2.5205325858444287,
      "grad_norm": 0.12196923792362213,
      "learning_rate": 4.919397273379574e-05,
      "loss": 0.0166,
      "step": 17984
    },
    {
      "epoch": 2.5206727400140156,
      "grad_norm": 0.16519244015216827,
      "learning_rate": 4.917962209997608e-05,
      "loss": 0.0276,
      "step": 17985
    },
    {
      "epoch": 2.520812894183602,
      "grad_norm": 0.2617371678352356,
      "learning_rate": 4.916527146615641e-05,
      "loss": 0.0593,
      "step": 17986
    },
    {
      "epoch": 2.5209530483531886,
      "grad_norm": 0.20458994805812836,
      "learning_rate": 4.915092083233676e-05,
      "loss": 0.032,
      "step": 17987
    },
    {
      "epoch": 2.521093202522775,
      "grad_norm": 0.3306627869606018,
      "learning_rate": 4.91365701985171e-05,
      "loss": 0.0532,
      "step": 17988
    },
    {
      "epoch": 2.5212333566923615,
      "grad_norm": 0.12184417992830276,
      "learning_rate": 4.912221956469743e-05,
      "loss": 0.0205,
      "step": 17989
    },
    {
      "epoch": 2.521373510861948,
      "grad_norm": 0.12734109163284302,
      "learning_rate": 4.9107868930877774e-05,
      "loss": 0.024,
      "step": 17990
    },
    {
      "epoch": 2.521513665031535,
      "grad_norm": 0.3743153214454651,
      "learning_rate": 4.9093518297058117e-05,
      "loss": 0.0393,
      "step": 17991
    },
    {
      "epoch": 2.5216538192011213,
      "grad_norm": 0.2873273491859436,
      "learning_rate": 4.907916766323846e-05,
      "loss": 0.0186,
      "step": 17992
    },
    {
      "epoch": 2.5217939733707078,
      "grad_norm": 0.1817651391029358,
      "learning_rate": 4.9064817029418794e-05,
      "loss": 0.0224,
      "step": 17993
    },
    {
      "epoch": 2.5219341275402942,
      "grad_norm": 0.2557279169559479,
      "learning_rate": 4.905046639559913e-05,
      "loss": 0.0426,
      "step": 17994
    },
    {
      "epoch": 2.522074281709881,
      "grad_norm": 0.4094601571559906,
      "learning_rate": 4.903611576177948e-05,
      "loss": 0.0023,
      "step": 17995
    },
    {
      "epoch": 2.5222144358794676,
      "grad_norm": 0.23652827739715576,
      "learning_rate": 4.9021765127959814e-05,
      "loss": 0.0182,
      "step": 17996
    },
    {
      "epoch": 2.522354590049054,
      "grad_norm": 0.18629562854766846,
      "learning_rate": 4.900741449414015e-05,
      "loss": 0.0234,
      "step": 17997
    },
    {
      "epoch": 2.5224947442186405,
      "grad_norm": 0.3949895203113556,
      "learning_rate": 4.89930638603205e-05,
      "loss": 0.0143,
      "step": 17998
    },
    {
      "epoch": 2.522634898388227,
      "grad_norm": 0.08356636017560959,
      "learning_rate": 4.897871322650083e-05,
      "loss": 0.0052,
      "step": 17999
    },
    {
      "epoch": 2.5227750525578134,
      "grad_norm": 0.49311429262161255,
      "learning_rate": 4.8964362592681175e-05,
      "loss": 0.0553,
      "step": 18000
    },
    {
      "epoch": 2.5229152067274,
      "grad_norm": 0.36115556955337524,
      "learning_rate": 4.895001195886151e-05,
      "loss": 0.0128,
      "step": 18001
    },
    {
      "epoch": 2.523055360896987,
      "grad_norm": 0.31049636006355286,
      "learning_rate": 4.893566132504185e-05,
      "loss": 0.0249,
      "step": 18002
    },
    {
      "epoch": 2.5231955150665732,
      "grad_norm": 0.04329279437661171,
      "learning_rate": 4.8921310691222195e-05,
      "loss": 0.0032,
      "step": 18003
    },
    {
      "epoch": 2.5233356692361597,
      "grad_norm": 0.15391989052295685,
      "learning_rate": 4.890696005740253e-05,
      "loss": 0.0198,
      "step": 18004
    },
    {
      "epoch": 2.523475823405746,
      "grad_norm": 1.2122329473495483,
      "learning_rate": 4.8892609423582865e-05,
      "loss": 0.0888,
      "step": 18005
    },
    {
      "epoch": 2.523615977575333,
      "grad_norm": 0.31095290184020996,
      "learning_rate": 4.8878258789763214e-05,
      "loss": 0.0145,
      "step": 18006
    },
    {
      "epoch": 2.5237561317449195,
      "grad_norm": 0.31254544854164124,
      "learning_rate": 4.886390815594355e-05,
      "loss": 0.0189,
      "step": 18007
    },
    {
      "epoch": 2.523896285914506,
      "grad_norm": 0.038663607090711594,
      "learning_rate": 4.884955752212389e-05,
      "loss": 0.0062,
      "step": 18008
    },
    {
      "epoch": 2.5240364400840924,
      "grad_norm": 0.0795617625117302,
      "learning_rate": 4.883520688830423e-05,
      "loss": 0.0058,
      "step": 18009
    },
    {
      "epoch": 2.524176594253679,
      "grad_norm": 0.17984271049499512,
      "learning_rate": 4.882085625448457e-05,
      "loss": 0.0046,
      "step": 18010
    },
    {
      "epoch": 2.5243167484232654,
      "grad_norm": 0.07299651950597763,
      "learning_rate": 4.880650562066491e-05,
      "loss": 0.0034,
      "step": 18011
    },
    {
      "epoch": 2.5244569025928523,
      "grad_norm": 0.17152342200279236,
      "learning_rate": 4.879215498684525e-05,
      "loss": 0.0039,
      "step": 18012
    },
    {
      "epoch": 2.5245970567624387,
      "grad_norm": 0.26020678877830505,
      "learning_rate": 4.877780435302558e-05,
      "loss": 0.0336,
      "step": 18013
    },
    {
      "epoch": 2.524737210932025,
      "grad_norm": 0.3551461696624756,
      "learning_rate": 4.876345371920593e-05,
      "loss": 0.0351,
      "step": 18014
    },
    {
      "epoch": 2.5248773651016116,
      "grad_norm": 0.11331973969936371,
      "learning_rate": 4.8749103085386266e-05,
      "loss": 0.006,
      "step": 18015
    },
    {
      "epoch": 2.5250175192711986,
      "grad_norm": 0.04525251314043999,
      "learning_rate": 4.873475245156661e-05,
      "loss": 0.0024,
      "step": 18016
    },
    {
      "epoch": 2.525157673440785,
      "grad_norm": 0.25383517146110535,
      "learning_rate": 4.8720401817746944e-05,
      "loss": 0.027,
      "step": 18017
    },
    {
      "epoch": 2.5252978276103715,
      "grad_norm": 0.03788610175251961,
      "learning_rate": 4.8706051183927286e-05,
      "loss": 0.0028,
      "step": 18018
    },
    {
      "epoch": 2.525437981779958,
      "grad_norm": 0.1253206431865692,
      "learning_rate": 4.869170055010763e-05,
      "loss": 0.0087,
      "step": 18019
    },
    {
      "epoch": 2.5255781359495444,
      "grad_norm": 2.796492338180542,
      "learning_rate": 4.867734991628796e-05,
      "loss": 0.121,
      "step": 18020
    },
    {
      "epoch": 2.525718290119131,
      "grad_norm": 0.3664858937263489,
      "learning_rate": 4.86629992824683e-05,
      "loss": 0.0085,
      "step": 18021
    },
    {
      "epoch": 2.5258584442887178,
      "grad_norm": 0.554985761642456,
      "learning_rate": 4.864864864864865e-05,
      "loss": 0.0246,
      "step": 18022
    },
    {
      "epoch": 2.525998598458304,
      "grad_norm": 0.20167531073093414,
      "learning_rate": 4.863429801482898e-05,
      "loss": 0.0337,
      "step": 18023
    },
    {
      "epoch": 2.5261387526278907,
      "grad_norm": 0.19294866919517517,
      "learning_rate": 4.8619947381009325e-05,
      "loss": 0.0194,
      "step": 18024
    },
    {
      "epoch": 2.526278906797477,
      "grad_norm": 0.3500731289386749,
      "learning_rate": 4.860559674718966e-05,
      "loss": 0.063,
      "step": 18025
    },
    {
      "epoch": 2.526419060967064,
      "grad_norm": 0.15809476375579834,
      "learning_rate": 4.859124611337e-05,
      "loss": 0.0398,
      "step": 18026
    },
    {
      "epoch": 2.5265592151366505,
      "grad_norm": 0.19352607429027557,
      "learning_rate": 4.8576895479550345e-05,
      "loss": 0.0374,
      "step": 18027
    },
    {
      "epoch": 2.526699369306237,
      "grad_norm": 0.16967914998531342,
      "learning_rate": 4.856254484573068e-05,
      "loss": 0.0256,
      "step": 18028
    },
    {
      "epoch": 2.5268395234758234,
      "grad_norm": 0.06481167674064636,
      "learning_rate": 4.8548194211911015e-05,
      "loss": 0.0089,
      "step": 18029
    },
    {
      "epoch": 2.52697967764541,
      "grad_norm": 0.18969322741031647,
      "learning_rate": 4.8533843578091364e-05,
      "loss": 0.0376,
      "step": 18030
    },
    {
      "epoch": 2.5271198318149963,
      "grad_norm": 0.2537195682525635,
      "learning_rate": 4.85194929442717e-05,
      "loss": 0.0085,
      "step": 18031
    },
    {
      "epoch": 2.527259985984583,
      "grad_norm": 0.36229178309440613,
      "learning_rate": 4.850514231045204e-05,
      "loss": 0.0561,
      "step": 18032
    },
    {
      "epoch": 2.5274001401541697,
      "grad_norm": 0.6040002703666687,
      "learning_rate": 4.849079167663238e-05,
      "loss": 0.0343,
      "step": 18033
    },
    {
      "epoch": 2.527540294323756,
      "grad_norm": 0.3086155652999878,
      "learning_rate": 4.847644104281272e-05,
      "loss": 0.0355,
      "step": 18034
    },
    {
      "epoch": 2.5276804484933426,
      "grad_norm": 0.12460111826658249,
      "learning_rate": 4.846209040899306e-05,
      "loss": 0.0133,
      "step": 18035
    },
    {
      "epoch": 2.527820602662929,
      "grad_norm": 0.15721264481544495,
      "learning_rate": 4.8447739775173396e-05,
      "loss": 0.0102,
      "step": 18036
    },
    {
      "epoch": 2.527960756832516,
      "grad_norm": 0.10478731244802475,
      "learning_rate": 4.8433389141353745e-05,
      "loss": 0.0083,
      "step": 18037
    },
    {
      "epoch": 2.5281009110021024,
      "grad_norm": 0.1951928734779358,
      "learning_rate": 4.841903850753408e-05,
      "loss": 0.0715,
      "step": 18038
    },
    {
      "epoch": 2.528241065171689,
      "grad_norm": 0.1319209337234497,
      "learning_rate": 4.8404687873714416e-05,
      "loss": 0.0159,
      "step": 18039
    },
    {
      "epoch": 2.5283812193412754,
      "grad_norm": 0.2013491839170456,
      "learning_rate": 4.839033723989476e-05,
      "loss": 0.018,
      "step": 18040
    },
    {
      "epoch": 2.528521373510862,
      "grad_norm": 0.506523072719574,
      "learning_rate": 4.83759866060751e-05,
      "loss": 0.0434,
      "step": 18041
    },
    {
      "epoch": 2.5286615276804483,
      "grad_norm": 0.1992463916540146,
      "learning_rate": 4.8361635972255436e-05,
      "loss": 0.0145,
      "step": 18042
    },
    {
      "epoch": 2.528801681850035,
      "grad_norm": 0.4441595673561096,
      "learning_rate": 4.834728533843578e-05,
      "loss": 0.0288,
      "step": 18043
    },
    {
      "epoch": 2.5289418360196216,
      "grad_norm": 0.61379075050354,
      "learning_rate": 4.833293470461611e-05,
      "loss": 0.0827,
      "step": 18044
    },
    {
      "epoch": 2.529081990189208,
      "grad_norm": 0.5615122318267822,
      "learning_rate": 4.831858407079646e-05,
      "loss": 0.0441,
      "step": 18045
    },
    {
      "epoch": 2.5292221443587946,
      "grad_norm": 0.0574890561401844,
      "learning_rate": 4.83042334369768e-05,
      "loss": 0.0045,
      "step": 18046
    },
    {
      "epoch": 2.5293622985283815,
      "grad_norm": 0.28991183638572693,
      "learning_rate": 4.828988280315713e-05,
      "loss": 0.0287,
      "step": 18047
    },
    {
      "epoch": 2.529502452697968,
      "grad_norm": 0.08361067622900009,
      "learning_rate": 4.8275532169337475e-05,
      "loss": 0.0134,
      "step": 18048
    },
    {
      "epoch": 2.5296426068675544,
      "grad_norm": 0.17233183979988098,
      "learning_rate": 4.826118153551782e-05,
      "loss": 0.0333,
      "step": 18049
    },
    {
      "epoch": 2.529782761037141,
      "grad_norm": 0.6347658634185791,
      "learning_rate": 4.824683090169815e-05,
      "loss": 0.0911,
      "step": 18050
    },
    {
      "epoch": 2.5299229152067273,
      "grad_norm": 0.3135128915309906,
      "learning_rate": 4.8232480267878494e-05,
      "loss": 0.0467,
      "step": 18051
    },
    {
      "epoch": 2.5300630693763138,
      "grad_norm": 0.0376332625746727,
      "learning_rate": 4.821812963405883e-05,
      "loss": 0.0174,
      "step": 18052
    },
    {
      "epoch": 2.5302032235459,
      "grad_norm": 0.5588517785072327,
      "learning_rate": 4.820377900023918e-05,
      "loss": 0.0273,
      "step": 18053
    },
    {
      "epoch": 2.530343377715487,
      "grad_norm": 0.12721534073352814,
      "learning_rate": 4.8189428366419514e-05,
      "loss": 0.0115,
      "step": 18054
    },
    {
      "epoch": 2.5304835318850736,
      "grad_norm": 0.2577660083770752,
      "learning_rate": 4.817507773259985e-05,
      "loss": 0.0115,
      "step": 18055
    },
    {
      "epoch": 2.53062368605466,
      "grad_norm": 0.15856869518756866,
      "learning_rate": 4.816072709878019e-05,
      "loss": 0.0131,
      "step": 18056
    },
    {
      "epoch": 2.530763840224247,
      "grad_norm": 0.08920245617628098,
      "learning_rate": 4.814637646496053e-05,
      "loss": 0.0038,
      "step": 18057
    },
    {
      "epoch": 2.5309039943938334,
      "grad_norm": 0.07169445604085922,
      "learning_rate": 4.813202583114087e-05,
      "loss": 0.0047,
      "step": 18058
    },
    {
      "epoch": 2.53104414856342,
      "grad_norm": 0.18557284772396088,
      "learning_rate": 4.811767519732121e-05,
      "loss": 0.0205,
      "step": 18059
    },
    {
      "epoch": 2.5311843027330063,
      "grad_norm": 0.09519461542367935,
      "learning_rate": 4.8103324563501546e-05,
      "loss": 0.005,
      "step": 18060
    },
    {
      "epoch": 2.531324456902593,
      "grad_norm": 0.39410871267318726,
      "learning_rate": 4.8088973929681895e-05,
      "loss": 0.0614,
      "step": 18061
    },
    {
      "epoch": 2.5314646110721792,
      "grad_norm": 0.06212591007351875,
      "learning_rate": 4.807462329586223e-05,
      "loss": 0.005,
      "step": 18062
    },
    {
      "epoch": 2.5316047652417657,
      "grad_norm": 0.25844892859458923,
      "learning_rate": 4.8060272662042566e-05,
      "loss": 0.0204,
      "step": 18063
    },
    {
      "epoch": 2.5317449194113526,
      "grad_norm": 0.09644581377506256,
      "learning_rate": 4.804592202822291e-05,
      "loss": 0.0053,
      "step": 18064
    },
    {
      "epoch": 2.531885073580939,
      "grad_norm": 0.01994510553777218,
      "learning_rate": 4.803157139440325e-05,
      "loss": 0.0012,
      "step": 18065
    },
    {
      "epoch": 2.5320252277505255,
      "grad_norm": 0.5027076601982117,
      "learning_rate": 4.8017220760583585e-05,
      "loss": 0.0173,
      "step": 18066
    },
    {
      "epoch": 2.532165381920112,
      "grad_norm": 0.016899367794394493,
      "learning_rate": 4.800287012676393e-05,
      "loss": 0.0013,
      "step": 18067
    },
    {
      "epoch": 2.532305536089699,
      "grad_norm": 0.6354995965957642,
      "learning_rate": 4.798851949294426e-05,
      "loss": 0.0346,
      "step": 18068
    },
    {
      "epoch": 2.5324456902592853,
      "grad_norm": 3.736849784851074,
      "learning_rate": 4.797416885912461e-05,
      "loss": 0.1438,
      "step": 18069
    },
    {
      "epoch": 2.532585844428872,
      "grad_norm": 1.1676243543624878,
      "learning_rate": 4.795981822530495e-05,
      "loss": 0.0203,
      "step": 18070
    },
    {
      "epoch": 2.5327259985984583,
      "grad_norm": 0.30121898651123047,
      "learning_rate": 4.794546759148528e-05,
      "loss": 0.0432,
      "step": 18071
    },
    {
      "epoch": 2.5328661527680447,
      "grad_norm": 0.19894465804100037,
      "learning_rate": 4.793111695766563e-05,
      "loss": 0.0234,
      "step": 18072
    },
    {
      "epoch": 2.533006306937631,
      "grad_norm": 0.12876781821250916,
      "learning_rate": 4.7916766323845966e-05,
      "loss": 0.0061,
      "step": 18073
    },
    {
      "epoch": 2.533146461107218,
      "grad_norm": 0.2891557812690735,
      "learning_rate": 4.79024156900263e-05,
      "loss": 0.04,
      "step": 18074
    },
    {
      "epoch": 2.5332866152768045,
      "grad_norm": 0.08656831085681915,
      "learning_rate": 4.7888065056206644e-05,
      "loss": 0.0345,
      "step": 18075
    },
    {
      "epoch": 2.533426769446391,
      "grad_norm": 0.28342166543006897,
      "learning_rate": 4.7873714422386986e-05,
      "loss": 0.0186,
      "step": 18076
    },
    {
      "epoch": 2.5335669236159775,
      "grad_norm": 0.17604508996009827,
      "learning_rate": 4.785936378856733e-05,
      "loss": 0.0559,
      "step": 18077
    },
    {
      "epoch": 2.5337070777855644,
      "grad_norm": 0.11375840753316879,
      "learning_rate": 4.7845013154747664e-05,
      "loss": 0.0132,
      "step": 18078
    },
    {
      "epoch": 2.533847231955151,
      "grad_norm": 0.07525541633367538,
      "learning_rate": 4.7830662520928e-05,
      "loss": 0.0057,
      "step": 18079
    },
    {
      "epoch": 2.5339873861247373,
      "grad_norm": 0.32763245701789856,
      "learning_rate": 4.781631188710835e-05,
      "loss": 0.0583,
      "step": 18080
    },
    {
      "epoch": 2.5341275402943237,
      "grad_norm": 0.19011011719703674,
      "learning_rate": 4.780196125328868e-05,
      "loss": 0.0181,
      "step": 18081
    },
    {
      "epoch": 2.53426769446391,
      "grad_norm": 0.31514057517051697,
      "learning_rate": 4.778761061946902e-05,
      "loss": 0.0513,
      "step": 18082
    },
    {
      "epoch": 2.5344078486334967,
      "grad_norm": 0.25016865134239197,
      "learning_rate": 4.777325998564936e-05,
      "loss": 0.0311,
      "step": 18083
    },
    {
      "epoch": 2.534548002803083,
      "grad_norm": 0.22858941555023193,
      "learning_rate": 4.77589093518297e-05,
      "loss": 0.0211,
      "step": 18084
    },
    {
      "epoch": 2.53468815697267,
      "grad_norm": 0.3425392210483551,
      "learning_rate": 4.7744558718010045e-05,
      "loss": 0.0392,
      "step": 18085
    },
    {
      "epoch": 2.5348283111422565,
      "grad_norm": 0.3620901107788086,
      "learning_rate": 4.773020808419038e-05,
      "loss": 0.0572,
      "step": 18086
    },
    {
      "epoch": 2.534968465311843,
      "grad_norm": 0.13005124032497406,
      "learning_rate": 4.7715857450370715e-05,
      "loss": 0.0198,
      "step": 18087
    },
    {
      "epoch": 2.53510861948143,
      "grad_norm": 0.06559021025896072,
      "learning_rate": 4.7701506816551064e-05,
      "loss": 0.0063,
      "step": 18088
    },
    {
      "epoch": 2.5352487736510163,
      "grad_norm": 0.42133620381355286,
      "learning_rate": 4.76871561827314e-05,
      "loss": 0.0396,
      "step": 18089
    },
    {
      "epoch": 2.5353889278206028,
      "grad_norm": 0.3147045969963074,
      "learning_rate": 4.7672805548911735e-05,
      "loss": 0.0328,
      "step": 18090
    },
    {
      "epoch": 2.5355290819901892,
      "grad_norm": 0.052514929324388504,
      "learning_rate": 4.765845491509208e-05,
      "loss": 0.0038,
      "step": 18091
    },
    {
      "epoch": 2.5356692361597757,
      "grad_norm": 0.389994740486145,
      "learning_rate": 4.764410428127242e-05,
      "loss": 0.0534,
      "step": 18092
    },
    {
      "epoch": 2.535809390329362,
      "grad_norm": 0.3380163311958313,
      "learning_rate": 4.762975364745276e-05,
      "loss": 0.0246,
      "step": 18093
    },
    {
      "epoch": 2.5359495444989486,
      "grad_norm": 0.19539874792099,
      "learning_rate": 4.76154030136331e-05,
      "loss": 0.0367,
      "step": 18094
    },
    {
      "epoch": 2.5360896986685355,
      "grad_norm": 0.2270340621471405,
      "learning_rate": 4.760105237981343e-05,
      "loss": 0.0361,
      "step": 18095
    },
    {
      "epoch": 2.536229852838122,
      "grad_norm": 0.11640356481075287,
      "learning_rate": 4.758670174599378e-05,
      "loss": 0.0119,
      "step": 18096
    },
    {
      "epoch": 2.5363700070077084,
      "grad_norm": 0.3057485520839691,
      "learning_rate": 4.7572351112174116e-05,
      "loss": 0.04,
      "step": 18097
    },
    {
      "epoch": 2.536510161177295,
      "grad_norm": 0.09914348274469376,
      "learning_rate": 4.755800047835445e-05,
      "loss": 0.0052,
      "step": 18098
    },
    {
      "epoch": 2.536650315346882,
      "grad_norm": 0.06515375524759293,
      "learning_rate": 4.7543649844534794e-05,
      "loss": 0.0027,
      "step": 18099
    },
    {
      "epoch": 2.5367904695164682,
      "grad_norm": 0.39961355924606323,
      "learning_rate": 4.7529299210715136e-05,
      "loss": 0.0522,
      "step": 18100
    },
    {
      "epoch": 2.5369306236860547,
      "grad_norm": 0.15189528465270996,
      "learning_rate": 4.751494857689548e-05,
      "loss": 0.0137,
      "step": 18101
    },
    {
      "epoch": 2.537070777855641,
      "grad_norm": 0.7765271663665771,
      "learning_rate": 4.750059794307581e-05,
      "loss": 0.0204,
      "step": 18102
    },
    {
      "epoch": 2.5372109320252276,
      "grad_norm": 0.34573015570640564,
      "learning_rate": 4.748624730925615e-05,
      "loss": 0.0395,
      "step": 18103
    },
    {
      "epoch": 2.537351086194814,
      "grad_norm": 0.07802074402570724,
      "learning_rate": 4.74718966754365e-05,
      "loss": 0.0077,
      "step": 18104
    },
    {
      "epoch": 2.537491240364401,
      "grad_norm": 0.23393578827381134,
      "learning_rate": 4.745754604161683e-05,
      "loss": 0.0405,
      "step": 18105
    },
    {
      "epoch": 2.5376313945339875,
      "grad_norm": 0.11811116337776184,
      "learning_rate": 4.7443195407797175e-05,
      "loss": 0.0095,
      "step": 18106
    },
    {
      "epoch": 2.537771548703574,
      "grad_norm": 0.09666913002729416,
      "learning_rate": 4.742884477397751e-05,
      "loss": 0.0024,
      "step": 18107
    },
    {
      "epoch": 2.5379117028731604,
      "grad_norm": 0.7012326717376709,
      "learning_rate": 4.741449414015785e-05,
      "loss": 0.0493,
      "step": 18108
    },
    {
      "epoch": 2.5380518570427473,
      "grad_norm": 0.03890467435121536,
      "learning_rate": 4.7400143506338194e-05,
      "loss": 0.002,
      "step": 18109
    },
    {
      "epoch": 2.5381920112123337,
      "grad_norm": 0.16483518481254578,
      "learning_rate": 4.738579287251853e-05,
      "loss": 0.013,
      "step": 18110
    },
    {
      "epoch": 2.53833216538192,
      "grad_norm": 0.5212472677230835,
      "learning_rate": 4.737144223869888e-05,
      "loss": 0.0165,
      "step": 18111
    },
    {
      "epoch": 2.5384723195515067,
      "grad_norm": 0.23199720680713654,
      "learning_rate": 4.7357091604879214e-05,
      "loss": 0.0593,
      "step": 18112
    },
    {
      "epoch": 2.538612473721093,
      "grad_norm": 0.5993015766143799,
      "learning_rate": 4.734274097105955e-05,
      "loss": 0.033,
      "step": 18113
    },
    {
      "epoch": 2.5387526278906796,
      "grad_norm": 0.8392219543457031,
      "learning_rate": 4.732839033723989e-05,
      "loss": 0.0327,
      "step": 18114
    },
    {
      "epoch": 2.538892782060266,
      "grad_norm": 0.24607329070568085,
      "learning_rate": 4.7314039703420234e-05,
      "loss": 0.0201,
      "step": 18115
    },
    {
      "epoch": 2.539032936229853,
      "grad_norm": 0.5014885067939758,
      "learning_rate": 4.729968906960057e-05,
      "loss": 0.0703,
      "step": 18116
    },
    {
      "epoch": 2.5391730903994394,
      "grad_norm": 0.5055301189422607,
      "learning_rate": 4.728533843578091e-05,
      "loss": 0.0266,
      "step": 18117
    },
    {
      "epoch": 2.539313244569026,
      "grad_norm": 0.6922755837440491,
      "learning_rate": 4.7270987801961246e-05,
      "loss": 0.0414,
      "step": 18118
    },
    {
      "epoch": 2.5394533987386123,
      "grad_norm": 0.5933761596679688,
      "learning_rate": 4.7256637168141595e-05,
      "loss": 0.0616,
      "step": 18119
    },
    {
      "epoch": 2.539593552908199,
      "grad_norm": 0.17315752804279327,
      "learning_rate": 4.724228653432193e-05,
      "loss": 0.0211,
      "step": 18120
    },
    {
      "epoch": 2.5397337070777857,
      "grad_norm": 0.32750681042671204,
      "learning_rate": 4.7227935900502266e-05,
      "loss": 0.042,
      "step": 18121
    },
    {
      "epoch": 2.539873861247372,
      "grad_norm": 0.1828962117433548,
      "learning_rate": 4.721358526668261e-05,
      "loss": 0.0345,
      "step": 18122
    },
    {
      "epoch": 2.5400140154169586,
      "grad_norm": 0.5407280325889587,
      "learning_rate": 4.719923463286295e-05,
      "loss": 0.0367,
      "step": 18123
    },
    {
      "epoch": 2.540154169586545,
      "grad_norm": 0.1347690224647522,
      "learning_rate": 4.7184883999043285e-05,
      "loss": 0.0329,
      "step": 18124
    },
    {
      "epoch": 2.5402943237561315,
      "grad_norm": 0.11175432801246643,
      "learning_rate": 4.717053336522363e-05,
      "loss": 0.0205,
      "step": 18125
    },
    {
      "epoch": 2.5404344779257184,
      "grad_norm": 0.1709836721420288,
      "learning_rate": 4.715618273140396e-05,
      "loss": 0.0295,
      "step": 18126
    },
    {
      "epoch": 2.540574632095305,
      "grad_norm": 0.24169591069221497,
      "learning_rate": 4.714183209758431e-05,
      "loss": 0.0325,
      "step": 18127
    },
    {
      "epoch": 2.5407147862648913,
      "grad_norm": 0.3713139593601227,
      "learning_rate": 4.712748146376465e-05,
      "loss": 0.0754,
      "step": 18128
    },
    {
      "epoch": 2.540854940434478,
      "grad_norm": 0.3695809841156006,
      "learning_rate": 4.711313082994498e-05,
      "loss": 0.0195,
      "step": 18129
    },
    {
      "epoch": 2.5409950946040647,
      "grad_norm": 0.3140918016433716,
      "learning_rate": 4.7098780196125325e-05,
      "loss": 0.0432,
      "step": 18130
    },
    {
      "epoch": 2.541135248773651,
      "grad_norm": 0.17927393317222595,
      "learning_rate": 4.708442956230567e-05,
      "loss": 0.0152,
      "step": 18131
    },
    {
      "epoch": 2.5412754029432376,
      "grad_norm": 0.19948315620422363,
      "learning_rate": 4.7070078928486e-05,
      "loss": 0.0067,
      "step": 18132
    },
    {
      "epoch": 2.541415557112824,
      "grad_norm": 0.5426936149597168,
      "learning_rate": 4.7055728294666344e-05,
      "loss": 0.0621,
      "step": 18133
    },
    {
      "epoch": 2.5415557112824105,
      "grad_norm": 0.1400577425956726,
      "learning_rate": 4.704137766084668e-05,
      "loss": 0.0249,
      "step": 18134
    },
    {
      "epoch": 2.541695865451997,
      "grad_norm": 0.1902443766593933,
      "learning_rate": 4.702702702702703e-05,
      "loss": 0.0123,
      "step": 18135
    },
    {
      "epoch": 2.541836019621584,
      "grad_norm": 1.3975757360458374,
      "learning_rate": 4.7012676393207364e-05,
      "loss": 0.0284,
      "step": 18136
    },
    {
      "epoch": 2.5419761737911704,
      "grad_norm": 0.17924724519252777,
      "learning_rate": 4.69983257593877e-05,
      "loss": 0.0285,
      "step": 18137
    },
    {
      "epoch": 2.542116327960757,
      "grad_norm": 0.21573258936405182,
      "learning_rate": 4.698397512556804e-05,
      "loss": 0.0047,
      "step": 18138
    },
    {
      "epoch": 2.5422564821303433,
      "grad_norm": 0.27791786193847656,
      "learning_rate": 4.696962449174838e-05,
      "loss": 0.0264,
      "step": 18139
    },
    {
      "epoch": 2.54239663629993,
      "grad_norm": 0.35525017976760864,
      "learning_rate": 4.695527385792872e-05,
      "loss": 0.0248,
      "step": 18140
    },
    {
      "epoch": 2.5425367904695166,
      "grad_norm": 0.17878513038158417,
      "learning_rate": 4.694092322410906e-05,
      "loss": 0.0207,
      "step": 18141
    },
    {
      "epoch": 2.542676944639103,
      "grad_norm": 0.527005672454834,
      "learning_rate": 4.6926572590289396e-05,
      "loss": 0.0286,
      "step": 18142
    },
    {
      "epoch": 2.5428170988086896,
      "grad_norm": 0.22158260643482208,
      "learning_rate": 4.6912221956469745e-05,
      "loss": 0.0187,
      "step": 18143
    },
    {
      "epoch": 2.542957252978276,
      "grad_norm": 0.1652626395225525,
      "learning_rate": 4.689787132265008e-05,
      "loss": 0.0115,
      "step": 18144
    },
    {
      "epoch": 2.5430974071478625,
      "grad_norm": 0.1802297681570053,
      "learning_rate": 4.6883520688830416e-05,
      "loss": 0.0165,
      "step": 18145
    },
    {
      "epoch": 2.543237561317449,
      "grad_norm": 0.09339119493961334,
      "learning_rate": 4.6869170055010765e-05,
      "loss": 0.015,
      "step": 18146
    },
    {
      "epoch": 2.543377715487036,
      "grad_norm": 0.12759867310523987,
      "learning_rate": 4.68548194211911e-05,
      "loss": 0.0146,
      "step": 18147
    },
    {
      "epoch": 2.5435178696566223,
      "grad_norm": 0.1906467080116272,
      "learning_rate": 4.6840468787371435e-05,
      "loss": 0.0101,
      "step": 18148
    },
    {
      "epoch": 2.5436580238262088,
      "grad_norm": 0.5868571400642395,
      "learning_rate": 4.682611815355178e-05,
      "loss": 0.034,
      "step": 18149
    },
    {
      "epoch": 2.543798177995795,
      "grad_norm": 0.0958685353398323,
      "learning_rate": 4.681176751973212e-05,
      "loss": 0.0129,
      "step": 18150
    },
    {
      "epoch": 2.543938332165382,
      "grad_norm": 0.31481021642684937,
      "learning_rate": 4.679741688591246e-05,
      "loss": 0.0549,
      "step": 18151
    },
    {
      "epoch": 2.5440784863349686,
      "grad_norm": 0.1680697351694107,
      "learning_rate": 4.67830662520928e-05,
      "loss": 0.0232,
      "step": 18152
    },
    {
      "epoch": 2.544218640504555,
      "grad_norm": 0.07928688079118729,
      "learning_rate": 4.676871561827313e-05,
      "loss": 0.0053,
      "step": 18153
    },
    {
      "epoch": 2.5443587946741415,
      "grad_norm": 0.11807805299758911,
      "learning_rate": 4.675436498445348e-05,
      "loss": 0.0226,
      "step": 18154
    },
    {
      "epoch": 2.544498948843728,
      "grad_norm": 0.31371504068374634,
      "learning_rate": 4.6740014350633816e-05,
      "loss": 0.0523,
      "step": 18155
    },
    {
      "epoch": 2.5446391030133144,
      "grad_norm": 0.1173236072063446,
      "learning_rate": 4.672566371681415e-05,
      "loss": 0.0055,
      "step": 18156
    },
    {
      "epoch": 2.5447792571829013,
      "grad_norm": 0.2525007724761963,
      "learning_rate": 4.6711313082994494e-05,
      "loss": 0.0292,
      "step": 18157
    },
    {
      "epoch": 2.544919411352488,
      "grad_norm": 0.31894198060035706,
      "learning_rate": 4.6696962449174836e-05,
      "loss": 0.0201,
      "step": 18158
    },
    {
      "epoch": 2.5450595655220742,
      "grad_norm": 0.30312296748161316,
      "learning_rate": 4.668261181535518e-05,
      "loss": 0.0072,
      "step": 18159
    },
    {
      "epoch": 2.5451997196916607,
      "grad_norm": 0.30419832468032837,
      "learning_rate": 4.6668261181535513e-05,
      "loss": 0.0206,
      "step": 18160
    },
    {
      "epoch": 2.5453398738612476,
      "grad_norm": 0.17003212869167328,
      "learning_rate": 4.665391054771585e-05,
      "loss": 0.0296,
      "step": 18161
    },
    {
      "epoch": 2.545480028030834,
      "grad_norm": 0.24882318079471588,
      "learning_rate": 4.66395599138962e-05,
      "loss": 0.0361,
      "step": 18162
    },
    {
      "epoch": 2.5456201822004205,
      "grad_norm": 0.03577204421162605,
      "learning_rate": 4.662520928007653e-05,
      "loss": 0.0022,
      "step": 18163
    },
    {
      "epoch": 2.545760336370007,
      "grad_norm": 0.17998141050338745,
      "learning_rate": 4.661085864625687e-05,
      "loss": 0.0872,
      "step": 18164
    },
    {
      "epoch": 2.5459004905395934,
      "grad_norm": 0.028033820912241936,
      "learning_rate": 4.659650801243721e-05,
      "loss": 0.0013,
      "step": 18165
    },
    {
      "epoch": 2.54604064470918,
      "grad_norm": 0.30691298842430115,
      "learning_rate": 4.658215737861755e-05,
      "loss": 0.0341,
      "step": 18166
    },
    {
      "epoch": 2.546180798878767,
      "grad_norm": 0.056820787489414215,
      "learning_rate": 4.6567806744797895e-05,
      "loss": 0.0032,
      "step": 18167
    },
    {
      "epoch": 2.5463209530483533,
      "grad_norm": 0.22577285766601562,
      "learning_rate": 4.655345611097823e-05,
      "loss": 0.0046,
      "step": 18168
    },
    {
      "epoch": 2.5464611072179397,
      "grad_norm": 0.5582699179649353,
      "learning_rate": 4.6539105477158565e-05,
      "loss": 0.0699,
      "step": 18169
    },
    {
      "epoch": 2.546601261387526,
      "grad_norm": 1.1404780149459839,
      "learning_rate": 4.6524754843338914e-05,
      "loss": 0.0952,
      "step": 18170
    },
    {
      "epoch": 2.546741415557113,
      "grad_norm": 0.13056795299053192,
      "learning_rate": 4.651040420951925e-05,
      "loss": 0.0105,
      "step": 18171
    },
    {
      "epoch": 2.5468815697266995,
      "grad_norm": 0.7744981646537781,
      "learning_rate": 4.6496053575699585e-05,
      "loss": 0.0711,
      "step": 18172
    },
    {
      "epoch": 2.547021723896286,
      "grad_norm": 0.19840537011623383,
      "learning_rate": 4.648170294187993e-05,
      "loss": 0.0429,
      "step": 18173
    },
    {
      "epoch": 2.5471618780658725,
      "grad_norm": 0.4157489836215973,
      "learning_rate": 4.646735230806027e-05,
      "loss": 0.0625,
      "step": 18174
    },
    {
      "epoch": 2.547302032235459,
      "grad_norm": 0.27275922894477844,
      "learning_rate": 4.645300167424061e-05,
      "loss": 0.024,
      "step": 18175
    },
    {
      "epoch": 2.5474421864050454,
      "grad_norm": 0.21879683434963226,
      "learning_rate": 4.6438651040420947e-05,
      "loss": 0.0624,
      "step": 18176
    },
    {
      "epoch": 2.547582340574632,
      "grad_norm": 0.1999858021736145,
      "learning_rate": 4.642430040660128e-05,
      "loss": 0.0219,
      "step": 18177
    },
    {
      "epoch": 2.5477224947442187,
      "grad_norm": 0.12946660816669464,
      "learning_rate": 4.640994977278163e-05,
      "loss": 0.0161,
      "step": 18178
    },
    {
      "epoch": 2.547862648913805,
      "grad_norm": 0.1165948137640953,
      "learning_rate": 4.6395599138961966e-05,
      "loss": 0.0158,
      "step": 18179
    },
    {
      "epoch": 2.5480028030833917,
      "grad_norm": 0.20548461377620697,
      "learning_rate": 4.63812485051423e-05,
      "loss": 0.0516,
      "step": 18180
    },
    {
      "epoch": 2.548142957252978,
      "grad_norm": 0.2472551017999649,
      "learning_rate": 4.6366897871322644e-05,
      "loss": 0.043,
      "step": 18181
    },
    {
      "epoch": 2.548283111422565,
      "grad_norm": 0.11097093671560287,
      "learning_rate": 4.6352547237502986e-05,
      "loss": 0.006,
      "step": 18182
    },
    {
      "epoch": 2.5484232655921515,
      "grad_norm": 0.133823961019516,
      "learning_rate": 4.633819660368333e-05,
      "loss": 0.0147,
      "step": 18183
    },
    {
      "epoch": 2.548563419761738,
      "grad_norm": 0.3769824504852295,
      "learning_rate": 4.632384596986366e-05,
      "loss": 0.0262,
      "step": 18184
    },
    {
      "epoch": 2.5487035739313244,
      "grad_norm": 0.43822044134140015,
      "learning_rate": 4.6309495336044005e-05,
      "loss": 0.0207,
      "step": 18185
    },
    {
      "epoch": 2.548843728100911,
      "grad_norm": 0.48032504320144653,
      "learning_rate": 4.629514470222435e-05,
      "loss": 0.0294,
      "step": 18186
    },
    {
      "epoch": 2.5489838822704973,
      "grad_norm": 0.18290066719055176,
      "learning_rate": 4.628079406840468e-05,
      "loss": 0.0134,
      "step": 18187
    },
    {
      "epoch": 2.5491240364400842,
      "grad_norm": 0.22822889685630798,
      "learning_rate": 4.626644343458502e-05,
      "loss": 0.0146,
      "step": 18188
    },
    {
      "epoch": 2.5492641906096707,
      "grad_norm": 0.49276286363601685,
      "learning_rate": 4.625209280076537e-05,
      "loss": 0.0585,
      "step": 18189
    },
    {
      "epoch": 2.549404344779257,
      "grad_norm": 0.29231491684913635,
      "learning_rate": 4.62377421669457e-05,
      "loss": 0.0173,
      "step": 18190
    },
    {
      "epoch": 2.5495444989488436,
      "grad_norm": 0.08550060540437698,
      "learning_rate": 4.6223391533126044e-05,
      "loss": 0.011,
      "step": 18191
    },
    {
      "epoch": 2.5496846531184305,
      "grad_norm": 0.10867733508348465,
      "learning_rate": 4.620904089930638e-05,
      "loss": 0.0049,
      "step": 18192
    },
    {
      "epoch": 2.549824807288017,
      "grad_norm": 0.17415931820869446,
      "learning_rate": 4.619469026548672e-05,
      "loss": 0.0437,
      "step": 18193
    },
    {
      "epoch": 2.5499649614576034,
      "grad_norm": 0.3211270868778229,
      "learning_rate": 4.6180339631667064e-05,
      "loss": 0.0439,
      "step": 18194
    },
    {
      "epoch": 2.55010511562719,
      "grad_norm": 0.09146318584680557,
      "learning_rate": 4.61659889978474e-05,
      "loss": 0.0078,
      "step": 18195
    },
    {
      "epoch": 2.5502452697967763,
      "grad_norm": 0.0607328936457634,
      "learning_rate": 4.6151638364027735e-05,
      "loss": 0.0043,
      "step": 18196
    },
    {
      "epoch": 2.550385423966363,
      "grad_norm": 0.5442698001861572,
      "learning_rate": 4.6137287730208084e-05,
      "loss": 0.0186,
      "step": 18197
    },
    {
      "epoch": 2.5505255781359493,
      "grad_norm": 0.25543537735939026,
      "learning_rate": 4.612293709638842e-05,
      "loss": 0.0295,
      "step": 18198
    },
    {
      "epoch": 2.550665732305536,
      "grad_norm": 0.4715573787689209,
      "learning_rate": 4.610858646256876e-05,
      "loss": 0.038,
      "step": 18199
    },
    {
      "epoch": 2.5508058864751226,
      "grad_norm": 0.23139479756355286,
      "learning_rate": 4.6094235828749096e-05,
      "loss": 0.0307,
      "step": 18200
    },
    {
      "epoch": 2.550946040644709,
      "grad_norm": 0.2581072747707367,
      "learning_rate": 4.607988519492944e-05,
      "loss": 0.0473,
      "step": 18201
    },
    {
      "epoch": 2.551086194814296,
      "grad_norm": 0.12375976890325546,
      "learning_rate": 4.606553456110978e-05,
      "loss": 0.0139,
      "step": 18202
    },
    {
      "epoch": 2.5512263489838825,
      "grad_norm": 0.4782208800315857,
      "learning_rate": 4.6051183927290116e-05,
      "loss": 0.038,
      "step": 18203
    },
    {
      "epoch": 2.551366503153469,
      "grad_norm": 0.24449339509010315,
      "learning_rate": 4.603683329347045e-05,
      "loss": 0.0269,
      "step": 18204
    },
    {
      "epoch": 2.5515066573230554,
      "grad_norm": 0.10471366345882416,
      "learning_rate": 4.60224826596508e-05,
      "loss": 0.0058,
      "step": 18205
    },
    {
      "epoch": 2.551646811492642,
      "grad_norm": 0.15542851388454437,
      "learning_rate": 4.6008132025831135e-05,
      "loss": 0.0176,
      "step": 18206
    },
    {
      "epoch": 2.5517869656622283,
      "grad_norm": 0.34025418758392334,
      "learning_rate": 4.599378139201148e-05,
      "loss": 0.024,
      "step": 18207
    },
    {
      "epoch": 2.5519271198318147,
      "grad_norm": 0.3624800145626068,
      "learning_rate": 4.597943075819181e-05,
      "loss": 0.0263,
      "step": 18208
    },
    {
      "epoch": 2.5520672740014017,
      "grad_norm": 0.49432995915412903,
      "learning_rate": 4.5965080124372155e-05,
      "loss": 0.0384,
      "step": 18209
    },
    {
      "epoch": 2.552207428170988,
      "grad_norm": 0.45202943682670593,
      "learning_rate": 4.59507294905525e-05,
      "loss": 0.092,
      "step": 18210
    },
    {
      "epoch": 2.5523475823405746,
      "grad_norm": 0.7757467031478882,
      "learning_rate": 4.593637885673283e-05,
      "loss": 0.0767,
      "step": 18211
    },
    {
      "epoch": 2.552487736510161,
      "grad_norm": 0.32253241539001465,
      "learning_rate": 4.592202822291317e-05,
      "loss": 0.0206,
      "step": 18212
    },
    {
      "epoch": 2.552627890679748,
      "grad_norm": 0.09224551916122437,
      "learning_rate": 4.590767758909352e-05,
      "loss": 0.0052,
      "step": 18213
    },
    {
      "epoch": 2.5527680448493344,
      "grad_norm": 0.5775570869445801,
      "learning_rate": 4.589332695527385e-05,
      "loss": 0.0883,
      "step": 18214
    },
    {
      "epoch": 2.552908199018921,
      "grad_norm": 0.15705999732017517,
      "learning_rate": 4.5878976321454194e-05,
      "loss": 0.0071,
      "step": 18215
    },
    {
      "epoch": 2.5530483531885073,
      "grad_norm": 0.5348058342933655,
      "learning_rate": 4.586462568763453e-05,
      "loss": 0.0362,
      "step": 18216
    },
    {
      "epoch": 2.5531885073580938,
      "grad_norm": 0.14788256585597992,
      "learning_rate": 4.585027505381488e-05,
      "loss": 0.0092,
      "step": 18217
    },
    {
      "epoch": 2.5533286615276802,
      "grad_norm": 0.2483576536178589,
      "learning_rate": 4.5835924419995214e-05,
      "loss": 0.0437,
      "step": 18218
    },
    {
      "epoch": 2.553468815697267,
      "grad_norm": 0.49410519003868103,
      "learning_rate": 4.582157378617555e-05,
      "loss": 0.0121,
      "step": 18219
    },
    {
      "epoch": 2.5536089698668536,
      "grad_norm": 2.821918249130249,
      "learning_rate": 4.58072231523559e-05,
      "loss": 0.0732,
      "step": 18220
    },
    {
      "epoch": 2.55374912403644,
      "grad_norm": 0.22442418336868286,
      "learning_rate": 4.579287251853623e-05,
      "loss": 0.0508,
      "step": 18221
    },
    {
      "epoch": 2.5538892782060265,
      "grad_norm": 0.09452279657125473,
      "learning_rate": 4.577852188471657e-05,
      "loss": 0.0037,
      "step": 18222
    },
    {
      "epoch": 2.5540294323756134,
      "grad_norm": 0.0503377839922905,
      "learning_rate": 4.576417125089691e-05,
      "loss": 0.0053,
      "step": 18223
    },
    {
      "epoch": 2.5541695865452,
      "grad_norm": 0.32724228501319885,
      "learning_rate": 4.574982061707725e-05,
      "loss": 0.0261,
      "step": 18224
    },
    {
      "epoch": 2.5543097407147863,
      "grad_norm": 0.1376713514328003,
      "learning_rate": 4.5735469983257595e-05,
      "loss": 0.0257,
      "step": 18225
    },
    {
      "epoch": 2.554449894884373,
      "grad_norm": 0.235674187541008,
      "learning_rate": 4.572111934943793e-05,
      "loss": 0.017,
      "step": 18226
    },
    {
      "epoch": 2.5545900490539593,
      "grad_norm": 0.12057920545339584,
      "learning_rate": 4.5706768715618266e-05,
      "loss": 0.0232,
      "step": 18227
    },
    {
      "epoch": 2.5547302032235457,
      "grad_norm": 1.7068744897842407,
      "learning_rate": 4.5692418081798614e-05,
      "loss": 0.0718,
      "step": 18228
    },
    {
      "epoch": 2.554870357393132,
      "grad_norm": 0.3147523105144501,
      "learning_rate": 4.567806744797895e-05,
      "loss": 0.0235,
      "step": 18229
    },
    {
      "epoch": 2.555010511562719,
      "grad_norm": 0.21371452510356903,
      "learning_rate": 4.5663716814159285e-05,
      "loss": 0.0316,
      "step": 18230
    },
    {
      "epoch": 2.5551506657323055,
      "grad_norm": 0.39048099517822266,
      "learning_rate": 4.564936618033963e-05,
      "loss": 0.0276,
      "step": 18231
    },
    {
      "epoch": 2.555290819901892,
      "grad_norm": 0.18305683135986328,
      "learning_rate": 4.563501554651997e-05,
      "loss": 0.0135,
      "step": 18232
    },
    {
      "epoch": 2.555430974071479,
      "grad_norm": 0.20129340887069702,
      "learning_rate": 4.562066491270031e-05,
      "loss": 0.0495,
      "step": 18233
    },
    {
      "epoch": 2.5555711282410654,
      "grad_norm": 0.21871869266033173,
      "learning_rate": 4.560631427888065e-05,
      "loss": 0.0162,
      "step": 18234
    },
    {
      "epoch": 2.555711282410652,
      "grad_norm": 0.23431149125099182,
      "learning_rate": 4.559196364506098e-05,
      "loss": 0.0483,
      "step": 18235
    },
    {
      "epoch": 2.5558514365802383,
      "grad_norm": 0.14710527658462524,
      "learning_rate": 4.557761301124133e-05,
      "loss": 0.0079,
      "step": 18236
    },
    {
      "epoch": 2.5559915907498247,
      "grad_norm": 0.19244778156280518,
      "learning_rate": 4.5563262377421666e-05,
      "loss": 0.0283,
      "step": 18237
    },
    {
      "epoch": 2.556131744919411,
      "grad_norm": 0.16671106219291687,
      "learning_rate": 4.5548911743602e-05,
      "loss": 0.0169,
      "step": 18238
    },
    {
      "epoch": 2.5562718990889977,
      "grad_norm": 0.09498811513185501,
      "learning_rate": 4.5534561109782344e-05,
      "loss": 0.0095,
      "step": 18239
    },
    {
      "epoch": 2.5564120532585846,
      "grad_norm": 0.37462374567985535,
      "learning_rate": 4.5520210475962686e-05,
      "loss": 0.0089,
      "step": 18240
    },
    {
      "epoch": 2.556552207428171,
      "grad_norm": 0.19506587088108063,
      "learning_rate": 4.550585984214303e-05,
      "loss": 0.0177,
      "step": 18241
    },
    {
      "epoch": 2.5566923615977575,
      "grad_norm": 0.2198793590068817,
      "learning_rate": 4.5491509208323363e-05,
      "loss": 0.0116,
      "step": 18242
    },
    {
      "epoch": 2.556832515767344,
      "grad_norm": 0.02570890448987484,
      "learning_rate": 4.54771585745037e-05,
      "loss": 0.0023,
      "step": 18243
    },
    {
      "epoch": 2.556972669936931,
      "grad_norm": 0.24092863500118256,
      "learning_rate": 4.546280794068405e-05,
      "loss": 0.0553,
      "step": 18244
    },
    {
      "epoch": 2.5571128241065173,
      "grad_norm": 0.1576327234506607,
      "learning_rate": 4.544845730686438e-05,
      "loss": 0.0168,
      "step": 18245
    },
    {
      "epoch": 2.5572529782761038,
      "grad_norm": 0.24204587936401367,
      "learning_rate": 4.543410667304472e-05,
      "loss": 0.0232,
      "step": 18246
    },
    {
      "epoch": 2.55739313244569,
      "grad_norm": 0.10460522025823593,
      "learning_rate": 4.541975603922506e-05,
      "loss": 0.0073,
      "step": 18247
    },
    {
      "epoch": 2.5575332866152767,
      "grad_norm": 0.15091226994991302,
      "learning_rate": 4.54054054054054e-05,
      "loss": 0.0056,
      "step": 18248
    },
    {
      "epoch": 2.557673440784863,
      "grad_norm": 0.13097040355205536,
      "learning_rate": 4.5391054771585745e-05,
      "loss": 0.0123,
      "step": 18249
    },
    {
      "epoch": 2.55781359495445,
      "grad_norm": 0.09134726971387863,
      "learning_rate": 4.537670413776608e-05,
      "loss": 0.0071,
      "step": 18250
    },
    {
      "epoch": 2.5579537491240365,
      "grad_norm": 0.05973028019070625,
      "learning_rate": 4.5362353503946415e-05,
      "loss": 0.0047,
      "step": 18251
    },
    {
      "epoch": 2.558093903293623,
      "grad_norm": 0.4121887683868408,
      "learning_rate": 4.5348002870126764e-05,
      "loss": 0.0248,
      "step": 18252
    },
    {
      "epoch": 2.5582340574632094,
      "grad_norm": 0.3629038631916046,
      "learning_rate": 4.53336522363071e-05,
      "loss": 0.0472,
      "step": 18253
    },
    {
      "epoch": 2.5583742116327963,
      "grad_norm": 0.8563442826271057,
      "learning_rate": 4.5319301602487435e-05,
      "loss": 0.0299,
      "step": 18254
    },
    {
      "epoch": 2.558514365802383,
      "grad_norm": 0.1299944818019867,
      "learning_rate": 4.530495096866778e-05,
      "loss": 0.0081,
      "step": 18255
    },
    {
      "epoch": 2.5586545199719692,
      "grad_norm": 0.3666844069957733,
      "learning_rate": 4.529060033484812e-05,
      "loss": 0.064,
      "step": 18256
    },
    {
      "epoch": 2.5587946741415557,
      "grad_norm": 0.18424613773822784,
      "learning_rate": 4.527624970102846e-05,
      "loss": 0.0087,
      "step": 18257
    },
    {
      "epoch": 2.558934828311142,
      "grad_norm": 0.06487855315208435,
      "learning_rate": 4.5261899067208797e-05,
      "loss": 0.0214,
      "step": 18258
    },
    {
      "epoch": 2.5590749824807286,
      "grad_norm": 0.193250373005867,
      "learning_rate": 4.524754843338914e-05,
      "loss": 0.0212,
      "step": 18259
    },
    {
      "epoch": 2.559215136650315,
      "grad_norm": 0.05787370726466179,
      "learning_rate": 4.523319779956948e-05,
      "loss": 0.0087,
      "step": 18260
    },
    {
      "epoch": 2.559355290819902,
      "grad_norm": 0.0944092720746994,
      "learning_rate": 4.5218847165749816e-05,
      "loss": 0.0035,
      "step": 18261
    },
    {
      "epoch": 2.5594954449894884,
      "grad_norm": 0.300245463848114,
      "learning_rate": 4.520449653193015e-05,
      "loss": 0.014,
      "step": 18262
    },
    {
      "epoch": 2.559635599159075,
      "grad_norm": 0.19826042652130127,
      "learning_rate": 4.51901458981105e-05,
      "loss": 0.008,
      "step": 18263
    },
    {
      "epoch": 2.5597757533286614,
      "grad_norm": 0.8386831879615784,
      "learning_rate": 4.5175795264290836e-05,
      "loss": 0.0447,
      "step": 18264
    },
    {
      "epoch": 2.5599159074982483,
      "grad_norm": 0.5162139534950256,
      "learning_rate": 4.516144463047118e-05,
      "loss": 0.0505,
      "step": 18265
    },
    {
      "epoch": 2.5600560616678347,
      "grad_norm": 0.18964841961860657,
      "learning_rate": 4.514709399665151e-05,
      "loss": 0.0378,
      "step": 18266
    },
    {
      "epoch": 2.560196215837421,
      "grad_norm": 0.5080700516700745,
      "learning_rate": 4.5132743362831855e-05,
      "loss": 0.0626,
      "step": 18267
    },
    {
      "epoch": 2.5603363700070076,
      "grad_norm": 0.2166011482477188,
      "learning_rate": 4.51183927290122e-05,
      "loss": 0.0186,
      "step": 18268
    },
    {
      "epoch": 2.560476524176594,
      "grad_norm": 0.7766144275665283,
      "learning_rate": 4.510404209519253e-05,
      "loss": 0.0774,
      "step": 18269
    },
    {
      "epoch": 2.5606166783461806,
      "grad_norm": 0.3797752857208252,
      "learning_rate": 4.508969146137287e-05,
      "loss": 0.0069,
      "step": 18270
    },
    {
      "epoch": 2.5607568325157675,
      "grad_norm": 0.7576006650924683,
      "learning_rate": 4.507534082755322e-05,
      "loss": 0.0343,
      "step": 18271
    },
    {
      "epoch": 2.560896986685354,
      "grad_norm": 0.544466495513916,
      "learning_rate": 4.506099019373355e-05,
      "loss": 0.0312,
      "step": 18272
    },
    {
      "epoch": 2.5610371408549404,
      "grad_norm": 0.38504084944725037,
      "learning_rate": 4.5046639559913894e-05,
      "loss": 0.043,
      "step": 18273
    },
    {
      "epoch": 2.561177295024527,
      "grad_norm": 0.19185112416744232,
      "learning_rate": 4.503228892609423e-05,
      "loss": 0.0166,
      "step": 18274
    },
    {
      "epoch": 2.5613174491941137,
      "grad_norm": 0.24532164633274078,
      "learning_rate": 4.501793829227457e-05,
      "loss": 0.0378,
      "step": 18275
    },
    {
      "epoch": 2.5614576033637,
      "grad_norm": 0.8023670315742493,
      "learning_rate": 4.5003587658454914e-05,
      "loss": 0.0269,
      "step": 18276
    },
    {
      "epoch": 2.5615977575332867,
      "grad_norm": 0.2732026278972626,
      "learning_rate": 4.498923702463525e-05,
      "loss": 0.0174,
      "step": 18277
    },
    {
      "epoch": 2.561737911702873,
      "grad_norm": 0.12387651950120926,
      "learning_rate": 4.4974886390815585e-05,
      "loss": 0.0114,
      "step": 18278
    },
    {
      "epoch": 2.5618780658724596,
      "grad_norm": 0.13737429678440094,
      "learning_rate": 4.4960535756995933e-05,
      "loss": 0.0188,
      "step": 18279
    },
    {
      "epoch": 2.562018220042046,
      "grad_norm": 0.31917428970336914,
      "learning_rate": 4.494618512317627e-05,
      "loss": 0.0626,
      "step": 18280
    },
    {
      "epoch": 2.562158374211633,
      "grad_norm": 0.08627309650182724,
      "learning_rate": 4.493183448935661e-05,
      "loss": 0.0042,
      "step": 18281
    },
    {
      "epoch": 2.5622985283812194,
      "grad_norm": 0.15085409581661224,
      "learning_rate": 4.4917483855536946e-05,
      "loss": 0.0275,
      "step": 18282
    },
    {
      "epoch": 2.562438682550806,
      "grad_norm": 0.15053768455982208,
      "learning_rate": 4.490313322171729e-05,
      "loss": 0.0086,
      "step": 18283
    },
    {
      "epoch": 2.5625788367203923,
      "grad_norm": 0.23506274819374084,
      "learning_rate": 4.488878258789763e-05,
      "loss": 0.0386,
      "step": 18284
    },
    {
      "epoch": 2.5627189908899792,
      "grad_norm": 0.23363982141017914,
      "learning_rate": 4.4874431954077966e-05,
      "loss": 0.0331,
      "step": 18285
    },
    {
      "epoch": 2.5628591450595657,
      "grad_norm": 0.2985607385635376,
      "learning_rate": 4.48600813202583e-05,
      "loss": 0.0441,
      "step": 18286
    },
    {
      "epoch": 2.562999299229152,
      "grad_norm": 0.15091802179813385,
      "learning_rate": 4.484573068643865e-05,
      "loss": 0.0104,
      "step": 18287
    },
    {
      "epoch": 2.5631394533987386,
      "grad_norm": 0.026273801922798157,
      "learning_rate": 4.4831380052618985e-05,
      "loss": 0.001,
      "step": 18288
    },
    {
      "epoch": 2.563279607568325,
      "grad_norm": 0.10379917919635773,
      "learning_rate": 4.481702941879933e-05,
      "loss": 0.0189,
      "step": 18289
    },
    {
      "epoch": 2.5634197617379115,
      "grad_norm": 0.33128872513771057,
      "learning_rate": 4.480267878497966e-05,
      "loss": 0.0465,
      "step": 18290
    },
    {
      "epoch": 2.563559915907498,
      "grad_norm": 0.3149373531341553,
      "learning_rate": 4.4788328151160005e-05,
      "loss": 0.0224,
      "step": 18291
    },
    {
      "epoch": 2.563700070077085,
      "grad_norm": 0.1765746921300888,
      "learning_rate": 4.477397751734035e-05,
      "loss": 0.0147,
      "step": 18292
    },
    {
      "epoch": 2.5638402242466714,
      "grad_norm": 0.06919504702091217,
      "learning_rate": 4.475962688352068e-05,
      "loss": 0.0072,
      "step": 18293
    },
    {
      "epoch": 2.563980378416258,
      "grad_norm": 0.51337069272995,
      "learning_rate": 4.474527624970103e-05,
      "loss": 0.0545,
      "step": 18294
    },
    {
      "epoch": 2.5641205325858443,
      "grad_norm": 0.16400989890098572,
      "learning_rate": 4.473092561588137e-05,
      "loss": 0.0521,
      "step": 18295
    },
    {
      "epoch": 2.564260686755431,
      "grad_norm": 0.19401909410953522,
      "learning_rate": 4.47165749820617e-05,
      "loss": 0.0347,
      "step": 18296
    },
    {
      "epoch": 2.5644008409250176,
      "grad_norm": 0.16718193888664246,
      "learning_rate": 4.4702224348242044e-05,
      "loss": 0.0113,
      "step": 18297
    },
    {
      "epoch": 2.564540995094604,
      "grad_norm": 0.3253571689128876,
      "learning_rate": 4.4687873714422386e-05,
      "loss": 0.0464,
      "step": 18298
    },
    {
      "epoch": 2.5646811492641906,
      "grad_norm": 0.6502189040184021,
      "learning_rate": 4.467352308060272e-05,
      "loss": 0.0357,
      "step": 18299
    },
    {
      "epoch": 2.564821303433777,
      "grad_norm": 0.3103705644607544,
      "learning_rate": 4.4659172446783064e-05,
      "loss": 0.0171,
      "step": 18300
    },
    {
      "epoch": 2.5649614576033635,
      "grad_norm": 0.9361165761947632,
      "learning_rate": 4.46448218129634e-05,
      "loss": 0.0355,
      "step": 18301
    },
    {
      "epoch": 2.5651016117729504,
      "grad_norm": 0.13929978013038635,
      "learning_rate": 4.463047117914375e-05,
      "loss": 0.0454,
      "step": 18302
    },
    {
      "epoch": 2.565241765942537,
      "grad_norm": 0.2941024601459503,
      "learning_rate": 4.461612054532408e-05,
      "loss": 0.0681,
      "step": 18303
    },
    {
      "epoch": 2.5653819201121233,
      "grad_norm": 0.3907836973667145,
      "learning_rate": 4.460176991150442e-05,
      "loss": 0.1028,
      "step": 18304
    },
    {
      "epoch": 2.5655220742817098,
      "grad_norm": 0.7705672979354858,
      "learning_rate": 4.458741927768476e-05,
      "loss": 0.0259,
      "step": 18305
    },
    {
      "epoch": 2.5656622284512967,
      "grad_norm": 0.164019376039505,
      "learning_rate": 4.45730686438651e-05,
      "loss": 0.0041,
      "step": 18306
    },
    {
      "epoch": 2.565802382620883,
      "grad_norm": 0.20495638251304626,
      "learning_rate": 4.455871801004544e-05,
      "loss": 0.0061,
      "step": 18307
    },
    {
      "epoch": 2.5659425367904696,
      "grad_norm": 0.24308723211288452,
      "learning_rate": 4.454436737622578e-05,
      "loss": 0.0391,
      "step": 18308
    },
    {
      "epoch": 2.566082690960056,
      "grad_norm": 0.1835632175207138,
      "learning_rate": 4.4530016742406116e-05,
      "loss": 0.0288,
      "step": 18309
    },
    {
      "epoch": 2.5662228451296425,
      "grad_norm": 0.233128622174263,
      "learning_rate": 4.4515666108586464e-05,
      "loss": 0.0534,
      "step": 18310
    },
    {
      "epoch": 2.566362999299229,
      "grad_norm": 0.2201002985239029,
      "learning_rate": 4.45013154747668e-05,
      "loss": 0.0131,
      "step": 18311
    },
    {
      "epoch": 2.566503153468816,
      "grad_norm": 0.6070578098297119,
      "learning_rate": 4.4486964840947135e-05,
      "loss": 0.0171,
      "step": 18312
    },
    {
      "epoch": 2.5666433076384023,
      "grad_norm": 0.3898150622844696,
      "learning_rate": 4.447261420712748e-05,
      "loss": 0.0749,
      "step": 18313
    },
    {
      "epoch": 2.5667834618079888,
      "grad_norm": 0.22572098672389984,
      "learning_rate": 4.445826357330782e-05,
      "loss": 0.0325,
      "step": 18314
    },
    {
      "epoch": 2.5669236159775752,
      "grad_norm": 0.2406264841556549,
      "learning_rate": 4.4443912939488155e-05,
      "loss": 0.0568,
      "step": 18315
    },
    {
      "epoch": 2.567063770147162,
      "grad_norm": 0.9114459753036499,
      "learning_rate": 4.44295623056685e-05,
      "loss": 0.0264,
      "step": 18316
    },
    {
      "epoch": 2.5672039243167486,
      "grad_norm": 0.1865604668855667,
      "learning_rate": 4.441521167184883e-05,
      "loss": 0.0163,
      "step": 18317
    },
    {
      "epoch": 2.567344078486335,
      "grad_norm": 0.24366647005081177,
      "learning_rate": 4.440086103802918e-05,
      "loss": 0.0222,
      "step": 18318
    },
    {
      "epoch": 2.5674842326559215,
      "grad_norm": 0.48044905066490173,
      "learning_rate": 4.4386510404209516e-05,
      "loss": 0.1255,
      "step": 18319
    },
    {
      "epoch": 2.567624386825508,
      "grad_norm": 0.4263973832130432,
      "learning_rate": 4.437215977038985e-05,
      "loss": 0.0369,
      "step": 18320
    },
    {
      "epoch": 2.5677645409950944,
      "grad_norm": 0.06407378613948822,
      "learning_rate": 4.4357809136570194e-05,
      "loss": 0.0057,
      "step": 18321
    },
    {
      "epoch": 2.567904695164681,
      "grad_norm": 0.12242628633975983,
      "learning_rate": 4.4343458502750536e-05,
      "loss": 0.0348,
      "step": 18322
    },
    {
      "epoch": 2.568044849334268,
      "grad_norm": 0.34747666120529175,
      "learning_rate": 4.432910786893087e-05,
      "loss": 0.0128,
      "step": 18323
    },
    {
      "epoch": 2.5681850035038543,
      "grad_norm": 0.04966302961111069,
      "learning_rate": 4.431475723511121e-05,
      "loss": 0.0039,
      "step": 18324
    },
    {
      "epoch": 2.5683251576734407,
      "grad_norm": 0.27620163559913635,
      "learning_rate": 4.430040660129155e-05,
      "loss": 0.0219,
      "step": 18325
    },
    {
      "epoch": 2.568465311843027,
      "grad_norm": 0.19051134586334229,
      "learning_rate": 4.42860559674719e-05,
      "loss": 0.0227,
      "step": 18326
    },
    {
      "epoch": 2.568605466012614,
      "grad_norm": 0.25438886880874634,
      "learning_rate": 4.427170533365223e-05,
      "loss": 0.0516,
      "step": 18327
    },
    {
      "epoch": 2.5687456201822005,
      "grad_norm": 0.2635815739631653,
      "learning_rate": 4.425735469983257e-05,
      "loss": 0.0095,
      "step": 18328
    },
    {
      "epoch": 2.568885774351787,
      "grad_norm": 0.3351733684539795,
      "learning_rate": 4.424300406601291e-05,
      "loss": 0.042,
      "step": 18329
    },
    {
      "epoch": 2.5690259285213735,
      "grad_norm": 0.2577472925186157,
      "learning_rate": 4.422865343219325e-05,
      "loss": 0.0472,
      "step": 18330
    },
    {
      "epoch": 2.56916608269096,
      "grad_norm": 0.10261719673871994,
      "learning_rate": 4.421430279837359e-05,
      "loss": 0.0105,
      "step": 18331
    },
    {
      "epoch": 2.5693062368605464,
      "grad_norm": 0.4056299328804016,
      "learning_rate": 4.419995216455393e-05,
      "loss": 0.0171,
      "step": 18332
    },
    {
      "epoch": 2.5694463910301333,
      "grad_norm": 0.14270593225955963,
      "learning_rate": 4.418560153073427e-05,
      "loss": 0.0279,
      "step": 18333
    },
    {
      "epoch": 2.5695865451997197,
      "grad_norm": 0.12293795496225357,
      "learning_rate": 4.4171250896914614e-05,
      "loss": 0.0206,
      "step": 18334
    },
    {
      "epoch": 2.569726699369306,
      "grad_norm": 0.24613048136234283,
      "learning_rate": 4.415690026309495e-05,
      "loss": 0.0178,
      "step": 18335
    },
    {
      "epoch": 2.5698668535388927,
      "grad_norm": 0.5383864045143127,
      "learning_rate": 4.4142549629275285e-05,
      "loss": 0.0654,
      "step": 18336
    },
    {
      "epoch": 2.5700070077084796,
      "grad_norm": 0.22100849449634552,
      "learning_rate": 4.4128198995455634e-05,
      "loss": 0.067,
      "step": 18337
    },
    {
      "epoch": 2.570147161878066,
      "grad_norm": 0.35696256160736084,
      "learning_rate": 4.411384836163597e-05,
      "loss": 0.0164,
      "step": 18338
    },
    {
      "epoch": 2.5702873160476525,
      "grad_norm": 0.18946433067321777,
      "learning_rate": 4.4099497727816304e-05,
      "loss": 0.0214,
      "step": 18339
    },
    {
      "epoch": 2.570427470217239,
      "grad_norm": 0.20125623047351837,
      "learning_rate": 4.4085147093996646e-05,
      "loss": 0.0298,
      "step": 18340
    },
    {
      "epoch": 2.5705676243868254,
      "grad_norm": 0.2019723355770111,
      "learning_rate": 4.407079646017699e-05,
      "loss": 0.0384,
      "step": 18341
    },
    {
      "epoch": 2.570707778556412,
      "grad_norm": 0.09664402902126312,
      "learning_rate": 4.405644582635733e-05,
      "loss": 0.0237,
      "step": 18342
    },
    {
      "epoch": 2.5708479327259988,
      "grad_norm": 0.10501237958669662,
      "learning_rate": 4.4042095192537666e-05,
      "loss": 0.006,
      "step": 18343
    },
    {
      "epoch": 2.5709880868955852,
      "grad_norm": 0.1476975679397583,
      "learning_rate": 4.4027744558718e-05,
      "loss": 0.0114,
      "step": 18344
    },
    {
      "epoch": 2.5711282410651717,
      "grad_norm": 0.7946990728378296,
      "learning_rate": 4.401339392489835e-05,
      "loss": 0.0607,
      "step": 18345
    },
    {
      "epoch": 2.571268395234758,
      "grad_norm": 0.08263041824102402,
      "learning_rate": 4.3999043291078686e-05,
      "loss": 0.0061,
      "step": 18346
    },
    {
      "epoch": 2.571408549404345,
      "grad_norm": 0.3564976453781128,
      "learning_rate": 4.398469265725902e-05,
      "loss": 0.0498,
      "step": 18347
    },
    {
      "epoch": 2.5715487035739315,
      "grad_norm": 0.18800511956214905,
      "learning_rate": 4.397034202343936e-05,
      "loss": 0.0357,
      "step": 18348
    },
    {
      "epoch": 2.571688857743518,
      "grad_norm": 0.05083693563938141,
      "learning_rate": 4.3955991389619705e-05,
      "loss": 0.0018,
      "step": 18349
    },
    {
      "epoch": 2.5718290119131044,
      "grad_norm": 0.10608505457639694,
      "learning_rate": 4.394164075580005e-05,
      "loss": 0.0098,
      "step": 18350
    },
    {
      "epoch": 2.571969166082691,
      "grad_norm": 0.5920299291610718,
      "learning_rate": 4.392729012198038e-05,
      "loss": 0.0245,
      "step": 18351
    },
    {
      "epoch": 2.5721093202522773,
      "grad_norm": 0.14916224777698517,
      "learning_rate": 4.391293948816072e-05,
      "loss": 0.0189,
      "step": 18352
    },
    {
      "epoch": 2.572249474421864,
      "grad_norm": 0.2072741836309433,
      "learning_rate": 4.389858885434107e-05,
      "loss": 0.0269,
      "step": 18353
    },
    {
      "epoch": 2.5723896285914507,
      "grad_norm": 0.13273829221725464,
      "learning_rate": 4.38842382205214e-05,
      "loss": 0.026,
      "step": 18354
    },
    {
      "epoch": 2.572529782761037,
      "grad_norm": 0.034441083669662476,
      "learning_rate": 4.386988758670174e-05,
      "loss": 0.0024,
      "step": 18355
    },
    {
      "epoch": 2.5726699369306236,
      "grad_norm": 0.27286863327026367,
      "learning_rate": 4.385553695288208e-05,
      "loss": 0.0217,
      "step": 18356
    },
    {
      "epoch": 2.57281009110021,
      "grad_norm": 0.16159188747406006,
      "learning_rate": 4.384118631906242e-05,
      "loss": 0.0298,
      "step": 18357
    },
    {
      "epoch": 2.572950245269797,
      "grad_norm": 0.19649852812290192,
      "learning_rate": 4.3826835685242764e-05,
      "loss": 0.0216,
      "step": 18358
    },
    {
      "epoch": 2.5730903994393834,
      "grad_norm": 0.340781033039093,
      "learning_rate": 4.38124850514231e-05,
      "loss": 0.03,
      "step": 18359
    },
    {
      "epoch": 2.57323055360897,
      "grad_norm": 0.21577833592891693,
      "learning_rate": 4.3798134417603435e-05,
      "loss": 0.0254,
      "step": 18360
    },
    {
      "epoch": 2.5733707077785564,
      "grad_norm": 0.19541950523853302,
      "learning_rate": 4.3783783783783783e-05,
      "loss": 0.0093,
      "step": 18361
    },
    {
      "epoch": 2.573510861948143,
      "grad_norm": 0.3279116451740265,
      "learning_rate": 4.376943314996412e-05,
      "loss": 0.0658,
      "step": 18362
    },
    {
      "epoch": 2.5736510161177293,
      "grad_norm": 0.18248826265335083,
      "learning_rate": 4.3755082516144454e-05,
      "loss": 0.0179,
      "step": 18363
    },
    {
      "epoch": 2.573791170287316,
      "grad_norm": 0.4545961022377014,
      "learning_rate": 4.3740731882324796e-05,
      "loss": 0.0204,
      "step": 18364
    },
    {
      "epoch": 2.5739313244569026,
      "grad_norm": 0.0790606439113617,
      "learning_rate": 4.372638124850514e-05,
      "loss": 0.0071,
      "step": 18365
    },
    {
      "epoch": 2.574071478626489,
      "grad_norm": 0.7308099269866943,
      "learning_rate": 4.371203061468548e-05,
      "loss": 0.0288,
      "step": 18366
    },
    {
      "epoch": 2.5742116327960756,
      "grad_norm": 0.3434159457683563,
      "learning_rate": 4.3697679980865816e-05,
      "loss": 0.0411,
      "step": 18367
    },
    {
      "epoch": 2.5743517869656625,
      "grad_norm": 1.4460663795471191,
      "learning_rate": 4.3683329347046165e-05,
      "loss": 0.0399,
      "step": 18368
    },
    {
      "epoch": 2.574491941135249,
      "grad_norm": 0.15944431722164154,
      "learning_rate": 4.36689787132265e-05,
      "loss": 0.0039,
      "step": 18369
    },
    {
      "epoch": 2.5746320953048354,
      "grad_norm": 0.6013672947883606,
      "learning_rate": 4.3654628079406835e-05,
      "loss": 0.0455,
      "step": 18370
    },
    {
      "epoch": 2.574772249474422,
      "grad_norm": 0.484423965215683,
      "learning_rate": 4.364027744558717e-05,
      "loss": 0.0522,
      "step": 18371
    },
    {
      "epoch": 2.5749124036440083,
      "grad_norm": 0.19116707146167755,
      "learning_rate": 4.362592681176752e-05,
      "loss": 0.0106,
      "step": 18372
    },
    {
      "epoch": 2.5750525578135948,
      "grad_norm": 0.2497573047876358,
      "learning_rate": 4.3611576177947855e-05,
      "loss": 0.0077,
      "step": 18373
    },
    {
      "epoch": 2.5751927119831812,
      "grad_norm": 0.5705140829086304,
      "learning_rate": 4.35972255441282e-05,
      "loss": 0.0719,
      "step": 18374
    },
    {
      "epoch": 2.575332866152768,
      "grad_norm": 0.19943994283676147,
      "learning_rate": 4.358287491030853e-05,
      "loss": 0.0142,
      "step": 18375
    },
    {
      "epoch": 2.5754730203223546,
      "grad_norm": 0.2182893007993698,
      "learning_rate": 4.356852427648888e-05,
      "loss": 0.0492,
      "step": 18376
    },
    {
      "epoch": 2.575613174491941,
      "grad_norm": 0.1173836961388588,
      "learning_rate": 4.3554173642669217e-05,
      "loss": 0.0075,
      "step": 18377
    },
    {
      "epoch": 2.575753328661528,
      "grad_norm": 0.10085103660821915,
      "learning_rate": 4.353982300884955e-05,
      "loss": 0.0092,
      "step": 18378
    },
    {
      "epoch": 2.5758934828311144,
      "grad_norm": 0.43157729506492615,
      "learning_rate": 4.352547237502989e-05,
      "loss": 0.0612,
      "step": 18379
    },
    {
      "epoch": 2.576033637000701,
      "grad_norm": 1.0196260213851929,
      "learning_rate": 4.3511121741210236e-05,
      "loss": 0.028,
      "step": 18380
    },
    {
      "epoch": 2.5761737911702873,
      "grad_norm": 0.16957086324691772,
      "learning_rate": 4.349677110739057e-05,
      "loss": 0.0153,
      "step": 18381
    },
    {
      "epoch": 2.576313945339874,
      "grad_norm": 0.3136398196220398,
      "learning_rate": 4.3482420473570914e-05,
      "loss": 0.0298,
      "step": 18382
    },
    {
      "epoch": 2.5764540995094602,
      "grad_norm": 0.21988998353481293,
      "learning_rate": 4.346806983975125e-05,
      "loss": 0.0917,
      "step": 18383
    },
    {
      "epoch": 2.5765942536790467,
      "grad_norm": 0.30825674533843994,
      "learning_rate": 4.34537192059316e-05,
      "loss": 0.0205,
      "step": 18384
    },
    {
      "epoch": 2.5767344078486336,
      "grad_norm": 0.24619002640247345,
      "learning_rate": 4.343936857211193e-05,
      "loss": 0.0204,
      "step": 18385
    },
    {
      "epoch": 2.57687456201822,
      "grad_norm": 0.11678573489189148,
      "learning_rate": 4.342501793829227e-05,
      "loss": 0.0043,
      "step": 18386
    },
    {
      "epoch": 2.5770147161878065,
      "grad_norm": 0.12242564558982849,
      "learning_rate": 4.341066730447261e-05,
      "loss": 0.0136,
      "step": 18387
    },
    {
      "epoch": 2.577154870357393,
      "grad_norm": 0.12053769826889038,
      "learning_rate": 4.339631667065295e-05,
      "loss": 0.0151,
      "step": 18388
    },
    {
      "epoch": 2.57729502452698,
      "grad_norm": 0.18951837718486786,
      "learning_rate": 4.338196603683329e-05,
      "loss": 0.0258,
      "step": 18389
    },
    {
      "epoch": 2.5774351786965664,
      "grad_norm": 0.2236301451921463,
      "learning_rate": 4.336761540301363e-05,
      "loss": 0.0367,
      "step": 18390
    },
    {
      "epoch": 2.577575332866153,
      "grad_norm": 0.09455155581235886,
      "learning_rate": 4.3353264769193965e-05,
      "loss": 0.006,
      "step": 18391
    },
    {
      "epoch": 2.5777154870357393,
      "grad_norm": 0.07828865945339203,
      "learning_rate": 4.3338914135374314e-05,
      "loss": 0.006,
      "step": 18392
    },
    {
      "epoch": 2.5778556412053257,
      "grad_norm": 0.04116327315568924,
      "learning_rate": 4.332456350155465e-05,
      "loss": 0.003,
      "step": 18393
    },
    {
      "epoch": 2.577995795374912,
      "grad_norm": 0.1679389625787735,
      "learning_rate": 4.3310212867734985e-05,
      "loss": 0.0265,
      "step": 18394
    },
    {
      "epoch": 2.578135949544499,
      "grad_norm": 0.13193699717521667,
      "learning_rate": 4.329586223391533e-05,
      "loss": 0.0159,
      "step": 18395
    },
    {
      "epoch": 2.5782761037140856,
      "grad_norm": 0.18403862416744232,
      "learning_rate": 4.328151160009567e-05,
      "loss": 0.0446,
      "step": 18396
    },
    {
      "epoch": 2.578416257883672,
      "grad_norm": 0.6315674781799316,
      "learning_rate": 4.3267160966276005e-05,
      "loss": 0.0248,
      "step": 18397
    },
    {
      "epoch": 2.5785564120532585,
      "grad_norm": 0.2854273319244385,
      "learning_rate": 4.325281033245635e-05,
      "loss": 0.0204,
      "step": 18398
    },
    {
      "epoch": 2.5786965662228454,
      "grad_norm": 0.4008709788322449,
      "learning_rate": 4.323845969863668e-05,
      "loss": 0.0351,
      "step": 18399
    },
    {
      "epoch": 2.578836720392432,
      "grad_norm": 0.10933005064725876,
      "learning_rate": 4.322410906481703e-05,
      "loss": 0.0105,
      "step": 18400
    },
    {
      "epoch": 2.5789768745620183,
      "grad_norm": 0.16192081570625305,
      "learning_rate": 4.3209758430997366e-05,
      "loss": 0.0438,
      "step": 18401
    },
    {
      "epoch": 2.5791170287316048,
      "grad_norm": 0.8346202969551086,
      "learning_rate": 4.31954077971777e-05,
      "loss": 0.0342,
      "step": 18402
    },
    {
      "epoch": 2.579257182901191,
      "grad_norm": 0.2391958087682724,
      "learning_rate": 4.3181057163358044e-05,
      "loss": 0.0236,
      "step": 18403
    },
    {
      "epoch": 2.5793973370707777,
      "grad_norm": 0.14135269820690155,
      "learning_rate": 4.3166706529538386e-05,
      "loss": 0.0381,
      "step": 18404
    },
    {
      "epoch": 2.579537491240364,
      "grad_norm": 0.17120225727558136,
      "learning_rate": 4.315235589571872e-05,
      "loss": 0.0109,
      "step": 18405
    },
    {
      "epoch": 2.579677645409951,
      "grad_norm": 0.04020866006612778,
      "learning_rate": 4.313800526189906e-05,
      "loss": 0.0034,
      "step": 18406
    },
    {
      "epoch": 2.5798177995795375,
      "grad_norm": 0.28193074464797974,
      "learning_rate": 4.3123654628079405e-05,
      "loss": 0.043,
      "step": 18407
    },
    {
      "epoch": 2.579957953749124,
      "grad_norm": 0.3373149037361145,
      "learning_rate": 4.310930399425975e-05,
      "loss": 0.0104,
      "step": 18408
    },
    {
      "epoch": 2.580098107918711,
      "grad_norm": 0.19258467853069305,
      "learning_rate": 4.309495336044008e-05,
      "loss": 0.0521,
      "step": 18409
    },
    {
      "epoch": 2.5802382620882973,
      "grad_norm": 0.5233100056648254,
      "learning_rate": 4.308060272662042e-05,
      "loss": 0.0181,
      "step": 18410
    },
    {
      "epoch": 2.580378416257884,
      "grad_norm": 0.2077948898077011,
      "learning_rate": 4.306625209280077e-05,
      "loss": 0.0166,
      "step": 18411
    },
    {
      "epoch": 2.5805185704274702,
      "grad_norm": 0.33574870228767395,
      "learning_rate": 4.30519014589811e-05,
      "loss": 0.0301,
      "step": 18412
    },
    {
      "epoch": 2.5806587245970567,
      "grad_norm": 0.38488245010375977,
      "learning_rate": 4.303755082516144e-05,
      "loss": 0.0343,
      "step": 18413
    },
    {
      "epoch": 2.580798878766643,
      "grad_norm": 0.021686121821403503,
      "learning_rate": 4.302320019134178e-05,
      "loss": 0.0018,
      "step": 18414
    },
    {
      "epoch": 2.5809390329362296,
      "grad_norm": 0.132170170545578,
      "learning_rate": 4.300884955752212e-05,
      "loss": 0.0143,
      "step": 18415
    },
    {
      "epoch": 2.5810791871058165,
      "grad_norm": 0.14774839580059052,
      "learning_rate": 4.2994498923702464e-05,
      "loss": 0.0422,
      "step": 18416
    },
    {
      "epoch": 2.581219341275403,
      "grad_norm": 0.1877659559249878,
      "learning_rate": 4.29801482898828e-05,
      "loss": 0.0149,
      "step": 18417
    },
    {
      "epoch": 2.5813594954449894,
      "grad_norm": 0.34519264101982117,
      "learning_rate": 4.2965797656063135e-05,
      "loss": 0.0131,
      "step": 18418
    },
    {
      "epoch": 2.581499649614576,
      "grad_norm": 0.8436042666435242,
      "learning_rate": 4.2951447022243484e-05,
      "loss": 0.0815,
      "step": 18419
    },
    {
      "epoch": 2.581639803784163,
      "grad_norm": 0.6240946650505066,
      "learning_rate": 4.293709638842382e-05,
      "loss": 0.045,
      "step": 18420
    },
    {
      "epoch": 2.5817799579537493,
      "grad_norm": 0.46625909209251404,
      "learning_rate": 4.2922745754604154e-05,
      "loss": 0.0383,
      "step": 18421
    },
    {
      "epoch": 2.5819201121233357,
      "grad_norm": 0.0793287381529808,
      "learning_rate": 4.2908395120784496e-05,
      "loss": 0.0046,
      "step": 18422
    },
    {
      "epoch": 2.582060266292922,
      "grad_norm": 0.3346347510814667,
      "learning_rate": 4.289404448696484e-05,
      "loss": 0.0344,
      "step": 18423
    },
    {
      "epoch": 2.5822004204625086,
      "grad_norm": 0.09946427494287491,
      "learning_rate": 4.287969385314518e-05,
      "loss": 0.0087,
      "step": 18424
    },
    {
      "epoch": 2.582340574632095,
      "grad_norm": 0.5931811332702637,
      "learning_rate": 4.2865343219325516e-05,
      "loss": 0.0555,
      "step": 18425
    },
    {
      "epoch": 2.582480728801682,
      "grad_norm": 0.32411035895347595,
      "learning_rate": 4.285099258550585e-05,
      "loss": 0.0374,
      "step": 18426
    },
    {
      "epoch": 2.5826208829712685,
      "grad_norm": 0.3806039094924927,
      "learning_rate": 4.28366419516862e-05,
      "loss": 0.0117,
      "step": 18427
    },
    {
      "epoch": 2.582761037140855,
      "grad_norm": 0.07168141007423401,
      "learning_rate": 4.2822291317866536e-05,
      "loss": 0.0059,
      "step": 18428
    },
    {
      "epoch": 2.5829011913104414,
      "grad_norm": 0.5293334722518921,
      "learning_rate": 4.280794068404687e-05,
      "loss": 0.0221,
      "step": 18429
    },
    {
      "epoch": 2.5830413454800283,
      "grad_norm": 0.3320784270763397,
      "learning_rate": 4.279359005022721e-05,
      "loss": 0.0319,
      "step": 18430
    },
    {
      "epoch": 2.5831814996496147,
      "grad_norm": 0.18998445570468903,
      "learning_rate": 4.2779239416407555e-05,
      "loss": 0.0256,
      "step": 18431
    },
    {
      "epoch": 2.583321653819201,
      "grad_norm": 0.2475634664297104,
      "learning_rate": 4.27648887825879e-05,
      "loss": 0.0488,
      "step": 18432
    },
    {
      "epoch": 2.5834618079887877,
      "grad_norm": 0.28841716051101685,
      "learning_rate": 4.275053814876823e-05,
      "loss": 0.0728,
      "step": 18433
    },
    {
      "epoch": 2.583601962158374,
      "grad_norm": 0.28521859645843506,
      "learning_rate": 4.273618751494857e-05,
      "loss": 0.0325,
      "step": 18434
    },
    {
      "epoch": 2.5837421163279606,
      "grad_norm": 0.16159546375274658,
      "learning_rate": 4.272183688112892e-05,
      "loss": 0.0066,
      "step": 18435
    },
    {
      "epoch": 2.583882270497547,
      "grad_norm": 0.22974339127540588,
      "learning_rate": 4.270748624730925e-05,
      "loss": 0.0209,
      "step": 18436
    },
    {
      "epoch": 2.584022424667134,
      "grad_norm": 0.5020043253898621,
      "learning_rate": 4.269313561348959e-05,
      "loss": 0.0565,
      "step": 18437
    },
    {
      "epoch": 2.5841625788367204,
      "grad_norm": 0.2730749547481537,
      "learning_rate": 4.267878497966993e-05,
      "loss": 0.0931,
      "step": 18438
    },
    {
      "epoch": 2.584302733006307,
      "grad_norm": 0.5322865843772888,
      "learning_rate": 4.266443434585027e-05,
      "loss": 0.1059,
      "step": 18439
    },
    {
      "epoch": 2.5844428871758933,
      "grad_norm": 0.07507546991109848,
      "learning_rate": 4.2650083712030614e-05,
      "loss": 0.0042,
      "step": 18440
    },
    {
      "epoch": 2.5845830413454802,
      "grad_norm": 0.2223496437072754,
      "learning_rate": 4.263573307821095e-05,
      "loss": 0.0175,
      "step": 18441
    },
    {
      "epoch": 2.5847231955150667,
      "grad_norm": 0.31748926639556885,
      "learning_rate": 4.262138244439129e-05,
      "loss": 0.0273,
      "step": 18442
    },
    {
      "epoch": 2.584863349684653,
      "grad_norm": 0.06157343089580536,
      "learning_rate": 4.260703181057163e-05,
      "loss": 0.0031,
      "step": 18443
    },
    {
      "epoch": 2.5850035038542396,
      "grad_norm": 0.08624120056629181,
      "learning_rate": 4.259268117675197e-05,
      "loss": 0.015,
      "step": 18444
    },
    {
      "epoch": 2.585143658023826,
      "grad_norm": 0.20628924667835236,
      "learning_rate": 4.2578330542932304e-05,
      "loss": 0.017,
      "step": 18445
    },
    {
      "epoch": 2.5852838121934125,
      "grad_norm": 0.21479536592960358,
      "learning_rate": 4.256397990911265e-05,
      "loss": 0.0175,
      "step": 18446
    },
    {
      "epoch": 2.5854239663629994,
      "grad_norm": 0.036246560513973236,
      "learning_rate": 4.254962927529299e-05,
      "loss": 0.0043,
      "step": 18447
    },
    {
      "epoch": 2.585564120532586,
      "grad_norm": 0.11516187340021133,
      "learning_rate": 4.253527864147333e-05,
      "loss": 0.0039,
      "step": 18448
    },
    {
      "epoch": 2.5857042747021723,
      "grad_norm": 0.12631452083587646,
      "learning_rate": 4.2520928007653666e-05,
      "loss": 0.0099,
      "step": 18449
    },
    {
      "epoch": 2.585844428871759,
      "grad_norm": 0.13929536938667297,
      "learning_rate": 4.250657737383401e-05,
      "loss": 0.0082,
      "step": 18450
    },
    {
      "epoch": 2.5859845830413457,
      "grad_norm": 0.20480823516845703,
      "learning_rate": 4.249222674001435e-05,
      "loss": 0.0349,
      "step": 18451
    },
    {
      "epoch": 2.586124737210932,
      "grad_norm": 0.1254032552242279,
      "learning_rate": 4.2477876106194685e-05,
      "loss": 0.0087,
      "step": 18452
    },
    {
      "epoch": 2.5862648913805186,
      "grad_norm": 1.004272699356079,
      "learning_rate": 4.246352547237502e-05,
      "loss": 0.0597,
      "step": 18453
    },
    {
      "epoch": 2.586405045550105,
      "grad_norm": 0.09287992864847183,
      "learning_rate": 4.244917483855537e-05,
      "loss": 0.0143,
      "step": 18454
    },
    {
      "epoch": 2.5865451997196915,
      "grad_norm": 0.27284035086631775,
      "learning_rate": 4.2434824204735705e-05,
      "loss": 0.0183,
      "step": 18455
    },
    {
      "epoch": 2.586685353889278,
      "grad_norm": 0.1848074048757553,
      "learning_rate": 4.242047357091605e-05,
      "loss": 0.0081,
      "step": 18456
    },
    {
      "epoch": 2.586825508058865,
      "grad_norm": 0.21825416386127472,
      "learning_rate": 4.240612293709638e-05,
      "loss": 0.0104,
      "step": 18457
    },
    {
      "epoch": 2.5869656622284514,
      "grad_norm": 0.22444355487823486,
      "learning_rate": 4.2391772303276724e-05,
      "loss": 0.0372,
      "step": 18458
    },
    {
      "epoch": 2.587105816398038,
      "grad_norm": 0.20439444482326508,
      "learning_rate": 4.2377421669457067e-05,
      "loss": 0.0215,
      "step": 18459
    },
    {
      "epoch": 2.5872459705676243,
      "grad_norm": 0.0767916887998581,
      "learning_rate": 4.23630710356374e-05,
      "loss": 0.0048,
      "step": 18460
    },
    {
      "epoch": 2.587386124737211,
      "grad_norm": 0.22043246030807495,
      "learning_rate": 4.234872040181774e-05,
      "loss": 0.0322,
      "step": 18461
    },
    {
      "epoch": 2.5875262789067976,
      "grad_norm": 0.1928377002477646,
      "learning_rate": 4.2334369767998086e-05,
      "loss": 0.0097,
      "step": 18462
    },
    {
      "epoch": 2.587666433076384,
      "grad_norm": 0.07479529082775116,
      "learning_rate": 4.232001913417842e-05,
      "loss": 0.0066,
      "step": 18463
    },
    {
      "epoch": 2.5878065872459706,
      "grad_norm": 0.0872916430234909,
      "learning_rate": 4.2305668500358764e-05,
      "loss": 0.0054,
      "step": 18464
    },
    {
      "epoch": 2.587946741415557,
      "grad_norm": 0.146233469247818,
      "learning_rate": 4.22913178665391e-05,
      "loss": 0.0138,
      "step": 18465
    },
    {
      "epoch": 2.5880868955851435,
      "grad_norm": 0.5427401661872864,
      "learning_rate": 4.227696723271944e-05,
      "loss": 0.0261,
      "step": 18466
    },
    {
      "epoch": 2.58822704975473,
      "grad_norm": 0.1921662539243698,
      "learning_rate": 4.226261659889978e-05,
      "loss": 0.0162,
      "step": 18467
    },
    {
      "epoch": 2.588367203924317,
      "grad_norm": 0.8048128485679626,
      "learning_rate": 4.224826596508012e-05,
      "loss": 0.1125,
      "step": 18468
    },
    {
      "epoch": 2.5885073580939033,
      "grad_norm": 0.24106605350971222,
      "learning_rate": 4.2233915331260454e-05,
      "loss": 0.0116,
      "step": 18469
    },
    {
      "epoch": 2.5886475122634898,
      "grad_norm": 1.096875786781311,
      "learning_rate": 4.22195646974408e-05,
      "loss": 0.058,
      "step": 18470
    },
    {
      "epoch": 2.5887876664330762,
      "grad_norm": 0.248779296875,
      "learning_rate": 4.220521406362114e-05,
      "loss": 0.0459,
      "step": 18471
    },
    {
      "epoch": 2.588927820602663,
      "grad_norm": 0.21327553689479828,
      "learning_rate": 4.219086342980148e-05,
      "loss": 0.0239,
      "step": 18472
    },
    {
      "epoch": 2.5890679747722496,
      "grad_norm": 0.21323801577091217,
      "learning_rate": 4.2176512795981815e-05,
      "loss": 0.0072,
      "step": 18473
    },
    {
      "epoch": 2.589208128941836,
      "grad_norm": 0.33114296197891235,
      "learning_rate": 4.216216216216216e-05,
      "loss": 0.0484,
      "step": 18474
    },
    {
      "epoch": 2.5893482831114225,
      "grad_norm": 0.09613249450922012,
      "learning_rate": 4.21478115283425e-05,
      "loss": 0.0065,
      "step": 18475
    },
    {
      "epoch": 2.589488437281009,
      "grad_norm": 0.5842782258987427,
      "learning_rate": 4.2133460894522835e-05,
      "loss": 0.0366,
      "step": 18476
    },
    {
      "epoch": 2.5896285914505954,
      "grad_norm": 0.5013179183006287,
      "learning_rate": 4.211911026070317e-05,
      "loss": 0.0446,
      "step": 18477
    },
    {
      "epoch": 2.5897687456201823,
      "grad_norm": 0.45787426829338074,
      "learning_rate": 4.210475962688352e-05,
      "loss": 0.0111,
      "step": 18478
    },
    {
      "epoch": 2.589908899789769,
      "grad_norm": 0.338286429643631,
      "learning_rate": 4.2090408993063855e-05,
      "loss": 0.0191,
      "step": 18479
    },
    {
      "epoch": 2.5900490539593553,
      "grad_norm": 0.3275730609893799,
      "learning_rate": 4.20760583592442e-05,
      "loss": 0.0562,
      "step": 18480
    },
    {
      "epoch": 2.5901892081289417,
      "grad_norm": 0.17356082797050476,
      "learning_rate": 4.206170772542454e-05,
      "loss": 0.0055,
      "step": 18481
    },
    {
      "epoch": 2.5903293622985286,
      "grad_norm": 0.29584771394729614,
      "learning_rate": 4.2047357091604874e-05,
      "loss": 0.0208,
      "step": 18482
    },
    {
      "epoch": 2.590469516468115,
      "grad_norm": 0.18936409056186676,
      "learning_rate": 4.2033006457785216e-05,
      "loss": 0.0288,
      "step": 18483
    },
    {
      "epoch": 2.5906096706377015,
      "grad_norm": 0.08536079525947571,
      "learning_rate": 4.201865582396555e-05,
      "loss": 0.0209,
      "step": 18484
    },
    {
      "epoch": 2.590749824807288,
      "grad_norm": 0.4714975655078888,
      "learning_rate": 4.20043051901459e-05,
      "loss": 0.0528,
      "step": 18485
    },
    {
      "epoch": 2.5908899789768745,
      "grad_norm": 0.2756427824497223,
      "learning_rate": 4.1989954556326236e-05,
      "loss": 0.0846,
      "step": 18486
    },
    {
      "epoch": 2.591030133146461,
      "grad_norm": 0.16883422434329987,
      "learning_rate": 4.197560392250657e-05,
      "loss": 0.0515,
      "step": 18487
    },
    {
      "epoch": 2.591170287316048,
      "grad_norm": 0.1769227385520935,
      "learning_rate": 4.196125328868691e-05,
      "loss": 0.0061,
      "step": 18488
    },
    {
      "epoch": 2.5913104414856343,
      "grad_norm": 0.05441201478242874,
      "learning_rate": 4.1946902654867255e-05,
      "loss": 0.0049,
      "step": 18489
    },
    {
      "epoch": 2.5914505956552207,
      "grad_norm": 0.13326485455036163,
      "learning_rate": 4.193255202104759e-05,
      "loss": 0.0157,
      "step": 18490
    },
    {
      "epoch": 2.591590749824807,
      "grad_norm": 0.5163449645042419,
      "learning_rate": 4.191820138722793e-05,
      "loss": 0.0679,
      "step": 18491
    },
    {
      "epoch": 2.591730903994394,
      "grad_norm": 0.09258285164833069,
      "learning_rate": 4.190385075340827e-05,
      "loss": 0.0045,
      "step": 18492
    },
    {
      "epoch": 2.5918710581639806,
      "grad_norm": 0.1363135576248169,
      "learning_rate": 4.188950011958862e-05,
      "loss": 0.0141,
      "step": 18493
    },
    {
      "epoch": 2.592011212333567,
      "grad_norm": 0.5469874143600464,
      "learning_rate": 4.187514948576895e-05,
      "loss": 0.0535,
      "step": 18494
    },
    {
      "epoch": 2.5921513665031535,
      "grad_norm": 0.08730471879243851,
      "learning_rate": 4.186079885194929e-05,
      "loss": 0.0067,
      "step": 18495
    },
    {
      "epoch": 2.59229152067274,
      "grad_norm": 0.1166330873966217,
      "learning_rate": 4.184644821812963e-05,
      "loss": 0.0095,
      "step": 18496
    },
    {
      "epoch": 2.5924316748423264,
      "grad_norm": 0.25809523463249207,
      "learning_rate": 4.183209758430997e-05,
      "loss": 0.0187,
      "step": 18497
    },
    {
      "epoch": 2.592571829011913,
      "grad_norm": 0.2569851577281952,
      "learning_rate": 4.181774695049031e-05,
      "loss": 0.0165,
      "step": 18498
    },
    {
      "epoch": 2.5927119831814998,
      "grad_norm": 0.37615805864334106,
      "learning_rate": 4.180339631667065e-05,
      "loss": 0.0385,
      "step": 18499
    },
    {
      "epoch": 2.592852137351086,
      "grad_norm": 0.3347506523132324,
      "learning_rate": 4.1789045682850985e-05,
      "loss": 0.0273,
      "step": 18500
    },
    {
      "epoch": 2.5929922915206727,
      "grad_norm": 0.23138432204723358,
      "learning_rate": 4.1774695049031334e-05,
      "loss": 0.0544,
      "step": 18501
    },
    {
      "epoch": 2.593132445690259,
      "grad_norm": 0.5268917679786682,
      "learning_rate": 4.176034441521167e-05,
      "loss": 0.0283,
      "step": 18502
    },
    {
      "epoch": 2.593272599859846,
      "grad_norm": 0.3320567309856415,
      "learning_rate": 4.1745993781392004e-05,
      "loss": 0.0662,
      "step": 18503
    },
    {
      "epoch": 2.5934127540294325,
      "grad_norm": 0.09726834297180176,
      "learning_rate": 4.1731643147572346e-05,
      "loss": 0.0319,
      "step": 18504
    },
    {
      "epoch": 2.593552908199019,
      "grad_norm": 0.30209630727767944,
      "learning_rate": 4.171729251375269e-05,
      "loss": 0.0306,
      "step": 18505
    },
    {
      "epoch": 2.5936930623686054,
      "grad_norm": 0.14331725239753723,
      "learning_rate": 4.170294187993303e-05,
      "loss": 0.0169,
      "step": 18506
    },
    {
      "epoch": 2.593833216538192,
      "grad_norm": 0.10237036645412445,
      "learning_rate": 4.1688591246113366e-05,
      "loss": 0.0041,
      "step": 18507
    },
    {
      "epoch": 2.5939733707077783,
      "grad_norm": 0.042850177735090256,
      "learning_rate": 4.16742406122937e-05,
      "loss": 0.0115,
      "step": 18508
    },
    {
      "epoch": 2.5941135248773652,
      "grad_norm": 0.8127962946891785,
      "learning_rate": 4.165988997847405e-05,
      "loss": 0.0365,
      "step": 18509
    },
    {
      "epoch": 2.5942536790469517,
      "grad_norm": 0.2180366963148117,
      "learning_rate": 4.1645539344654386e-05,
      "loss": 0.0496,
      "step": 18510
    },
    {
      "epoch": 2.594393833216538,
      "grad_norm": 0.16415521502494812,
      "learning_rate": 4.163118871083472e-05,
      "loss": 0.0174,
      "step": 18511
    },
    {
      "epoch": 2.5945339873861246,
      "grad_norm": 0.09151072055101395,
      "learning_rate": 4.161683807701506e-05,
      "loss": 0.0084,
      "step": 18512
    },
    {
      "epoch": 2.5946741415557115,
      "grad_norm": 0.0869913399219513,
      "learning_rate": 4.1602487443195405e-05,
      "loss": 0.0215,
      "step": 18513
    },
    {
      "epoch": 2.594814295725298,
      "grad_norm": 0.15585070848464966,
      "learning_rate": 4.158813680937575e-05,
      "loss": 0.0211,
      "step": 18514
    },
    {
      "epoch": 2.5949544498948844,
      "grad_norm": 0.9072529673576355,
      "learning_rate": 4.157378617555608e-05,
      "loss": 0.0596,
      "step": 18515
    },
    {
      "epoch": 2.595094604064471,
      "grad_norm": 0.07980072498321533,
      "learning_rate": 4.1559435541736425e-05,
      "loss": 0.0095,
      "step": 18516
    },
    {
      "epoch": 2.5952347582340574,
      "grad_norm": 0.326193630695343,
      "learning_rate": 4.154508490791677e-05,
      "loss": 0.0382,
      "step": 18517
    },
    {
      "epoch": 2.595374912403644,
      "grad_norm": 0.08973927050828934,
      "learning_rate": 4.15307342740971e-05,
      "loss": 0.008,
      "step": 18518
    },
    {
      "epoch": 2.5955150665732303,
      "grad_norm": 0.12881870567798615,
      "learning_rate": 4.151638364027744e-05,
      "loss": 0.0047,
      "step": 18519
    },
    {
      "epoch": 2.595655220742817,
      "grad_norm": 0.2405724823474884,
      "learning_rate": 4.1502033006457786e-05,
      "loss": 0.0127,
      "step": 18520
    },
    {
      "epoch": 2.5957953749124036,
      "grad_norm": 0.6358450055122375,
      "learning_rate": 4.148768237263812e-05,
      "loss": 0.0218,
      "step": 18521
    },
    {
      "epoch": 2.59593552908199,
      "grad_norm": 0.21287740767002106,
      "learning_rate": 4.1473331738818464e-05,
      "loss": 0.0379,
      "step": 18522
    },
    {
      "epoch": 2.596075683251577,
      "grad_norm": 0.17231370508670807,
      "learning_rate": 4.14589811049988e-05,
      "loss": 0.0427,
      "step": 18523
    },
    {
      "epoch": 2.5962158374211635,
      "grad_norm": 0.22845718264579773,
      "learning_rate": 4.144463047117914e-05,
      "loss": 0.0356,
      "step": 18524
    },
    {
      "epoch": 2.59635599159075,
      "grad_norm": 0.17676453292369843,
      "learning_rate": 4.143027983735948e-05,
      "loss": 0.0143,
      "step": 18525
    },
    {
      "epoch": 2.5964961457603364,
      "grad_norm": 0.13186609745025635,
      "learning_rate": 4.141592920353982e-05,
      "loss": 0.0188,
      "step": 18526
    },
    {
      "epoch": 2.596636299929923,
      "grad_norm": 0.08770882338285446,
      "learning_rate": 4.1401578569720154e-05,
      "loss": 0.0121,
      "step": 18527
    },
    {
      "epoch": 2.5967764540995093,
      "grad_norm": 0.3491726517677307,
      "learning_rate": 4.13872279359005e-05,
      "loss": 0.0033,
      "step": 18528
    },
    {
      "epoch": 2.5969166082690958,
      "grad_norm": 0.16179046034812927,
      "learning_rate": 4.137287730208084e-05,
      "loss": 0.0381,
      "step": 18529
    },
    {
      "epoch": 2.5970567624386827,
      "grad_norm": 0.20397137105464935,
      "learning_rate": 4.135852666826118e-05,
      "loss": 0.0407,
      "step": 18530
    },
    {
      "epoch": 2.597196916608269,
      "grad_norm": 0.20023071765899658,
      "learning_rate": 4.1344176034441516e-05,
      "loss": 0.0257,
      "step": 18531
    },
    {
      "epoch": 2.5973370707778556,
      "grad_norm": 1.0694224834442139,
      "learning_rate": 4.132982540062186e-05,
      "loss": 0.031,
      "step": 18532
    },
    {
      "epoch": 2.597477224947442,
      "grad_norm": 0.06407024711370468,
      "learning_rate": 4.13154747668022e-05,
      "loss": 0.0044,
      "step": 18533
    },
    {
      "epoch": 2.597617379117029,
      "grad_norm": 0.15482138097286224,
      "learning_rate": 4.1301124132982535e-05,
      "loss": 0.0386,
      "step": 18534
    },
    {
      "epoch": 2.5977575332866154,
      "grad_norm": 0.23148861527442932,
      "learning_rate": 4.128677349916287e-05,
      "loss": 0.0328,
      "step": 18535
    },
    {
      "epoch": 2.597897687456202,
      "grad_norm": 0.3446612060070038,
      "learning_rate": 4.127242286534322e-05,
      "loss": 0.0612,
      "step": 18536
    },
    {
      "epoch": 2.5980378416257883,
      "grad_norm": 0.08672413975000381,
      "learning_rate": 4.1258072231523555e-05,
      "loss": 0.02,
      "step": 18537
    },
    {
      "epoch": 2.598177995795375,
      "grad_norm": 0.16013412177562714,
      "learning_rate": 4.12437215977039e-05,
      "loss": 0.0175,
      "step": 18538
    },
    {
      "epoch": 2.5983181499649612,
      "grad_norm": 0.1819830983877182,
      "learning_rate": 4.122937096388423e-05,
      "loss": 0.0077,
      "step": 18539
    },
    {
      "epoch": 2.598458304134548,
      "grad_norm": 0.47628769278526306,
      "learning_rate": 4.1215020330064574e-05,
      "loss": 0.0579,
      "step": 18540
    },
    {
      "epoch": 2.5985984583041346,
      "grad_norm": 0.19871847331523895,
      "learning_rate": 4.1200669696244916e-05,
      "loss": 0.0325,
      "step": 18541
    },
    {
      "epoch": 2.598738612473721,
      "grad_norm": 0.07882734388113022,
      "learning_rate": 4.118631906242525e-05,
      "loss": 0.0166,
      "step": 18542
    },
    {
      "epoch": 2.5988787666433075,
      "grad_norm": 0.2993088662624359,
      "learning_rate": 4.117196842860559e-05,
      "loss": 0.0462,
      "step": 18543
    },
    {
      "epoch": 2.5990189208128944,
      "grad_norm": 0.17911703884601593,
      "learning_rate": 4.1157617794785936e-05,
      "loss": 0.0238,
      "step": 18544
    },
    {
      "epoch": 2.599159074982481,
      "grad_norm": 0.44249698519706726,
      "learning_rate": 4.114326716096627e-05,
      "loss": 0.0751,
      "step": 18545
    },
    {
      "epoch": 2.5992992291520673,
      "grad_norm": 0.20505116879940033,
      "learning_rate": 4.1128916527146613e-05,
      "loss": 0.0458,
      "step": 18546
    },
    {
      "epoch": 2.599439383321654,
      "grad_norm": 0.08975695818662643,
      "learning_rate": 4.111456589332695e-05,
      "loss": 0.0156,
      "step": 18547
    },
    {
      "epoch": 2.5995795374912403,
      "grad_norm": 0.20133747160434723,
      "learning_rate": 4.110021525950729e-05,
      "loss": 0.0305,
      "step": 18548
    },
    {
      "epoch": 2.5997196916608267,
      "grad_norm": 1.546108603477478,
      "learning_rate": 4.108586462568763e-05,
      "loss": 0.0526,
      "step": 18549
    },
    {
      "epoch": 2.599859845830413,
      "grad_norm": 0.43777474761009216,
      "learning_rate": 4.107151399186797e-05,
      "loss": 0.0155,
      "step": 18550
    },
    {
      "epoch": 2.6,
      "grad_norm": 0.40382590889930725,
      "learning_rate": 4.1057163358048304e-05,
      "loss": 0.0321,
      "step": 18551
    },
    {
      "epoch": 2.6001401541695865,
      "grad_norm": 0.10036035627126694,
      "learning_rate": 4.104281272422865e-05,
      "loss": 0.0088,
      "step": 18552
    },
    {
      "epoch": 2.600280308339173,
      "grad_norm": 0.22109805047512054,
      "learning_rate": 4.102846209040899e-05,
      "loss": 0.0095,
      "step": 18553
    },
    {
      "epoch": 2.60042046250876,
      "grad_norm": 0.18864211440086365,
      "learning_rate": 4.101411145658933e-05,
      "loss": 0.0372,
      "step": 18554
    },
    {
      "epoch": 2.6005606166783464,
      "grad_norm": 0.07249768078327179,
      "learning_rate": 4.099976082276967e-05,
      "loss": 0.0086,
      "step": 18555
    },
    {
      "epoch": 2.600700770847933,
      "grad_norm": 0.22700278460979462,
      "learning_rate": 4.098541018895001e-05,
      "loss": 0.0253,
      "step": 18556
    },
    {
      "epoch": 2.6008409250175193,
      "grad_norm": 0.320683091878891,
      "learning_rate": 4.097105955513035e-05,
      "loss": 0.0244,
      "step": 18557
    },
    {
      "epoch": 2.6009810791871057,
      "grad_norm": 0.2824430465698242,
      "learning_rate": 4.0956708921310685e-05,
      "loss": 0.0182,
      "step": 18558
    },
    {
      "epoch": 2.601121233356692,
      "grad_norm": 0.30952298641204834,
      "learning_rate": 4.0942358287491034e-05,
      "loss": 0.0147,
      "step": 18559
    },
    {
      "epoch": 2.6012613875262787,
      "grad_norm": 0.3556981384754181,
      "learning_rate": 4.092800765367137e-05,
      "loss": 0.016,
      "step": 18560
    },
    {
      "epoch": 2.6014015416958656,
      "grad_norm": 0.11541136354207993,
      "learning_rate": 4.0913657019851705e-05,
      "loss": 0.0079,
      "step": 18561
    },
    {
      "epoch": 2.601541695865452,
      "grad_norm": 0.22843077778816223,
      "learning_rate": 4.089930638603205e-05,
      "loss": 0.0136,
      "step": 18562
    },
    {
      "epoch": 2.6016818500350385,
      "grad_norm": 0.2954002916812897,
      "learning_rate": 4.088495575221239e-05,
      "loss": 0.0216,
      "step": 18563
    },
    {
      "epoch": 2.601822004204625,
      "grad_norm": 0.10958950221538544,
      "learning_rate": 4.0870605118392724e-05,
      "loss": 0.0087,
      "step": 18564
    },
    {
      "epoch": 2.601962158374212,
      "grad_norm": 0.32158029079437256,
      "learning_rate": 4.0856254484573066e-05,
      "loss": 0.0713,
      "step": 18565
    },
    {
      "epoch": 2.6021023125437983,
      "grad_norm": 0.2884770333766937,
      "learning_rate": 4.08419038507534e-05,
      "loss": 0.0246,
      "step": 18566
    },
    {
      "epoch": 2.6022424667133848,
      "grad_norm": 1.6252360343933105,
      "learning_rate": 4.082755321693375e-05,
      "loss": 0.0548,
      "step": 18567
    },
    {
      "epoch": 2.6023826208829712,
      "grad_norm": 0.5166999101638794,
      "learning_rate": 4.0813202583114086e-05,
      "loss": 0.0323,
      "step": 18568
    },
    {
      "epoch": 2.6025227750525577,
      "grad_norm": 0.11942706257104874,
      "learning_rate": 4.079885194929442e-05,
      "loss": 0.0047,
      "step": 18569
    },
    {
      "epoch": 2.602662929222144,
      "grad_norm": 0.48290205001831055,
      "learning_rate": 4.078450131547476e-05,
      "loss": 0.0317,
      "step": 18570
    },
    {
      "epoch": 2.602803083391731,
      "grad_norm": 0.09516703337430954,
      "learning_rate": 4.0770150681655105e-05,
      "loss": 0.0188,
      "step": 18571
    },
    {
      "epoch": 2.6029432375613175,
      "grad_norm": 0.2316620796918869,
      "learning_rate": 4.075580004783544e-05,
      "loss": 0.0252,
      "step": 18572
    },
    {
      "epoch": 2.603083391730904,
      "grad_norm": 0.19330072402954102,
      "learning_rate": 4.074144941401578e-05,
      "loss": 0.0305,
      "step": 18573
    },
    {
      "epoch": 2.6032235459004904,
      "grad_norm": 0.11013898253440857,
      "learning_rate": 4.072709878019612e-05,
      "loss": 0.0053,
      "step": 18574
    },
    {
      "epoch": 2.6033637000700773,
      "grad_norm": 0.8182455897331238,
      "learning_rate": 4.071274814637647e-05,
      "loss": 0.0403,
      "step": 18575
    },
    {
      "epoch": 2.603503854239664,
      "grad_norm": 0.16656769812107086,
      "learning_rate": 4.06983975125568e-05,
      "loss": 0.0151,
      "step": 18576
    },
    {
      "epoch": 2.6036440084092503,
      "grad_norm": 0.18454699218273163,
      "learning_rate": 4.068404687873714e-05,
      "loss": 0.0169,
      "step": 18577
    },
    {
      "epoch": 2.6037841625788367,
      "grad_norm": 0.08318546414375305,
      "learning_rate": 4.066969624491748e-05,
      "loss": 0.0194,
      "step": 18578
    },
    {
      "epoch": 2.603924316748423,
      "grad_norm": 0.4396114945411682,
      "learning_rate": 4.065534561109782e-05,
      "loss": 0.0598,
      "step": 18579
    },
    {
      "epoch": 2.6040644709180096,
      "grad_norm": 0.42915040254592896,
      "learning_rate": 4.064099497727816e-05,
      "loss": 0.0548,
      "step": 18580
    },
    {
      "epoch": 2.604204625087596,
      "grad_norm": 0.24726633727550507,
      "learning_rate": 4.06266443434585e-05,
      "loss": 0.0242,
      "step": 18581
    },
    {
      "epoch": 2.604344779257183,
      "grad_norm": 0.8310969471931458,
      "learning_rate": 4.0612293709638835e-05,
      "loss": 0.048,
      "step": 18582
    },
    {
      "epoch": 2.6044849334267695,
      "grad_norm": 0.07173267006874084,
      "learning_rate": 4.0597943075819184e-05,
      "loss": 0.0038,
      "step": 18583
    },
    {
      "epoch": 2.604625087596356,
      "grad_norm": 0.16220510005950928,
      "learning_rate": 4.058359244199952e-05,
      "loss": 0.0093,
      "step": 18584
    },
    {
      "epoch": 2.6047652417659424,
      "grad_norm": 0.07501126825809479,
      "learning_rate": 4.0569241808179854e-05,
      "loss": 0.0063,
      "step": 18585
    },
    {
      "epoch": 2.6049053959355293,
      "grad_norm": 0.24349640309810638,
      "learning_rate": 4.0554891174360196e-05,
      "loss": 0.0324,
      "step": 18586
    },
    {
      "epoch": 2.6050455501051157,
      "grad_norm": 0.2243170142173767,
      "learning_rate": 4.054054054054054e-05,
      "loss": 0.0111,
      "step": 18587
    },
    {
      "epoch": 2.605185704274702,
      "grad_norm": 0.5562153458595276,
      "learning_rate": 4.0526189906720874e-05,
      "loss": 0.0147,
      "step": 18588
    },
    {
      "epoch": 2.6053258584442887,
      "grad_norm": 0.1636039763689041,
      "learning_rate": 4.0511839272901216e-05,
      "loss": 0.0143,
      "step": 18589
    },
    {
      "epoch": 2.605466012613875,
      "grad_norm": 0.11264914274215698,
      "learning_rate": 4.049748863908155e-05,
      "loss": 0.0128,
      "step": 18590
    },
    {
      "epoch": 2.6056061667834616,
      "grad_norm": 0.11894587427377701,
      "learning_rate": 4.04831380052619e-05,
      "loss": 0.0047,
      "step": 18591
    },
    {
      "epoch": 2.6057463209530485,
      "grad_norm": 0.02416457235813141,
      "learning_rate": 4.0468787371442235e-05,
      "loss": 0.002,
      "step": 18592
    },
    {
      "epoch": 2.605886475122635,
      "grad_norm": 0.7101230621337891,
      "learning_rate": 4.045443673762257e-05,
      "loss": 0.0363,
      "step": 18593
    },
    {
      "epoch": 2.6060266292922214,
      "grad_norm": 0.17755533754825592,
      "learning_rate": 4.044008610380292e-05,
      "loss": 0.0487,
      "step": 18594
    },
    {
      "epoch": 2.606166783461808,
      "grad_norm": 0.23735153675079346,
      "learning_rate": 4.0425735469983255e-05,
      "loss": 0.0578,
      "step": 18595
    },
    {
      "epoch": 2.6063069376313948,
      "grad_norm": 0.291685551404953,
      "learning_rate": 4.041138483616359e-05,
      "loss": 0.0483,
      "step": 18596
    },
    {
      "epoch": 2.606447091800981,
      "grad_norm": 0.24020987749099731,
      "learning_rate": 4.039703420234393e-05,
      "loss": 0.0485,
      "step": 18597
    },
    {
      "epoch": 2.6065872459705677,
      "grad_norm": 0.08242785185575485,
      "learning_rate": 4.0382683568524275e-05,
      "loss": 0.0116,
      "step": 18598
    },
    {
      "epoch": 2.606727400140154,
      "grad_norm": 0.38203585147857666,
      "learning_rate": 4.036833293470462e-05,
      "loss": 0.1042,
      "step": 18599
    },
    {
      "epoch": 2.6068675543097406,
      "grad_norm": 0.3935910761356354,
      "learning_rate": 4.035398230088495e-05,
      "loss": 0.0196,
      "step": 18600
    },
    {
      "epoch": 2.607007708479327,
      "grad_norm": 0.2833174765110016,
      "learning_rate": 4.033963166706529e-05,
      "loss": 0.0122,
      "step": 18601
    },
    {
      "epoch": 2.607147862648914,
      "grad_norm": 0.24325448274612427,
      "learning_rate": 4.0325281033245636e-05,
      "loss": 0.0135,
      "step": 18602
    },
    {
      "epoch": 2.6072880168185004,
      "grad_norm": 0.11215915530920029,
      "learning_rate": 4.031093039942597e-05,
      "loss": 0.0371,
      "step": 18603
    },
    {
      "epoch": 2.607428170988087,
      "grad_norm": 0.1178319975733757,
      "learning_rate": 4.029657976560631e-05,
      "loss": 0.0117,
      "step": 18604
    },
    {
      "epoch": 2.6075683251576733,
      "grad_norm": 0.2862848937511444,
      "learning_rate": 4.028222913178665e-05,
      "loss": 0.0282,
      "step": 18605
    },
    {
      "epoch": 2.6077084793272602,
      "grad_norm": 0.3154328167438507,
      "learning_rate": 4.026787849796699e-05,
      "loss": 0.0112,
      "step": 18606
    },
    {
      "epoch": 2.6078486334968467,
      "grad_norm": 0.6282109618186951,
      "learning_rate": 4.025352786414733e-05,
      "loss": 0.0476,
      "step": 18607
    },
    {
      "epoch": 2.607988787666433,
      "grad_norm": 0.10051321983337402,
      "learning_rate": 4.023917723032767e-05,
      "loss": 0.0043,
      "step": 18608
    },
    {
      "epoch": 2.6081289418360196,
      "grad_norm": 0.38662955164909363,
      "learning_rate": 4.0224826596508004e-05,
      "loss": 0.0236,
      "step": 18609
    },
    {
      "epoch": 2.608269096005606,
      "grad_norm": 0.2803041934967041,
      "learning_rate": 4.021047596268835e-05,
      "loss": 0.0352,
      "step": 18610
    },
    {
      "epoch": 2.6084092501751925,
      "grad_norm": 0.6029271483421326,
      "learning_rate": 4.019612532886869e-05,
      "loss": 0.0226,
      "step": 18611
    },
    {
      "epoch": 2.608549404344779,
      "grad_norm": 0.13210083544254303,
      "learning_rate": 4.0181774695049024e-05,
      "loss": 0.0351,
      "step": 18612
    },
    {
      "epoch": 2.608689558514366,
      "grad_norm": 0.7515329122543335,
      "learning_rate": 4.0167424061229366e-05,
      "loss": 0.1048,
      "step": 18613
    },
    {
      "epoch": 2.6088297126839524,
      "grad_norm": 0.5831995010375977,
      "learning_rate": 4.015307342740971e-05,
      "loss": 0.0375,
      "step": 18614
    },
    {
      "epoch": 2.608969866853539,
      "grad_norm": 0.47365301847457886,
      "learning_rate": 4.013872279359005e-05,
      "loss": 0.0609,
      "step": 18615
    },
    {
      "epoch": 2.6091100210231253,
      "grad_norm": 0.11963936686515808,
      "learning_rate": 4.0124372159770385e-05,
      "loss": 0.0053,
      "step": 18616
    },
    {
      "epoch": 2.609250175192712,
      "grad_norm": 0.013404007069766521,
      "learning_rate": 4.011002152595072e-05,
      "loss": 0.0013,
      "step": 18617
    },
    {
      "epoch": 2.6093903293622986,
      "grad_norm": 0.08783210068941116,
      "learning_rate": 4.009567089213107e-05,
      "loss": 0.0036,
      "step": 18618
    },
    {
      "epoch": 2.609530483531885,
      "grad_norm": 0.8851631879806519,
      "learning_rate": 4.0081320258311405e-05,
      "loss": 0.0453,
      "step": 18619
    },
    {
      "epoch": 2.6096706377014716,
      "grad_norm": 0.1214151605963707,
      "learning_rate": 4.006696962449174e-05,
      "loss": 0.0102,
      "step": 18620
    },
    {
      "epoch": 2.609810791871058,
      "grad_norm": 0.22432951629161835,
      "learning_rate": 4.005261899067208e-05,
      "loss": 0.022,
      "step": 18621
    },
    {
      "epoch": 2.6099509460406445,
      "grad_norm": 0.20865321159362793,
      "learning_rate": 4.0038268356852424e-05,
      "loss": 0.0103,
      "step": 18622
    },
    {
      "epoch": 2.6100911002102314,
      "grad_norm": 0.14837253093719482,
      "learning_rate": 4.0023917723032766e-05,
      "loss": 0.0069,
      "step": 18623
    },
    {
      "epoch": 2.610231254379818,
      "grad_norm": 0.2372204214334488,
      "learning_rate": 4.00095670892131e-05,
      "loss": 0.0409,
      "step": 18624
    },
    {
      "epoch": 2.6103714085494043,
      "grad_norm": 0.1609046459197998,
      "learning_rate": 3.999521645539344e-05,
      "loss": 0.013,
      "step": 18625
    },
    {
      "epoch": 2.6105115627189908,
      "grad_norm": 0.16017426550388336,
      "learning_rate": 3.9980865821573786e-05,
      "loss": 0.0169,
      "step": 18626
    },
    {
      "epoch": 2.6106517168885777,
      "grad_norm": 0.238394096493721,
      "learning_rate": 3.996651518775412e-05,
      "loss": 0.051,
      "step": 18627
    },
    {
      "epoch": 2.610791871058164,
      "grad_norm": 0.43685412406921387,
      "learning_rate": 3.995216455393446e-05,
      "loss": 0.0388,
      "step": 18628
    },
    {
      "epoch": 2.6109320252277506,
      "grad_norm": 0.16761700809001923,
      "learning_rate": 3.9937813920114806e-05,
      "loss": 0.036,
      "step": 18629
    },
    {
      "epoch": 2.611072179397337,
      "grad_norm": 0.09796120971441269,
      "learning_rate": 3.992346328629514e-05,
      "loss": 0.006,
      "step": 18630
    },
    {
      "epoch": 2.6112123335669235,
      "grad_norm": 0.22367095947265625,
      "learning_rate": 3.990911265247548e-05,
      "loss": 0.028,
      "step": 18631
    },
    {
      "epoch": 2.61135248773651,
      "grad_norm": 1.1693377494812012,
      "learning_rate": 3.989476201865582e-05,
      "loss": 0.0534,
      "step": 18632
    },
    {
      "epoch": 2.611492641906097,
      "grad_norm": 0.39454886317253113,
      "learning_rate": 3.988041138483617e-05,
      "loss": 0.0451,
      "step": 18633
    },
    {
      "epoch": 2.6116327960756833,
      "grad_norm": 0.21093155443668365,
      "learning_rate": 3.98660607510165e-05,
      "loss": 0.0146,
      "step": 18634
    },
    {
      "epoch": 2.61177295024527,
      "grad_norm": 0.12873238325119019,
      "learning_rate": 3.985171011719684e-05,
      "loss": 0.025,
      "step": 18635
    },
    {
      "epoch": 2.6119131044148562,
      "grad_norm": 0.29310110211372375,
      "learning_rate": 3.983735948337717e-05,
      "loss": 0.028,
      "step": 18636
    },
    {
      "epoch": 2.612053258584443,
      "grad_norm": 0.5587449073791504,
      "learning_rate": 3.982300884955752e-05,
      "loss": 0.0232,
      "step": 18637
    },
    {
      "epoch": 2.6121934127540296,
      "grad_norm": 0.09433801472187042,
      "learning_rate": 3.980865821573786e-05,
      "loss": 0.0081,
      "step": 18638
    },
    {
      "epoch": 2.612333566923616,
      "grad_norm": 0.5015447735786438,
      "learning_rate": 3.97943075819182e-05,
      "loss": 0.0544,
      "step": 18639
    },
    {
      "epoch": 2.6124737210932025,
      "grad_norm": 0.10308688133955002,
      "learning_rate": 3.9779956948098535e-05,
      "loss": 0.0051,
      "step": 18640
    },
    {
      "epoch": 2.612613875262789,
      "grad_norm": 0.2075146734714508,
      "learning_rate": 3.9765606314278884e-05,
      "loss": 0.0152,
      "step": 18641
    },
    {
      "epoch": 2.6127540294323754,
      "grad_norm": 0.1745568811893463,
      "learning_rate": 3.975125568045922e-05,
      "loss": 0.0278,
      "step": 18642
    },
    {
      "epoch": 2.612894183601962,
      "grad_norm": 0.21518918871879578,
      "learning_rate": 3.9736905046639554e-05,
      "loss": 0.0353,
      "step": 18643
    },
    {
      "epoch": 2.613034337771549,
      "grad_norm": 0.056774020195007324,
      "learning_rate": 3.972255441281989e-05,
      "loss": 0.01,
      "step": 18644
    },
    {
      "epoch": 2.6131744919411353,
      "grad_norm": 0.12873075902462006,
      "learning_rate": 3.970820377900024e-05,
      "loss": 0.0155,
      "step": 18645
    },
    {
      "epoch": 2.6133146461107217,
      "grad_norm": 0.17562609910964966,
      "learning_rate": 3.9693853145180574e-05,
      "loss": 0.0199,
      "step": 18646
    },
    {
      "epoch": 2.613454800280308,
      "grad_norm": 0.17403732240200043,
      "learning_rate": 3.9679502511360916e-05,
      "loss": 0.0287,
      "step": 18647
    },
    {
      "epoch": 2.613594954449895,
      "grad_norm": 0.12795758247375488,
      "learning_rate": 3.966515187754125e-05,
      "loss": 0.0201,
      "step": 18648
    },
    {
      "epoch": 2.6137351086194816,
      "grad_norm": 0.13197122514247894,
      "learning_rate": 3.96508012437216e-05,
      "loss": 0.0143,
      "step": 18649
    },
    {
      "epoch": 2.613875262789068,
      "grad_norm": 0.21328748762607574,
      "learning_rate": 3.9636450609901936e-05,
      "loss": 0.0169,
      "step": 18650
    },
    {
      "epoch": 2.6140154169586545,
      "grad_norm": 0.40437453985214233,
      "learning_rate": 3.962209997608227e-05,
      "loss": 0.0665,
      "step": 18651
    },
    {
      "epoch": 2.614155571128241,
      "grad_norm": 0.1974000632762909,
      "learning_rate": 3.9607749342262606e-05,
      "loss": 0.0148,
      "step": 18652
    },
    {
      "epoch": 2.6142957252978274,
      "grad_norm": 0.14026904106140137,
      "learning_rate": 3.9593398708442955e-05,
      "loss": 0.0178,
      "step": 18653
    },
    {
      "epoch": 2.6144358794674143,
      "grad_norm": 0.5113092660903931,
      "learning_rate": 3.957904807462329e-05,
      "loss": 0.0232,
      "step": 18654
    },
    {
      "epoch": 2.6145760336370008,
      "grad_norm": 0.02748430334031582,
      "learning_rate": 3.956469744080363e-05,
      "loss": 0.0013,
      "step": 18655
    },
    {
      "epoch": 2.614716187806587,
      "grad_norm": 0.20271700620651245,
      "learning_rate": 3.955034680698397e-05,
      "loss": 0.0124,
      "step": 18656
    },
    {
      "epoch": 2.6148563419761737,
      "grad_norm": 0.06992552429437637,
      "learning_rate": 3.953599617316432e-05,
      "loss": 0.0056,
      "step": 18657
    },
    {
      "epoch": 2.6149964961457606,
      "grad_norm": 0.31607091426849365,
      "learning_rate": 3.952164553934465e-05,
      "loss": 0.0243,
      "step": 18658
    },
    {
      "epoch": 2.615136650315347,
      "grad_norm": 0.25952139496803284,
      "learning_rate": 3.950729490552499e-05,
      "loss": 0.0455,
      "step": 18659
    },
    {
      "epoch": 2.6152768044849335,
      "grad_norm": 0.12047473341226578,
      "learning_rate": 3.949294427170532e-05,
      "loss": 0.021,
      "step": 18660
    },
    {
      "epoch": 2.61541695865452,
      "grad_norm": 0.15446336567401886,
      "learning_rate": 3.947859363788567e-05,
      "loss": 0.0146,
      "step": 18661
    },
    {
      "epoch": 2.6155571128241064,
      "grad_norm": 0.14092274010181427,
      "learning_rate": 3.946424300406601e-05,
      "loss": 0.0048,
      "step": 18662
    },
    {
      "epoch": 2.615697266993693,
      "grad_norm": 0.5055900812149048,
      "learning_rate": 3.944989237024635e-05,
      "loss": 0.0464,
      "step": 18663
    },
    {
      "epoch": 2.6158374211632793,
      "grad_norm": 0.012625631876289845,
      "learning_rate": 3.9435541736426685e-05,
      "loss": 0.0009,
      "step": 18664
    },
    {
      "epoch": 2.6159775753328662,
      "grad_norm": 0.09986929595470428,
      "learning_rate": 3.9421191102607034e-05,
      "loss": 0.0028,
      "step": 18665
    },
    {
      "epoch": 2.6161177295024527,
      "grad_norm": 0.16719025373458862,
      "learning_rate": 3.940684046878737e-05,
      "loss": 0.0228,
      "step": 18666
    },
    {
      "epoch": 2.616257883672039,
      "grad_norm": 1.1687332391738892,
      "learning_rate": 3.9392489834967704e-05,
      "loss": 0.1238,
      "step": 18667
    },
    {
      "epoch": 2.616398037841626,
      "grad_norm": 0.24654161930084229,
      "learning_rate": 3.937813920114805e-05,
      "loss": 0.0354,
      "step": 18668
    },
    {
      "epoch": 2.6165381920112125,
      "grad_norm": 2.5124268531799316,
      "learning_rate": 3.936378856732839e-05,
      "loss": 0.1255,
      "step": 18669
    },
    {
      "epoch": 2.616678346180799,
      "grad_norm": 0.21716395020484924,
      "learning_rate": 3.9349437933508724e-05,
      "loss": 0.0075,
      "step": 18670
    },
    {
      "epoch": 2.6168185003503854,
      "grad_norm": 0.2228481024503708,
      "learning_rate": 3.9335087299689066e-05,
      "loss": 0.0212,
      "step": 18671
    },
    {
      "epoch": 2.616958654519972,
      "grad_norm": 0.1531447321176529,
      "learning_rate": 3.932073666586941e-05,
      "loss": 0.0167,
      "step": 18672
    },
    {
      "epoch": 2.6170988086895584,
      "grad_norm": 0.16984587907791138,
      "learning_rate": 3.930638603204975e-05,
      "loss": 0.0126,
      "step": 18673
    },
    {
      "epoch": 2.617238962859145,
      "grad_norm": 0.18887673318386078,
      "learning_rate": 3.9292035398230085e-05,
      "loss": 0.0371,
      "step": 18674
    },
    {
      "epoch": 2.6173791170287317,
      "grad_norm": 0.4719323217868805,
      "learning_rate": 3.927768476441042e-05,
      "loss": 0.0087,
      "step": 18675
    },
    {
      "epoch": 2.617519271198318,
      "grad_norm": 0.27098533511161804,
      "learning_rate": 3.926333413059077e-05,
      "loss": 0.0621,
      "step": 18676
    },
    {
      "epoch": 2.6176594253679046,
      "grad_norm": 0.2018071413040161,
      "learning_rate": 3.9248983496771105e-05,
      "loss": 0.0248,
      "step": 18677
    },
    {
      "epoch": 2.617799579537491,
      "grad_norm": 0.15955694019794464,
      "learning_rate": 3.923463286295144e-05,
      "loss": 0.0343,
      "step": 18678
    },
    {
      "epoch": 2.617939733707078,
      "grad_norm": 0.24343745410442352,
      "learning_rate": 3.922028222913178e-05,
      "loss": 0.0619,
      "step": 18679
    },
    {
      "epoch": 2.6180798878766645,
      "grad_norm": 0.3357413411140442,
      "learning_rate": 3.9205931595312125e-05,
      "loss": 0.0876,
      "step": 18680
    },
    {
      "epoch": 2.618220042046251,
      "grad_norm": 0.18207237124443054,
      "learning_rate": 3.919158096149247e-05,
      "loss": 0.0172,
      "step": 18681
    },
    {
      "epoch": 2.6183601962158374,
      "grad_norm": 0.4779229164123535,
      "learning_rate": 3.91772303276728e-05,
      "loss": 0.0238,
      "step": 18682
    },
    {
      "epoch": 2.618500350385424,
      "grad_norm": 0.45011502504348755,
      "learning_rate": 3.916287969385314e-05,
      "loss": 0.0597,
      "step": 18683
    },
    {
      "epoch": 2.6186405045550103,
      "grad_norm": 0.2848474681377411,
      "learning_rate": 3.9148529060033486e-05,
      "loss": 0.0169,
      "step": 18684
    },
    {
      "epoch": 2.618780658724597,
      "grad_norm": 0.12286489456892014,
      "learning_rate": 3.913417842621382e-05,
      "loss": 0.0121,
      "step": 18685
    },
    {
      "epoch": 2.6189208128941837,
      "grad_norm": 0.24249719083309174,
      "learning_rate": 3.911982779239416e-05,
      "loss": 0.0357,
      "step": 18686
    },
    {
      "epoch": 2.61906096706377,
      "grad_norm": 0.12054488062858582,
      "learning_rate": 3.91054771585745e-05,
      "loss": 0.0209,
      "step": 18687
    },
    {
      "epoch": 2.6192011212333566,
      "grad_norm": 0.08669152855873108,
      "learning_rate": 3.909112652475484e-05,
      "loss": 0.0076,
      "step": 18688
    },
    {
      "epoch": 2.6193412754029435,
      "grad_norm": 0.10875049978494644,
      "learning_rate": 3.907677589093518e-05,
      "loss": 0.0105,
      "step": 18689
    },
    {
      "epoch": 2.61948142957253,
      "grad_norm": 0.11482339352369308,
      "learning_rate": 3.906242525711552e-05,
      "loss": 0.0117,
      "step": 18690
    },
    {
      "epoch": 2.6196215837421164,
      "grad_norm": 0.21303151547908783,
      "learning_rate": 3.9048074623295854e-05,
      "loss": 0.0158,
      "step": 18691
    },
    {
      "epoch": 2.619761737911703,
      "grad_norm": 0.28482940793037415,
      "learning_rate": 3.90337239894762e-05,
      "loss": 0.01,
      "step": 18692
    },
    {
      "epoch": 2.6199018920812893,
      "grad_norm": 0.07731757313013077,
      "learning_rate": 3.901937335565654e-05,
      "loss": 0.0072,
      "step": 18693
    },
    {
      "epoch": 2.620042046250876,
      "grad_norm": 0.2117357701063156,
      "learning_rate": 3.9005022721836873e-05,
      "loss": 0.0472,
      "step": 18694
    },
    {
      "epoch": 2.6201822004204622,
      "grad_norm": 0.3398694097995758,
      "learning_rate": 3.8990672088017216e-05,
      "loss": 0.0201,
      "step": 18695
    },
    {
      "epoch": 2.620322354590049,
      "grad_norm": 0.08713182061910629,
      "learning_rate": 3.897632145419756e-05,
      "loss": 0.0152,
      "step": 18696
    },
    {
      "epoch": 2.6204625087596356,
      "grad_norm": 0.18486535549163818,
      "learning_rate": 3.89619708203779e-05,
      "loss": 0.0146,
      "step": 18697
    },
    {
      "epoch": 2.620602662929222,
      "grad_norm": 0.04910740628838539,
      "learning_rate": 3.8947620186558235e-05,
      "loss": 0.0058,
      "step": 18698
    },
    {
      "epoch": 2.620742817098809,
      "grad_norm": 0.29267776012420654,
      "learning_rate": 3.893326955273857e-05,
      "loss": 0.019,
      "step": 18699
    },
    {
      "epoch": 2.6208829712683954,
      "grad_norm": 0.06982609629631042,
      "learning_rate": 3.891891891891892e-05,
      "loss": 0.0037,
      "step": 18700
    },
    {
      "epoch": 2.621023125437982,
      "grad_norm": 0.10264930874109268,
      "learning_rate": 3.8904568285099255e-05,
      "loss": 0.0144,
      "step": 18701
    },
    {
      "epoch": 2.6211632796075683,
      "grad_norm": 0.11372557282447815,
      "learning_rate": 3.889021765127959e-05,
      "loss": 0.0167,
      "step": 18702
    },
    {
      "epoch": 2.621303433777155,
      "grad_norm": 0.02014012448489666,
      "learning_rate": 3.887586701745994e-05,
      "loss": 0.0017,
      "step": 18703
    },
    {
      "epoch": 2.6214435879467413,
      "grad_norm": 0.13954277336597443,
      "learning_rate": 3.8861516383640274e-05,
      "loss": 0.0057,
      "step": 18704
    },
    {
      "epoch": 2.6215837421163277,
      "grad_norm": 0.16644546389579773,
      "learning_rate": 3.8847165749820616e-05,
      "loss": 0.0066,
      "step": 18705
    },
    {
      "epoch": 2.6217238962859146,
      "grad_norm": 0.09715601801872253,
      "learning_rate": 3.883281511600095e-05,
      "loss": 0.0036,
      "step": 18706
    },
    {
      "epoch": 2.621864050455501,
      "grad_norm": 0.1313941925764084,
      "learning_rate": 3.8818464482181294e-05,
      "loss": 0.0086,
      "step": 18707
    },
    {
      "epoch": 2.6220042046250875,
      "grad_norm": 0.09903451800346375,
      "learning_rate": 3.8804113848361636e-05,
      "loss": 0.0118,
      "step": 18708
    },
    {
      "epoch": 2.622144358794674,
      "grad_norm": 0.35303157567977905,
      "learning_rate": 3.878976321454197e-05,
      "loss": 0.0416,
      "step": 18709
    },
    {
      "epoch": 2.622284512964261,
      "grad_norm": 0.1517418920993805,
      "learning_rate": 3.8775412580722307e-05,
      "loss": 0.0069,
      "step": 18710
    },
    {
      "epoch": 2.6224246671338474,
      "grad_norm": 0.1451348066329956,
      "learning_rate": 3.8761061946902655e-05,
      "loss": 0.0192,
      "step": 18711
    },
    {
      "epoch": 2.622564821303434,
      "grad_norm": 0.517396092414856,
      "learning_rate": 3.874671131308299e-05,
      "loss": 0.0586,
      "step": 18712
    },
    {
      "epoch": 2.6227049754730203,
      "grad_norm": 0.5663255453109741,
      "learning_rate": 3.873236067926333e-05,
      "loss": 0.0961,
      "step": 18713
    },
    {
      "epoch": 2.6228451296426067,
      "grad_norm": 0.330269455909729,
      "learning_rate": 3.871801004544367e-05,
      "loss": 0.0318,
      "step": 18714
    },
    {
      "epoch": 2.622985283812193,
      "grad_norm": 1.0959969758987427,
      "learning_rate": 3.870365941162401e-05,
      "loss": 0.0655,
      "step": 18715
    },
    {
      "epoch": 2.62312543798178,
      "grad_norm": 0.17817942798137665,
      "learning_rate": 3.868930877780435e-05,
      "loss": 0.0281,
      "step": 18716
    },
    {
      "epoch": 2.6232655921513666,
      "grad_norm": 1.015141487121582,
      "learning_rate": 3.867495814398469e-05,
      "loss": 0.1967,
      "step": 18717
    },
    {
      "epoch": 2.623405746320953,
      "grad_norm": 1.9474964141845703,
      "learning_rate": 3.866060751016502e-05,
      "loss": 0.2112,
      "step": 18718
    },
    {
      "epoch": 2.6235459004905395,
      "grad_norm": 0.7807974219322205,
      "learning_rate": 3.864625687634537e-05,
      "loss": 0.0294,
      "step": 18719
    },
    {
      "epoch": 2.6236860546601264,
      "grad_norm": 2.3920202255249023,
      "learning_rate": 3.863190624252571e-05,
      "loss": 0.0499,
      "step": 18720
    },
    {
      "epoch": 2.623826208829713,
      "grad_norm": 0.2157333642244339,
      "learning_rate": 3.861755560870605e-05,
      "loss": 0.0317,
      "step": 18721
    },
    {
      "epoch": 2.6239663629992993,
      "grad_norm": 0.19947752356529236,
      "learning_rate": 3.8603204974886385e-05,
      "loss": 0.0358,
      "step": 18722
    },
    {
      "epoch": 2.6241065171688858,
      "grad_norm": 0.1280020773410797,
      "learning_rate": 3.858885434106673e-05,
      "loss": 0.0184,
      "step": 18723
    },
    {
      "epoch": 2.6242466713384722,
      "grad_norm": 0.7770083546638489,
      "learning_rate": 3.857450370724707e-05,
      "loss": 0.0459,
      "step": 18724
    },
    {
      "epoch": 2.6243868255080587,
      "grad_norm": 0.23050451278686523,
      "learning_rate": 3.8560153073427404e-05,
      "loss": 0.016,
      "step": 18725
    },
    {
      "epoch": 2.624526979677645,
      "grad_norm": 0.28075748682022095,
      "learning_rate": 3.854580243960774e-05,
      "loss": 0.0149,
      "step": 18726
    },
    {
      "epoch": 2.624667133847232,
      "grad_norm": 0.1706288903951645,
      "learning_rate": 3.853145180578809e-05,
      "loss": 0.0152,
      "step": 18727
    },
    {
      "epoch": 2.6248072880168185,
      "grad_norm": 0.2784833312034607,
      "learning_rate": 3.8517101171968424e-05,
      "loss": 0.038,
      "step": 18728
    },
    {
      "epoch": 2.624947442186405,
      "grad_norm": 0.06164579465985298,
      "learning_rate": 3.8502750538148766e-05,
      "loss": 0.006,
      "step": 18729
    },
    {
      "epoch": 2.6250875963559914,
      "grad_norm": 0.195857971906662,
      "learning_rate": 3.84883999043291e-05,
      "loss": 0.0145,
      "step": 18730
    },
    {
      "epoch": 2.6252277505255783,
      "grad_norm": 0.11202902346849442,
      "learning_rate": 3.8474049270509444e-05,
      "loss": 0.0127,
      "step": 18731
    },
    {
      "epoch": 2.625367904695165,
      "grad_norm": 0.1263100653886795,
      "learning_rate": 3.8459698636689786e-05,
      "loss": 0.0194,
      "step": 18732
    },
    {
      "epoch": 2.6255080588647512,
      "grad_norm": 0.2511688768863678,
      "learning_rate": 3.844534800287012e-05,
      "loss": 0.0348,
      "step": 18733
    },
    {
      "epoch": 2.6256482130343377,
      "grad_norm": 0.40503019094467163,
      "learning_rate": 3.8430997369050456e-05,
      "loss": 0.0438,
      "step": 18734
    },
    {
      "epoch": 2.625788367203924,
      "grad_norm": 0.3158361315727234,
      "learning_rate": 3.8416646735230805e-05,
      "loss": 0.0304,
      "step": 18735
    },
    {
      "epoch": 2.6259285213735106,
      "grad_norm": 0.06601721048355103,
      "learning_rate": 3.840229610141114e-05,
      "loss": 0.0044,
      "step": 18736
    },
    {
      "epoch": 2.6260686755430975,
      "grad_norm": 0.41442951560020447,
      "learning_rate": 3.838794546759148e-05,
      "loss": 0.0316,
      "step": 18737
    },
    {
      "epoch": 2.626208829712684,
      "grad_norm": 0.4343923032283783,
      "learning_rate": 3.837359483377182e-05,
      "loss": 0.05,
      "step": 18738
    },
    {
      "epoch": 2.6263489838822704,
      "grad_norm": 0.442106157541275,
      "learning_rate": 3.835924419995216e-05,
      "loss": 0.0406,
      "step": 18739
    },
    {
      "epoch": 2.626489138051857,
      "grad_norm": 0.192957803606987,
      "learning_rate": 3.83448935661325e-05,
      "loss": 0.0375,
      "step": 18740
    },
    {
      "epoch": 2.626629292221444,
      "grad_norm": 0.08421743661165237,
      "learning_rate": 3.833054293231284e-05,
      "loss": 0.0216,
      "step": 18741
    },
    {
      "epoch": 2.6267694463910303,
      "grad_norm": 0.4264732599258423,
      "learning_rate": 3.8316192298493186e-05,
      "loss": 0.0542,
      "step": 18742
    },
    {
      "epoch": 2.6269096005606167,
      "grad_norm": 0.25863268971443176,
      "learning_rate": 3.830184166467352e-05,
      "loss": 0.0968,
      "step": 18743
    },
    {
      "epoch": 2.627049754730203,
      "grad_norm": 0.3869832158088684,
      "learning_rate": 3.828749103085386e-05,
      "loss": 0.0299,
      "step": 18744
    },
    {
      "epoch": 2.6271899088997896,
      "grad_norm": 0.3391312062740326,
      "learning_rate": 3.82731403970342e-05,
      "loss": 0.0109,
      "step": 18745
    },
    {
      "epoch": 2.627330063069376,
      "grad_norm": 0.06701905280351639,
      "learning_rate": 3.825878976321454e-05,
      "loss": 0.0074,
      "step": 18746
    },
    {
      "epoch": 2.627470217238963,
      "grad_norm": 0.22616274654865265,
      "learning_rate": 3.824443912939488e-05,
      "loss": 0.0399,
      "step": 18747
    },
    {
      "epoch": 2.6276103714085495,
      "grad_norm": 0.06368976086378098,
      "learning_rate": 3.823008849557522e-05,
      "loss": 0.006,
      "step": 18748
    },
    {
      "epoch": 2.627750525578136,
      "grad_norm": 0.6950241327285767,
      "learning_rate": 3.8215737861755554e-05,
      "loss": 0.0348,
      "step": 18749
    },
    {
      "epoch": 2.6278906797477224,
      "grad_norm": 0.11261796206235886,
      "learning_rate": 3.82013872279359e-05,
      "loss": 0.009,
      "step": 18750
    },
    {
      "epoch": 2.6280308339173093,
      "grad_norm": 0.4362614154815674,
      "learning_rate": 3.818703659411624e-05,
      "loss": 0.0283,
      "step": 18751
    },
    {
      "epoch": 2.6281709880868958,
      "grad_norm": 0.0697779431939125,
      "learning_rate": 3.8172685960296574e-05,
      "loss": 0.0057,
      "step": 18752
    },
    {
      "epoch": 2.628311142256482,
      "grad_norm": 0.1416267305612564,
      "learning_rate": 3.8158335326476916e-05,
      "loss": 0.0106,
      "step": 18753
    },
    {
      "epoch": 2.6284512964260687,
      "grad_norm": 0.20650041103363037,
      "learning_rate": 3.814398469265726e-05,
      "loss": 0.0452,
      "step": 18754
    },
    {
      "epoch": 2.628591450595655,
      "grad_norm": 0.15064755082130432,
      "learning_rate": 3.812963405883759e-05,
      "loss": 0.0106,
      "step": 18755
    },
    {
      "epoch": 2.6287316047652416,
      "grad_norm": 0.11406257748603821,
      "learning_rate": 3.8115283425017935e-05,
      "loss": 0.0039,
      "step": 18756
    },
    {
      "epoch": 2.628871758934828,
      "grad_norm": 0.396472692489624,
      "learning_rate": 3.810093279119827e-05,
      "loss": 0.0222,
      "step": 18757
    },
    {
      "epoch": 2.629011913104415,
      "grad_norm": 1.1222608089447021,
      "learning_rate": 3.808658215737862e-05,
      "loss": 0.0595,
      "step": 18758
    },
    {
      "epoch": 2.6291520672740014,
      "grad_norm": 0.6270762085914612,
      "learning_rate": 3.8072231523558955e-05,
      "loss": 0.0762,
      "step": 18759
    },
    {
      "epoch": 2.629292221443588,
      "grad_norm": 0.13180424273014069,
      "learning_rate": 3.805788088973929e-05,
      "loss": 0.0071,
      "step": 18760
    },
    {
      "epoch": 2.6294323756131743,
      "grad_norm": 0.11908093839883804,
      "learning_rate": 3.804353025591963e-05,
      "loss": 0.0165,
      "step": 18761
    },
    {
      "epoch": 2.6295725297827612,
      "grad_norm": 0.09246785193681717,
      "learning_rate": 3.8029179622099974e-05,
      "loss": 0.0083,
      "step": 18762
    },
    {
      "epoch": 2.6297126839523477,
      "grad_norm": 0.16399389505386353,
      "learning_rate": 3.801482898828031e-05,
      "loss": 0.0075,
      "step": 18763
    },
    {
      "epoch": 2.629852838121934,
      "grad_norm": 0.1467788815498352,
      "learning_rate": 3.800047835446065e-05,
      "loss": 0.0048,
      "step": 18764
    },
    {
      "epoch": 2.6299929922915206,
      "grad_norm": 0.09992936253547668,
      "learning_rate": 3.798612772064099e-05,
      "loss": 0.0111,
      "step": 18765
    },
    {
      "epoch": 2.630133146461107,
      "grad_norm": 0.6389560103416443,
      "learning_rate": 3.7971777086821336e-05,
      "loss": 0.0302,
      "step": 18766
    },
    {
      "epoch": 2.6302733006306935,
      "grad_norm": 1.1857097148895264,
      "learning_rate": 3.795742645300167e-05,
      "loss": 0.0768,
      "step": 18767
    },
    {
      "epoch": 2.6304134548002804,
      "grad_norm": 0.26087599992752075,
      "learning_rate": 3.794307581918201e-05,
      "loss": 0.0396,
      "step": 18768
    },
    {
      "epoch": 2.630553608969867,
      "grad_norm": 1.3678200244903564,
      "learning_rate": 3.792872518536235e-05,
      "loss": 0.2118,
      "step": 18769
    },
    {
      "epoch": 2.6306937631394534,
      "grad_norm": 0.25882554054260254,
      "learning_rate": 3.791437455154269e-05,
      "loss": 0.0101,
      "step": 18770
    },
    {
      "epoch": 2.63083391730904,
      "grad_norm": 0.1423613727092743,
      "learning_rate": 3.7900023917723026e-05,
      "loss": 0.0283,
      "step": 18771
    },
    {
      "epoch": 2.6309740714786267,
      "grad_norm": 0.44040006399154663,
      "learning_rate": 3.788567328390337e-05,
      "loss": 0.0138,
      "step": 18772
    },
    {
      "epoch": 2.631114225648213,
      "grad_norm": 0.15163441002368927,
      "learning_rate": 3.7871322650083704e-05,
      "loss": 0.0107,
      "step": 18773
    },
    {
      "epoch": 2.6312543798177996,
      "grad_norm": 0.10793764889240265,
      "learning_rate": 3.785697201626405e-05,
      "loss": 0.0205,
      "step": 18774
    },
    {
      "epoch": 2.631394533987386,
      "grad_norm": 0.2825261354446411,
      "learning_rate": 3.784262138244439e-05,
      "loss": 0.0296,
      "step": 18775
    },
    {
      "epoch": 2.6315346881569726,
      "grad_norm": 0.09201250970363617,
      "learning_rate": 3.7828270748624723e-05,
      "loss": 0.0113,
      "step": 18776
    },
    {
      "epoch": 2.631674842326559,
      "grad_norm": 0.14555513858795166,
      "learning_rate": 3.781392011480507e-05,
      "loss": 0.0194,
      "step": 18777
    },
    {
      "epoch": 2.631814996496146,
      "grad_norm": 0.08499991148710251,
      "learning_rate": 3.779956948098541e-05,
      "loss": 0.0087,
      "step": 18778
    },
    {
      "epoch": 2.6319551506657324,
      "grad_norm": 0.13297484815120697,
      "learning_rate": 3.778521884716574e-05,
      "loss": 0.0136,
      "step": 18779
    },
    {
      "epoch": 2.632095304835319,
      "grad_norm": 0.1272265464067459,
      "learning_rate": 3.7770868213346085e-05,
      "loss": 0.0214,
      "step": 18780
    },
    {
      "epoch": 2.6322354590049053,
      "grad_norm": 0.16664016246795654,
      "learning_rate": 3.775651757952643e-05,
      "loss": 0.0353,
      "step": 18781
    },
    {
      "epoch": 2.632375613174492,
      "grad_norm": 0.29483941197395325,
      "learning_rate": 3.774216694570677e-05,
      "loss": 0.04,
      "step": 18782
    },
    {
      "epoch": 2.6325157673440787,
      "grad_norm": 0.25430095195770264,
      "learning_rate": 3.7727816311887105e-05,
      "loss": 0.0246,
      "step": 18783
    },
    {
      "epoch": 2.632655921513665,
      "grad_norm": 0.25163504481315613,
      "learning_rate": 3.771346567806744e-05,
      "loss": 0.0327,
      "step": 18784
    },
    {
      "epoch": 2.6327960756832516,
      "grad_norm": 0.18371450901031494,
      "learning_rate": 3.769911504424779e-05,
      "loss": 0.0095,
      "step": 18785
    },
    {
      "epoch": 2.632936229852838,
      "grad_norm": 0.16136924922466278,
      "learning_rate": 3.7684764410428124e-05,
      "loss": 0.0102,
      "step": 18786
    },
    {
      "epoch": 2.6330763840224245,
      "grad_norm": 0.06175116449594498,
      "learning_rate": 3.7670413776608466e-05,
      "loss": 0.0142,
      "step": 18787
    },
    {
      "epoch": 2.633216538192011,
      "grad_norm": 0.24549566209316254,
      "learning_rate": 3.76560631427888e-05,
      "loss": 0.0119,
      "step": 18788
    },
    {
      "epoch": 2.633356692361598,
      "grad_norm": 1.073024868965149,
      "learning_rate": 3.7641712508969144e-05,
      "loss": 0.0601,
      "step": 18789
    },
    {
      "epoch": 2.6334968465311843,
      "grad_norm": 0.22789622843265533,
      "learning_rate": 3.7627361875149486e-05,
      "loss": 0.0248,
      "step": 18790
    },
    {
      "epoch": 2.633637000700771,
      "grad_norm": 0.1780734807252884,
      "learning_rate": 3.761301124132982e-05,
      "loss": 0.0276,
      "step": 18791
    },
    {
      "epoch": 2.6337771548703572,
      "grad_norm": 0.3170514702796936,
      "learning_rate": 3.7598660607510157e-05,
      "loss": 0.0279,
      "step": 18792
    },
    {
      "epoch": 2.633917309039944,
      "grad_norm": 0.16251447796821594,
      "learning_rate": 3.7584309973690505e-05,
      "loss": 0.0117,
      "step": 18793
    },
    {
      "epoch": 2.6340574632095306,
      "grad_norm": 0.12397683411836624,
      "learning_rate": 3.756995933987084e-05,
      "loss": 0.0201,
      "step": 18794
    },
    {
      "epoch": 2.634197617379117,
      "grad_norm": 0.1930229663848877,
      "learning_rate": 3.755560870605118e-05,
      "loss": 0.0352,
      "step": 18795
    },
    {
      "epoch": 2.6343377715487035,
      "grad_norm": 0.12706762552261353,
      "learning_rate": 3.754125807223152e-05,
      "loss": 0.0079,
      "step": 18796
    },
    {
      "epoch": 2.63447792571829,
      "grad_norm": 0.22937016189098358,
      "learning_rate": 3.752690743841186e-05,
      "loss": 0.0267,
      "step": 18797
    },
    {
      "epoch": 2.6346180798878764,
      "grad_norm": 0.2612133324146271,
      "learning_rate": 3.75125568045922e-05,
      "loss": 0.0341,
      "step": 18798
    },
    {
      "epoch": 2.6347582340574633,
      "grad_norm": 0.49783846735954285,
      "learning_rate": 3.749820617077254e-05,
      "loss": 0.0503,
      "step": 18799
    },
    {
      "epoch": 2.63489838822705,
      "grad_norm": 0.11377884447574615,
      "learning_rate": 3.748385553695288e-05,
      "loss": 0.0123,
      "step": 18800
    },
    {
      "epoch": 2.6350385423966363,
      "grad_norm": 0.2304898053407669,
      "learning_rate": 3.7469504903133215e-05,
      "loss": 0.0338,
      "step": 18801
    },
    {
      "epoch": 2.6351786965662227,
      "grad_norm": 0.1135595515370369,
      "learning_rate": 3.745515426931356e-05,
      "loss": 0.0177,
      "step": 18802
    },
    {
      "epoch": 2.6353188507358096,
      "grad_norm": 0.28614211082458496,
      "learning_rate": 3.74408036354939e-05,
      "loss": 0.0416,
      "step": 18803
    },
    {
      "epoch": 2.635459004905396,
      "grad_norm": 0.13080421090126038,
      "learning_rate": 3.7426453001674235e-05,
      "loss": 0.0049,
      "step": 18804
    },
    {
      "epoch": 2.6355991590749825,
      "grad_norm": 0.2835003733634949,
      "learning_rate": 3.741210236785458e-05,
      "loss": 0.0254,
      "step": 18805
    },
    {
      "epoch": 2.635739313244569,
      "grad_norm": 0.2274160236120224,
      "learning_rate": 3.739775173403492e-05,
      "loss": 0.0224,
      "step": 18806
    },
    {
      "epoch": 2.6358794674141555,
      "grad_norm": 0.2237512171268463,
      "learning_rate": 3.7383401100215254e-05,
      "loss": 0.0221,
      "step": 18807
    },
    {
      "epoch": 2.636019621583742,
      "grad_norm": 0.07804173231124878,
      "learning_rate": 3.7369050466395596e-05,
      "loss": 0.0044,
      "step": 18808
    },
    {
      "epoch": 2.636159775753329,
      "grad_norm": 0.7695476412773132,
      "learning_rate": 3.735469983257593e-05,
      "loss": 0.045,
      "step": 18809
    },
    {
      "epoch": 2.6362999299229153,
      "grad_norm": 0.23624801635742188,
      "learning_rate": 3.7340349198756274e-05,
      "loss": 0.0135,
      "step": 18810
    },
    {
      "epoch": 2.6364400840925017,
      "grad_norm": 0.5543002486228943,
      "learning_rate": 3.7325998564936616e-05,
      "loss": 0.0667,
      "step": 18811
    },
    {
      "epoch": 2.636580238262088,
      "grad_norm": 0.05600513517856598,
      "learning_rate": 3.731164793111695e-05,
      "loss": 0.0031,
      "step": 18812
    },
    {
      "epoch": 2.636720392431675,
      "grad_norm": 0.15577149391174316,
      "learning_rate": 3.7297297297297293e-05,
      "loss": 0.0285,
      "step": 18813
    },
    {
      "epoch": 2.6368605466012616,
      "grad_norm": 0.23244145512580872,
      "learning_rate": 3.7282946663477636e-05,
      "loss": 0.0474,
      "step": 18814
    },
    {
      "epoch": 2.637000700770848,
      "grad_norm": 0.6145548224449158,
      "learning_rate": 3.726859602965798e-05,
      "loss": 0.1077,
      "step": 18815
    },
    {
      "epoch": 2.6371408549404345,
      "grad_norm": 0.5801859498023987,
      "learning_rate": 3.725424539583831e-05,
      "loss": 0.0227,
      "step": 18816
    },
    {
      "epoch": 2.637281009110021,
      "grad_norm": 0.6869896650314331,
      "learning_rate": 3.7239894762018655e-05,
      "loss": 0.055,
      "step": 18817
    },
    {
      "epoch": 2.6374211632796074,
      "grad_norm": 1.5788154602050781,
      "learning_rate": 3.722554412819899e-05,
      "loss": 0.076,
      "step": 18818
    },
    {
      "epoch": 2.637561317449194,
      "grad_norm": 1.0737547874450684,
      "learning_rate": 3.721119349437933e-05,
      "loss": 0.1191,
      "step": 18819
    },
    {
      "epoch": 2.6377014716187808,
      "grad_norm": 1.3908734321594238,
      "learning_rate": 3.719684286055967e-05,
      "loss": 0.5036,
      "step": 18820
    },
    {
      "epoch": 2.6378416257883672,
      "grad_norm": 0.34114110469818115,
      "learning_rate": 3.718249222674001e-05,
      "loss": 0.0173,
      "step": 18821
    },
    {
      "epoch": 2.6379817799579537,
      "grad_norm": 0.15989378094673157,
      "learning_rate": 3.716814159292035e-05,
      "loss": 0.0214,
      "step": 18822
    },
    {
      "epoch": 2.63812193412754,
      "grad_norm": 0.2138173133134842,
      "learning_rate": 3.7153790959100694e-05,
      "loss": 0.0193,
      "step": 18823
    },
    {
      "epoch": 2.638262088297127,
      "grad_norm": 0.10950767248868942,
      "learning_rate": 3.713944032528103e-05,
      "loss": 0.0073,
      "step": 18824
    },
    {
      "epoch": 2.6384022424667135,
      "grad_norm": 0.09944452345371246,
      "learning_rate": 3.712508969146137e-05,
      "loss": 0.0087,
      "step": 18825
    },
    {
      "epoch": 2.6385423966363,
      "grad_norm": 0.09011691808700562,
      "learning_rate": 3.711073905764171e-05,
      "loss": 0.0089,
      "step": 18826
    },
    {
      "epoch": 2.6386825508058864,
      "grad_norm": 0.19053417444229126,
      "learning_rate": 3.709638842382205e-05,
      "loss": 0.0316,
      "step": 18827
    },
    {
      "epoch": 2.638822704975473,
      "grad_norm": 0.27169933915138245,
      "learning_rate": 3.7082037790002385e-05,
      "loss": 0.0382,
      "step": 18828
    },
    {
      "epoch": 2.6389628591450593,
      "grad_norm": 0.09622090309858322,
      "learning_rate": 3.706768715618273e-05,
      "loss": 0.0122,
      "step": 18829
    },
    {
      "epoch": 2.6391030133146463,
      "grad_norm": 0.12218036502599716,
      "learning_rate": 3.705333652236307e-05,
      "loss": 0.0116,
      "step": 18830
    },
    {
      "epoch": 2.6392431674842327,
      "grad_norm": 0.2866056561470032,
      "learning_rate": 3.703898588854341e-05,
      "loss": 0.0121,
      "step": 18831
    },
    {
      "epoch": 2.639383321653819,
      "grad_norm": 0.22964531183242798,
      "learning_rate": 3.7024635254723746e-05,
      "loss": 0.0377,
      "step": 18832
    },
    {
      "epoch": 2.6395234758234056,
      "grad_norm": 0.17243830859661102,
      "learning_rate": 3.701028462090409e-05,
      "loss": 0.0237,
      "step": 18833
    },
    {
      "epoch": 2.6396636299929925,
      "grad_norm": 0.37671300768852234,
      "learning_rate": 3.6995933987084424e-05,
      "loss": 0.0224,
      "step": 18834
    },
    {
      "epoch": 2.639803784162579,
      "grad_norm": 0.09349383413791656,
      "learning_rate": 3.6981583353264766e-05,
      "loss": 0.0097,
      "step": 18835
    },
    {
      "epoch": 2.6399439383321655,
      "grad_norm": 0.10590095818042755,
      "learning_rate": 3.69672327194451e-05,
      "loss": 0.0064,
      "step": 18836
    },
    {
      "epoch": 2.640084092501752,
      "grad_norm": 0.10722266882658005,
      "learning_rate": 3.695288208562544e-05,
      "loss": 0.0092,
      "step": 18837
    },
    {
      "epoch": 2.6402242466713384,
      "grad_norm": 0.14299264550209045,
      "learning_rate": 3.6938531451805785e-05,
      "loss": 0.0223,
      "step": 18838
    },
    {
      "epoch": 2.640364400840925,
      "grad_norm": 0.12215137481689453,
      "learning_rate": 3.692418081798613e-05,
      "loss": 0.0136,
      "step": 18839
    },
    {
      "epoch": 2.6405045550105113,
      "grad_norm": 0.06626320630311966,
      "learning_rate": 3.690983018416646e-05,
      "loss": 0.0076,
      "step": 18840
    },
    {
      "epoch": 2.640644709180098,
      "grad_norm": 0.6944338083267212,
      "learning_rate": 3.6895479550346805e-05,
      "loss": 0.0332,
      "step": 18841
    },
    {
      "epoch": 2.6407848633496847,
      "grad_norm": 0.15840215981006622,
      "learning_rate": 3.688112891652714e-05,
      "loss": 0.018,
      "step": 18842
    },
    {
      "epoch": 2.640925017519271,
      "grad_norm": 0.2623918056488037,
      "learning_rate": 3.686677828270748e-05,
      "loss": 0.0415,
      "step": 18843
    },
    {
      "epoch": 2.641065171688858,
      "grad_norm": 0.0757504254579544,
      "learning_rate": 3.685242764888782e-05,
      "loss": 0.0146,
      "step": 18844
    },
    {
      "epoch": 2.6412053258584445,
      "grad_norm": 0.31935518980026245,
      "learning_rate": 3.683807701506816e-05,
      "loss": 0.0379,
      "step": 18845
    },
    {
      "epoch": 2.641345480028031,
      "grad_norm": 0.25937655568122864,
      "learning_rate": 3.68237263812485e-05,
      "loss": 0.0172,
      "step": 18846
    },
    {
      "epoch": 2.6414856341976174,
      "grad_norm": 0.28963372111320496,
      "learning_rate": 3.6809375747428844e-05,
      "loss": 0.0318,
      "step": 18847
    },
    {
      "epoch": 2.641625788367204,
      "grad_norm": 0.07449734210968018,
      "learning_rate": 3.679502511360918e-05,
      "loss": 0.0048,
      "step": 18848
    },
    {
      "epoch": 2.6417659425367903,
      "grad_norm": 0.08503377437591553,
      "learning_rate": 3.678067447978952e-05,
      "loss": 0.0249,
      "step": 18849
    },
    {
      "epoch": 2.6419060967063768,
      "grad_norm": 0.08904992789030075,
      "learning_rate": 3.6766323845969864e-05,
      "loss": 0.0075,
      "step": 18850
    },
    {
      "epoch": 2.6420462508759637,
      "grad_norm": 0.24954663217067719,
      "learning_rate": 3.67519732121502e-05,
      "loss": 0.0104,
      "step": 18851
    },
    {
      "epoch": 2.64218640504555,
      "grad_norm": 0.44495853781700134,
      "learning_rate": 3.673762257833054e-05,
      "loss": 0.0364,
      "step": 18852
    },
    {
      "epoch": 2.6423265592151366,
      "grad_norm": 0.17904818058013916,
      "learning_rate": 3.6723271944510876e-05,
      "loss": 0.0063,
      "step": 18853
    },
    {
      "epoch": 2.642466713384723,
      "grad_norm": 0.10657403618097305,
      "learning_rate": 3.670892131069122e-05,
      "loss": 0.0073,
      "step": 18854
    },
    {
      "epoch": 2.64260686755431,
      "grad_norm": 0.21306875348091125,
      "learning_rate": 3.669457067687156e-05,
      "loss": 0.0294,
      "step": 18855
    },
    {
      "epoch": 2.6427470217238964,
      "grad_norm": 0.19244778156280518,
      "learning_rate": 3.66802200430519e-05,
      "loss": 0.0608,
      "step": 18856
    },
    {
      "epoch": 2.642887175893483,
      "grad_norm": 0.26421770453453064,
      "learning_rate": 3.666586940923224e-05,
      "loss": 0.0353,
      "step": 18857
    },
    {
      "epoch": 2.6430273300630693,
      "grad_norm": 0.4158807396888733,
      "learning_rate": 3.665151877541258e-05,
      "loss": 0.0814,
      "step": 18858
    },
    {
      "epoch": 2.643167484232656,
      "grad_norm": 0.15251097083091736,
      "learning_rate": 3.6637168141592915e-05,
      "loss": 0.0068,
      "step": 18859
    },
    {
      "epoch": 2.6433076384022423,
      "grad_norm": 0.2577818036079407,
      "learning_rate": 3.662281750777326e-05,
      "loss": 0.033,
      "step": 18860
    },
    {
      "epoch": 2.643447792571829,
      "grad_norm": 0.10909423232078552,
      "learning_rate": 3.660846687395359e-05,
      "loss": 0.0065,
      "step": 18861
    },
    {
      "epoch": 2.6435879467414156,
      "grad_norm": 0.05739521607756615,
      "learning_rate": 3.6594116240133935e-05,
      "loss": 0.0049,
      "step": 18862
    },
    {
      "epoch": 2.643728100911002,
      "grad_norm": 0.41375911235809326,
      "learning_rate": 3.657976560631428e-05,
      "loss": 0.0592,
      "step": 18863
    },
    {
      "epoch": 2.6438682550805885,
      "grad_norm": 0.49567365646362305,
      "learning_rate": 3.656541497249462e-05,
      "loss": 0.0492,
      "step": 18864
    },
    {
      "epoch": 2.6440084092501754,
      "grad_norm": 0.03446665406227112,
      "learning_rate": 3.6551064338674955e-05,
      "loss": 0.0037,
      "step": 18865
    },
    {
      "epoch": 2.644148563419762,
      "grad_norm": 0.16851094365119934,
      "learning_rate": 3.65367137048553e-05,
      "loss": 0.0263,
      "step": 18866
    },
    {
      "epoch": 2.6442887175893484,
      "grad_norm": 0.11972197145223618,
      "learning_rate": 3.652236307103563e-05,
      "loss": 0.0022,
      "step": 18867
    },
    {
      "epoch": 2.644428871758935,
      "grad_norm": 0.4065820276737213,
      "learning_rate": 3.6508012437215974e-05,
      "loss": 0.0624,
      "step": 18868
    },
    {
      "epoch": 2.6445690259285213,
      "grad_norm": 0.9345421195030212,
      "learning_rate": 3.649366180339631e-05,
      "loss": 0.1679,
      "step": 18869
    },
    {
      "epoch": 2.6447091800981077,
      "grad_norm": 0.2692727744579315,
      "learning_rate": 3.647931116957665e-05,
      "loss": 0.0184,
      "step": 18870
    },
    {
      "epoch": 2.644849334267694,
      "grad_norm": 0.1302371621131897,
      "learning_rate": 3.6464960535756994e-05,
      "loss": 0.0375,
      "step": 18871
    },
    {
      "epoch": 2.644989488437281,
      "grad_norm": 0.12588495016098022,
      "learning_rate": 3.6450609901937336e-05,
      "loss": 0.0141,
      "step": 18872
    },
    {
      "epoch": 2.6451296426068676,
      "grad_norm": 0.11748453229665756,
      "learning_rate": 3.643625926811767e-05,
      "loss": 0.0147,
      "step": 18873
    },
    {
      "epoch": 2.645269796776454,
      "grad_norm": 0.3150537610054016,
      "learning_rate": 3.642190863429801e-05,
      "loss": 0.0471,
      "step": 18874
    },
    {
      "epoch": 2.6454099509460405,
      "grad_norm": 0.18696613609790802,
      "learning_rate": 3.640755800047835e-05,
      "loss": 0.0531,
      "step": 18875
    },
    {
      "epoch": 2.6455501051156274,
      "grad_norm": 0.21966686844825745,
      "learning_rate": 3.639320736665869e-05,
      "loss": 0.0565,
      "step": 18876
    },
    {
      "epoch": 2.645690259285214,
      "grad_norm": 0.13831181824207306,
      "learning_rate": 3.6378856732839026e-05,
      "loss": 0.0246,
      "step": 18877
    },
    {
      "epoch": 2.6458304134548003,
      "grad_norm": 0.19923293590545654,
      "learning_rate": 3.636450609901937e-05,
      "loss": 0.0288,
      "step": 18878
    },
    {
      "epoch": 2.6459705676243868,
      "grad_norm": 0.2263757288455963,
      "learning_rate": 3.635015546519971e-05,
      "loss": 0.016,
      "step": 18879
    },
    {
      "epoch": 2.646110721793973,
      "grad_norm": 0.21622996032238007,
      "learning_rate": 3.633580483138005e-05,
      "loss": 0.0273,
      "step": 18880
    },
    {
      "epoch": 2.6462508759635597,
      "grad_norm": 0.12332558631896973,
      "learning_rate": 3.632145419756039e-05,
      "loss": 0.0155,
      "step": 18881
    },
    {
      "epoch": 2.6463910301331466,
      "grad_norm": 0.32262420654296875,
      "learning_rate": 3.630710356374073e-05,
      "loss": 0.0402,
      "step": 18882
    },
    {
      "epoch": 2.646531184302733,
      "grad_norm": 0.24925532937049866,
      "learning_rate": 3.6292752929921065e-05,
      "loss": 0.0213,
      "step": 18883
    },
    {
      "epoch": 2.6466713384723195,
      "grad_norm": 0.09134061634540558,
      "learning_rate": 3.627840229610141e-05,
      "loss": 0.0158,
      "step": 18884
    },
    {
      "epoch": 2.646811492641906,
      "grad_norm": 0.3863559663295746,
      "learning_rate": 3.626405166228174e-05,
      "loss": 0.0115,
      "step": 18885
    },
    {
      "epoch": 2.646951646811493,
      "grad_norm": 0.3700583577156067,
      "learning_rate": 3.6249701028462085e-05,
      "loss": 0.0163,
      "step": 18886
    },
    {
      "epoch": 2.6470918009810793,
      "grad_norm": 0.13881954550743103,
      "learning_rate": 3.623535039464243e-05,
      "loss": 0.0134,
      "step": 18887
    },
    {
      "epoch": 2.647231955150666,
      "grad_norm": 0.26932060718536377,
      "learning_rate": 3.622099976082277e-05,
      "loss": 0.0228,
      "step": 18888
    },
    {
      "epoch": 2.6473721093202522,
      "grad_norm": 0.4892718493938446,
      "learning_rate": 3.620664912700311e-05,
      "loss": 0.0533,
      "step": 18889
    },
    {
      "epoch": 2.6475122634898387,
      "grad_norm": 0.09212004393339157,
      "learning_rate": 3.6192298493183446e-05,
      "loss": 0.0074,
      "step": 18890
    },
    {
      "epoch": 2.647652417659425,
      "grad_norm": 0.24756762385368347,
      "learning_rate": 3.617794785936379e-05,
      "loss": 0.0409,
      "step": 18891
    },
    {
      "epoch": 2.647792571829012,
      "grad_norm": 0.3749455511569977,
      "learning_rate": 3.6163597225544124e-05,
      "loss": 0.0596,
      "step": 18892
    },
    {
      "epoch": 2.6479327259985985,
      "grad_norm": 0.08964864909648895,
      "learning_rate": 3.6149246591724466e-05,
      "loss": 0.0133,
      "step": 18893
    },
    {
      "epoch": 2.648072880168185,
      "grad_norm": 0.09138436615467072,
      "learning_rate": 3.61348959579048e-05,
      "loss": 0.0145,
      "step": 18894
    },
    {
      "epoch": 2.6482130343377714,
      "grad_norm": 0.18257910013198853,
      "learning_rate": 3.6120545324085143e-05,
      "loss": 0.0803,
      "step": 18895
    },
    {
      "epoch": 2.6483531885073583,
      "grad_norm": 0.12781164050102234,
      "learning_rate": 3.6106194690265486e-05,
      "loss": 0.0103,
      "step": 18896
    },
    {
      "epoch": 2.648493342676945,
      "grad_norm": 0.19195881485939026,
      "learning_rate": 3.609184405644583e-05,
      "loss": 0.0345,
      "step": 18897
    },
    {
      "epoch": 2.6486334968465313,
      "grad_norm": 0.14529934525489807,
      "learning_rate": 3.607749342262616e-05,
      "loss": 0.0095,
      "step": 18898
    },
    {
      "epoch": 2.6487736510161177,
      "grad_norm": 0.2866417467594147,
      "learning_rate": 3.6063142788806505e-05,
      "loss": 0.0187,
      "step": 18899
    },
    {
      "epoch": 2.648913805185704,
      "grad_norm": 0.23872049152851105,
      "learning_rate": 3.604879215498684e-05,
      "loss": 0.036,
      "step": 18900
    },
    {
      "epoch": 2.6490539593552906,
      "grad_norm": 0.057637810707092285,
      "learning_rate": 3.603444152116718e-05,
      "loss": 0.0016,
      "step": 18901
    },
    {
      "epoch": 2.649194113524877,
      "grad_norm": 0.2614864110946655,
      "learning_rate": 3.602009088734752e-05,
      "loss": 0.0321,
      "step": 18902
    },
    {
      "epoch": 2.649334267694464,
      "grad_norm": 1.463587999343872,
      "learning_rate": 3.600574025352786e-05,
      "loss": 0.0429,
      "step": 18903
    },
    {
      "epoch": 2.6494744218640505,
      "grad_norm": 0.12550930678844452,
      "learning_rate": 3.59913896197082e-05,
      "loss": 0.0084,
      "step": 18904
    },
    {
      "epoch": 2.649614576033637,
      "grad_norm": 0.18181069195270538,
      "learning_rate": 3.5977038985888544e-05,
      "loss": 0.0258,
      "step": 18905
    },
    {
      "epoch": 2.6497547302032234,
      "grad_norm": 0.28609520196914673,
      "learning_rate": 3.596268835206888e-05,
      "loss": 0.0352,
      "step": 18906
    },
    {
      "epoch": 2.6498948843728103,
      "grad_norm": 0.1043582409620285,
      "learning_rate": 3.594833771824922e-05,
      "loss": 0.0095,
      "step": 18907
    },
    {
      "epoch": 2.6500350385423967,
      "grad_norm": 0.727829098701477,
      "learning_rate": 3.593398708442956e-05,
      "loss": 0.0274,
      "step": 18908
    },
    {
      "epoch": 2.650175192711983,
      "grad_norm": 0.038930390030145645,
      "learning_rate": 3.59196364506099e-05,
      "loss": 0.0022,
      "step": 18909
    },
    {
      "epoch": 2.6503153468815697,
      "grad_norm": 1.0560684204101562,
      "learning_rate": 3.5905285816790234e-05,
      "loss": 0.0451,
      "step": 18910
    },
    {
      "epoch": 2.650455501051156,
      "grad_norm": 0.2329055815935135,
      "learning_rate": 3.5890935182970577e-05,
      "loss": 0.0215,
      "step": 18911
    },
    {
      "epoch": 2.6505956552207426,
      "grad_norm": 0.2704979479312897,
      "learning_rate": 3.587658454915092e-05,
      "loss": 0.0506,
      "step": 18912
    },
    {
      "epoch": 2.6507358093903295,
      "grad_norm": 0.016580017283558846,
      "learning_rate": 3.586223391533126e-05,
      "loss": 0.0013,
      "step": 18913
    },
    {
      "epoch": 2.650875963559916,
      "grad_norm": 0.2812025249004364,
      "learning_rate": 3.5847883281511596e-05,
      "loss": 0.031,
      "step": 18914
    },
    {
      "epoch": 2.6510161177295024,
      "grad_norm": 0.1254044771194458,
      "learning_rate": 3.583353264769194e-05,
      "loss": 0.0068,
      "step": 18915
    },
    {
      "epoch": 2.651156271899089,
      "grad_norm": 0.8288163542747498,
      "learning_rate": 3.5819182013872274e-05,
      "loss": 0.0716,
      "step": 18916
    },
    {
      "epoch": 2.6512964260686758,
      "grad_norm": 0.1282484382390976,
      "learning_rate": 3.5804831380052616e-05,
      "loss": 0.0208,
      "step": 18917
    },
    {
      "epoch": 2.6514365802382622,
      "grad_norm": 0.585984468460083,
      "learning_rate": 3.579048074623295e-05,
      "loss": 0.0235,
      "step": 18918
    },
    {
      "epoch": 2.6515767344078487,
      "grad_norm": 1.3142791986465454,
      "learning_rate": 3.577613011241329e-05,
      "loss": 0.0439,
      "step": 18919
    },
    {
      "epoch": 2.651716888577435,
      "grad_norm": 1.7962381839752197,
      "learning_rate": 3.5761779478593635e-05,
      "loss": 0.1308,
      "step": 18920
    },
    {
      "epoch": 2.6518570427470216,
      "grad_norm": 0.0776958018541336,
      "learning_rate": 3.574742884477398e-05,
      "loss": 0.0034,
      "step": 18921
    },
    {
      "epoch": 2.651997196916608,
      "grad_norm": 0.1606481671333313,
      "learning_rate": 3.573307821095431e-05,
      "loss": 0.015,
      "step": 18922
    },
    {
      "epoch": 2.652137351086195,
      "grad_norm": 0.34216776490211487,
      "learning_rate": 3.5718727577134655e-05,
      "loss": 0.0096,
      "step": 18923
    },
    {
      "epoch": 2.6522775052557814,
      "grad_norm": 0.14637905359268188,
      "learning_rate": 3.570437694331499e-05,
      "loss": 0.0392,
      "step": 18924
    },
    {
      "epoch": 2.652417659425368,
      "grad_norm": 0.1564028412103653,
      "learning_rate": 3.569002630949533e-05,
      "loss": 0.0086,
      "step": 18925
    },
    {
      "epoch": 2.6525578135949544,
      "grad_norm": 0.07537053525447845,
      "learning_rate": 3.5675675675675674e-05,
      "loss": 0.0125,
      "step": 18926
    },
    {
      "epoch": 2.6526979677645413,
      "grad_norm": 0.16018956899642944,
      "learning_rate": 3.566132504185601e-05,
      "loss": 0.0168,
      "step": 18927
    },
    {
      "epoch": 2.6528381219341277,
      "grad_norm": 0.09596879035234451,
      "learning_rate": 3.564697440803635e-05,
      "loss": 0.0085,
      "step": 18928
    },
    {
      "epoch": 2.652978276103714,
      "grad_norm": 0.049494411796331406,
      "learning_rate": 3.5632623774216694e-05,
      "loss": 0.007,
      "step": 18929
    },
    {
      "epoch": 2.6531184302733006,
      "grad_norm": 0.592934787273407,
      "learning_rate": 3.5618273140397036e-05,
      "loss": 0.0602,
      "step": 18930
    },
    {
      "epoch": 2.653258584442887,
      "grad_norm": 0.10772743076086044,
      "learning_rate": 3.560392250657737e-05,
      "loss": 0.0049,
      "step": 18931
    },
    {
      "epoch": 2.6533987386124736,
      "grad_norm": 0.2497989684343338,
      "learning_rate": 3.5589571872757714e-05,
      "loss": 0.0705,
      "step": 18932
    },
    {
      "epoch": 2.65353889278206,
      "grad_norm": 0.15350478887557983,
      "learning_rate": 3.557522123893805e-05,
      "loss": 0.013,
      "step": 18933
    },
    {
      "epoch": 2.653679046951647,
      "grad_norm": 0.1272919774055481,
      "learning_rate": 3.556087060511839e-05,
      "loss": 0.0194,
      "step": 18934
    },
    {
      "epoch": 2.6538192011212334,
      "grad_norm": 0.2402794361114502,
      "learning_rate": 3.5546519971298726e-05,
      "loss": 0.0426,
      "step": 18935
    },
    {
      "epoch": 2.65395935529082,
      "grad_norm": 0.24647396802902222,
      "learning_rate": 3.553216933747907e-05,
      "loss": 0.0515,
      "step": 18936
    },
    {
      "epoch": 2.6540995094604063,
      "grad_norm": 0.06253121793270111,
      "learning_rate": 3.551781870365941e-05,
      "loss": 0.0162,
      "step": 18937
    },
    {
      "epoch": 2.654239663629993,
      "grad_norm": 0.20771507918834686,
      "learning_rate": 3.550346806983975e-05,
      "loss": 0.0326,
      "step": 18938
    },
    {
      "epoch": 2.6543798177995797,
      "grad_norm": 0.16264566779136658,
      "learning_rate": 3.548911743602009e-05,
      "loss": 0.029,
      "step": 18939
    },
    {
      "epoch": 2.654519971969166,
      "grad_norm": 0.19247594475746155,
      "learning_rate": 3.547476680220043e-05,
      "loss": 0.0081,
      "step": 18940
    },
    {
      "epoch": 2.6546601261387526,
      "grad_norm": 0.17784127593040466,
      "learning_rate": 3.5460416168380765e-05,
      "loss": 0.0198,
      "step": 18941
    },
    {
      "epoch": 2.654800280308339,
      "grad_norm": 0.17636115849018097,
      "learning_rate": 3.544606553456111e-05,
      "loss": 0.0379,
      "step": 18942
    },
    {
      "epoch": 2.6549404344779255,
      "grad_norm": 0.09767672419548035,
      "learning_rate": 3.543171490074144e-05,
      "loss": 0.0221,
      "step": 18943
    },
    {
      "epoch": 2.6550805886475124,
      "grad_norm": 0.25687310099601746,
      "learning_rate": 3.5417364266921785e-05,
      "loss": 0.0299,
      "step": 18944
    },
    {
      "epoch": 2.655220742817099,
      "grad_norm": 0.13880766928195953,
      "learning_rate": 3.540301363310213e-05,
      "loss": 0.0227,
      "step": 18945
    },
    {
      "epoch": 2.6553608969866853,
      "grad_norm": 0.12838076055049896,
      "learning_rate": 3.538866299928247e-05,
      "loss": 0.0248,
      "step": 18946
    },
    {
      "epoch": 2.6555010511562718,
      "grad_norm": 0.542231023311615,
      "learning_rate": 3.5374312365462805e-05,
      "loss": 0.0443,
      "step": 18947
    },
    {
      "epoch": 2.6556412053258587,
      "grad_norm": 0.3959196209907532,
      "learning_rate": 3.535996173164315e-05,
      "loss": 0.0409,
      "step": 18948
    },
    {
      "epoch": 2.655781359495445,
      "grad_norm": 0.1578100323677063,
      "learning_rate": 3.534561109782348e-05,
      "loss": 0.0273,
      "step": 18949
    },
    {
      "epoch": 2.6559215136650316,
      "grad_norm": 0.20516900718212128,
      "learning_rate": 3.5331260464003824e-05,
      "loss": 0.0151,
      "step": 18950
    },
    {
      "epoch": 2.656061667834618,
      "grad_norm": 0.3041323125362396,
      "learning_rate": 3.531690983018416e-05,
      "loss": 0.046,
      "step": 18951
    },
    {
      "epoch": 2.6562018220042045,
      "grad_norm": 0.04740258306264877,
      "learning_rate": 3.53025591963645e-05,
      "loss": 0.0037,
      "step": 18952
    },
    {
      "epoch": 2.656341976173791,
      "grad_norm": 0.15366479754447937,
      "learning_rate": 3.5288208562544844e-05,
      "loss": 0.0269,
      "step": 18953
    },
    {
      "epoch": 2.656482130343378,
      "grad_norm": 0.07111267745494843,
      "learning_rate": 3.5273857928725186e-05,
      "loss": 0.0037,
      "step": 18954
    },
    {
      "epoch": 2.6566222845129643,
      "grad_norm": 0.3880583941936493,
      "learning_rate": 3.525950729490552e-05,
      "loss": 0.026,
      "step": 18955
    },
    {
      "epoch": 2.656762438682551,
      "grad_norm": 0.3831045925617218,
      "learning_rate": 3.524515666108586e-05,
      "loss": 0.068,
      "step": 18956
    },
    {
      "epoch": 2.6569025928521373,
      "grad_norm": 0.14486099779605865,
      "learning_rate": 3.52308060272662e-05,
      "loss": 0.008,
      "step": 18957
    },
    {
      "epoch": 2.657042747021724,
      "grad_norm": 0.14572961628437042,
      "learning_rate": 3.521645539344654e-05,
      "loss": 0.0173,
      "step": 18958
    },
    {
      "epoch": 2.6571829011913106,
      "grad_norm": 0.19364328682422638,
      "learning_rate": 3.5202104759626876e-05,
      "loss": 0.0154,
      "step": 18959
    },
    {
      "epoch": 2.657323055360897,
      "grad_norm": 0.5254031419754028,
      "learning_rate": 3.518775412580722e-05,
      "loss": 0.0923,
      "step": 18960
    },
    {
      "epoch": 2.6574632095304835,
      "grad_norm": 0.39478737115859985,
      "learning_rate": 3.517340349198756e-05,
      "loss": 0.016,
      "step": 18961
    },
    {
      "epoch": 2.65760336370007,
      "grad_norm": 0.3078412413597107,
      "learning_rate": 3.51590528581679e-05,
      "loss": 0.0094,
      "step": 18962
    },
    {
      "epoch": 2.6577435178696565,
      "grad_norm": 0.14455798268318176,
      "learning_rate": 3.514470222434824e-05,
      "loss": 0.0263,
      "step": 18963
    },
    {
      "epoch": 2.657883672039243,
      "grad_norm": 0.12683892250061035,
      "learning_rate": 3.513035159052858e-05,
      "loss": 0.0291,
      "step": 18964
    },
    {
      "epoch": 2.65802382620883,
      "grad_norm": 0.14042271673679352,
      "learning_rate": 3.511600095670892e-05,
      "loss": 0.007,
      "step": 18965
    },
    {
      "epoch": 2.6581639803784163,
      "grad_norm": 0.3700561821460724,
      "learning_rate": 3.510165032288926e-05,
      "loss": 0.0184,
      "step": 18966
    },
    {
      "epoch": 2.6583041345480027,
      "grad_norm": 1.007859706878662,
      "learning_rate": 3.50872996890696e-05,
      "loss": 0.0415,
      "step": 18967
    },
    {
      "epoch": 2.658444288717589,
      "grad_norm": 1.218210220336914,
      "learning_rate": 3.5072949055249935e-05,
      "loss": 0.0565,
      "step": 18968
    },
    {
      "epoch": 2.658584442887176,
      "grad_norm": 0.162925586104393,
      "learning_rate": 3.505859842143028e-05,
      "loss": 0.0196,
      "step": 18969
    },
    {
      "epoch": 2.6587245970567626,
      "grad_norm": 1.554841160774231,
      "learning_rate": 3.504424778761062e-05,
      "loss": 0.0312,
      "step": 18970
    },
    {
      "epoch": 2.658864751226349,
      "grad_norm": 0.19761858880519867,
      "learning_rate": 3.5029897153790954e-05,
      "loss": 0.0311,
      "step": 18971
    },
    {
      "epoch": 2.6590049053959355,
      "grad_norm": 0.14827671647071838,
      "learning_rate": 3.5015546519971296e-05,
      "loss": 0.0048,
      "step": 18972
    },
    {
      "epoch": 2.659145059565522,
      "grad_norm": 0.49627920985221863,
      "learning_rate": 3.500119588615164e-05,
      "loss": 0.0123,
      "step": 18973
    },
    {
      "epoch": 2.6592852137351084,
      "grad_norm": 0.10764946043491364,
      "learning_rate": 3.4986845252331974e-05,
      "loss": 0.0227,
      "step": 18974
    },
    {
      "epoch": 2.6594253679046953,
      "grad_norm": 0.28822797536849976,
      "learning_rate": 3.4972494618512316e-05,
      "loss": 0.0395,
      "step": 18975
    },
    {
      "epoch": 2.6595655220742818,
      "grad_norm": 0.9102768898010254,
      "learning_rate": 3.495814398469265e-05,
      "loss": 0.0145,
      "step": 18976
    },
    {
      "epoch": 2.659705676243868,
      "grad_norm": 0.42475634813308716,
      "learning_rate": 3.494379335087299e-05,
      "loss": 0.0128,
      "step": 18977
    },
    {
      "epoch": 2.6598458304134547,
      "grad_norm": 0.16807851195335388,
      "learning_rate": 3.4929442717053335e-05,
      "loss": 0.0225,
      "step": 18978
    },
    {
      "epoch": 2.6599859845830416,
      "grad_norm": 0.1106436625123024,
      "learning_rate": 3.491509208323367e-05,
      "loss": 0.0176,
      "step": 18979
    },
    {
      "epoch": 2.660126138752628,
      "grad_norm": 0.3467520475387573,
      "learning_rate": 3.490074144941401e-05,
      "loss": 0.0243,
      "step": 18980
    },
    {
      "epoch": 2.6602662929222145,
      "grad_norm": 0.24657103419303894,
      "learning_rate": 3.4886390815594355e-05,
      "loss": 0.0474,
      "step": 18981
    },
    {
      "epoch": 2.660406447091801,
      "grad_norm": 0.9045045971870422,
      "learning_rate": 3.487204018177469e-05,
      "loss": 0.0224,
      "step": 18982
    },
    {
      "epoch": 2.6605466012613874,
      "grad_norm": 0.3485991656780243,
      "learning_rate": 3.485768954795503e-05,
      "loss": 0.0605,
      "step": 18983
    },
    {
      "epoch": 2.660686755430974,
      "grad_norm": 0.1156155914068222,
      "learning_rate": 3.484333891413537e-05,
      "loss": 0.0116,
      "step": 18984
    },
    {
      "epoch": 2.6608269096005603,
      "grad_norm": 0.06602780520915985,
      "learning_rate": 3.482898828031571e-05,
      "loss": 0.006,
      "step": 18985
    },
    {
      "epoch": 2.6609670637701472,
      "grad_norm": 0.18632598221302032,
      "learning_rate": 3.481463764649605e-05,
      "loss": 0.0201,
      "step": 18986
    },
    {
      "epoch": 2.6611072179397337,
      "grad_norm": 0.779121994972229,
      "learning_rate": 3.4800287012676394e-05,
      "loss": 0.0585,
      "step": 18987
    },
    {
      "epoch": 2.66124737210932,
      "grad_norm": 0.39528995752334595,
      "learning_rate": 3.478593637885673e-05,
      "loss": 0.0339,
      "step": 18988
    },
    {
      "epoch": 2.661387526278907,
      "grad_norm": 0.23643599450588226,
      "learning_rate": 3.477158574503707e-05,
      "loss": 0.0309,
      "step": 18989
    },
    {
      "epoch": 2.6615276804484935,
      "grad_norm": 0.31903865933418274,
      "learning_rate": 3.475723511121741e-05,
      "loss": 0.0072,
      "step": 18990
    },
    {
      "epoch": 2.66166783461808,
      "grad_norm": 0.21672050654888153,
      "learning_rate": 3.474288447739775e-05,
      "loss": 0.0276,
      "step": 18991
    },
    {
      "epoch": 2.6618079887876664,
      "grad_norm": 0.1100042313337326,
      "learning_rate": 3.4728533843578084e-05,
      "loss": 0.0167,
      "step": 18992
    },
    {
      "epoch": 2.661948142957253,
      "grad_norm": 0.39871272444725037,
      "learning_rate": 3.4714183209758427e-05,
      "loss": 0.1247,
      "step": 18993
    },
    {
      "epoch": 2.6620882971268394,
      "grad_norm": 0.17371678352355957,
      "learning_rate": 3.469983257593877e-05,
      "loss": 0.0036,
      "step": 18994
    },
    {
      "epoch": 2.662228451296426,
      "grad_norm": 0.15228840708732605,
      "learning_rate": 3.468548194211911e-05,
      "loss": 0.0129,
      "step": 18995
    },
    {
      "epoch": 2.6623686054660127,
      "grad_norm": 0.04049418866634369,
      "learning_rate": 3.4671131308299446e-05,
      "loss": 0.0048,
      "step": 18996
    },
    {
      "epoch": 2.662508759635599,
      "grad_norm": 0.14255087077617645,
      "learning_rate": 3.465678067447979e-05,
      "loss": 0.0247,
      "step": 18997
    },
    {
      "epoch": 2.6626489138051856,
      "grad_norm": 0.41856247186660767,
      "learning_rate": 3.4642430040660124e-05,
      "loss": 0.0308,
      "step": 18998
    },
    {
      "epoch": 2.662789067974772,
      "grad_norm": 0.04924283176660538,
      "learning_rate": 3.4628079406840466e-05,
      "loss": 0.004,
      "step": 18999
    },
    {
      "epoch": 2.662929222144359,
      "grad_norm": 0.1426495611667633,
      "learning_rate": 3.461372877302081e-05,
      "loss": 0.0105,
      "step": 19000
    },
    {
      "epoch": 2.6630693763139455,
      "grad_norm": 0.2284073829650879,
      "learning_rate": 3.459937813920114e-05,
      "loss": 0.0187,
      "step": 19001
    },
    {
      "epoch": 2.663209530483532,
      "grad_norm": 0.24322715401649475,
      "learning_rate": 3.4585027505381485e-05,
      "loss": 0.0265,
      "step": 19002
    },
    {
      "epoch": 2.6633496846531184,
      "grad_norm": 0.2997440695762634,
      "learning_rate": 3.457067687156183e-05,
      "loss": 0.036,
      "step": 19003
    },
    {
      "epoch": 2.663489838822705,
      "grad_norm": 0.26688557863235474,
      "learning_rate": 3.455632623774216e-05,
      "loss": 0.0265,
      "step": 19004
    },
    {
      "epoch": 2.6636299929922913,
      "grad_norm": 0.30681562423706055,
      "learning_rate": 3.4541975603922505e-05,
      "loss": 0.0183,
      "step": 19005
    },
    {
      "epoch": 2.663770147161878,
      "grad_norm": 0.29359689354896545,
      "learning_rate": 3.452762497010285e-05,
      "loss": 0.011,
      "step": 19006
    },
    {
      "epoch": 2.6639103013314647,
      "grad_norm": 0.13415510952472687,
      "learning_rate": 3.451327433628318e-05,
      "loss": 0.0162,
      "step": 19007
    },
    {
      "epoch": 2.664050455501051,
      "grad_norm": 0.20264385640621185,
      "learning_rate": 3.4498923702463524e-05,
      "loss": 0.0212,
      "step": 19008
    },
    {
      "epoch": 2.6641906096706376,
      "grad_norm": 0.23709861934185028,
      "learning_rate": 3.448457306864386e-05,
      "loss": 0.0202,
      "step": 19009
    },
    {
      "epoch": 2.6643307638402245,
      "grad_norm": 0.18399900197982788,
      "learning_rate": 3.44702224348242e-05,
      "loss": 0.0311,
      "step": 19010
    },
    {
      "epoch": 2.664470918009811,
      "grad_norm": 1.5087790489196777,
      "learning_rate": 3.4455871801004544e-05,
      "loss": 0.0328,
      "step": 19011
    },
    {
      "epoch": 2.6646110721793974,
      "grad_norm": 0.1305650919675827,
      "learning_rate": 3.444152116718488e-05,
      "loss": 0.0064,
      "step": 19012
    },
    {
      "epoch": 2.664751226348984,
      "grad_norm": 1.2339916229248047,
      "learning_rate": 3.442717053336522e-05,
      "loss": 0.0511,
      "step": 19013
    },
    {
      "epoch": 2.6648913805185703,
      "grad_norm": 0.10763300955295563,
      "learning_rate": 3.4412819899545563e-05,
      "loss": 0.0063,
      "step": 19014
    },
    {
      "epoch": 2.665031534688157,
      "grad_norm": 0.4558926522731781,
      "learning_rate": 3.43984692657259e-05,
      "loss": 0.0089,
      "step": 19015
    },
    {
      "epoch": 2.6651716888577432,
      "grad_norm": 0.847318172454834,
      "learning_rate": 3.438411863190624e-05,
      "loss": 0.0591,
      "step": 19016
    },
    {
      "epoch": 2.66531184302733,
      "grad_norm": 0.03884310647845268,
      "learning_rate": 3.4369767998086576e-05,
      "loss": 0.0021,
      "step": 19017
    },
    {
      "epoch": 2.6654519971969166,
      "grad_norm": 0.06958263367414474,
      "learning_rate": 3.435541736426692e-05,
      "loss": 0.002,
      "step": 19018
    },
    {
      "epoch": 2.665592151366503,
      "grad_norm": 0.1667638123035431,
      "learning_rate": 3.434106673044726e-05,
      "loss": 0.0426,
      "step": 19019
    },
    {
      "epoch": 2.66573230553609,
      "grad_norm": 0.27648887038230896,
      "learning_rate": 3.4326716096627596e-05,
      "loss": 0.0108,
      "step": 19020
    },
    {
      "epoch": 2.6658724597056764,
      "grad_norm": 0.09245189279317856,
      "learning_rate": 3.431236546280794e-05,
      "loss": 0.0136,
      "step": 19021
    },
    {
      "epoch": 2.666012613875263,
      "grad_norm": 0.5118001699447632,
      "learning_rate": 3.429801482898828e-05,
      "loss": 0.0276,
      "step": 19022
    },
    {
      "epoch": 2.6661527680448494,
      "grad_norm": 0.26874810457229614,
      "learning_rate": 3.4283664195168615e-05,
      "loss": 0.0943,
      "step": 19023
    },
    {
      "epoch": 2.666292922214436,
      "grad_norm": 0.23096001148223877,
      "learning_rate": 3.426931356134896e-05,
      "loss": 0.0228,
      "step": 19024
    },
    {
      "epoch": 2.6664330763840223,
      "grad_norm": 0.06933944672346115,
      "learning_rate": 3.425496292752929e-05,
      "loss": 0.0034,
      "step": 19025
    },
    {
      "epoch": 2.6665732305536087,
      "grad_norm": 0.1690792739391327,
      "learning_rate": 3.4240612293709635e-05,
      "loss": 0.0155,
      "step": 19026
    },
    {
      "epoch": 2.6667133847231956,
      "grad_norm": 0.33580997586250305,
      "learning_rate": 3.422626165988998e-05,
      "loss": 0.0108,
      "step": 19027
    },
    {
      "epoch": 2.666853538892782,
      "grad_norm": 0.5261475443840027,
      "learning_rate": 3.421191102607031e-05,
      "loss": 0.0769,
      "step": 19028
    },
    {
      "epoch": 2.6669936930623686,
      "grad_norm": 0.21120865643024445,
      "learning_rate": 3.4197560392250654e-05,
      "loss": 0.0145,
      "step": 19029
    },
    {
      "epoch": 2.667133847231955,
      "grad_norm": 0.18853791058063507,
      "learning_rate": 3.4183209758430997e-05,
      "loss": 0.0171,
      "step": 19030
    },
    {
      "epoch": 2.667274001401542,
      "grad_norm": 0.12237367779016495,
      "learning_rate": 3.416885912461133e-05,
      "loss": 0.017,
      "step": 19031
    },
    {
      "epoch": 2.6674141555711284,
      "grad_norm": 0.12582413852214813,
      "learning_rate": 3.4154508490791674e-05,
      "loss": 0.0478,
      "step": 19032
    },
    {
      "epoch": 2.667554309740715,
      "grad_norm": 0.18743738532066345,
      "learning_rate": 3.414015785697201e-05,
      "loss": 0.0122,
      "step": 19033
    },
    {
      "epoch": 2.6676944639103013,
      "grad_norm": 0.07217022776603699,
      "learning_rate": 3.412580722315235e-05,
      "loss": 0.007,
      "step": 19034
    },
    {
      "epoch": 2.6678346180798878,
      "grad_norm": 0.29194602370262146,
      "learning_rate": 3.4111456589332694e-05,
      "loss": 0.0165,
      "step": 19035
    },
    {
      "epoch": 2.667974772249474,
      "grad_norm": 0.11871635168790817,
      "learning_rate": 3.409710595551303e-05,
      "loss": 0.0184,
      "step": 19036
    },
    {
      "epoch": 2.668114926419061,
      "grad_norm": 0.3665285110473633,
      "learning_rate": 3.408275532169337e-05,
      "loss": 0.0171,
      "step": 19037
    },
    {
      "epoch": 2.6682550805886476,
      "grad_norm": 0.27697038650512695,
      "learning_rate": 3.406840468787371e-05,
      "loss": 0.0245,
      "step": 19038
    },
    {
      "epoch": 2.668395234758234,
      "grad_norm": 0.1697680801153183,
      "learning_rate": 3.4054054054054055e-05,
      "loss": 0.0144,
      "step": 19039
    },
    {
      "epoch": 2.6685353889278205,
      "grad_norm": 0.21182020008563995,
      "learning_rate": 3.403970342023439e-05,
      "loss": 0.0184,
      "step": 19040
    },
    {
      "epoch": 2.6686755430974074,
      "grad_norm": 0.026870721951127052,
      "learning_rate": 3.402535278641473e-05,
      "loss": 0.0018,
      "step": 19041
    },
    {
      "epoch": 2.668815697266994,
      "grad_norm": 0.3477993309497833,
      "learning_rate": 3.401100215259507e-05,
      "loss": 0.051,
      "step": 19042
    },
    {
      "epoch": 2.6689558514365803,
      "grad_norm": 0.23562243580818176,
      "learning_rate": 3.399665151877541e-05,
      "loss": 0.044,
      "step": 19043
    },
    {
      "epoch": 2.6690960056061668,
      "grad_norm": 0.16737982630729675,
      "learning_rate": 3.3982300884955746e-05,
      "loss": 0.0108,
      "step": 19044
    },
    {
      "epoch": 2.6692361597757532,
      "grad_norm": 0.10520761460065842,
      "learning_rate": 3.396795025113609e-05,
      "loss": 0.0085,
      "step": 19045
    },
    {
      "epoch": 2.6693763139453397,
      "grad_norm": 0.23197759687900543,
      "learning_rate": 3.395359961731643e-05,
      "loss": 0.037,
      "step": 19046
    },
    {
      "epoch": 2.669516468114926,
      "grad_norm": 0.46111539006233215,
      "learning_rate": 3.393924898349677e-05,
      "loss": 0.0416,
      "step": 19047
    },
    {
      "epoch": 2.669656622284513,
      "grad_norm": 0.19760540127754211,
      "learning_rate": 3.392489834967711e-05,
      "loss": 0.0232,
      "step": 19048
    },
    {
      "epoch": 2.6697967764540995,
      "grad_norm": 0.07133524864912033,
      "learning_rate": 3.391054771585745e-05,
      "loss": 0.0304,
      "step": 19049
    },
    {
      "epoch": 2.669936930623686,
      "grad_norm": 0.3377271890640259,
      "learning_rate": 3.3896197082037785e-05,
      "loss": 0.0378,
      "step": 19050
    },
    {
      "epoch": 2.6700770847932724,
      "grad_norm": 0.3027612864971161,
      "learning_rate": 3.388184644821813e-05,
      "loss": 0.072,
      "step": 19051
    },
    {
      "epoch": 2.6702172389628593,
      "grad_norm": 0.30920156836509705,
      "learning_rate": 3.386749581439846e-05,
      "loss": 0.012,
      "step": 19052
    },
    {
      "epoch": 2.670357393132446,
      "grad_norm": 0.2439681589603424,
      "learning_rate": 3.3853145180578804e-05,
      "loss": 0.0158,
      "step": 19053
    },
    {
      "epoch": 2.6704975473020323,
      "grad_norm": 0.04542482644319534,
      "learning_rate": 3.3838794546759146e-05,
      "loss": 0.0021,
      "step": 19054
    },
    {
      "epoch": 2.6706377014716187,
      "grad_norm": 0.2715564966201782,
      "learning_rate": 3.382444391293949e-05,
      "loss": 0.0152,
      "step": 19055
    },
    {
      "epoch": 2.670777855641205,
      "grad_norm": 0.08921615779399872,
      "learning_rate": 3.3810093279119824e-05,
      "loss": 0.0061,
      "step": 19056
    },
    {
      "epoch": 2.6709180098107916,
      "grad_norm": 0.18041425943374634,
      "learning_rate": 3.3795742645300166e-05,
      "loss": 0.0159,
      "step": 19057
    },
    {
      "epoch": 2.6710581639803785,
      "grad_norm": 0.44078949093818665,
      "learning_rate": 3.37813920114805e-05,
      "loss": 0.0239,
      "step": 19058
    },
    {
      "epoch": 2.671198318149965,
      "grad_norm": 0.2560601532459259,
      "learning_rate": 3.376704137766084e-05,
      "loss": 0.0183,
      "step": 19059
    },
    {
      "epoch": 2.6713384723195515,
      "grad_norm": 0.7539780735969543,
      "learning_rate": 3.375269074384118e-05,
      "loss": 0.0201,
      "step": 19060
    },
    {
      "epoch": 2.671478626489138,
      "grad_norm": 0.13882705569267273,
      "learning_rate": 3.373834011002152e-05,
      "loss": 0.0134,
      "step": 19061
    },
    {
      "epoch": 2.671618780658725,
      "grad_norm": 0.917358934879303,
      "learning_rate": 3.372398947620186e-05,
      "loss": 0.0388,
      "step": 19062
    },
    {
      "epoch": 2.6717589348283113,
      "grad_norm": 0.5067930221557617,
      "learning_rate": 3.3709638842382205e-05,
      "loss": 0.0262,
      "step": 19063
    },
    {
      "epoch": 2.6718990889978977,
      "grad_norm": 0.3193817138671875,
      "learning_rate": 3.369528820856254e-05,
      "loss": 0.0144,
      "step": 19064
    },
    {
      "epoch": 2.672039243167484,
      "grad_norm": 0.1851431280374527,
      "learning_rate": 3.368093757474288e-05,
      "loss": 0.0057,
      "step": 19065
    },
    {
      "epoch": 2.6721793973370707,
      "grad_norm": 0.06776344776153564,
      "learning_rate": 3.366658694092322e-05,
      "loss": 0.0072,
      "step": 19066
    },
    {
      "epoch": 2.672319551506657,
      "grad_norm": 0.4747142195701599,
      "learning_rate": 3.365223630710356e-05,
      "loss": 0.0135,
      "step": 19067
    },
    {
      "epoch": 2.672459705676244,
      "grad_norm": 0.839833676815033,
      "learning_rate": 3.3637885673283895e-05,
      "loss": 0.0418,
      "step": 19068
    },
    {
      "epoch": 2.6725998598458305,
      "grad_norm": 2.1092984676361084,
      "learning_rate": 3.362353503946424e-05,
      "loss": 0.1103,
      "step": 19069
    },
    {
      "epoch": 2.672740014015417,
      "grad_norm": 0.06647612154483795,
      "learning_rate": 3.360918440564458e-05,
      "loss": 0.0029,
      "step": 19070
    },
    {
      "epoch": 2.6728801681850034,
      "grad_norm": 0.4686107039451599,
      "learning_rate": 3.359483377182492e-05,
      "loss": 0.032,
      "step": 19071
    },
    {
      "epoch": 2.6730203223545903,
      "grad_norm": 0.0698034018278122,
      "learning_rate": 3.358048313800526e-05,
      "loss": 0.0037,
      "step": 19072
    },
    {
      "epoch": 2.6731604765241768,
      "grad_norm": 0.15699800848960876,
      "learning_rate": 3.35661325041856e-05,
      "loss": 0.0282,
      "step": 19073
    },
    {
      "epoch": 2.6733006306937632,
      "grad_norm": 0.37768080830574036,
      "learning_rate": 3.355178187036594e-05,
      "loss": 0.0261,
      "step": 19074
    },
    {
      "epoch": 2.6734407848633497,
      "grad_norm": 0.581476628780365,
      "learning_rate": 3.3537431236546276e-05,
      "loss": 0.0865,
      "step": 19075
    },
    {
      "epoch": 2.673580939032936,
      "grad_norm": 0.13296441733837128,
      "learning_rate": 3.352308060272662e-05,
      "loss": 0.0227,
      "step": 19076
    },
    {
      "epoch": 2.6737210932025226,
      "grad_norm": 0.2900911569595337,
      "learning_rate": 3.3508729968906954e-05,
      "loss": 0.0482,
      "step": 19077
    },
    {
      "epoch": 2.673861247372109,
      "grad_norm": 0.478647381067276,
      "learning_rate": 3.3494379335087296e-05,
      "loss": 0.0749,
      "step": 19078
    },
    {
      "epoch": 2.674001401541696,
      "grad_norm": 0.34488147497177124,
      "learning_rate": 3.348002870126764e-05,
      "loss": 0.0203,
      "step": 19079
    },
    {
      "epoch": 2.6741415557112824,
      "grad_norm": 0.07160986959934235,
      "learning_rate": 3.346567806744798e-05,
      "loss": 0.0163,
      "step": 19080
    },
    {
      "epoch": 2.674281709880869,
      "grad_norm": 0.15958480536937714,
      "learning_rate": 3.3451327433628316e-05,
      "loss": 0.0323,
      "step": 19081
    },
    {
      "epoch": 2.6744218640504553,
      "grad_norm": 0.06173427402973175,
      "learning_rate": 3.343697679980866e-05,
      "loss": 0.0047,
      "step": 19082
    },
    {
      "epoch": 2.6745620182200422,
      "grad_norm": 0.1466919630765915,
      "learning_rate": 3.342262616598899e-05,
      "loss": 0.0191,
      "step": 19083
    },
    {
      "epoch": 2.6747021723896287,
      "grad_norm": 0.3475992977619171,
      "learning_rate": 3.3408275532169335e-05,
      "loss": 0.0249,
      "step": 19084
    },
    {
      "epoch": 2.674842326559215,
      "grad_norm": 0.24090787768363953,
      "learning_rate": 3.339392489834967e-05,
      "loss": 0.0382,
      "step": 19085
    },
    {
      "epoch": 2.6749824807288016,
      "grad_norm": 0.11143588274717331,
      "learning_rate": 3.337957426453001e-05,
      "loss": 0.029,
      "step": 19086
    },
    {
      "epoch": 2.675122634898388,
      "grad_norm": 0.18509972095489502,
      "learning_rate": 3.3365223630710355e-05,
      "loss": 0.0354,
      "step": 19087
    },
    {
      "epoch": 2.6752627890679745,
      "grad_norm": 0.27674421668052673,
      "learning_rate": 3.33508729968907e-05,
      "loss": 0.0399,
      "step": 19088
    },
    {
      "epoch": 2.6754029432375614,
      "grad_norm": 0.09937164187431335,
      "learning_rate": 3.333652236307103e-05,
      "loss": 0.008,
      "step": 19089
    },
    {
      "epoch": 2.675543097407148,
      "grad_norm": 0.0679984912276268,
      "learning_rate": 3.3322171729251374e-05,
      "loss": 0.0084,
      "step": 19090
    },
    {
      "epoch": 2.6756832515767344,
      "grad_norm": 0.5039976239204407,
      "learning_rate": 3.330782109543171e-05,
      "loss": 0.0264,
      "step": 19091
    },
    {
      "epoch": 2.675823405746321,
      "grad_norm": 0.25375527143478394,
      "learning_rate": 3.329347046161205e-05,
      "loss": 0.0044,
      "step": 19092
    },
    {
      "epoch": 2.6759635599159077,
      "grad_norm": 0.6165251135826111,
      "learning_rate": 3.327911982779239e-05,
      "loss": 0.0322,
      "step": 19093
    },
    {
      "epoch": 2.676103714085494,
      "grad_norm": 0.2337779402732849,
      "learning_rate": 3.326476919397273e-05,
      "loss": 0.0282,
      "step": 19094
    },
    {
      "epoch": 2.6762438682550806,
      "grad_norm": 0.2859530448913574,
      "learning_rate": 3.325041856015307e-05,
      "loss": 0.0093,
      "step": 19095
    },
    {
      "epoch": 2.676384022424667,
      "grad_norm": 0.027607286348938942,
      "learning_rate": 3.3236067926333413e-05,
      "loss": 0.0022,
      "step": 19096
    },
    {
      "epoch": 2.6765241765942536,
      "grad_norm": 0.18494337797164917,
      "learning_rate": 3.322171729251375e-05,
      "loss": 0.0131,
      "step": 19097
    },
    {
      "epoch": 2.67666433076384,
      "grad_norm": 0.24858082830905914,
      "learning_rate": 3.320736665869409e-05,
      "loss": 0.0138,
      "step": 19098
    },
    {
      "epoch": 2.676804484933427,
      "grad_norm": 0.18532218039035797,
      "learning_rate": 3.3193016024874426e-05,
      "loss": 0.0355,
      "step": 19099
    },
    {
      "epoch": 2.6769446391030134,
      "grad_norm": 0.16465339064598083,
      "learning_rate": 3.317866539105477e-05,
      "loss": 0.0205,
      "step": 19100
    },
    {
      "epoch": 2.6770847932726,
      "grad_norm": 0.24401308596134186,
      "learning_rate": 3.3164314757235104e-05,
      "loss": 0.0489,
      "step": 19101
    },
    {
      "epoch": 2.6772249474421863,
      "grad_norm": 0.1126483604311943,
      "learning_rate": 3.3149964123415446e-05,
      "loss": 0.0137,
      "step": 19102
    },
    {
      "epoch": 2.677365101611773,
      "grad_norm": 0.3556588292121887,
      "learning_rate": 3.313561348959579e-05,
      "loss": 0.0661,
      "step": 19103
    },
    {
      "epoch": 2.6775052557813597,
      "grad_norm": 0.09374068677425385,
      "learning_rate": 3.312126285577613e-05,
      "loss": 0.0091,
      "step": 19104
    },
    {
      "epoch": 2.677645409950946,
      "grad_norm": 0.6063767671585083,
      "learning_rate": 3.3106912221956465e-05,
      "loss": 0.0505,
      "step": 19105
    },
    {
      "epoch": 2.6777855641205326,
      "grad_norm": 0.3379063606262207,
      "learning_rate": 3.309256158813681e-05,
      "loss": 0.0353,
      "step": 19106
    },
    {
      "epoch": 2.677925718290119,
      "grad_norm": 0.4798426628112793,
      "learning_rate": 3.307821095431714e-05,
      "loss": 0.0106,
      "step": 19107
    },
    {
      "epoch": 2.6780658724597055,
      "grad_norm": 0.41608569025993347,
      "learning_rate": 3.3063860320497485e-05,
      "loss": 0.1626,
      "step": 19108
    },
    {
      "epoch": 2.678206026629292,
      "grad_norm": 0.8105961680412292,
      "learning_rate": 3.304950968667782e-05,
      "loss": 0.0207,
      "step": 19109
    },
    {
      "epoch": 2.678346180798879,
      "grad_norm": 0.08732123672962189,
      "learning_rate": 3.303515905285816e-05,
      "loss": 0.0046,
      "step": 19110
    },
    {
      "epoch": 2.6784863349684653,
      "grad_norm": 0.3657243251800537,
      "learning_rate": 3.3020808419038504e-05,
      "loss": 0.0884,
      "step": 19111
    },
    {
      "epoch": 2.678626489138052,
      "grad_norm": 0.003782531712204218,
      "learning_rate": 3.3006457785218847e-05,
      "loss": 0.0004,
      "step": 19112
    },
    {
      "epoch": 2.6787666433076383,
      "grad_norm": 0.04383417218923569,
      "learning_rate": 3.299210715139919e-05,
      "loss": 0.0044,
      "step": 19113
    },
    {
      "epoch": 2.678906797477225,
      "grad_norm": 0.14292454719543457,
      "learning_rate": 3.2977756517579524e-05,
      "loss": 0.007,
      "step": 19114
    },
    {
      "epoch": 2.6790469516468116,
      "grad_norm": 0.27917948365211487,
      "learning_rate": 3.2963405883759866e-05,
      "loss": 0.0135,
      "step": 19115
    },
    {
      "epoch": 2.679187105816398,
      "grad_norm": 0.16126903891563416,
      "learning_rate": 3.29490552499402e-05,
      "loss": 0.0092,
      "step": 19116
    },
    {
      "epoch": 2.6793272599859845,
      "grad_norm": 0.16355693340301514,
      "learning_rate": 3.2934704616120544e-05,
      "loss": 0.0198,
      "step": 19117
    },
    {
      "epoch": 2.679467414155571,
      "grad_norm": 0.01838826760649681,
      "learning_rate": 3.292035398230088e-05,
      "loss": 0.0009,
      "step": 19118
    },
    {
      "epoch": 2.6796075683251575,
      "grad_norm": 0.05736461654305458,
      "learning_rate": 3.290600334848122e-05,
      "loss": 0.0038,
      "step": 19119
    },
    {
      "epoch": 2.6797477224947444,
      "grad_norm": 0.8242635130882263,
      "learning_rate": 3.289165271466156e-05,
      "loss": 0.0687,
      "step": 19120
    },
    {
      "epoch": 2.679887876664331,
      "grad_norm": 0.3225947618484497,
      "learning_rate": 3.2877302080841905e-05,
      "loss": 0.0911,
      "step": 19121
    },
    {
      "epoch": 2.6800280308339173,
      "grad_norm": 0.10817845165729523,
      "learning_rate": 3.286295144702224e-05,
      "loss": 0.0085,
      "step": 19122
    },
    {
      "epoch": 2.6801681850035037,
      "grad_norm": 0.11754167079925537,
      "learning_rate": 3.284860081320258e-05,
      "loss": 0.0123,
      "step": 19123
    },
    {
      "epoch": 2.6803083391730906,
      "grad_norm": 0.09311207383871078,
      "learning_rate": 3.283425017938292e-05,
      "loss": 0.0107,
      "step": 19124
    },
    {
      "epoch": 2.680448493342677,
      "grad_norm": 0.25137677788734436,
      "learning_rate": 3.281989954556326e-05,
      "loss": 0.0247,
      "step": 19125
    },
    {
      "epoch": 2.6805886475122636,
      "grad_norm": 0.27546167373657227,
      "learning_rate": 3.2805548911743595e-05,
      "loss": 0.0708,
      "step": 19126
    },
    {
      "epoch": 2.68072880168185,
      "grad_norm": 0.17241717875003815,
      "learning_rate": 3.279119827792394e-05,
      "loss": 0.0371,
      "step": 19127
    },
    {
      "epoch": 2.6808689558514365,
      "grad_norm": 0.3461940586566925,
      "learning_rate": 3.277684764410428e-05,
      "loss": 0.0465,
      "step": 19128
    },
    {
      "epoch": 2.681009110021023,
      "grad_norm": 0.23759038746356964,
      "learning_rate": 3.276249701028462e-05,
      "loss": 0.0414,
      "step": 19129
    },
    {
      "epoch": 2.6811492641906094,
      "grad_norm": 0.3206912875175476,
      "learning_rate": 3.274814637646496e-05,
      "loss": 0.0292,
      "step": 19130
    },
    {
      "epoch": 2.6812894183601963,
      "grad_norm": 0.636391282081604,
      "learning_rate": 3.27337957426453e-05,
      "loss": 0.0376,
      "step": 19131
    },
    {
      "epoch": 2.6814295725297828,
      "grad_norm": 0.29441094398498535,
      "learning_rate": 3.2719445108825635e-05,
      "loss": 0.0202,
      "step": 19132
    },
    {
      "epoch": 2.681569726699369,
      "grad_norm": 0.20151387155056,
      "learning_rate": 3.270509447500598e-05,
      "loss": 0.025,
      "step": 19133
    },
    {
      "epoch": 2.681709880868956,
      "grad_norm": 0.1554955244064331,
      "learning_rate": 3.269074384118631e-05,
      "loss": 0.0208,
      "step": 19134
    },
    {
      "epoch": 2.6818500350385426,
      "grad_norm": 0.04337377846240997,
      "learning_rate": 3.2676393207366654e-05,
      "loss": 0.0072,
      "step": 19135
    },
    {
      "epoch": 2.681990189208129,
      "grad_norm": 0.33683398365974426,
      "learning_rate": 3.2662042573546996e-05,
      "loss": 0.0131,
      "step": 19136
    },
    {
      "epoch": 2.6821303433777155,
      "grad_norm": 0.3223258852958679,
      "learning_rate": 3.264769193972734e-05,
      "loss": 0.0299,
      "step": 19137
    },
    {
      "epoch": 2.682270497547302,
      "grad_norm": 0.36297956109046936,
      "learning_rate": 3.2633341305907674e-05,
      "loss": 0.033,
      "step": 19138
    },
    {
      "epoch": 2.6824106517168884,
      "grad_norm": 0.10239852964878082,
      "learning_rate": 3.2618990672088016e-05,
      "loss": 0.0022,
      "step": 19139
    },
    {
      "epoch": 2.682550805886475,
      "grad_norm": 0.6311405897140503,
      "learning_rate": 3.260464003826835e-05,
      "loss": 0.0903,
      "step": 19140
    },
    {
      "epoch": 2.682690960056062,
      "grad_norm": 0.017206447198987007,
      "learning_rate": 3.259028940444869e-05,
      "loss": 0.0012,
      "step": 19141
    },
    {
      "epoch": 2.6828311142256482,
      "grad_norm": 0.43335095047950745,
      "learning_rate": 3.257593877062903e-05,
      "loss": 0.0715,
      "step": 19142
    },
    {
      "epoch": 2.6829712683952347,
      "grad_norm": 0.21185848116874695,
      "learning_rate": 3.256158813680937e-05,
      "loss": 0.04,
      "step": 19143
    },
    {
      "epoch": 2.683111422564821,
      "grad_norm": 0.19532856345176697,
      "learning_rate": 3.254723750298971e-05,
      "loss": 0.0214,
      "step": 19144
    },
    {
      "epoch": 2.683251576734408,
      "grad_norm": 0.22247779369354248,
      "learning_rate": 3.2532886869170055e-05,
      "loss": 0.0195,
      "step": 19145
    },
    {
      "epoch": 2.6833917309039945,
      "grad_norm": 0.061474330723285675,
      "learning_rate": 3.251853623535039e-05,
      "loss": 0.0085,
      "step": 19146
    },
    {
      "epoch": 2.683531885073581,
      "grad_norm": 0.0926731750369072,
      "learning_rate": 3.250418560153073e-05,
      "loss": 0.0067,
      "step": 19147
    },
    {
      "epoch": 2.6836720392431674,
      "grad_norm": 0.3298453986644745,
      "learning_rate": 3.2489834967711075e-05,
      "loss": 0.0444,
      "step": 19148
    },
    {
      "epoch": 2.683812193412754,
      "grad_norm": 0.12941385805606842,
      "learning_rate": 3.247548433389141e-05,
      "loss": 0.0104,
      "step": 19149
    },
    {
      "epoch": 2.6839523475823404,
      "grad_norm": 0.2862327992916107,
      "learning_rate": 3.246113370007175e-05,
      "loss": 0.0326,
      "step": 19150
    },
    {
      "epoch": 2.6840925017519273,
      "grad_norm": 0.29845452308654785,
      "learning_rate": 3.244678306625209e-05,
      "loss": 0.0243,
      "step": 19151
    },
    {
      "epoch": 2.6842326559215137,
      "grad_norm": 0.09152431786060333,
      "learning_rate": 3.243243243243243e-05,
      "loss": 0.0112,
      "step": 19152
    },
    {
      "epoch": 2.6843728100911,
      "grad_norm": 0.25802135467529297,
      "learning_rate": 3.241808179861277e-05,
      "loss": 0.0094,
      "step": 19153
    },
    {
      "epoch": 2.6845129642606866,
      "grad_norm": 0.2895890772342682,
      "learning_rate": 3.2403731164793114e-05,
      "loss": 0.035,
      "step": 19154
    },
    {
      "epoch": 2.6846531184302735,
      "grad_norm": 0.2526528537273407,
      "learning_rate": 3.238938053097345e-05,
      "loss": 0.0169,
      "step": 19155
    },
    {
      "epoch": 2.68479327259986,
      "grad_norm": 0.4358956515789032,
      "learning_rate": 3.237502989715379e-05,
      "loss": 0.0283,
      "step": 19156
    },
    {
      "epoch": 2.6849334267694465,
      "grad_norm": 0.4859786927700043,
      "learning_rate": 3.2360679263334126e-05,
      "loss": 0.0231,
      "step": 19157
    },
    {
      "epoch": 2.685073580939033,
      "grad_norm": 0.19008463621139526,
      "learning_rate": 3.234632862951447e-05,
      "loss": 0.0139,
      "step": 19158
    },
    {
      "epoch": 2.6852137351086194,
      "grad_norm": 0.11303292214870453,
      "learning_rate": 3.2331977995694804e-05,
      "loss": 0.021,
      "step": 19159
    },
    {
      "epoch": 2.685353889278206,
      "grad_norm": 0.09639807790517807,
      "learning_rate": 3.2317627361875146e-05,
      "loss": 0.0063,
      "step": 19160
    },
    {
      "epoch": 2.6854940434477923,
      "grad_norm": 0.04655229672789574,
      "learning_rate": 3.230327672805549e-05,
      "loss": 0.0024,
      "step": 19161
    },
    {
      "epoch": 2.685634197617379,
      "grad_norm": 0.1499675065279007,
      "learning_rate": 3.228892609423583e-05,
      "loss": 0.0236,
      "step": 19162
    },
    {
      "epoch": 2.6857743517869657,
      "grad_norm": 0.9135861396789551,
      "learning_rate": 3.2274575460416166e-05,
      "loss": 0.0334,
      "step": 19163
    },
    {
      "epoch": 2.685914505956552,
      "grad_norm": 0.119294673204422,
      "learning_rate": 3.226022482659651e-05,
      "loss": 0.0022,
      "step": 19164
    },
    {
      "epoch": 2.686054660126139,
      "grad_norm": 0.17394724488258362,
      "learning_rate": 3.224587419277684e-05,
      "loss": 0.0071,
      "step": 19165
    },
    {
      "epoch": 2.6861948142957255,
      "grad_norm": 0.19721810519695282,
      "learning_rate": 3.2231523558957185e-05,
      "loss": 0.0104,
      "step": 19166
    },
    {
      "epoch": 2.686334968465312,
      "grad_norm": 0.35553625226020813,
      "learning_rate": 3.221717292513752e-05,
      "loss": 0.0156,
      "step": 19167
    },
    {
      "epoch": 2.6864751226348984,
      "grad_norm": 0.5617117881774902,
      "learning_rate": 3.220282229131786e-05,
      "loss": 0.0291,
      "step": 19168
    },
    {
      "epoch": 2.686615276804485,
      "grad_norm": 0.2425331026315689,
      "learning_rate": 3.2188471657498205e-05,
      "loss": 0.0059,
      "step": 19169
    },
    {
      "epoch": 2.6867554309740713,
      "grad_norm": 1.1779361963272095,
      "learning_rate": 3.217412102367855e-05,
      "loss": 0.0563,
      "step": 19170
    },
    {
      "epoch": 2.686895585143658,
      "grad_norm": 0.23621971905231476,
      "learning_rate": 3.215977038985888e-05,
      "loss": 0.028,
      "step": 19171
    },
    {
      "epoch": 2.6870357393132447,
      "grad_norm": 0.22600258886814117,
      "learning_rate": 3.2145419756039224e-05,
      "loss": 0.0118,
      "step": 19172
    },
    {
      "epoch": 2.687175893482831,
      "grad_norm": 0.047861602157354355,
      "learning_rate": 3.213106912221956e-05,
      "loss": 0.0043,
      "step": 19173
    },
    {
      "epoch": 2.6873160476524176,
      "grad_norm": 0.6564404368400574,
      "learning_rate": 3.21167184883999e-05,
      "loss": 0.0305,
      "step": 19174
    },
    {
      "epoch": 2.687456201822004,
      "grad_norm": 1.4879180192947388,
      "learning_rate": 3.210236785458024e-05,
      "loss": 0.0705,
      "step": 19175
    },
    {
      "epoch": 2.687596355991591,
      "grad_norm": 0.47716811299324036,
      "learning_rate": 3.208801722076058e-05,
      "loss": 0.0518,
      "step": 19176
    },
    {
      "epoch": 2.6877365101611774,
      "grad_norm": 0.03959793597459793,
      "learning_rate": 3.207366658694092e-05,
      "loss": 0.0028,
      "step": 19177
    },
    {
      "epoch": 2.687876664330764,
      "grad_norm": 0.15340334177017212,
      "learning_rate": 3.205931595312126e-05,
      "loss": 0.0222,
      "step": 19178
    },
    {
      "epoch": 2.6880168185003503,
      "grad_norm": 0.3232771158218384,
      "learning_rate": 3.20449653193016e-05,
      "loss": 0.0252,
      "step": 19179
    },
    {
      "epoch": 2.688156972669937,
      "grad_norm": 0.2110404521226883,
      "learning_rate": 3.203061468548194e-05,
      "loss": 0.041,
      "step": 19180
    },
    {
      "epoch": 2.6882971268395233,
      "grad_norm": 0.12288319319486618,
      "learning_rate": 3.2016264051662276e-05,
      "loss": 0.0067,
      "step": 19181
    },
    {
      "epoch": 2.68843728100911,
      "grad_norm": 0.5726479291915894,
      "learning_rate": 3.200191341784262e-05,
      "loss": 0.037,
      "step": 19182
    },
    {
      "epoch": 2.6885774351786966,
      "grad_norm": 0.07861670851707458,
      "learning_rate": 3.1987562784022954e-05,
      "loss": 0.0073,
      "step": 19183
    },
    {
      "epoch": 2.688717589348283,
      "grad_norm": 0.15803228318691254,
      "learning_rate": 3.1973212150203296e-05,
      "loss": 0.0109,
      "step": 19184
    },
    {
      "epoch": 2.6888577435178695,
      "grad_norm": 0.1511002779006958,
      "learning_rate": 3.195886151638364e-05,
      "loss": 0.014,
      "step": 19185
    },
    {
      "epoch": 2.6889978976874565,
      "grad_norm": 0.17733974754810333,
      "learning_rate": 3.194451088256398e-05,
      "loss": 0.0551,
      "step": 19186
    },
    {
      "epoch": 2.689138051857043,
      "grad_norm": 0.19843867421150208,
      "learning_rate": 3.193016024874432e-05,
      "loss": 0.0173,
      "step": 19187
    },
    {
      "epoch": 2.6892782060266294,
      "grad_norm": 0.09071072936058044,
      "learning_rate": 3.191580961492466e-05,
      "loss": 0.0026,
      "step": 19188
    },
    {
      "epoch": 2.689418360196216,
      "grad_norm": 0.3565714955329895,
      "learning_rate": 3.1901458981105e-05,
      "loss": 0.0221,
      "step": 19189
    },
    {
      "epoch": 2.6895585143658023,
      "grad_norm": 0.12883344292640686,
      "learning_rate": 3.1887108347285335e-05,
      "loss": 0.0174,
      "step": 19190
    },
    {
      "epoch": 2.6896986685353887,
      "grad_norm": 1.3392831087112427,
      "learning_rate": 3.187275771346568e-05,
      "loss": 0.0258,
      "step": 19191
    },
    {
      "epoch": 2.689838822704975,
      "grad_norm": 0.223307803273201,
      "learning_rate": 3.185840707964601e-05,
      "loss": 0.0091,
      "step": 19192
    },
    {
      "epoch": 2.689978976874562,
      "grad_norm": 0.20991235971450806,
      "learning_rate": 3.1844056445826354e-05,
      "loss": 0.0199,
      "step": 19193
    },
    {
      "epoch": 2.6901191310441486,
      "grad_norm": 0.876562774181366,
      "learning_rate": 3.1829705812006696e-05,
      "loss": 0.0532,
      "step": 19194
    },
    {
      "epoch": 2.690259285213735,
      "grad_norm": 0.1591513454914093,
      "learning_rate": 3.181535517818704e-05,
      "loss": 0.0146,
      "step": 19195
    },
    {
      "epoch": 2.6903994393833215,
      "grad_norm": 0.17315897345542908,
      "learning_rate": 3.1801004544367374e-05,
      "loss": 0.0407,
      "step": 19196
    },
    {
      "epoch": 2.6905395935529084,
      "grad_norm": 0.06539208441972733,
      "learning_rate": 3.1786653910547716e-05,
      "loss": 0.0057,
      "step": 19197
    },
    {
      "epoch": 2.690679747722495,
      "grad_norm": 0.10878187417984009,
      "learning_rate": 3.177230327672805e-05,
      "loss": 0.0052,
      "step": 19198
    },
    {
      "epoch": 2.6908199018920813,
      "grad_norm": 0.06214870885014534,
      "learning_rate": 3.1757952642908394e-05,
      "loss": 0.0044,
      "step": 19199
    },
    {
      "epoch": 2.6909600560616678,
      "grad_norm": 0.04756125062704086,
      "learning_rate": 3.174360200908873e-05,
      "loss": 0.0043,
      "step": 19200
    },
    {
      "epoch": 2.6911002102312542,
      "grad_norm": 0.05723460763692856,
      "learning_rate": 3.172925137526907e-05,
      "loss": 0.0052,
      "step": 19201
    },
    {
      "epoch": 2.6912403644008407,
      "grad_norm": 0.11539015173912048,
      "learning_rate": 3.171490074144941e-05,
      "loss": 0.0157,
      "step": 19202
    },
    {
      "epoch": 2.6913805185704276,
      "grad_norm": 0.20039841532707214,
      "learning_rate": 3.1700550107629755e-05,
      "loss": 0.0251,
      "step": 19203
    },
    {
      "epoch": 2.691520672740014,
      "grad_norm": 0.11434441804885864,
      "learning_rate": 3.168619947381009e-05,
      "loss": 0.0183,
      "step": 19204
    },
    {
      "epoch": 2.6916608269096005,
      "grad_norm": 0.058986179530620575,
      "learning_rate": 3.167184883999043e-05,
      "loss": 0.0035,
      "step": 19205
    },
    {
      "epoch": 2.691800981079187,
      "grad_norm": 0.23571625351905823,
      "learning_rate": 3.165749820617077e-05,
      "loss": 0.0214,
      "step": 19206
    },
    {
      "epoch": 2.691941135248774,
      "grad_norm": 0.7661653161048889,
      "learning_rate": 3.164314757235111e-05,
      "loss": 0.0616,
      "step": 19207
    },
    {
      "epoch": 2.6920812894183603,
      "grad_norm": 0.22897180914878845,
      "learning_rate": 3.1628796938531445e-05,
      "loss": 0.0092,
      "step": 19208
    },
    {
      "epoch": 2.692221443587947,
      "grad_norm": 0.03336496278643608,
      "learning_rate": 3.161444630471179e-05,
      "loss": 0.0013,
      "step": 19209
    },
    {
      "epoch": 2.6923615977575333,
      "grad_norm": 0.4114571213722229,
      "learning_rate": 3.160009567089213e-05,
      "loss": 0.062,
      "step": 19210
    },
    {
      "epoch": 2.6925017519271197,
      "grad_norm": 0.0611628033220768,
      "learning_rate": 3.158574503707247e-05,
      "loss": 0.003,
      "step": 19211
    },
    {
      "epoch": 2.692641906096706,
      "grad_norm": 0.39568981528282166,
      "learning_rate": 3.157139440325281e-05,
      "loss": 0.0392,
      "step": 19212
    },
    {
      "epoch": 2.692782060266293,
      "grad_norm": 0.3336886167526245,
      "learning_rate": 3.155704376943315e-05,
      "loss": 0.0385,
      "step": 19213
    },
    {
      "epoch": 2.6929222144358795,
      "grad_norm": 0.10875049978494644,
      "learning_rate": 3.1542693135613485e-05,
      "loss": 0.0033,
      "step": 19214
    },
    {
      "epoch": 2.693062368605466,
      "grad_norm": 0.7708019614219666,
      "learning_rate": 3.152834250179383e-05,
      "loss": 0.0257,
      "step": 19215
    },
    {
      "epoch": 2.6932025227750525,
      "grad_norm": 0.5931612849235535,
      "learning_rate": 3.151399186797416e-05,
      "loss": 0.0474,
      "step": 19216
    },
    {
      "epoch": 2.6933426769446394,
      "grad_norm": 0.2632920742034912,
      "learning_rate": 3.1499641234154504e-05,
      "loss": 0.0043,
      "step": 19217
    },
    {
      "epoch": 2.693482831114226,
      "grad_norm": 0.8514915108680725,
      "learning_rate": 3.1485290600334846e-05,
      "loss": 0.1476,
      "step": 19218
    },
    {
      "epoch": 2.6936229852838123,
      "grad_norm": 1.1172393560409546,
      "learning_rate": 3.147093996651519e-05,
      "loss": 0.0936,
      "step": 19219
    },
    {
      "epoch": 2.6937631394533987,
      "grad_norm": 1.2590055465698242,
      "learning_rate": 3.1456589332695524e-05,
      "loss": 0.039,
      "step": 19220
    },
    {
      "epoch": 2.693903293622985,
      "grad_norm": 0.3902694582939148,
      "learning_rate": 3.1442238698875866e-05,
      "loss": 0.0467,
      "step": 19221
    },
    {
      "epoch": 2.6940434477925717,
      "grad_norm": 0.2569301724433899,
      "learning_rate": 3.142788806505621e-05,
      "loss": 0.0304,
      "step": 19222
    },
    {
      "epoch": 2.694183601962158,
      "grad_norm": 0.18170863389968872,
      "learning_rate": 3.141353743123654e-05,
      "loss": 0.0422,
      "step": 19223
    },
    {
      "epoch": 2.694323756131745,
      "grad_norm": 0.2346627414226532,
      "learning_rate": 3.1399186797416885e-05,
      "loss": 0.0117,
      "step": 19224
    },
    {
      "epoch": 2.6944639103013315,
      "grad_norm": 0.13657331466674805,
      "learning_rate": 3.138483616359722e-05,
      "loss": 0.0203,
      "step": 19225
    },
    {
      "epoch": 2.694604064470918,
      "grad_norm": 0.2002572864294052,
      "learning_rate": 3.137048552977756e-05,
      "loss": 0.0259,
      "step": 19226
    },
    {
      "epoch": 2.6947442186405044,
      "grad_norm": 0.2764512598514557,
      "learning_rate": 3.1356134895957905e-05,
      "loss": 0.0433,
      "step": 19227
    },
    {
      "epoch": 2.6948843728100913,
      "grad_norm": 0.15916398167610168,
      "learning_rate": 3.134178426213824e-05,
      "loss": 0.0288,
      "step": 19228
    },
    {
      "epoch": 2.6950245269796778,
      "grad_norm": 0.2500961720943451,
      "learning_rate": 3.132743362831858e-05,
      "loss": 0.016,
      "step": 19229
    },
    {
      "epoch": 2.695164681149264,
      "grad_norm": 0.16817136108875275,
      "learning_rate": 3.1313082994498924e-05,
      "loss": 0.0101,
      "step": 19230
    },
    {
      "epoch": 2.6953048353188507,
      "grad_norm": 0.062395401298999786,
      "learning_rate": 3.129873236067926e-05,
      "loss": 0.0063,
      "step": 19231
    },
    {
      "epoch": 2.695444989488437,
      "grad_norm": 0.08706668764352798,
      "learning_rate": 3.12843817268596e-05,
      "loss": 0.0137,
      "step": 19232
    },
    {
      "epoch": 2.6955851436580236,
      "grad_norm": 0.35383933782577515,
      "learning_rate": 3.127003109303994e-05,
      "loss": 0.01,
      "step": 19233
    },
    {
      "epoch": 2.6957252978276105,
      "grad_norm": 0.27063995599746704,
      "learning_rate": 3.125568045922028e-05,
      "loss": 0.0689,
      "step": 19234
    },
    {
      "epoch": 2.695865451997197,
      "grad_norm": 0.09047505259513855,
      "learning_rate": 3.124132982540062e-05,
      "loss": 0.0137,
      "step": 19235
    },
    {
      "epoch": 2.6960056061667834,
      "grad_norm": 0.15810434520244598,
      "learning_rate": 3.122697919158096e-05,
      "loss": 0.0112,
      "step": 19236
    },
    {
      "epoch": 2.69614576033637,
      "grad_norm": 0.4797719120979309,
      "learning_rate": 3.12126285577613e-05,
      "loss": 0.0589,
      "step": 19237
    },
    {
      "epoch": 2.696285914505957,
      "grad_norm": 0.07274297624826431,
      "learning_rate": 3.119827792394164e-05,
      "loss": 0.005,
      "step": 19238
    },
    {
      "epoch": 2.6964260686755432,
      "grad_norm": 0.23896664381027222,
      "learning_rate": 3.1183927290121976e-05,
      "loss": 0.0228,
      "step": 19239
    },
    {
      "epoch": 2.6965662228451297,
      "grad_norm": 0.7867727875709534,
      "learning_rate": 3.116957665630232e-05,
      "loss": 0.0166,
      "step": 19240
    },
    {
      "epoch": 2.696706377014716,
      "grad_norm": 0.5129566192626953,
      "learning_rate": 3.1155226022482654e-05,
      "loss": 0.0249,
      "step": 19241
    },
    {
      "epoch": 2.6968465311843026,
      "grad_norm": 0.17058336734771729,
      "learning_rate": 3.1140875388662996e-05,
      "loss": 0.0267,
      "step": 19242
    },
    {
      "epoch": 2.696986685353889,
      "grad_norm": 0.5474103093147278,
      "learning_rate": 3.112652475484334e-05,
      "loss": 0.0419,
      "step": 19243
    },
    {
      "epoch": 2.697126839523476,
      "grad_norm": 0.38460883498191833,
      "learning_rate": 3.111217412102367e-05,
      "loss": 0.0222,
      "step": 19244
    },
    {
      "epoch": 2.6972669936930624,
      "grad_norm": 0.08524461835622787,
      "learning_rate": 3.1097823487204015e-05,
      "loss": 0.0271,
      "step": 19245
    },
    {
      "epoch": 2.697407147862649,
      "grad_norm": 0.40569591522216797,
      "learning_rate": 3.108347285338436e-05,
      "loss": 0.058,
      "step": 19246
    },
    {
      "epoch": 2.6975473020322354,
      "grad_norm": 0.13875296711921692,
      "learning_rate": 3.106912221956469e-05,
      "loss": 0.0246,
      "step": 19247
    },
    {
      "epoch": 2.6976874562018223,
      "grad_norm": 0.23318496346473694,
      "learning_rate": 3.1054771585745035e-05,
      "loss": 0.0077,
      "step": 19248
    },
    {
      "epoch": 2.6978276103714087,
      "grad_norm": 1.1501868963241577,
      "learning_rate": 3.104042095192537e-05,
      "loss": 0.1671,
      "step": 19249
    },
    {
      "epoch": 2.697967764540995,
      "grad_norm": 0.014810135588049889,
      "learning_rate": 3.102607031810571e-05,
      "loss": 0.001,
      "step": 19250
    },
    {
      "epoch": 2.6981079187105816,
      "grad_norm": 0.0525469109416008,
      "learning_rate": 3.1011719684286055e-05,
      "loss": 0.0033,
      "step": 19251
    },
    {
      "epoch": 2.698248072880168,
      "grad_norm": 0.23817744851112366,
      "learning_rate": 3.099736905046639e-05,
      "loss": 0.0592,
      "step": 19252
    },
    {
      "epoch": 2.6983882270497546,
      "grad_norm": 0.07781829684972763,
      "learning_rate": 3.098301841664673e-05,
      "loss": 0.0032,
      "step": 19253
    },
    {
      "epoch": 2.698528381219341,
      "grad_norm": 0.3349219262599945,
      "learning_rate": 3.0968667782827074e-05,
      "loss": 0.0109,
      "step": 19254
    },
    {
      "epoch": 2.698668535388928,
      "grad_norm": 0.22210752964019775,
      "learning_rate": 3.095431714900741e-05,
      "loss": 0.0499,
      "step": 19255
    },
    {
      "epoch": 2.6988086895585144,
      "grad_norm": 0.3338225185871124,
      "learning_rate": 3.093996651518775e-05,
      "loss": 0.0446,
      "step": 19256
    },
    {
      "epoch": 2.698948843728101,
      "grad_norm": 0.21556487679481506,
      "learning_rate": 3.092561588136809e-05,
      "loss": 0.0247,
      "step": 19257
    },
    {
      "epoch": 2.6990889978976873,
      "grad_norm": 0.5140829682350159,
      "learning_rate": 3.091126524754843e-05,
      "loss": 0.0318,
      "step": 19258
    },
    {
      "epoch": 2.699229152067274,
      "grad_norm": 0.3025587797164917,
      "learning_rate": 3.089691461372877e-05,
      "loss": 0.0488,
      "step": 19259
    },
    {
      "epoch": 2.6993693062368607,
      "grad_norm": 0.12738296389579773,
      "learning_rate": 3.0882563979909107e-05,
      "loss": 0.0294,
      "step": 19260
    },
    {
      "epoch": 2.699509460406447,
      "grad_norm": 0.213106170296669,
      "learning_rate": 3.086821334608945e-05,
      "loss": 0.0328,
      "step": 19261
    },
    {
      "epoch": 2.6996496145760336,
      "grad_norm": 0.21545083820819855,
      "learning_rate": 3.085386271226979e-05,
      "loss": 0.0413,
      "step": 19262
    },
    {
      "epoch": 2.69978976874562,
      "grad_norm": 0.1640138030052185,
      "learning_rate": 3.083951207845013e-05,
      "loss": 0.0152,
      "step": 19263
    },
    {
      "epoch": 2.6999299229152065,
      "grad_norm": 0.16837556660175323,
      "learning_rate": 3.082516144463047e-05,
      "loss": 0.0168,
      "step": 19264
    },
    {
      "epoch": 2.7000700770847934,
      "grad_norm": 0.46164238452911377,
      "learning_rate": 3.081081081081081e-05,
      "loss": 0.0303,
      "step": 19265
    },
    {
      "epoch": 2.70021023125438,
      "grad_norm": 0.8670220375061035,
      "learning_rate": 3.0796460176991146e-05,
      "loss": 0.043,
      "step": 19266
    },
    {
      "epoch": 2.7003503854239663,
      "grad_norm": 0.09610386192798615,
      "learning_rate": 3.078210954317149e-05,
      "loss": 0.0036,
      "step": 19267
    },
    {
      "epoch": 2.700490539593553,
      "grad_norm": 0.4493335485458374,
      "learning_rate": 3.076775890935182e-05,
      "loss": 0.0501,
      "step": 19268
    },
    {
      "epoch": 2.7006306937631397,
      "grad_norm": 3.75961971282959,
      "learning_rate": 3.0753408275532165e-05,
      "loss": 0.1498,
      "step": 19269
    },
    {
      "epoch": 2.700770847932726,
      "grad_norm": 0.05502821132540703,
      "learning_rate": 3.073905764171251e-05,
      "loss": 0.0024,
      "step": 19270
    },
    {
      "epoch": 2.7009110021023126,
      "grad_norm": 0.09672386199235916,
      "learning_rate": 3.072470700789285e-05,
      "loss": 0.0138,
      "step": 19271
    },
    {
      "epoch": 2.701051156271899,
      "grad_norm": 0.2971091866493225,
      "learning_rate": 3.0710356374073185e-05,
      "loss": 0.0293,
      "step": 19272
    },
    {
      "epoch": 2.7011913104414855,
      "grad_norm": 0.11204875260591507,
      "learning_rate": 3.069600574025353e-05,
      "loss": 0.0107,
      "step": 19273
    },
    {
      "epoch": 2.701331464611072,
      "grad_norm": 0.41986027359962463,
      "learning_rate": 3.068165510643386e-05,
      "loss": 0.0729,
      "step": 19274
    },
    {
      "epoch": 2.701471618780659,
      "grad_norm": 0.12055579572916031,
      "learning_rate": 3.0667304472614204e-05,
      "loss": 0.0126,
      "step": 19275
    },
    {
      "epoch": 2.7016117729502453,
      "grad_norm": 0.05585943162441254,
      "learning_rate": 3.0652953838794546e-05,
      "loss": 0.0029,
      "step": 19276
    },
    {
      "epoch": 2.701751927119832,
      "grad_norm": 0.33906131982803345,
      "learning_rate": 3.063860320497488e-05,
      "loss": 0.0778,
      "step": 19277
    },
    {
      "epoch": 2.7018920812894183,
      "grad_norm": 0.10691893845796585,
      "learning_rate": 3.0624252571155224e-05,
      "loss": 0.0082,
      "step": 19278
    },
    {
      "epoch": 2.702032235459005,
      "grad_norm": 0.33289530873298645,
      "learning_rate": 3.0609901937335566e-05,
      "loss": 0.022,
      "step": 19279
    },
    {
      "epoch": 2.7021723896285916,
      "grad_norm": 0.39636731147766113,
      "learning_rate": 3.05955513035159e-05,
      "loss": 0.0236,
      "step": 19280
    },
    {
      "epoch": 2.702312543798178,
      "grad_norm": 0.20659658312797546,
      "learning_rate": 3.0581200669696243e-05,
      "loss": 0.0144,
      "step": 19281
    },
    {
      "epoch": 2.7024526979677645,
      "grad_norm": 0.0468139685690403,
      "learning_rate": 3.056685003587658e-05,
      "loss": 0.0038,
      "step": 19282
    },
    {
      "epoch": 2.702592852137351,
      "grad_norm": 0.2533717751502991,
      "learning_rate": 3.055249940205692e-05,
      "loss": 0.0265,
      "step": 19283
    },
    {
      "epoch": 2.7027330063069375,
      "grad_norm": 0.2451319545507431,
      "learning_rate": 3.053814876823726e-05,
      "loss": 0.0331,
      "step": 19284
    },
    {
      "epoch": 2.702873160476524,
      "grad_norm": 0.2609651982784271,
      "learning_rate": 3.05237981344176e-05,
      "loss": 0.0315,
      "step": 19285
    },
    {
      "epoch": 2.703013314646111,
      "grad_norm": 0.18486066162586212,
      "learning_rate": 3.050944750059794e-05,
      "loss": 0.0124,
      "step": 19286
    },
    {
      "epoch": 2.7031534688156973,
      "grad_norm": 0.3572184145450592,
      "learning_rate": 3.049509686677828e-05,
      "loss": 0.0164,
      "step": 19287
    },
    {
      "epoch": 2.7032936229852838,
      "grad_norm": 0.1174364909529686,
      "learning_rate": 3.0480746232958618e-05,
      "loss": 0.011,
      "step": 19288
    },
    {
      "epoch": 2.70343377715487,
      "grad_norm": 0.04312127083539963,
      "learning_rate": 3.046639559913896e-05,
      "loss": 0.0041,
      "step": 19289
    },
    {
      "epoch": 2.703573931324457,
      "grad_norm": 0.062024567276239395,
      "learning_rate": 3.04520449653193e-05,
      "loss": 0.0072,
      "step": 19290
    },
    {
      "epoch": 2.7037140854940436,
      "grad_norm": 0.18893611431121826,
      "learning_rate": 3.0437694331499637e-05,
      "loss": 0.0124,
      "step": 19291
    },
    {
      "epoch": 2.70385423966363,
      "grad_norm": 0.030310414731502533,
      "learning_rate": 3.0423343697679976e-05,
      "loss": 0.0022,
      "step": 19292
    },
    {
      "epoch": 2.7039943938332165,
      "grad_norm": 0.06617215275764465,
      "learning_rate": 3.040899306386032e-05,
      "loss": 0.0066,
      "step": 19293
    },
    {
      "epoch": 2.704134548002803,
      "grad_norm": 0.17179124057292938,
      "learning_rate": 3.0394642430040657e-05,
      "loss": 0.0059,
      "step": 19294
    },
    {
      "epoch": 2.7042747021723894,
      "grad_norm": 0.16624535620212555,
      "learning_rate": 3.0380291796220996e-05,
      "loss": 0.0197,
      "step": 19295
    },
    {
      "epoch": 2.7044148563419763,
      "grad_norm": 0.05562801659107208,
      "learning_rate": 3.0365941162401338e-05,
      "loss": 0.0058,
      "step": 19296
    },
    {
      "epoch": 2.7045550105115628,
      "grad_norm": 0.30680719017982483,
      "learning_rate": 3.0351590528581677e-05,
      "loss": 0.0167,
      "step": 19297
    },
    {
      "epoch": 2.7046951646811492,
      "grad_norm": 0.14756862819194794,
      "learning_rate": 3.033723989476202e-05,
      "loss": 0.0266,
      "step": 19298
    },
    {
      "epoch": 2.7048353188507357,
      "grad_norm": 0.055031850934028625,
      "learning_rate": 3.0322889260942354e-05,
      "loss": 0.0214,
      "step": 19299
    },
    {
      "epoch": 2.7049754730203226,
      "grad_norm": 0.30849936604499817,
      "learning_rate": 3.0308538627122696e-05,
      "loss": 0.0775,
      "step": 19300
    },
    {
      "epoch": 2.705115627189909,
      "grad_norm": 0.16849109530448914,
      "learning_rate": 3.0294187993303035e-05,
      "loss": 0.0196,
      "step": 19301
    },
    {
      "epoch": 2.7052557813594955,
      "grad_norm": 0.09167186170816422,
      "learning_rate": 3.0279837359483377e-05,
      "loss": 0.0031,
      "step": 19302
    },
    {
      "epoch": 2.705395935529082,
      "grad_norm": 0.2093467265367508,
      "learning_rate": 3.0265486725663712e-05,
      "loss": 0.0201,
      "step": 19303
    },
    {
      "epoch": 2.7055360896986684,
      "grad_norm": 0.24855487048625946,
      "learning_rate": 3.0251136091844054e-05,
      "loss": 0.0116,
      "step": 19304
    },
    {
      "epoch": 2.705676243868255,
      "grad_norm": 0.32233527302742004,
      "learning_rate": 3.0236785458024393e-05,
      "loss": 0.0423,
      "step": 19305
    },
    {
      "epoch": 2.7058163980378414,
      "grad_norm": 0.09075789898633957,
      "learning_rate": 3.0222434824204735e-05,
      "loss": 0.0092,
      "step": 19306
    },
    {
      "epoch": 2.7059565522074283,
      "grad_norm": 0.22996492683887482,
      "learning_rate": 3.020808419038507e-05,
      "loss": 0.0194,
      "step": 19307
    },
    {
      "epoch": 2.7060967063770147,
      "grad_norm": 0.20749510824680328,
      "learning_rate": 3.0193733556565413e-05,
      "loss": 0.0167,
      "step": 19308
    },
    {
      "epoch": 2.706236860546601,
      "grad_norm": 0.36879536509513855,
      "learning_rate": 3.017938292274575e-05,
      "loss": 0.0506,
      "step": 19309
    },
    {
      "epoch": 2.706377014716188,
      "grad_norm": 0.2779279947280884,
      "learning_rate": 3.0165032288926094e-05,
      "loss": 0.0493,
      "step": 19310
    },
    {
      "epoch": 2.7065171688857745,
      "grad_norm": 0.3724881112575531,
      "learning_rate": 3.015068165510643e-05,
      "loss": 0.067,
      "step": 19311
    },
    {
      "epoch": 2.706657323055361,
      "grad_norm": 0.21186470985412598,
      "learning_rate": 3.013633102128677e-05,
      "loss": 0.0411,
      "step": 19312
    },
    {
      "epoch": 2.7067974772249475,
      "grad_norm": 0.10183316469192505,
      "learning_rate": 3.012198038746711e-05,
      "loss": 0.0087,
      "step": 19313
    },
    {
      "epoch": 2.706937631394534,
      "grad_norm": 0.48984479904174805,
      "learning_rate": 3.0107629753647452e-05,
      "loss": 0.1369,
      "step": 19314
    },
    {
      "epoch": 2.7070777855641204,
      "grad_norm": 0.016151679679751396,
      "learning_rate": 3.0093279119827787e-05,
      "loss": 0.0014,
      "step": 19315
    },
    {
      "epoch": 2.707217939733707,
      "grad_norm": 0.34023815393447876,
      "learning_rate": 3.007892848600813e-05,
      "loss": 0.037,
      "step": 19316
    },
    {
      "epoch": 2.7073580939032937,
      "grad_norm": 0.661260724067688,
      "learning_rate": 3.0064577852188468e-05,
      "loss": 0.1406,
      "step": 19317
    },
    {
      "epoch": 2.70749824807288,
      "grad_norm": 0.2402743399143219,
      "learning_rate": 3.005022721836881e-05,
      "loss": 0.0063,
      "step": 19318
    },
    {
      "epoch": 2.7076384022424667,
      "grad_norm": 2.0923216342926025,
      "learning_rate": 3.0035876584549145e-05,
      "loss": 0.1607,
      "step": 19319
    },
    {
      "epoch": 2.707778556412053,
      "grad_norm": NaN,
      "learning_rate": 3.0021525950729488e-05,
      "loss": 0.4982,
      "step": 19320
    },
    {
      "epoch": 2.70791871058164,
      "grad_norm": 0.2181277573108673,
      "learning_rate": 3.0021525950729488e-05,
      "loss": 0.022,
      "step": 19321
    },
    {
      "epoch": 2.7080588647512265,
      "grad_norm": 0.2151375114917755,
      "learning_rate": 3.0007175316909826e-05,
      "loss": 0.0246,
      "step": 19322
    },
    {
      "epoch": 2.708199018920813,
      "grad_norm": 0.8480563759803772,
      "learning_rate": 2.999282468309017e-05,
      "loss": 0.0288,
      "step": 19323
    },
    {
      "epoch": 2.7083391730903994,
      "grad_norm": 0.35201698541641235,
      "learning_rate": 2.9978474049270504e-05,
      "loss": 0.057,
      "step": 19324
    },
    {
      "epoch": 2.708479327259986,
      "grad_norm": 0.204260915517807,
      "learning_rate": 2.9964123415450846e-05,
      "loss": 0.0136,
      "step": 19325
    },
    {
      "epoch": 2.7086194814295723,
      "grad_norm": 0.0723210871219635,
      "learning_rate": 2.9949772781631185e-05,
      "loss": 0.0117,
      "step": 19326
    },
    {
      "epoch": 2.708759635599159,
      "grad_norm": 0.1000300869345665,
      "learning_rate": 2.9935422147811527e-05,
      "loss": 0.006,
      "step": 19327
    },
    {
      "epoch": 2.7088997897687457,
      "grad_norm": 0.15267686545848846,
      "learning_rate": 2.9921071513991862e-05,
      "loss": 0.0169,
      "step": 19328
    },
    {
      "epoch": 2.709039943938332,
      "grad_norm": 0.14752452075481415,
      "learning_rate": 2.9906720880172204e-05,
      "loss": 0.017,
      "step": 19329
    },
    {
      "epoch": 2.7091800981079186,
      "grad_norm": 0.4122331738471985,
      "learning_rate": 2.9892370246352543e-05,
      "loss": 0.0351,
      "step": 19330
    },
    {
      "epoch": 2.7093202522775055,
      "grad_norm": 0.2442162185907364,
      "learning_rate": 2.9878019612532885e-05,
      "loss": 0.0484,
      "step": 19331
    },
    {
      "epoch": 2.709460406447092,
      "grad_norm": 0.20503857731819153,
      "learning_rate": 2.986366897871322e-05,
      "loss": 0.0168,
      "step": 19332
    },
    {
      "epoch": 2.7096005606166784,
      "grad_norm": 0.1015247330069542,
      "learning_rate": 2.9849318344893562e-05,
      "loss": 0.0238,
      "step": 19333
    },
    {
      "epoch": 2.709740714786265,
      "grad_norm": 0.062133099883794785,
      "learning_rate": 2.9834967711073905e-05,
      "loss": 0.0032,
      "step": 19334
    },
    {
      "epoch": 2.7098808689558513,
      "grad_norm": 0.39330267906188965,
      "learning_rate": 2.9820617077254243e-05,
      "loss": 0.0164,
      "step": 19335
    },
    {
      "epoch": 2.710021023125438,
      "grad_norm": 0.0634106993675232,
      "learning_rate": 2.9806266443434585e-05,
      "loss": 0.0027,
      "step": 19336
    },
    {
      "epoch": 2.7101611772950243,
      "grad_norm": 0.3063826262950897,
      "learning_rate": 2.979191580961492e-05,
      "loss": 0.1063,
      "step": 19337
    },
    {
      "epoch": 2.710301331464611,
      "grad_norm": 0.18283236026763916,
      "learning_rate": 2.9777565175795263e-05,
      "loss": 0.0339,
      "step": 19338
    },
    {
      "epoch": 2.7104414856341976,
      "grad_norm": 0.1700955033302307,
      "learning_rate": 2.97632145419756e-05,
      "loss": 0.0223,
      "step": 19339
    },
    {
      "epoch": 2.710581639803784,
      "grad_norm": 0.5060061812400818,
      "learning_rate": 2.9748863908155944e-05,
      "loss": 0.018,
      "step": 19340
    },
    {
      "epoch": 2.7107217939733705,
      "grad_norm": 0.17611917853355408,
      "learning_rate": 2.973451327433628e-05,
      "loss": 0.02,
      "step": 19341
    },
    {
      "epoch": 2.7108619481429574,
      "grad_norm": 0.2941989004611969,
      "learning_rate": 2.972016264051662e-05,
      "loss": 0.0182,
      "step": 19342
    },
    {
      "epoch": 2.711002102312544,
      "grad_norm": 0.26543599367141724,
      "learning_rate": 2.970581200669696e-05,
      "loss": 0.0341,
      "step": 19343
    },
    {
      "epoch": 2.7111422564821304,
      "grad_norm": 0.39066413044929504,
      "learning_rate": 2.9691461372877302e-05,
      "loss": 0.0108,
      "step": 19344
    },
    {
      "epoch": 2.711282410651717,
      "grad_norm": 0.14799782633781433,
      "learning_rate": 2.9677110739057637e-05,
      "loss": 0.0062,
      "step": 19345
    },
    {
      "epoch": 2.7114225648213033,
      "grad_norm": 0.38191646337509155,
      "learning_rate": 2.966276010523798e-05,
      "loss": 0.0157,
      "step": 19346
    },
    {
      "epoch": 2.7115627189908897,
      "grad_norm": 0.02129509672522545,
      "learning_rate": 2.9648409471418318e-05,
      "loss": 0.0011,
      "step": 19347
    },
    {
      "epoch": 2.7117028731604766,
      "grad_norm": 0.04881303384900093,
      "learning_rate": 2.963405883759866e-05,
      "loss": 0.0207,
      "step": 19348
    },
    {
      "epoch": 2.711843027330063,
      "grad_norm": 0.10351043939590454,
      "learning_rate": 2.9619708203778996e-05,
      "loss": 0.009,
      "step": 19349
    },
    {
      "epoch": 2.7119831814996496,
      "grad_norm": 0.34045377373695374,
      "learning_rate": 2.9605357569959338e-05,
      "loss": 0.0208,
      "step": 19350
    },
    {
      "epoch": 2.712123335669236,
      "grad_norm": 0.11457183957099915,
      "learning_rate": 2.9591006936139676e-05,
      "loss": 0.0192,
      "step": 19351
    },
    {
      "epoch": 2.712263489838823,
      "grad_norm": 0.33656391501426697,
      "learning_rate": 2.957665630232002e-05,
      "loss": 0.0303,
      "step": 19352
    },
    {
      "epoch": 2.7124036440084094,
      "grad_norm": 1.1371277570724487,
      "learning_rate": 2.9562305668500354e-05,
      "loss": 0.0078,
      "step": 19353
    },
    {
      "epoch": 2.712543798177996,
      "grad_norm": 0.6659761071205139,
      "learning_rate": 2.9547955034680696e-05,
      "loss": 0.1001,
      "step": 19354
    },
    {
      "epoch": 2.7126839523475823,
      "grad_norm": 0.1783275604248047,
      "learning_rate": 2.9533604400861035e-05,
      "loss": 0.0468,
      "step": 19355
    },
    {
      "epoch": 2.7128241065171688,
      "grad_norm": 0.4399384558200836,
      "learning_rate": 2.9519253767041377e-05,
      "loss": 0.0187,
      "step": 19356
    },
    {
      "epoch": 2.7129642606867552,
      "grad_norm": 0.525000274181366,
      "learning_rate": 2.9504903133221712e-05,
      "loss": 0.0304,
      "step": 19357
    },
    {
      "epoch": 2.713104414856342,
      "grad_norm": 1.0218911170959473,
      "learning_rate": 2.9490552499402054e-05,
      "loss": 0.0432,
      "step": 19358
    },
    {
      "epoch": 2.7132445690259286,
      "grad_norm": 0.08774565905332565,
      "learning_rate": 2.9476201865582393e-05,
      "loss": 0.0138,
      "step": 19359
    },
    {
      "epoch": 2.713384723195515,
      "grad_norm": 0.10079289972782135,
      "learning_rate": 2.9461851231762735e-05,
      "loss": 0.0064,
      "step": 19360
    },
    {
      "epoch": 2.7135248773651015,
      "grad_norm": 0.036775000393390656,
      "learning_rate": 2.944750059794307e-05,
      "loss": 0.0026,
      "step": 19361
    },
    {
      "epoch": 2.7136650315346884,
      "grad_norm": 0.3501421809196472,
      "learning_rate": 2.9433149964123413e-05,
      "loss": 0.0465,
      "step": 19362
    },
    {
      "epoch": 2.713805185704275,
      "grad_norm": 0.43776899576187134,
      "learning_rate": 2.941879933030375e-05,
      "loss": 0.0389,
      "step": 19363
    },
    {
      "epoch": 2.7139453398738613,
      "grad_norm": 0.3469582796096802,
      "learning_rate": 2.9404448696484093e-05,
      "loss": 0.0692,
      "step": 19364
    },
    {
      "epoch": 2.714085494043448,
      "grad_norm": 0.0735086128115654,
      "learning_rate": 2.939009806266443e-05,
      "loss": 0.0021,
      "step": 19365
    },
    {
      "epoch": 2.7142256482130342,
      "grad_norm": 0.9639247059822083,
      "learning_rate": 2.937574742884477e-05,
      "loss": 0.0616,
      "step": 19366
    },
    {
      "epoch": 2.7143658023826207,
      "grad_norm": 0.43536242842674255,
      "learning_rate": 2.936139679502511e-05,
      "loss": 0.0122,
      "step": 19367
    },
    {
      "epoch": 2.714505956552207,
      "grad_norm": 1.8804329633712769,
      "learning_rate": 2.934704616120545e-05,
      "loss": 0.0964,
      "step": 19368
    },
    {
      "epoch": 2.714646110721794,
      "grad_norm": 1.5284156799316406,
      "learning_rate": 2.9332695527385787e-05,
      "loss": 0.1079,
      "step": 19369
    },
    {
      "epoch": 2.7147862648913805,
      "grad_norm": 0.1590013951063156,
      "learning_rate": 2.931834489356613e-05,
      "loss": 0.0052,
      "step": 19370
    },
    {
      "epoch": 2.714926419060967,
      "grad_norm": 0.06793346256017685,
      "learning_rate": 2.930399425974647e-05,
      "loss": 0.0054,
      "step": 19371
    },
    {
      "epoch": 2.7150665732305534,
      "grad_norm": 0.3141545355319977,
      "learning_rate": 2.928964362592681e-05,
      "loss": 0.0726,
      "step": 19372
    },
    {
      "epoch": 2.7152067274001404,
      "grad_norm": 0.17232657968997955,
      "learning_rate": 2.9275292992107152e-05,
      "loss": 0.0149,
      "step": 19373
    },
    {
      "epoch": 2.715346881569727,
      "grad_norm": 0.2903800904750824,
      "learning_rate": 2.9260942358287487e-05,
      "loss": 0.0317,
      "step": 19374
    },
    {
      "epoch": 2.7154870357393133,
      "grad_norm": 0.16670174896717072,
      "learning_rate": 2.924659172446783e-05,
      "loss": 0.0294,
      "step": 19375
    },
    {
      "epoch": 2.7156271899088997,
      "grad_norm": 0.10735027492046356,
      "learning_rate": 2.9232241090648168e-05,
      "loss": 0.0103,
      "step": 19376
    },
    {
      "epoch": 2.715767344078486,
      "grad_norm": 0.1174519807100296,
      "learning_rate": 2.921789045682851e-05,
      "loss": 0.0133,
      "step": 19377
    },
    {
      "epoch": 2.7159074982480726,
      "grad_norm": 0.13293814659118652,
      "learning_rate": 2.9203539823008846e-05,
      "loss": 0.0106,
      "step": 19378
    },
    {
      "epoch": 2.7160476524176596,
      "grad_norm": 0.3962964415550232,
      "learning_rate": 2.9189189189189188e-05,
      "loss": 0.0211,
      "step": 19379
    },
    {
      "epoch": 2.716187806587246,
      "grad_norm": 0.3265494108200073,
      "learning_rate": 2.9174838555369527e-05,
      "loss": 0.0294,
      "step": 19380
    },
    {
      "epoch": 2.7163279607568325,
      "grad_norm": 0.15156802535057068,
      "learning_rate": 2.916048792154987e-05,
      "loss": 0.0295,
      "step": 19381
    },
    {
      "epoch": 2.716468114926419,
      "grad_norm": 0.1435842663049698,
      "learning_rate": 2.9146137287730204e-05,
      "loss": 0.015,
      "step": 19382
    },
    {
      "epoch": 2.716608269096006,
      "grad_norm": 0.15818379819393158,
      "learning_rate": 2.9131786653910546e-05,
      "loss": 0.012,
      "step": 19383
    },
    {
      "epoch": 2.7167484232655923,
      "grad_norm": 0.16083475947380066,
      "learning_rate": 2.9117436020090885e-05,
      "loss": 0.0409,
      "step": 19384
    },
    {
      "epoch": 2.7168885774351788,
      "grad_norm": 0.309542715549469,
      "learning_rate": 2.9103085386271227e-05,
      "loss": 0.0185,
      "step": 19385
    },
    {
      "epoch": 2.717028731604765,
      "grad_norm": 0.07071244716644287,
      "learning_rate": 2.9088734752451562e-05,
      "loss": 0.0142,
      "step": 19386
    },
    {
      "epoch": 2.7171688857743517,
      "grad_norm": 0.23412373661994934,
      "learning_rate": 2.9074384118631904e-05,
      "loss": 0.0458,
      "step": 19387
    },
    {
      "epoch": 2.717309039943938,
      "grad_norm": 0.38587841391563416,
      "learning_rate": 2.9060033484812243e-05,
      "loss": 0.0154,
      "step": 19388
    },
    {
      "epoch": 2.717449194113525,
      "grad_norm": 0.6501622796058655,
      "learning_rate": 2.9045682850992585e-05,
      "loss": 0.0994,
      "step": 19389
    },
    {
      "epoch": 2.7175893482831115,
      "grad_norm": 0.9390354752540588,
      "learning_rate": 2.903133221717292e-05,
      "loss": 0.0542,
      "step": 19390
    },
    {
      "epoch": 2.717729502452698,
      "grad_norm": 0.14710909128189087,
      "learning_rate": 2.9016981583353263e-05,
      "loss": 0.0295,
      "step": 19391
    },
    {
      "epoch": 2.7178696566222844,
      "grad_norm": 0.25121939182281494,
      "learning_rate": 2.90026309495336e-05,
      "loss": 0.0239,
      "step": 19392
    },
    {
      "epoch": 2.7180098107918713,
      "grad_norm": 0.20317435264587402,
      "learning_rate": 2.8988280315713944e-05,
      "loss": 0.0301,
      "step": 19393
    },
    {
      "epoch": 2.7181499649614578,
      "grad_norm": 0.4621746242046356,
      "learning_rate": 2.897392968189428e-05,
      "loss": 0.1095,
      "step": 19394
    },
    {
      "epoch": 2.7182901191310442,
      "grad_norm": 0.18017859756946564,
      "learning_rate": 2.895957904807462e-05,
      "loss": 0.0306,
      "step": 19395
    },
    {
      "epoch": 2.7184302733006307,
      "grad_norm": 0.11911261081695557,
      "learning_rate": 2.894522841425496e-05,
      "loss": 0.0195,
      "step": 19396
    },
    {
      "epoch": 2.718570427470217,
      "grad_norm": 0.14756834506988525,
      "learning_rate": 2.8930877780435302e-05,
      "loss": 0.0186,
      "step": 19397
    },
    {
      "epoch": 2.7187105816398036,
      "grad_norm": 0.13270531594753265,
      "learning_rate": 2.8916527146615637e-05,
      "loss": 0.0104,
      "step": 19398
    },
    {
      "epoch": 2.71885073580939,
      "grad_norm": 0.1459481418132782,
      "learning_rate": 2.890217651279598e-05,
      "loss": 0.0469,
      "step": 19399
    },
    {
      "epoch": 2.718990889978977,
      "grad_norm": 0.12475838512182236,
      "learning_rate": 2.8887825878976318e-05,
      "loss": 0.0101,
      "step": 19400
    },
    {
      "epoch": 2.7191310441485634,
      "grad_norm": 0.3245266377925873,
      "learning_rate": 2.887347524515666e-05,
      "loss": 0.0066,
      "step": 19401
    },
    {
      "epoch": 2.71927119831815,
      "grad_norm": 0.20143072307109833,
      "learning_rate": 2.8859124611336995e-05,
      "loss": 0.0232,
      "step": 19402
    },
    {
      "epoch": 2.7194113524877364,
      "grad_norm": 0.36088934540748596,
      "learning_rate": 2.8844773977517338e-05,
      "loss": 0.039,
      "step": 19403
    },
    {
      "epoch": 2.7195515066573233,
      "grad_norm": 0.29269424080848694,
      "learning_rate": 2.8830423343697676e-05,
      "loss": 0.0201,
      "step": 19404
    },
    {
      "epoch": 2.7196916608269097,
      "grad_norm": 0.12134575843811035,
      "learning_rate": 2.881607270987802e-05,
      "loss": 0.0258,
      "step": 19405
    },
    {
      "epoch": 2.719831814996496,
      "grad_norm": 0.039690855890512466,
      "learning_rate": 2.8801722076058354e-05,
      "loss": 0.0021,
      "step": 19406
    },
    {
      "epoch": 2.7199719691660826,
      "grad_norm": 0.4401889443397522,
      "learning_rate": 2.8787371442238696e-05,
      "loss": 0.0412,
      "step": 19407
    },
    {
      "epoch": 2.720112123335669,
      "grad_norm": 0.2721652686595917,
      "learning_rate": 2.8773020808419038e-05,
      "loss": 0.0287,
      "step": 19408
    },
    {
      "epoch": 2.7202522775052556,
      "grad_norm": 0.06766214966773987,
      "learning_rate": 2.8758670174599377e-05,
      "loss": 0.0057,
      "step": 19409
    },
    {
      "epoch": 2.7203924316748425,
      "grad_norm": 0.09282040596008301,
      "learning_rate": 2.874431954077972e-05,
      "loss": 0.0104,
      "step": 19410
    },
    {
      "epoch": 2.720532585844429,
      "grad_norm": 0.138297900557518,
      "learning_rate": 2.8729968906960054e-05,
      "loss": 0.0057,
      "step": 19411
    },
    {
      "epoch": 2.7206727400140154,
      "grad_norm": 0.6147881746292114,
      "learning_rate": 2.8715618273140396e-05,
      "loss": 0.0248,
      "step": 19412
    },
    {
      "epoch": 2.720812894183602,
      "grad_norm": 0.5212564468383789,
      "learning_rate": 2.8701267639320735e-05,
      "loss": 0.0346,
      "step": 19413
    },
    {
      "epoch": 2.7209530483531887,
      "grad_norm": 0.2303069680929184,
      "learning_rate": 2.8686917005501077e-05,
      "loss": 0.0304,
      "step": 19414
    },
    {
      "epoch": 2.721093202522775,
      "grad_norm": 0.17975178360939026,
      "learning_rate": 2.8672566371681412e-05,
      "loss": 0.0126,
      "step": 19415
    },
    {
      "epoch": 2.7212333566923617,
      "grad_norm": 0.08367469906806946,
      "learning_rate": 2.8658215737861755e-05,
      "loss": 0.0195,
      "step": 19416
    },
    {
      "epoch": 2.721373510861948,
      "grad_norm": 0.14639873802661896,
      "learning_rate": 2.8643865104042093e-05,
      "loss": 0.0135,
      "step": 19417
    },
    {
      "epoch": 2.7215136650315346,
      "grad_norm": 0.4908272325992584,
      "learning_rate": 2.8629514470222435e-05,
      "loss": 0.1071,
      "step": 19418
    },
    {
      "epoch": 2.721653819201121,
      "grad_norm": 1.3779081106185913,
      "learning_rate": 2.861516383640277e-05,
      "loss": 0.147,
      "step": 19419
    },
    {
      "epoch": 2.721793973370708,
      "grad_norm": 0.8610438704490662,
      "learning_rate": 2.8600813202583113e-05,
      "loss": 0.0349,
      "step": 19420
    },
    {
      "epoch": 2.7219341275402944,
      "grad_norm": 0.03710014745593071,
      "learning_rate": 2.858646256876345e-05,
      "loss": 0.0034,
      "step": 19421
    },
    {
      "epoch": 2.722074281709881,
      "grad_norm": 0.20834094285964966,
      "learning_rate": 2.8572111934943794e-05,
      "loss": 0.0125,
      "step": 19422
    },
    {
      "epoch": 2.7222144358794673,
      "grad_norm": 0.23193572461605072,
      "learning_rate": 2.855776130112413e-05,
      "loss": 0.0425,
      "step": 19423
    },
    {
      "epoch": 2.7223545900490542,
      "grad_norm": 0.11113123595714569,
      "learning_rate": 2.854341066730447e-05,
      "loss": 0.0143,
      "step": 19424
    },
    {
      "epoch": 2.7224947442186407,
      "grad_norm": 0.17334316670894623,
      "learning_rate": 2.852906003348481e-05,
      "loss": 0.0234,
      "step": 19425
    },
    {
      "epoch": 2.722634898388227,
      "grad_norm": 0.19656060636043549,
      "learning_rate": 2.8514709399665152e-05,
      "loss": 0.0195,
      "step": 19426
    },
    {
      "epoch": 2.7227750525578136,
      "grad_norm": 0.06525545567274094,
      "learning_rate": 2.8500358765845487e-05,
      "loss": 0.0035,
      "step": 19427
    },
    {
      "epoch": 2.7229152067274,
      "grad_norm": 0.38396257162094116,
      "learning_rate": 2.848600813202583e-05,
      "loss": 0.0428,
      "step": 19428
    },
    {
      "epoch": 2.7230553608969865,
      "grad_norm": 0.05099687725305557,
      "learning_rate": 2.8471657498206168e-05,
      "loss": 0.006,
      "step": 19429
    },
    {
      "epoch": 2.723195515066573,
      "grad_norm": 0.42549929022789,
      "learning_rate": 2.845730686438651e-05,
      "loss": 0.044,
      "step": 19430
    },
    {
      "epoch": 2.72333566923616,
      "grad_norm": 0.1808493584394455,
      "learning_rate": 2.8442956230566846e-05,
      "loss": 0.0393,
      "step": 19431
    },
    {
      "epoch": 2.7234758234057463,
      "grad_norm": 0.4539145231246948,
      "learning_rate": 2.8428605596747188e-05,
      "loss": 0.0637,
      "step": 19432
    },
    {
      "epoch": 2.723615977575333,
      "grad_norm": 0.8421884775161743,
      "learning_rate": 2.8414254962927526e-05,
      "loss": 0.0475,
      "step": 19433
    },
    {
      "epoch": 2.7237561317449193,
      "grad_norm": 0.09060514718294144,
      "learning_rate": 2.839990432910787e-05,
      "loss": 0.0052,
      "step": 19434
    },
    {
      "epoch": 2.723896285914506,
      "grad_norm": 0.07637181133031845,
      "learning_rate": 2.8385553695288204e-05,
      "loss": 0.0041,
      "step": 19435
    },
    {
      "epoch": 2.7240364400840926,
      "grad_norm": 0.3477778434753418,
      "learning_rate": 2.8371203061468546e-05,
      "loss": 0.0473,
      "step": 19436
    },
    {
      "epoch": 2.724176594253679,
      "grad_norm": 0.20549605786800385,
      "learning_rate": 2.8356852427648885e-05,
      "loss": 0.0164,
      "step": 19437
    },
    {
      "epoch": 2.7243167484232655,
      "grad_norm": 0.6651625633239746,
      "learning_rate": 2.8342501793829227e-05,
      "loss": 0.0511,
      "step": 19438
    },
    {
      "epoch": 2.724456902592852,
      "grad_norm": 0.09272904694080353,
      "learning_rate": 2.8328151160009562e-05,
      "loss": 0.0039,
      "step": 19439
    },
    {
      "epoch": 2.7245970567624385,
      "grad_norm": 0.6257189512252808,
      "learning_rate": 2.8313800526189904e-05,
      "loss": 0.0117,
      "step": 19440
    },
    {
      "epoch": 2.7247372109320254,
      "grad_norm": 0.11424887180328369,
      "learning_rate": 2.8299449892370243e-05,
      "loss": 0.0105,
      "step": 19441
    },
    {
      "epoch": 2.724877365101612,
      "grad_norm": 0.38900700211524963,
      "learning_rate": 2.8285099258550585e-05,
      "loss": 0.0264,
      "step": 19442
    },
    {
      "epoch": 2.7250175192711983,
      "grad_norm": 0.21545597910881042,
      "learning_rate": 2.827074862473092e-05,
      "loss": 0.0491,
      "step": 19443
    },
    {
      "epoch": 2.7251576734407847,
      "grad_norm": 0.45535576343536377,
      "learning_rate": 2.8256397990911263e-05,
      "loss": 0.0425,
      "step": 19444
    },
    {
      "epoch": 2.7252978276103716,
      "grad_norm": 0.21709613502025604,
      "learning_rate": 2.82420473570916e-05,
      "loss": 0.0213,
      "step": 19445
    },
    {
      "epoch": 2.725437981779958,
      "grad_norm": 0.14636696875095367,
      "learning_rate": 2.8227696723271943e-05,
      "loss": 0.0122,
      "step": 19446
    },
    {
      "epoch": 2.7255781359495446,
      "grad_norm": 0.42629554867744446,
      "learning_rate": 2.8213346089452282e-05,
      "loss": 0.0853,
      "step": 19447
    },
    {
      "epoch": 2.725718290119131,
      "grad_norm": 0.12266921252012253,
      "learning_rate": 2.819899545563262e-05,
      "loss": 0.006,
      "step": 19448
    },
    {
      "epoch": 2.7258584442887175,
      "grad_norm": 0.42401838302612305,
      "learning_rate": 2.8184644821812963e-05,
      "loss": 0.0406,
      "step": 19449
    },
    {
      "epoch": 2.725998598458304,
      "grad_norm": 0.23997937142848969,
      "learning_rate": 2.81702941879933e-05,
      "loss": 0.062,
      "step": 19450
    },
    {
      "epoch": 2.7261387526278904,
      "grad_norm": 0.10570783168077469,
      "learning_rate": 2.815594355417364e-05,
      "loss": 0.0276,
      "step": 19451
    },
    {
      "epoch": 2.7262789067974773,
      "grad_norm": 0.11220171302556992,
      "learning_rate": 2.814159292035398e-05,
      "loss": 0.0216,
      "step": 19452
    },
    {
      "epoch": 2.7264190609670638,
      "grad_norm": 0.31235256791114807,
      "learning_rate": 2.812724228653432e-05,
      "loss": 0.0213,
      "step": 19453
    },
    {
      "epoch": 2.7265592151366502,
      "grad_norm": 0.24962976574897766,
      "learning_rate": 2.811289165271466e-05,
      "loss": 0.0076,
      "step": 19454
    },
    {
      "epoch": 2.726699369306237,
      "grad_norm": 0.18489417433738708,
      "learning_rate": 2.8098541018895e-05,
      "loss": 0.0116,
      "step": 19455
    },
    {
      "epoch": 2.7268395234758236,
      "grad_norm": 0.4950798451900482,
      "learning_rate": 2.8084190385075337e-05,
      "loss": 0.0215,
      "step": 19456
    },
    {
      "epoch": 2.72697967764541,
      "grad_norm": 0.21324247121810913,
      "learning_rate": 2.806983975125568e-05,
      "loss": 0.0071,
      "step": 19457
    },
    {
      "epoch": 2.7271198318149965,
      "grad_norm": 0.18768100440502167,
      "learning_rate": 2.8055489117436018e-05,
      "loss": 0.018,
      "step": 19458
    },
    {
      "epoch": 2.727259985984583,
      "grad_norm": 0.25995275378227234,
      "learning_rate": 2.8041138483616357e-05,
      "loss": 0.0169,
      "step": 19459
    },
    {
      "epoch": 2.7274001401541694,
      "grad_norm": 0.13752292096614838,
      "learning_rate": 2.8026787849796696e-05,
      "loss": 0.0114,
      "step": 19460
    },
    {
      "epoch": 2.727540294323756,
      "grad_norm": 0.11174220591783524,
      "learning_rate": 2.8012437215977038e-05,
      "loss": 0.0064,
      "step": 19461
    },
    {
      "epoch": 2.727680448493343,
      "grad_norm": 0.13980048894882202,
      "learning_rate": 2.7998086582157376e-05,
      "loss": 0.0047,
      "step": 19462
    },
    {
      "epoch": 2.7278206026629293,
      "grad_norm": 0.7803621888160706,
      "learning_rate": 2.7983735948337715e-05,
      "loss": 0.053,
      "step": 19463
    },
    {
      "epoch": 2.7279607568325157,
      "grad_norm": 0.12612923979759216,
      "learning_rate": 2.7969385314518054e-05,
      "loss": 0.0048,
      "step": 19464
    },
    {
      "epoch": 2.728100911002102,
      "grad_norm": 0.08793405443429947,
      "learning_rate": 2.7955034680698396e-05,
      "loss": 0.0038,
      "step": 19465
    },
    {
      "epoch": 2.728241065171689,
      "grad_norm": 0.37049227952957153,
      "learning_rate": 2.7940684046878735e-05,
      "loss": 0.047,
      "step": 19466
    },
    {
      "epoch": 2.7283812193412755,
      "grad_norm": 0.162235289812088,
      "learning_rate": 2.7926333413059074e-05,
      "loss": 0.006,
      "step": 19467
    },
    {
      "epoch": 2.728521373510862,
      "grad_norm": 0.33643218874931335,
      "learning_rate": 2.7911982779239412e-05,
      "loss": 0.0157,
      "step": 19468
    },
    {
      "epoch": 2.7286615276804485,
      "grad_norm": 0.6997774839401245,
      "learning_rate": 2.7897632145419754e-05,
      "loss": 0.0382,
      "step": 19469
    },
    {
      "epoch": 2.728801681850035,
      "grad_norm": 0.3014971613883972,
      "learning_rate": 2.7883281511600093e-05,
      "loss": 0.0365,
      "step": 19470
    },
    {
      "epoch": 2.7289418360196214,
      "grad_norm": 0.1825840175151825,
      "learning_rate": 2.7868930877780435e-05,
      "loss": 0.0711,
      "step": 19471
    },
    {
      "epoch": 2.7290819901892083,
      "grad_norm": 0.21778908371925354,
      "learning_rate": 2.785458024396077e-05,
      "loss": 0.0427,
      "step": 19472
    },
    {
      "epoch": 2.7292221443587947,
      "grad_norm": 0.03799497336149216,
      "learning_rate": 2.7840229610141113e-05,
      "loss": 0.002,
      "step": 19473
    },
    {
      "epoch": 2.729362298528381,
      "grad_norm": 0.4807928502559662,
      "learning_rate": 2.782587897632145e-05,
      "loss": 0.0521,
      "step": 19474
    },
    {
      "epoch": 2.7295024526979677,
      "grad_norm": 0.12992073595523834,
      "learning_rate": 2.7811528342501793e-05,
      "loss": 0.009,
      "step": 19475
    },
    {
      "epoch": 2.7296426068675546,
      "grad_norm": 0.06080075725913048,
      "learning_rate": 2.779717770868213e-05,
      "loss": 0.0053,
      "step": 19476
    },
    {
      "epoch": 2.729782761037141,
      "grad_norm": 0.28789106011390686,
      "learning_rate": 2.778282707486247e-05,
      "loss": 0.0379,
      "step": 19477
    },
    {
      "epoch": 2.7299229152067275,
      "grad_norm": 0.15006516873836517,
      "learning_rate": 2.776847644104281e-05,
      "loss": 0.022,
      "step": 19478
    },
    {
      "epoch": 2.730063069376314,
      "grad_norm": 0.30781152844429016,
      "learning_rate": 2.7754125807223152e-05,
      "loss": 0.0378,
      "step": 19479
    },
    {
      "epoch": 2.7302032235459004,
      "grad_norm": 0.14483267068862915,
      "learning_rate": 2.7739775173403487e-05,
      "loss": 0.0376,
      "step": 19480
    },
    {
      "epoch": 2.730343377715487,
      "grad_norm": 0.19391140341758728,
      "learning_rate": 2.772542453958383e-05,
      "loss": 0.0471,
      "step": 19481
    },
    {
      "epoch": 2.7304835318850733,
      "grad_norm": 0.25918322801589966,
      "learning_rate": 2.7711073905764168e-05,
      "loss": 0.0124,
      "step": 19482
    },
    {
      "epoch": 2.73062368605466,
      "grad_norm": 0.1158665269613266,
      "learning_rate": 2.769672327194451e-05,
      "loss": 0.0037,
      "step": 19483
    },
    {
      "epoch": 2.7307638402242467,
      "grad_norm": 0.22808733582496643,
      "learning_rate": 2.768237263812485e-05,
      "loss": 0.0263,
      "step": 19484
    },
    {
      "epoch": 2.730903994393833,
      "grad_norm": 0.11032707989215851,
      "learning_rate": 2.7668022004305187e-05,
      "loss": 0.005,
      "step": 19485
    },
    {
      "epoch": 2.73104414856342,
      "grad_norm": 0.12492655962705612,
      "learning_rate": 2.765367137048553e-05,
      "loss": 0.0191,
      "step": 19486
    },
    {
      "epoch": 2.7311843027330065,
      "grad_norm": 0.31816771626472473,
      "learning_rate": 2.763932073666587e-05,
      "loss": 0.036,
      "step": 19487
    },
    {
      "epoch": 2.731324456902593,
      "grad_norm": 0.12014422565698624,
      "learning_rate": 2.7624970102846207e-05,
      "loss": 0.0103,
      "step": 19488
    },
    {
      "epoch": 2.7314646110721794,
      "grad_norm": 0.45320242643356323,
      "learning_rate": 2.7610619469026546e-05,
      "loss": 0.0326,
      "step": 19489
    },
    {
      "epoch": 2.731604765241766,
      "grad_norm": 0.6622133851051331,
      "learning_rate": 2.7596268835206888e-05,
      "loss": 0.0164,
      "step": 19490
    },
    {
      "epoch": 2.7317449194113523,
      "grad_norm": 0.08448745310306549,
      "learning_rate": 2.7581918201387227e-05,
      "loss": 0.0118,
      "step": 19491
    },
    {
      "epoch": 2.731885073580939,
      "grad_norm": 0.310040146112442,
      "learning_rate": 2.7567567567567565e-05,
      "loss": 0.0263,
      "step": 19492
    },
    {
      "epoch": 2.7320252277505257,
      "grad_norm": 0.4742746651172638,
      "learning_rate": 2.7553216933747904e-05,
      "loss": 0.053,
      "step": 19493
    },
    {
      "epoch": 2.732165381920112,
      "grad_norm": 0.21682065725326538,
      "learning_rate": 2.7538866299928246e-05,
      "loss": 0.0373,
      "step": 19494
    },
    {
      "epoch": 2.7323055360896986,
      "grad_norm": 0.09455765783786774,
      "learning_rate": 2.7524515666108585e-05,
      "loss": 0.0059,
      "step": 19495
    },
    {
      "epoch": 2.732445690259285,
      "grad_norm": 0.3978339731693268,
      "learning_rate": 2.7510165032288924e-05,
      "loss": 0.0318,
      "step": 19496
    },
    {
      "epoch": 2.732585844428872,
      "grad_norm": 0.10412567853927612,
      "learning_rate": 2.7495814398469262e-05,
      "loss": 0.0036,
      "step": 19497
    },
    {
      "epoch": 2.7327259985984584,
      "grad_norm": 0.8110059499740601,
      "learning_rate": 2.7481463764649604e-05,
      "loss": 0.0743,
      "step": 19498
    },
    {
      "epoch": 2.732866152768045,
      "grad_norm": 0.16204354166984558,
      "learning_rate": 2.7467113130829943e-05,
      "loss": 0.0188,
      "step": 19499
    },
    {
      "epoch": 2.7330063069376314,
      "grad_norm": 0.2942333519458771,
      "learning_rate": 2.7452762497010282e-05,
      "loss": 0.045,
      "step": 19500
    },
    {
      "epoch": 2.733146461107218,
      "grad_norm": 0.18157266080379486,
      "learning_rate": 2.743841186319062e-05,
      "loss": 0.0182,
      "step": 19501
    },
    {
      "epoch": 2.7332866152768043,
      "grad_norm": 0.3237326145172119,
      "learning_rate": 2.7424061229370963e-05,
      "loss": 0.0108,
      "step": 19502
    },
    {
      "epoch": 2.733426769446391,
      "grad_norm": 0.44361165165901184,
      "learning_rate": 2.74097105955513e-05,
      "loss": 0.027,
      "step": 19503
    },
    {
      "epoch": 2.7335669236159776,
      "grad_norm": 0.10324309766292572,
      "learning_rate": 2.739535996173164e-05,
      "loss": 0.0255,
      "step": 19504
    },
    {
      "epoch": 2.733707077785564,
      "grad_norm": 0.11034010350704193,
      "learning_rate": 2.738100932791198e-05,
      "loss": 0.0128,
      "step": 19505
    },
    {
      "epoch": 2.7338472319551506,
      "grad_norm": 0.20787382125854492,
      "learning_rate": 2.736665869409232e-05,
      "loss": 0.0386,
      "step": 19506
    },
    {
      "epoch": 2.7339873861247375,
      "grad_norm": 0.3326590955257416,
      "learning_rate": 2.735230806027266e-05,
      "loss": 0.0183,
      "step": 19507
    },
    {
      "epoch": 2.734127540294324,
      "grad_norm": 0.10108204931020737,
      "learning_rate": 2.7337957426453e-05,
      "loss": 0.0105,
      "step": 19508
    },
    {
      "epoch": 2.7342676944639104,
      "grad_norm": 0.5957496762275696,
      "learning_rate": 2.7323606792633337e-05,
      "loss": 0.0349,
      "step": 19509
    },
    {
      "epoch": 2.734407848633497,
      "grad_norm": 0.5165349245071411,
      "learning_rate": 2.730925615881368e-05,
      "loss": 0.0333,
      "step": 19510
    },
    {
      "epoch": 2.7345480028030833,
      "grad_norm": 0.13919594883918762,
      "learning_rate": 2.7294905524994018e-05,
      "loss": 0.0181,
      "step": 19511
    },
    {
      "epoch": 2.7346881569726698,
      "grad_norm": 0.1279531866312027,
      "learning_rate": 2.7280554891174357e-05,
      "loss": 0.0053,
      "step": 19512
    },
    {
      "epoch": 2.734828311142256,
      "grad_norm": 0.14738792181015015,
      "learning_rate": 2.7266204257354695e-05,
      "loss": 0.0057,
      "step": 19513
    },
    {
      "epoch": 2.734968465311843,
      "grad_norm": 0.25338512659072876,
      "learning_rate": 2.7251853623535038e-05,
      "loss": 0.0196,
      "step": 19514
    },
    {
      "epoch": 2.7351086194814296,
      "grad_norm": 0.8143048286437988,
      "learning_rate": 2.7237502989715376e-05,
      "loss": 0.0313,
      "step": 19515
    },
    {
      "epoch": 2.735248773651016,
      "grad_norm": 0.21566462516784668,
      "learning_rate": 2.7223152355895715e-05,
      "loss": 0.0226,
      "step": 19516
    },
    {
      "epoch": 2.7353889278206025,
      "grad_norm": 0.2890811562538147,
      "learning_rate": 2.7208801722076054e-05,
      "loss": 0.0164,
      "step": 19517
    },
    {
      "epoch": 2.7355290819901894,
      "grad_norm": 0.29207444190979004,
      "learning_rate": 2.7194451088256396e-05,
      "loss": 0.0065,
      "step": 19518
    },
    {
      "epoch": 2.735669236159776,
      "grad_norm": 0.029932493343949318,
      "learning_rate": 2.7180100454436735e-05,
      "loss": 0.002,
      "step": 19519
    },
    {
      "epoch": 2.7358093903293623,
      "grad_norm": 0.5298171639442444,
      "learning_rate": 2.7165749820617073e-05,
      "loss": 0.0271,
      "step": 19520
    },
    {
      "epoch": 2.735949544498949,
      "grad_norm": 0.3158225417137146,
      "learning_rate": 2.7151399186797415e-05,
      "loss": 0.0733,
      "step": 19521
    },
    {
      "epoch": 2.7360896986685352,
      "grad_norm": 0.10514859855175018,
      "learning_rate": 2.7137048552977754e-05,
      "loss": 0.033,
      "step": 19522
    },
    {
      "epoch": 2.7362298528381217,
      "grad_norm": 0.048539139330387115,
      "learning_rate": 2.7122697919158096e-05,
      "loss": 0.0043,
      "step": 19523
    },
    {
      "epoch": 2.7363700070077086,
      "grad_norm": 0.07124984264373779,
      "learning_rate": 2.710834728533843e-05,
      "loss": 0.003,
      "step": 19524
    },
    {
      "epoch": 2.736510161177295,
      "grad_norm": 0.25148114562034607,
      "learning_rate": 2.7093996651518774e-05,
      "loss": 0.0235,
      "step": 19525
    },
    {
      "epoch": 2.7366503153468815,
      "grad_norm": 0.1240663155913353,
      "learning_rate": 2.7079646017699112e-05,
      "loss": 0.013,
      "step": 19526
    },
    {
      "epoch": 2.736790469516468,
      "grad_norm": 0.2573774755001068,
      "learning_rate": 2.7065295383879455e-05,
      "loss": 0.0109,
      "step": 19527
    },
    {
      "epoch": 2.736930623686055,
      "grad_norm": 0.20868085324764252,
      "learning_rate": 2.705094475005979e-05,
      "loss": 0.0175,
      "step": 19528
    },
    {
      "epoch": 2.7370707778556413,
      "grad_norm": 0.21531789004802704,
      "learning_rate": 2.7036594116240132e-05,
      "loss": 0.0249,
      "step": 19529
    },
    {
      "epoch": 2.737210932025228,
      "grad_norm": 0.13642530143260956,
      "learning_rate": 2.702224348242047e-05,
      "loss": 0.0083,
      "step": 19530
    },
    {
      "epoch": 2.7373510861948143,
      "grad_norm": 0.41284242272377014,
      "learning_rate": 2.7007892848600813e-05,
      "loss": 0.0351,
      "step": 19531
    },
    {
      "epoch": 2.7374912403644007,
      "grad_norm": 0.22113879024982452,
      "learning_rate": 2.6993542214781148e-05,
      "loss": 0.0303,
      "step": 19532
    },
    {
      "epoch": 2.737631394533987,
      "grad_norm": 0.11116727441549301,
      "learning_rate": 2.697919158096149e-05,
      "loss": 0.0159,
      "step": 19533
    },
    {
      "epoch": 2.737771548703574,
      "grad_norm": 0.5387252569198608,
      "learning_rate": 2.696484094714183e-05,
      "loss": 0.0781,
      "step": 19534
    },
    {
      "epoch": 2.7379117028731605,
      "grad_norm": 0.371222585439682,
      "learning_rate": 2.695049031332217e-05,
      "loss": 0.0074,
      "step": 19535
    },
    {
      "epoch": 2.738051857042747,
      "grad_norm": 0.07178059965372086,
      "learning_rate": 2.6936139679502506e-05,
      "loss": 0.0063,
      "step": 19536
    },
    {
      "epoch": 2.7381920112123335,
      "grad_norm": 0.17377156019210815,
      "learning_rate": 2.692178904568285e-05,
      "loss": 0.0134,
      "step": 19537
    },
    {
      "epoch": 2.7383321653819204,
      "grad_norm": 0.38265857100486755,
      "learning_rate": 2.6907438411863187e-05,
      "loss": 0.0277,
      "step": 19538
    },
    {
      "epoch": 2.738472319551507,
      "grad_norm": 0.3630073070526123,
      "learning_rate": 2.689308777804353e-05,
      "loss": 0.0433,
      "step": 19539
    },
    {
      "epoch": 2.7386124737210933,
      "grad_norm": 0.21383531391620636,
      "learning_rate": 2.6878737144223865e-05,
      "loss": 0.0236,
      "step": 19540
    },
    {
      "epoch": 2.7387526278906797,
      "grad_norm": 0.07050150632858276,
      "learning_rate": 2.6864386510404207e-05,
      "loss": 0.0222,
      "step": 19541
    },
    {
      "epoch": 2.738892782060266,
      "grad_norm": 0.09447526186704636,
      "learning_rate": 2.6850035876584546e-05,
      "loss": 0.0091,
      "step": 19542
    },
    {
      "epoch": 2.7390329362298527,
      "grad_norm": 0.2507941424846649,
      "learning_rate": 2.6835685242764888e-05,
      "loss": 0.0291,
      "step": 19543
    },
    {
      "epoch": 2.739173090399439,
      "grad_norm": 0.2393520623445511,
      "learning_rate": 2.6821334608945223e-05,
      "loss": 0.0177,
      "step": 19544
    },
    {
      "epoch": 2.739313244569026,
      "grad_norm": 0.013296181336045265,
      "learning_rate": 2.6806983975125565e-05,
      "loss": 0.001,
      "step": 19545
    },
    {
      "epoch": 2.7394533987386125,
      "grad_norm": 0.7778105139732361,
      "learning_rate": 2.6792633341305904e-05,
      "loss": 0.0713,
      "step": 19546
    },
    {
      "epoch": 2.739593552908199,
      "grad_norm": 0.28636693954467773,
      "learning_rate": 2.6778282707486246e-05,
      "loss": 0.0175,
      "step": 19547
    },
    {
      "epoch": 2.7397337070777854,
      "grad_norm": 0.06413032859563828,
      "learning_rate": 2.676393207366658e-05,
      "loss": 0.0094,
      "step": 19548
    },
    {
      "epoch": 2.7398738612473723,
      "grad_norm": 0.17904096841812134,
      "learning_rate": 2.6749581439846923e-05,
      "loss": 0.0315,
      "step": 19549
    },
    {
      "epoch": 2.7400140154169588,
      "grad_norm": 0.19152651727199554,
      "learning_rate": 2.6735230806027262e-05,
      "loss": 0.016,
      "step": 19550
    },
    {
      "epoch": 2.7401541695865452,
      "grad_norm": 0.2007782757282257,
      "learning_rate": 2.6720880172207604e-05,
      "loss": 0.0209,
      "step": 19551
    },
    {
      "epoch": 2.7402943237561317,
      "grad_norm": 0.1621219962835312,
      "learning_rate": 2.670652953838794e-05,
      "loss": 0.0282,
      "step": 19552
    },
    {
      "epoch": 2.740434477925718,
      "grad_norm": 0.2676963210105896,
      "learning_rate": 2.6692178904568282e-05,
      "loss": 0.0234,
      "step": 19553
    },
    {
      "epoch": 2.7405746320953046,
      "grad_norm": 0.16784681379795074,
      "learning_rate": 2.667782827074862e-05,
      "loss": 0.0221,
      "step": 19554
    },
    {
      "epoch": 2.7407147862648915,
      "grad_norm": 0.12461447715759277,
      "learning_rate": 2.6663477636928963e-05,
      "loss": 0.0222,
      "step": 19555
    },
    {
      "epoch": 2.740854940434478,
      "grad_norm": 0.09514413774013519,
      "learning_rate": 2.6649127003109298e-05,
      "loss": 0.0061,
      "step": 19556
    },
    {
      "epoch": 2.7409950946040644,
      "grad_norm": 0.18024393916130066,
      "learning_rate": 2.663477636928964e-05,
      "loss": 0.0421,
      "step": 19557
    },
    {
      "epoch": 2.741135248773651,
      "grad_norm": 0.1847964972257614,
      "learning_rate": 2.6620425735469982e-05,
      "loss": 0.0467,
      "step": 19558
    },
    {
      "epoch": 2.741275402943238,
      "grad_norm": 0.12194634974002838,
      "learning_rate": 2.660607510165032e-05,
      "loss": 0.0069,
      "step": 19559
    },
    {
      "epoch": 2.7414155571128243,
      "grad_norm": 0.07754049450159073,
      "learning_rate": 2.6591724467830663e-05,
      "loss": 0.0153,
      "step": 19560
    },
    {
      "epoch": 2.7415557112824107,
      "grad_norm": 0.14624209702014923,
      "learning_rate": 2.6577373834011e-05,
      "loss": 0.0159,
      "step": 19561
    },
    {
      "epoch": 2.741695865451997,
      "grad_norm": 0.18295606970787048,
      "learning_rate": 2.656302320019134e-05,
      "loss": 0.0112,
      "step": 19562
    },
    {
      "epoch": 2.7418360196215836,
      "grad_norm": 0.15413790941238403,
      "learning_rate": 2.654867256637168e-05,
      "loss": 0.0424,
      "step": 19563
    },
    {
      "epoch": 2.74197617379117,
      "grad_norm": 0.1537507176399231,
      "learning_rate": 2.653432193255202e-05,
      "loss": 0.0227,
      "step": 19564
    },
    {
      "epoch": 2.742116327960757,
      "grad_norm": 0.1616927981376648,
      "learning_rate": 2.6519971298732357e-05,
      "loss": 0.0075,
      "step": 19565
    },
    {
      "epoch": 2.7422564821303435,
      "grad_norm": 0.3945559859275818,
      "learning_rate": 2.65056206649127e-05,
      "loss": 0.0398,
      "step": 19566
    },
    {
      "epoch": 2.74239663629993,
      "grad_norm": 0.45897382497787476,
      "learning_rate": 2.6491270031093037e-05,
      "loss": 0.0563,
      "step": 19567
    },
    {
      "epoch": 2.7425367904695164,
      "grad_norm": 0.3488963842391968,
      "learning_rate": 2.647691939727338e-05,
      "loss": 0.068,
      "step": 19568
    },
    {
      "epoch": 2.7426769446391033,
      "grad_norm": 0.6251013875007629,
      "learning_rate": 2.6462568763453715e-05,
      "loss": 0.0507,
      "step": 19569
    },
    {
      "epoch": 2.7428170988086897,
      "grad_norm": 4.094577789306641,
      "learning_rate": 2.6448218129634057e-05,
      "loss": 0.5689,
      "step": 19570
    },
    {
      "epoch": 2.742957252978276,
      "grad_norm": 0.027309728786349297,
      "learning_rate": 2.6433867495814396e-05,
      "loss": 0.0026,
      "step": 19571
    },
    {
      "epoch": 2.7430974071478627,
      "grad_norm": 0.1594122350215912,
      "learning_rate": 2.6419516861994738e-05,
      "loss": 0.0135,
      "step": 19572
    },
    {
      "epoch": 2.743237561317449,
      "grad_norm": 0.5485442876815796,
      "learning_rate": 2.6405166228175073e-05,
      "loss": 0.0243,
      "step": 19573
    },
    {
      "epoch": 2.7433777154870356,
      "grad_norm": 0.1151149645447731,
      "learning_rate": 2.6390815594355415e-05,
      "loss": 0.0158,
      "step": 19574
    },
    {
      "epoch": 2.743517869656622,
      "grad_norm": 0.2660016119480133,
      "learning_rate": 2.6376464960535754e-05,
      "loss": 0.0292,
      "step": 19575
    },
    {
      "epoch": 2.743658023826209,
      "grad_norm": 0.08330038189888,
      "learning_rate": 2.6362114326716096e-05,
      "loss": 0.0115,
      "step": 19576
    },
    {
      "epoch": 2.7437981779957954,
      "grad_norm": 0.13475893437862396,
      "learning_rate": 2.634776369289643e-05,
      "loss": 0.0287,
      "step": 19577
    },
    {
      "epoch": 2.743938332165382,
      "grad_norm": 0.3418802320957184,
      "learning_rate": 2.6333413059076774e-05,
      "loss": 0.0768,
      "step": 19578
    },
    {
      "epoch": 2.7440784863349683,
      "grad_norm": 0.30021992325782776,
      "learning_rate": 2.6319062425257112e-05,
      "loss": 0.054,
      "step": 19579
    },
    {
      "epoch": 2.744218640504555,
      "grad_norm": 0.5679717659950256,
      "learning_rate": 2.6304711791437454e-05,
      "loss": 0.1236,
      "step": 19580
    },
    {
      "epoch": 2.7443587946741417,
      "grad_norm": 0.23690758645534515,
      "learning_rate": 2.629036115761779e-05,
      "loss": 0.0233,
      "step": 19581
    },
    {
      "epoch": 2.744498948843728,
      "grad_norm": 0.24494487047195435,
      "learning_rate": 2.6276010523798132e-05,
      "loss": 0.0371,
      "step": 19582
    },
    {
      "epoch": 2.7446391030133146,
      "grad_norm": 0.11789540201425552,
      "learning_rate": 2.626165988997847e-05,
      "loss": 0.0199,
      "step": 19583
    },
    {
      "epoch": 2.744779257182901,
      "grad_norm": 0.06834021955728531,
      "learning_rate": 2.6247309256158813e-05,
      "loss": 0.0032,
      "step": 19584
    },
    {
      "epoch": 2.7449194113524875,
      "grad_norm": 0.26725703477859497,
      "learning_rate": 2.6232958622339148e-05,
      "loss": 0.0179,
      "step": 19585
    },
    {
      "epoch": 2.7450595655220744,
      "grad_norm": 0.01768028549849987,
      "learning_rate": 2.621860798851949e-05,
      "loss": 0.0016,
      "step": 19586
    },
    {
      "epoch": 2.745199719691661,
      "grad_norm": 0.049925513565540314,
      "learning_rate": 2.620425735469983e-05,
      "loss": 0.0025,
      "step": 19587
    },
    {
      "epoch": 2.7453398738612473,
      "grad_norm": 0.2147846668958664,
      "learning_rate": 2.618990672088017e-05,
      "loss": 0.0337,
      "step": 19588
    },
    {
      "epoch": 2.745480028030834,
      "grad_norm": 0.37037205696105957,
      "learning_rate": 2.6175556087060506e-05,
      "loss": 0.0367,
      "step": 19589
    },
    {
      "epoch": 2.7456201822004207,
      "grad_norm": 0.08151174336671829,
      "learning_rate": 2.616120545324085e-05,
      "loss": 0.0095,
      "step": 19590
    },
    {
      "epoch": 2.745760336370007,
      "grad_norm": 0.5276138186454773,
      "learning_rate": 2.6146854819421187e-05,
      "loss": 0.0396,
      "step": 19591
    },
    {
      "epoch": 2.7459004905395936,
      "grad_norm": 0.08152516931295395,
      "learning_rate": 2.613250418560153e-05,
      "loss": 0.0076,
      "step": 19592
    },
    {
      "epoch": 2.74604064470918,
      "grad_norm": 0.052304986864328384,
      "learning_rate": 2.6118153551781865e-05,
      "loss": 0.0182,
      "step": 19593
    },
    {
      "epoch": 2.7461807988787665,
      "grad_norm": 0.18098177015781403,
      "learning_rate": 2.6103802917962207e-05,
      "loss": 0.0083,
      "step": 19594
    },
    {
      "epoch": 2.746320953048353,
      "grad_norm": 0.4572729170322418,
      "learning_rate": 2.608945228414255e-05,
      "loss": 0.0622,
      "step": 19595
    },
    {
      "epoch": 2.7464611072179395,
      "grad_norm": 0.153750479221344,
      "learning_rate": 2.6075101650322888e-05,
      "loss": 0.0077,
      "step": 19596
    },
    {
      "epoch": 2.7466012613875264,
      "grad_norm": 0.19995661079883575,
      "learning_rate": 2.606075101650323e-05,
      "loss": 0.0219,
      "step": 19597
    },
    {
      "epoch": 2.746741415557113,
      "grad_norm": 0.19544510543346405,
      "learning_rate": 2.6046400382683565e-05,
      "loss": 0.0139,
      "step": 19598
    },
    {
      "epoch": 2.7468815697266993,
      "grad_norm": 0.14776603877544403,
      "learning_rate": 2.6032049748863907e-05,
      "loss": 0.014,
      "step": 19599
    },
    {
      "epoch": 2.747021723896286,
      "grad_norm": 0.21596087515354156,
      "learning_rate": 2.6017699115044246e-05,
      "loss": 0.0251,
      "step": 19600
    },
    {
      "epoch": 2.7471618780658726,
      "grad_norm": 0.25250551104545593,
      "learning_rate": 2.6003348481224588e-05,
      "loss": 0.0269,
      "step": 19601
    },
    {
      "epoch": 2.747302032235459,
      "grad_norm": 0.11572522670030594,
      "learning_rate": 2.5988997847404923e-05,
      "loss": 0.0182,
      "step": 19602
    },
    {
      "epoch": 2.7474421864050456,
      "grad_norm": 0.22012758255004883,
      "learning_rate": 2.5974647213585265e-05,
      "loss": 0.0333,
      "step": 19603
    },
    {
      "epoch": 2.747582340574632,
      "grad_norm": 0.23778599500656128,
      "learning_rate": 2.5960296579765604e-05,
      "loss": 0.0196,
      "step": 19604
    },
    {
      "epoch": 2.7477224947442185,
      "grad_norm": 0.11386556923389435,
      "learning_rate": 2.5945945945945946e-05,
      "loss": 0.0332,
      "step": 19605
    },
    {
      "epoch": 2.747862648913805,
      "grad_norm": 0.42700856924057007,
      "learning_rate": 2.593159531212628e-05,
      "loss": 0.0526,
      "step": 19606
    },
    {
      "epoch": 2.748002803083392,
      "grad_norm": 0.18638870120048523,
      "learning_rate": 2.5917244678306624e-05,
      "loss": 0.0254,
      "step": 19607
    },
    {
      "epoch": 2.7481429572529783,
      "grad_norm": 0.1494918018579483,
      "learning_rate": 2.5902894044486962e-05,
      "loss": 0.0355,
      "step": 19608
    },
    {
      "epoch": 2.7482831114225648,
      "grad_norm": 0.20779912173748016,
      "learning_rate": 2.5888543410667305e-05,
      "loss": 0.0084,
      "step": 19609
    },
    {
      "epoch": 2.748423265592151,
      "grad_norm": 0.20620432496070862,
      "learning_rate": 2.587419277684764e-05,
      "loss": 0.0187,
      "step": 19610
    },
    {
      "epoch": 2.748563419761738,
      "grad_norm": 0.43751242756843567,
      "learning_rate": 2.5859842143027982e-05,
      "loss": 0.043,
      "step": 19611
    },
    {
      "epoch": 2.7487035739313246,
      "grad_norm": 0.22197231650352478,
      "learning_rate": 2.584549150920832e-05,
      "loss": 0.0135,
      "step": 19612
    },
    {
      "epoch": 2.748843728100911,
      "grad_norm": 0.26453399658203125,
      "learning_rate": 2.5831140875388663e-05,
      "loss": 0.0148,
      "step": 19613
    },
    {
      "epoch": 2.7489838822704975,
      "grad_norm": 0.2887434661388397,
      "learning_rate": 2.5816790241568998e-05,
      "loss": 0.0331,
      "step": 19614
    },
    {
      "epoch": 2.749124036440084,
      "grad_norm": 0.03805162012577057,
      "learning_rate": 2.580243960774934e-05,
      "loss": 0.002,
      "step": 19615
    },
    {
      "epoch": 2.7492641906096704,
      "grad_norm": 0.010262798517942429,
      "learning_rate": 2.578808897392968e-05,
      "loss": 0.001,
      "step": 19616
    },
    {
      "epoch": 2.7494043447792573,
      "grad_norm": 0.2911091148853302,
      "learning_rate": 2.577373834011002e-05,
      "loss": 0.0155,
      "step": 19617
    },
    {
      "epoch": 2.749544498948844,
      "grad_norm": 0.1620534211397171,
      "learning_rate": 2.5759387706290356e-05,
      "loss": 0.0367,
      "step": 19618
    },
    {
      "epoch": 2.7496846531184302,
      "grad_norm": 1.4123246669769287,
      "learning_rate": 2.57450370724707e-05,
      "loss": 0.0814,
      "step": 19619
    },
    {
      "epoch": 2.7498248072880167,
      "grad_norm": 0.48660263419151306,
      "learning_rate": 2.5730686438651037e-05,
      "loss": 0.0499,
      "step": 19620
    },
    {
      "epoch": 2.7499649614576036,
      "grad_norm": 0.3663142919540405,
      "learning_rate": 2.571633580483138e-05,
      "loss": 0.0244,
      "step": 19621
    },
    {
      "epoch": 2.75010511562719,
      "grad_norm": 0.05115147680044174,
      "learning_rate": 2.5701985171011715e-05,
      "loss": 0.0036,
      "step": 19622
    },
    {
      "epoch": 2.7502452697967765,
      "grad_norm": 0.042254868894815445,
      "learning_rate": 2.5687634537192057e-05,
      "loss": 0.0036,
      "step": 19623
    },
    {
      "epoch": 2.750385423966363,
      "grad_norm": 0.203120157122612,
      "learning_rate": 2.5673283903372396e-05,
      "loss": 0.0615,
      "step": 19624
    },
    {
      "epoch": 2.7505255781359494,
      "grad_norm": 0.37443286180496216,
      "learning_rate": 2.5658933269552738e-05,
      "loss": 0.0494,
      "step": 19625
    },
    {
      "epoch": 2.750665732305536,
      "grad_norm": 0.09348617494106293,
      "learning_rate": 2.5644582635733073e-05,
      "loss": 0.0104,
      "step": 19626
    },
    {
      "epoch": 2.7508058864751224,
      "grad_norm": 0.14447706937789917,
      "learning_rate": 2.5630232001913415e-05,
      "loss": 0.0166,
      "step": 19627
    },
    {
      "epoch": 2.7509460406447093,
      "grad_norm": 0.5320804119110107,
      "learning_rate": 2.5615881368093754e-05,
      "loss": 0.0558,
      "step": 19628
    },
    {
      "epoch": 2.7510861948142957,
      "grad_norm": 0.16138537228107452,
      "learning_rate": 2.5601530734274096e-05,
      "loss": 0.0102,
      "step": 19629
    },
    {
      "epoch": 2.751226348983882,
      "grad_norm": 0.11705881357192993,
      "learning_rate": 2.558718010045443e-05,
      "loss": 0.0167,
      "step": 19630
    },
    {
      "epoch": 2.751366503153469,
      "grad_norm": 0.3916154205799103,
      "learning_rate": 2.5572829466634773e-05,
      "loss": 0.0172,
      "step": 19631
    },
    {
      "epoch": 2.7515066573230555,
      "grad_norm": 0.09149889647960663,
      "learning_rate": 2.5558478832815116e-05,
      "loss": 0.0075,
      "step": 19632
    },
    {
      "epoch": 2.751646811492642,
      "grad_norm": 0.09864673763513565,
      "learning_rate": 2.5544128198995454e-05,
      "loss": 0.0088,
      "step": 19633
    },
    {
      "epoch": 2.7517869656622285,
      "grad_norm": 0.19561338424682617,
      "learning_rate": 2.5529777565175796e-05,
      "loss": 0.013,
      "step": 19634
    },
    {
      "epoch": 2.751927119831815,
      "grad_norm": 0.1205207109451294,
      "learning_rate": 2.551542693135613e-05,
      "loss": 0.0184,
      "step": 19635
    },
    {
      "epoch": 2.7520672740014014,
      "grad_norm": 0.08013292402029037,
      "learning_rate": 2.5501076297536474e-05,
      "loss": 0.0057,
      "step": 19636
    },
    {
      "epoch": 2.752207428170988,
      "grad_norm": 0.06723947077989578,
      "learning_rate": 2.5486725663716813e-05,
      "loss": 0.0068,
      "step": 19637
    },
    {
      "epoch": 2.7523475823405747,
      "grad_norm": 0.13616110384464264,
      "learning_rate": 2.5472375029897155e-05,
      "loss": 0.0211,
      "step": 19638
    },
    {
      "epoch": 2.752487736510161,
      "grad_norm": 0.10861470550298691,
      "learning_rate": 2.545802439607749e-05,
      "loss": 0.0091,
      "step": 19639
    },
    {
      "epoch": 2.7526278906797477,
      "grad_norm": 0.7247406244277954,
      "learning_rate": 2.5443673762257832e-05,
      "loss": 0.0639,
      "step": 19640
    },
    {
      "epoch": 2.752768044849334,
      "grad_norm": 0.12205429375171661,
      "learning_rate": 2.542932312843817e-05,
      "loss": 0.0163,
      "step": 19641
    },
    {
      "epoch": 2.752908199018921,
      "grad_norm": 0.2109237015247345,
      "learning_rate": 2.5414972494618513e-05,
      "loss": 0.0303,
      "step": 19642
    },
    {
      "epoch": 2.7530483531885075,
      "grad_norm": 0.23513998091220856,
      "learning_rate": 2.5400621860798848e-05,
      "loss": 0.0118,
      "step": 19643
    },
    {
      "epoch": 2.753188507358094,
      "grad_norm": 0.07528147101402283,
      "learning_rate": 2.538627122697919e-05,
      "loss": 0.0023,
      "step": 19644
    },
    {
      "epoch": 2.7533286615276804,
      "grad_norm": 0.14794763922691345,
      "learning_rate": 2.537192059315953e-05,
      "loss": 0.0041,
      "step": 19645
    },
    {
      "epoch": 2.753468815697267,
      "grad_norm": 0.2268216907978058,
      "learning_rate": 2.535756995933987e-05,
      "loss": 0.0253,
      "step": 19646
    },
    {
      "epoch": 2.7536089698668533,
      "grad_norm": 0.3463696539402008,
      "learning_rate": 2.5343219325520207e-05,
      "loss": 0.0326,
      "step": 19647
    },
    {
      "epoch": 2.7537491240364402,
      "grad_norm": 0.09093203395605087,
      "learning_rate": 2.532886869170055e-05,
      "loss": 0.0234,
      "step": 19648
    },
    {
      "epoch": 2.7538892782060267,
      "grad_norm": 0.311085045337677,
      "learning_rate": 2.5314518057880887e-05,
      "loss": 0.0422,
      "step": 19649
    },
    {
      "epoch": 2.754029432375613,
      "grad_norm": 0.05444694310426712,
      "learning_rate": 2.530016742406123e-05,
      "loss": 0.0051,
      "step": 19650
    },
    {
      "epoch": 2.7541695865451996,
      "grad_norm": 0.20184659957885742,
      "learning_rate": 2.5285816790241565e-05,
      "loss": 0.0106,
      "step": 19651
    },
    {
      "epoch": 2.7543097407147865,
      "grad_norm": 0.1727100908756256,
      "learning_rate": 2.5271466156421907e-05,
      "loss": 0.0161,
      "step": 19652
    },
    {
      "epoch": 2.754449894884373,
      "grad_norm": 0.22122488915920258,
      "learning_rate": 2.5257115522602246e-05,
      "loss": 0.0396,
      "step": 19653
    },
    {
      "epoch": 2.7545900490539594,
      "grad_norm": 0.22743038833141327,
      "learning_rate": 2.5242764888782588e-05,
      "loss": 0.0196,
      "step": 19654
    },
    {
      "epoch": 2.754730203223546,
      "grad_norm": 0.14631937444210052,
      "learning_rate": 2.5228414254962923e-05,
      "loss": 0.0208,
      "step": 19655
    },
    {
      "epoch": 2.7548703573931324,
      "grad_norm": 0.12212645262479782,
      "learning_rate": 2.5214063621143265e-05,
      "loss": 0.0037,
      "step": 19656
    },
    {
      "epoch": 2.755010511562719,
      "grad_norm": 0.6375158429145813,
      "learning_rate": 2.5199712987323604e-05,
      "loss": 0.0292,
      "step": 19657
    },
    {
      "epoch": 2.7551506657323053,
      "grad_norm": 0.2148681879043579,
      "learning_rate": 2.5185362353503946e-05,
      "loss": 0.0342,
      "step": 19658
    },
    {
      "epoch": 2.755290819901892,
      "grad_norm": 0.6318073868751526,
      "learning_rate": 2.517101171968428e-05,
      "loss": 0.0207,
      "step": 19659
    },
    {
      "epoch": 2.7554309740714786,
      "grad_norm": 0.26880159974098206,
      "learning_rate": 2.5156661085864624e-05,
      "loss": 0.0309,
      "step": 19660
    },
    {
      "epoch": 2.755571128241065,
      "grad_norm": 0.11923572421073914,
      "learning_rate": 2.5142310452044962e-05,
      "loss": 0.0184,
      "step": 19661
    },
    {
      "epoch": 2.7557112824106516,
      "grad_norm": 0.8590844869613647,
      "learning_rate": 2.5127959818225304e-05,
      "loss": 0.0572,
      "step": 19662
    },
    {
      "epoch": 2.7558514365802385,
      "grad_norm": 0.1506556272506714,
      "learning_rate": 2.511360918440564e-05,
      "loss": 0.0048,
      "step": 19663
    },
    {
      "epoch": 2.755991590749825,
      "grad_norm": 0.12805749475955963,
      "learning_rate": 2.5099258550585982e-05,
      "loss": 0.0102,
      "step": 19664
    },
    {
      "epoch": 2.7561317449194114,
      "grad_norm": 0.43791666626930237,
      "learning_rate": 2.508490791676632e-05,
      "loss": 0.034,
      "step": 19665
    },
    {
      "epoch": 2.756271899088998,
      "grad_norm": 1.111950397491455,
      "learning_rate": 2.5070557282946663e-05,
      "loss": 0.0515,
      "step": 19666
    },
    {
      "epoch": 2.7564120532585843,
      "grad_norm": 0.5381291508674622,
      "learning_rate": 2.5056206649126998e-05,
      "loss": 0.076,
      "step": 19667
    },
    {
      "epoch": 2.7565522074281708,
      "grad_norm": 0.3691125810146332,
      "learning_rate": 2.504185601530734e-05,
      "loss": 0.0138,
      "step": 19668
    },
    {
      "epoch": 2.7566923615977577,
      "grad_norm": 0.31792792677879333,
      "learning_rate": 2.5027505381487682e-05,
      "loss": 0.022,
      "step": 19669
    },
    {
      "epoch": 2.756832515767344,
      "grad_norm": 0.9472073912620544,
      "learning_rate": 2.501315474766802e-05,
      "loss": 0.0327,
      "step": 19670
    },
    {
      "epoch": 2.7569726699369306,
      "grad_norm": 0.02514302171766758,
      "learning_rate": 2.4998804113848363e-05,
      "loss": 0.0025,
      "step": 19671
    },
    {
      "epoch": 2.757112824106517,
      "grad_norm": 0.14029261469841003,
      "learning_rate": 2.49844534800287e-05,
      "loss": 0.0085,
      "step": 19672
    },
    {
      "epoch": 2.757252978276104,
      "grad_norm": 0.2055148184299469,
      "learning_rate": 2.497010284620904e-05,
      "loss": 0.011,
      "step": 19673
    },
    {
      "epoch": 2.7573931324456904,
      "grad_norm": 0.40623387694358826,
      "learning_rate": 2.495575221238938e-05,
      "loss": 0.0211,
      "step": 19674
    },
    {
      "epoch": 2.757533286615277,
      "grad_norm": 0.1334078311920166,
      "learning_rate": 2.494140157856972e-05,
      "loss": 0.0162,
      "step": 19675
    },
    {
      "epoch": 2.7576734407848633,
      "grad_norm": 0.21310746669769287,
      "learning_rate": 2.4927050944750057e-05,
      "loss": 0.0188,
      "step": 19676
    },
    {
      "epoch": 2.7578135949544498,
      "grad_norm": 0.21902373433113098,
      "learning_rate": 2.49127003109304e-05,
      "loss": 0.0439,
      "step": 19677
    },
    {
      "epoch": 2.7579537491240362,
      "grad_norm": 0.1510096937417984,
      "learning_rate": 2.4898349677110738e-05,
      "loss": 0.0392,
      "step": 19678
    },
    {
      "epoch": 2.758093903293623,
      "grad_norm": 0.13296210765838623,
      "learning_rate": 2.488399904329108e-05,
      "loss": 0.0029,
      "step": 19679
    },
    {
      "epoch": 2.7582340574632096,
      "grad_norm": 0.08638874441385269,
      "learning_rate": 2.4869648409471415e-05,
      "loss": 0.0075,
      "step": 19680
    },
    {
      "epoch": 2.758374211632796,
      "grad_norm": 0.543943464756012,
      "learning_rate": 2.4855297775651757e-05,
      "loss": 0.0472,
      "step": 19681
    },
    {
      "epoch": 2.7585143658023825,
      "grad_norm": 0.18065424263477325,
      "learning_rate": 2.4840947141832096e-05,
      "loss": 0.0141,
      "step": 19682
    },
    {
      "epoch": 2.7586545199719694,
      "grad_norm": 0.203375443816185,
      "learning_rate": 2.4826596508012438e-05,
      "loss": 0.005,
      "step": 19683
    },
    {
      "epoch": 2.758794674141556,
      "grad_norm": 0.10661984980106354,
      "learning_rate": 2.4812245874192773e-05,
      "loss": 0.0232,
      "step": 19684
    },
    {
      "epoch": 2.7589348283111423,
      "grad_norm": 0.15223032236099243,
      "learning_rate": 2.4797895240373115e-05,
      "loss": 0.0329,
      "step": 19685
    },
    {
      "epoch": 2.759074982480729,
      "grad_norm": 0.3048984408378601,
      "learning_rate": 2.4783544606553454e-05,
      "loss": 0.0201,
      "step": 19686
    },
    {
      "epoch": 2.7592151366503153,
      "grad_norm": 0.15029452741146088,
      "learning_rate": 2.4769193972733796e-05,
      "loss": 0.0088,
      "step": 19687
    },
    {
      "epoch": 2.7593552908199017,
      "grad_norm": 0.1745430827140808,
      "learning_rate": 2.475484333891413e-05,
      "loss": 0.0177,
      "step": 19688
    },
    {
      "epoch": 2.759495444989488,
      "grad_norm": 0.06062167137861252,
      "learning_rate": 2.4740492705094474e-05,
      "loss": 0.0073,
      "step": 19689
    },
    {
      "epoch": 2.759635599159075,
      "grad_norm": 0.28899627923965454,
      "learning_rate": 2.4726142071274812e-05,
      "loss": 0.0242,
      "step": 19690
    },
    {
      "epoch": 2.7597757533286615,
      "grad_norm": 0.1996884047985077,
      "learning_rate": 2.4711791437455154e-05,
      "loss": 0.0511,
      "step": 19691
    },
    {
      "epoch": 2.759915907498248,
      "grad_norm": 0.13133123517036438,
      "learning_rate": 2.469744080363549e-05,
      "loss": 0.031,
      "step": 19692
    },
    {
      "epoch": 2.7600560616678345,
      "grad_norm": 0.274278849363327,
      "learning_rate": 2.4683090169815832e-05,
      "loss": 0.0329,
      "step": 19693
    },
    {
      "epoch": 2.7601962158374214,
      "grad_norm": 0.6064431071281433,
      "learning_rate": 2.466873953599617e-05,
      "loss": 0.0255,
      "step": 19694
    },
    {
      "epoch": 2.760336370007008,
      "grad_norm": 0.9422585368156433,
      "learning_rate": 2.4654388902176513e-05,
      "loss": 0.0993,
      "step": 19695
    },
    {
      "epoch": 2.7604765241765943,
      "grad_norm": 0.3376786708831787,
      "learning_rate": 2.4640038268356848e-05,
      "loss": 0.0288,
      "step": 19696
    },
    {
      "epoch": 2.7606166783461807,
      "grad_norm": 0.1295783519744873,
      "learning_rate": 2.462568763453719e-05,
      "loss": 0.0093,
      "step": 19697
    },
    {
      "epoch": 2.760756832515767,
      "grad_norm": 0.18326438963413239,
      "learning_rate": 2.461133700071753e-05,
      "loss": 0.0136,
      "step": 19698
    },
    {
      "epoch": 2.7608969866853537,
      "grad_norm": 0.05002005025744438,
      "learning_rate": 2.459698636689787e-05,
      "loss": 0.0052,
      "step": 19699
    },
    {
      "epoch": 2.7610371408549406,
      "grad_norm": 0.29818522930145264,
      "learning_rate": 2.4582635733078206e-05,
      "loss": 0.0064,
      "step": 19700
    },
    {
      "epoch": 2.761177295024527,
      "grad_norm": 0.2192409634590149,
      "learning_rate": 2.456828509925855e-05,
      "loss": 0.0128,
      "step": 19701
    },
    {
      "epoch": 2.7613174491941135,
      "grad_norm": 0.10857821255922318,
      "learning_rate": 2.4553934465438887e-05,
      "loss": 0.0139,
      "step": 19702
    },
    {
      "epoch": 2.7614576033637,
      "grad_norm": 0.047642312943935394,
      "learning_rate": 2.453958383161923e-05,
      "loss": 0.0023,
      "step": 19703
    },
    {
      "epoch": 2.761597757533287,
      "grad_norm": 0.11918731033802032,
      "learning_rate": 2.4525233197799565e-05,
      "loss": 0.021,
      "step": 19704
    },
    {
      "epoch": 2.7617379117028733,
      "grad_norm": 0.14333125948905945,
      "learning_rate": 2.4510882563979907e-05,
      "loss": 0.0241,
      "step": 19705
    },
    {
      "epoch": 2.7618780658724598,
      "grad_norm": 0.03244173899292946,
      "learning_rate": 2.449653193016025e-05,
      "loss": 0.0016,
      "step": 19706
    },
    {
      "epoch": 2.7620182200420462,
      "grad_norm": 0.4298053979873657,
      "learning_rate": 2.4482181296340588e-05,
      "loss": 0.0354,
      "step": 19707
    },
    {
      "epoch": 2.7621583742116327,
      "grad_norm": 0.3704368472099304,
      "learning_rate": 2.4467830662520926e-05,
      "loss": 0.039,
      "step": 19708
    },
    {
      "epoch": 2.762298528381219,
      "grad_norm": 0.12969982624053955,
      "learning_rate": 2.4453480028701265e-05,
      "loss": 0.0095,
      "step": 19709
    },
    {
      "epoch": 2.762438682550806,
      "grad_norm": 0.1285260021686554,
      "learning_rate": 2.4439129394881607e-05,
      "loss": 0.0077,
      "step": 19710
    },
    {
      "epoch": 2.7625788367203925,
      "grad_norm": 0.07477578520774841,
      "learning_rate": 2.4424778761061946e-05,
      "loss": 0.0063,
      "step": 19711
    },
    {
      "epoch": 2.762718990889979,
      "grad_norm": 0.9807720184326172,
      "learning_rate": 2.4410428127242285e-05,
      "loss": 0.0771,
      "step": 19712
    },
    {
      "epoch": 2.7628591450595654,
      "grad_norm": 0.09606026113033295,
      "learning_rate": 2.4396077493422623e-05,
      "loss": 0.0191,
      "step": 19713
    },
    {
      "epoch": 2.7629992992291523,
      "grad_norm": 0.027572304010391235,
      "learning_rate": 2.4381726859602965e-05,
      "loss": 0.0023,
      "step": 19714
    },
    {
      "epoch": 2.763139453398739,
      "grad_norm": 0.10843207687139511,
      "learning_rate": 2.4367376225783304e-05,
      "loss": 0.0062,
      "step": 19715
    },
    {
      "epoch": 2.7632796075683252,
      "grad_norm": 0.5103393197059631,
      "learning_rate": 2.4353025591963643e-05,
      "loss": 0.0377,
      "step": 19716
    },
    {
      "epoch": 2.7634197617379117,
      "grad_norm": 0.15288607776165009,
      "learning_rate": 2.433867495814398e-05,
      "loss": 0.0122,
      "step": 19717
    },
    {
      "epoch": 2.763559915907498,
      "grad_norm": 0.3308166265487671,
      "learning_rate": 2.4324324324324324e-05,
      "loss": 0.0082,
      "step": 19718
    },
    {
      "epoch": 2.7637000700770846,
      "grad_norm": 0.6372894644737244,
      "learning_rate": 2.4309973690504662e-05,
      "loss": 0.0976,
      "step": 19719
    },
    {
      "epoch": 2.763840224246671,
      "grad_norm": 0.4351816475391388,
      "learning_rate": 2.4295623056685e-05,
      "loss": 0.0325,
      "step": 19720
    },
    {
      "epoch": 2.763980378416258,
      "grad_norm": 0.7362777590751648,
      "learning_rate": 2.428127242286534e-05,
      "loss": 0.1098,
      "step": 19721
    },
    {
      "epoch": 2.7641205325858444,
      "grad_norm": 0.24201729893684387,
      "learning_rate": 2.4266921789045682e-05,
      "loss": 0.025,
      "step": 19722
    },
    {
      "epoch": 2.764260686755431,
      "grad_norm": 0.171593576669693,
      "learning_rate": 2.425257115522602e-05,
      "loss": 0.0139,
      "step": 19723
    },
    {
      "epoch": 2.7644008409250174,
      "grad_norm": 0.25515106320381165,
      "learning_rate": 2.423822052140636e-05,
      "loss": 0.0599,
      "step": 19724
    },
    {
      "epoch": 2.7645409950946043,
      "grad_norm": 0.19472874701023102,
      "learning_rate": 2.4223869887586698e-05,
      "loss": 0.0332,
      "step": 19725
    },
    {
      "epoch": 2.7646811492641907,
      "grad_norm": 0.10033785551786423,
      "learning_rate": 2.420951925376704e-05,
      "loss": 0.0111,
      "step": 19726
    },
    {
      "epoch": 2.764821303433777,
      "grad_norm": 0.36555203795433044,
      "learning_rate": 2.419516861994738e-05,
      "loss": 0.0106,
      "step": 19727
    },
    {
      "epoch": 2.7649614576033636,
      "grad_norm": 0.26554426550865173,
      "learning_rate": 2.4180817986127718e-05,
      "loss": 0.0148,
      "step": 19728
    },
    {
      "epoch": 2.76510161177295,
      "grad_norm": 0.23013059794902802,
      "learning_rate": 2.4166467352308056e-05,
      "loss": 0.0356,
      "step": 19729
    },
    {
      "epoch": 2.7652417659425366,
      "grad_norm": 0.16837471723556519,
      "learning_rate": 2.41521167184884e-05,
      "loss": 0.018,
      "step": 19730
    },
    {
      "epoch": 2.7653819201121235,
      "grad_norm": 0.44855231046676636,
      "learning_rate": 2.4137766084668737e-05,
      "loss": 0.0143,
      "step": 19731
    },
    {
      "epoch": 2.76552207428171,
      "grad_norm": 0.13251405954360962,
      "learning_rate": 2.4123415450849076e-05,
      "loss": 0.0071,
      "step": 19732
    },
    {
      "epoch": 2.7656622284512964,
      "grad_norm": 0.1331344097852707,
      "learning_rate": 2.4109064817029415e-05,
      "loss": 0.0239,
      "step": 19733
    },
    {
      "epoch": 2.765802382620883,
      "grad_norm": 0.08691603690385818,
      "learning_rate": 2.4094714183209757e-05,
      "loss": 0.0061,
      "step": 19734
    },
    {
      "epoch": 2.7659425367904698,
      "grad_norm": 0.2996131479740143,
      "learning_rate": 2.4080363549390096e-05,
      "loss": 0.0088,
      "step": 19735
    },
    {
      "epoch": 2.766082690960056,
      "grad_norm": 0.11712979525327682,
      "learning_rate": 2.4066012915570434e-05,
      "loss": 0.0281,
      "step": 19736
    },
    {
      "epoch": 2.7662228451296427,
      "grad_norm": 0.15780794620513916,
      "learning_rate": 2.4051662281750773e-05,
      "loss": 0.0216,
      "step": 19737
    },
    {
      "epoch": 2.766362999299229,
      "grad_norm": 0.28268617391586304,
      "learning_rate": 2.4037311647931115e-05,
      "loss": 0.0268,
      "step": 19738
    },
    {
      "epoch": 2.7665031534688156,
      "grad_norm": 0.2745468020439148,
      "learning_rate": 2.4022961014111454e-05,
      "loss": 0.0127,
      "step": 19739
    },
    {
      "epoch": 2.766643307638402,
      "grad_norm": 0.12582933902740479,
      "learning_rate": 2.4008610380291793e-05,
      "loss": 0.0155,
      "step": 19740
    },
    {
      "epoch": 2.766783461807989,
      "grad_norm": 0.07017385214567184,
      "learning_rate": 2.399425974647213e-05,
      "loss": 0.0069,
      "step": 19741
    },
    {
      "epoch": 2.7669236159775754,
      "grad_norm": 0.22924700379371643,
      "learning_rate": 2.3979909112652473e-05,
      "loss": 0.0289,
      "step": 19742
    },
    {
      "epoch": 2.767063770147162,
      "grad_norm": 0.26382994651794434,
      "learning_rate": 2.3965558478832816e-05,
      "loss": 0.012,
      "step": 19743
    },
    {
      "epoch": 2.7672039243167483,
      "grad_norm": 0.1538844257593155,
      "learning_rate": 2.395120784501315e-05,
      "loss": 0.0222,
      "step": 19744
    },
    {
      "epoch": 2.7673440784863352,
      "grad_norm": 0.12456761300563812,
      "learning_rate": 2.3936857211193493e-05,
      "loss": 0.0084,
      "step": 19745
    },
    {
      "epoch": 2.7674842326559217,
      "grad_norm": 0.11433926969766617,
      "learning_rate": 2.3922506577373832e-05,
      "loss": 0.0156,
      "step": 19746
    },
    {
      "epoch": 2.767624386825508,
      "grad_norm": 0.037178389728069305,
      "learning_rate": 2.3908155943554174e-05,
      "loss": 0.0023,
      "step": 19747
    },
    {
      "epoch": 2.7677645409950946,
      "grad_norm": 0.12188324332237244,
      "learning_rate": 2.389380530973451e-05,
      "loss": 0.0287,
      "step": 19748
    },
    {
      "epoch": 2.767904695164681,
      "grad_norm": 0.46723467111587524,
      "learning_rate": 2.387945467591485e-05,
      "loss": 0.0385,
      "step": 19749
    },
    {
      "epoch": 2.7680448493342675,
      "grad_norm": 0.0709342509508133,
      "learning_rate": 2.386510404209519e-05,
      "loss": 0.0055,
      "step": 19750
    },
    {
      "epoch": 2.768185003503854,
      "grad_norm": 0.17604003846645355,
      "learning_rate": 2.3850753408275532e-05,
      "loss": 0.0237,
      "step": 19751
    },
    {
      "epoch": 2.768325157673441,
      "grad_norm": 0.21704339981079102,
      "learning_rate": 2.3836402774455867e-05,
      "loss": 0.0267,
      "step": 19752
    },
    {
      "epoch": 2.7684653118430274,
      "grad_norm": 0.09651315957307816,
      "learning_rate": 2.382205214063621e-05,
      "loss": 0.007,
      "step": 19753
    },
    {
      "epoch": 2.768605466012614,
      "grad_norm": 0.14743298292160034,
      "learning_rate": 2.380770150681655e-05,
      "loss": 0.0199,
      "step": 19754
    },
    {
      "epoch": 2.7687456201822003,
      "grad_norm": 0.20465560257434845,
      "learning_rate": 2.379335087299689e-05,
      "loss": 0.0086,
      "step": 19755
    },
    {
      "epoch": 2.768885774351787,
      "grad_norm": 0.1472039520740509,
      "learning_rate": 2.3779000239177226e-05,
      "loss": 0.0346,
      "step": 19756
    },
    {
      "epoch": 2.7690259285213736,
      "grad_norm": 0.02799607440829277,
      "learning_rate": 2.3764649605357568e-05,
      "loss": 0.0014,
      "step": 19757
    },
    {
      "epoch": 2.76916608269096,
      "grad_norm": 0.3255635201931,
      "learning_rate": 2.3750298971537907e-05,
      "loss": 0.0657,
      "step": 19758
    },
    {
      "epoch": 2.7693062368605466,
      "grad_norm": 0.3716498911380768,
      "learning_rate": 2.373594833771825e-05,
      "loss": 0.0741,
      "step": 19759
    },
    {
      "epoch": 2.769446391030133,
      "grad_norm": 0.10132478922605515,
      "learning_rate": 2.3721597703898587e-05,
      "loss": 0.0133,
      "step": 19760
    },
    {
      "epoch": 2.7695865451997195,
      "grad_norm": 0.09269966185092926,
      "learning_rate": 2.3707247070078926e-05,
      "loss": 0.007,
      "step": 19761
    },
    {
      "epoch": 2.7697266993693064,
      "grad_norm": 0.26648542284965515,
      "learning_rate": 2.3692896436259265e-05,
      "loss": 0.0169,
      "step": 19762
    },
    {
      "epoch": 2.769866853538893,
      "grad_norm": 0.14521369338035583,
      "learning_rate": 2.3678545802439607e-05,
      "loss": 0.0173,
      "step": 19763
    },
    {
      "epoch": 2.7700070077084793,
      "grad_norm": 0.09749699383974075,
      "learning_rate": 2.3664195168619946e-05,
      "loss": 0.0077,
      "step": 19764
    },
    {
      "epoch": 2.7701471618780658,
      "grad_norm": 0.32288554310798645,
      "learning_rate": 2.3649844534800284e-05,
      "loss": 0.0511,
      "step": 19765
    },
    {
      "epoch": 2.7702873160476527,
      "grad_norm": 0.03599060699343681,
      "learning_rate": 2.3635493900980623e-05,
      "loss": 0.003,
      "step": 19766
    },
    {
      "epoch": 2.770427470217239,
      "grad_norm": 1.1522167921066284,
      "learning_rate": 2.3621143267160965e-05,
      "loss": 0.0552,
      "step": 19767
    },
    {
      "epoch": 2.7705676243868256,
      "grad_norm": 0.09315192699432373,
      "learning_rate": 2.3606792633341304e-05,
      "loss": 0.0053,
      "step": 19768
    },
    {
      "epoch": 2.770707778556412,
      "grad_norm": 0.019055582582950592,
      "learning_rate": 2.3592441999521643e-05,
      "loss": 0.001,
      "step": 19769
    },
    {
      "epoch": 2.7708479327259985,
      "grad_norm": 0.11070076376199722,
      "learning_rate": 2.357809136570198e-05,
      "loss": 0.0045,
      "step": 19770
    },
    {
      "epoch": 2.770988086895585,
      "grad_norm": 0.06582643836736679,
      "learning_rate": 2.3563740731882324e-05,
      "loss": 0.0023,
      "step": 19771
    },
    {
      "epoch": 2.7711282410651714,
      "grad_norm": 0.3027840852737427,
      "learning_rate": 2.3549390098062662e-05,
      "loss": 0.0275,
      "step": 19772
    },
    {
      "epoch": 2.7712683952347583,
      "grad_norm": 0.09555289149284363,
      "learning_rate": 2.3535039464243e-05,
      "loss": 0.0163,
      "step": 19773
    },
    {
      "epoch": 2.771408549404345,
      "grad_norm": 0.21223174035549164,
      "learning_rate": 2.352068883042334e-05,
      "loss": 0.034,
      "step": 19774
    },
    {
      "epoch": 2.7715487035739312,
      "grad_norm": 0.11695139855146408,
      "learning_rate": 2.3506338196603682e-05,
      "loss": 0.0114,
      "step": 19775
    },
    {
      "epoch": 2.771688857743518,
      "grad_norm": 0.15395647287368774,
      "learning_rate": 2.349198756278402e-05,
      "loss": 0.0164,
      "step": 19776
    },
    {
      "epoch": 2.7718290119131046,
      "grad_norm": 0.2660770118236542,
      "learning_rate": 2.347763692896436e-05,
      "loss": 0.0808,
      "step": 19777
    },
    {
      "epoch": 2.771969166082691,
      "grad_norm": 0.21667182445526123,
      "learning_rate": 2.3463286295144698e-05,
      "loss": 0.0144,
      "step": 19778
    },
    {
      "epoch": 2.7721093202522775,
      "grad_norm": 0.44012385606765747,
      "learning_rate": 2.344893566132504e-05,
      "loss": 0.0255,
      "step": 19779
    },
    {
      "epoch": 2.772249474421864,
      "grad_norm": 0.18145763874053955,
      "learning_rate": 2.3434585027505382e-05,
      "loss": 0.0166,
      "step": 19780
    },
    {
      "epoch": 2.7723896285914504,
      "grad_norm": 0.4543394446372986,
      "learning_rate": 2.3420234393685718e-05,
      "loss": 0.0246,
      "step": 19781
    },
    {
      "epoch": 2.772529782761037,
      "grad_norm": 0.08987830579280853,
      "learning_rate": 2.340588375986606e-05,
      "loss": 0.0074,
      "step": 19782
    },
    {
      "epoch": 2.772669936930624,
      "grad_norm": 0.6360204219818115,
      "learning_rate": 2.33915331260464e-05,
      "loss": 0.0788,
      "step": 19783
    },
    {
      "epoch": 2.7728100911002103,
      "grad_norm": 0.3392009735107422,
      "learning_rate": 2.337718249222674e-05,
      "loss": 0.0352,
      "step": 19784
    },
    {
      "epoch": 2.7729502452697967,
      "grad_norm": 0.17291054129600525,
      "learning_rate": 2.3362831858407076e-05,
      "loss": 0.0164,
      "step": 19785
    },
    {
      "epoch": 2.773090399439383,
      "grad_norm": 0.2199353277683258,
      "learning_rate": 2.3348481224587418e-05,
      "loss": 0.0226,
      "step": 19786
    },
    {
      "epoch": 2.77323055360897,
      "grad_norm": 0.5193815231323242,
      "learning_rate": 2.3334130590767757e-05,
      "loss": 0.0349,
      "step": 19787
    },
    {
      "epoch": 2.7733707077785565,
      "grad_norm": 0.006943926215171814,
      "learning_rate": 2.33197799569481e-05,
      "loss": 0.0007,
      "step": 19788
    },
    {
      "epoch": 2.773510861948143,
      "grad_norm": 0.4154582917690277,
      "learning_rate": 2.3305429323128434e-05,
      "loss": 0.0661,
      "step": 19789
    },
    {
      "epoch": 2.7736510161177295,
      "grad_norm": 0.3527981638908386,
      "learning_rate": 2.3291078689308776e-05,
      "loss": 0.0773,
      "step": 19790
    },
    {
      "epoch": 2.773791170287316,
      "grad_norm": 0.0982864573597908,
      "learning_rate": 2.3276728055489115e-05,
      "loss": 0.0107,
      "step": 19791
    },
    {
      "epoch": 2.7739313244569024,
      "grad_norm": 0.27146583795547485,
      "learning_rate": 2.3262377421669457e-05,
      "loss": 0.0248,
      "step": 19792
    },
    {
      "epoch": 2.7740714786264893,
      "grad_norm": 0.13046716153621674,
      "learning_rate": 2.3248026787849792e-05,
      "loss": 0.01,
      "step": 19793
    },
    {
      "epoch": 2.7742116327960757,
      "grad_norm": 0.14189113676548004,
      "learning_rate": 2.3233676154030135e-05,
      "loss": 0.0084,
      "step": 19794
    },
    {
      "epoch": 2.774351786965662,
      "grad_norm": 0.5580766201019287,
      "learning_rate": 2.3219325520210473e-05,
      "loss": 0.0285,
      "step": 19795
    },
    {
      "epoch": 2.7744919411352487,
      "grad_norm": 0.18864187598228455,
      "learning_rate": 2.3204974886390815e-05,
      "loss": 0.0188,
      "step": 19796
    },
    {
      "epoch": 2.7746320953048356,
      "grad_norm": 0.15269257128238678,
      "learning_rate": 2.319062425257115e-05,
      "loss": 0.019,
      "step": 19797
    },
    {
      "epoch": 2.774772249474422,
      "grad_norm": 0.12125962972640991,
      "learning_rate": 2.3176273618751493e-05,
      "loss": 0.007,
      "step": 19798
    },
    {
      "epoch": 2.7749124036440085,
      "grad_norm": 0.34460729360580444,
      "learning_rate": 2.316192298493183e-05,
      "loss": 0.0524,
      "step": 19799
    },
    {
      "epoch": 2.775052557813595,
      "grad_norm": 0.20926158130168915,
      "learning_rate": 2.3147572351112174e-05,
      "loss": 0.0363,
      "step": 19800
    },
    {
      "epoch": 2.7751927119831814,
      "grad_norm": 0.20499832928180695,
      "learning_rate": 2.313322171729251e-05,
      "loss": 0.0291,
      "step": 19801
    },
    {
      "epoch": 2.775332866152768,
      "grad_norm": 0.0962163656949997,
      "learning_rate": 2.311887108347285e-05,
      "loss": 0.0066,
      "step": 19802
    },
    {
      "epoch": 2.7754730203223543,
      "grad_norm": 0.20544691383838654,
      "learning_rate": 2.310452044965319e-05,
      "loss": 0.014,
      "step": 19803
    },
    {
      "epoch": 2.7756131744919412,
      "grad_norm": 0.12060816586017609,
      "learning_rate": 2.3090169815833532e-05,
      "loss": 0.0097,
      "step": 19804
    },
    {
      "epoch": 2.7757533286615277,
      "grad_norm": 0.20883211493492126,
      "learning_rate": 2.3075819182013867e-05,
      "loss": 0.0266,
      "step": 19805
    },
    {
      "epoch": 2.775893482831114,
      "grad_norm": 0.25741854310035706,
      "learning_rate": 2.306146854819421e-05,
      "loss": 0.0689,
      "step": 19806
    },
    {
      "epoch": 2.7760336370007006,
      "grad_norm": 0.18505950272083282,
      "learning_rate": 2.3047117914374548e-05,
      "loss": 0.0306,
      "step": 19807
    },
    {
      "epoch": 2.7761737911702875,
      "grad_norm": 0.184142604470253,
      "learning_rate": 2.303276728055489e-05,
      "loss": 0.0088,
      "step": 19808
    },
    {
      "epoch": 2.776313945339874,
      "grad_norm": 0.06502658873796463,
      "learning_rate": 2.3018416646735226e-05,
      "loss": 0.0069,
      "step": 19809
    },
    {
      "epoch": 2.7764540995094604,
      "grad_norm": 0.2792748510837555,
      "learning_rate": 2.3004066012915568e-05,
      "loss": 0.0188,
      "step": 19810
    },
    {
      "epoch": 2.776594253679047,
      "grad_norm": 3.139632225036621,
      "learning_rate": 2.2989715379095906e-05,
      "loss": 0.0537,
      "step": 19811
    },
    {
      "epoch": 2.7767344078486333,
      "grad_norm": 0.012282947078347206,
      "learning_rate": 2.297536474527625e-05,
      "loss": 0.0008,
      "step": 19812
    },
    {
      "epoch": 2.77687456201822,
      "grad_norm": 0.30013370513916016,
      "learning_rate": 2.2961014111456584e-05,
      "loss": 0.067,
      "step": 19813
    },
    {
      "epoch": 2.7770147161878067,
      "grad_norm": 0.2789977192878723,
      "learning_rate": 2.2946663477636926e-05,
      "loss": 0.0562,
      "step": 19814
    },
    {
      "epoch": 2.777154870357393,
      "grad_norm": 0.5803288221359253,
      "learning_rate": 2.2932312843817265e-05,
      "loss": 0.0197,
      "step": 19815
    },
    {
      "epoch": 2.7772950245269796,
      "grad_norm": 0.019407566636800766,
      "learning_rate": 2.2917962209997607e-05,
      "loss": 0.0012,
      "step": 19816
    },
    {
      "epoch": 2.777435178696566,
      "grad_norm": 0.2308964878320694,
      "learning_rate": 2.290361157617795e-05,
      "loss": 0.0546,
      "step": 19817
    },
    {
      "epoch": 2.777575332866153,
      "grad_norm": 0.13067297637462616,
      "learning_rate": 2.2889260942358284e-05,
      "loss": 0.0044,
      "step": 19818
    },
    {
      "epoch": 2.7777154870357395,
      "grad_norm": 0.026862701401114464,
      "learning_rate": 2.2874910308538626e-05,
      "loss": 0.0017,
      "step": 19819
    },
    {
      "epoch": 2.777855641205326,
      "grad_norm": 2.3220887184143066,
      "learning_rate": 2.2860559674718965e-05,
      "loss": 0.134,
      "step": 19820
    },
    {
      "epoch": 2.7779957953749124,
      "grad_norm": 0.06867286562919617,
      "learning_rate": 2.2846209040899307e-05,
      "loss": 0.003,
      "step": 19821
    },
    {
      "epoch": 2.778135949544499,
      "grad_norm": 0.07390697300434113,
      "learning_rate": 2.2831858407079643e-05,
      "loss": 0.0054,
      "step": 19822
    },
    {
      "epoch": 2.7782761037140853,
      "grad_norm": 0.16000312566757202,
      "learning_rate": 2.2817507773259985e-05,
      "loss": 0.0297,
      "step": 19823
    },
    {
      "epoch": 2.778416257883672,
      "grad_norm": 0.17961719632148743,
      "learning_rate": 2.2803157139440323e-05,
      "loss": 0.0147,
      "step": 19824
    },
    {
      "epoch": 2.7785564120532587,
      "grad_norm": 0.2783864736557007,
      "learning_rate": 2.2788806505620666e-05,
      "loss": 0.0337,
      "step": 19825
    },
    {
      "epoch": 2.778696566222845,
      "grad_norm": 0.08844500035047531,
      "learning_rate": 2.2774455871801e-05,
      "loss": 0.0186,
      "step": 19826
    },
    {
      "epoch": 2.7788367203924316,
      "grad_norm": 0.3968968689441681,
      "learning_rate": 2.2760105237981343e-05,
      "loss": 0.0693,
      "step": 19827
    },
    {
      "epoch": 2.7789768745620185,
      "grad_norm": 0.1678866446018219,
      "learning_rate": 2.2745754604161682e-05,
      "loss": 0.0092,
      "step": 19828
    },
    {
      "epoch": 2.779117028731605,
      "grad_norm": 0.18766486644744873,
      "learning_rate": 2.2731403970342024e-05,
      "loss": 0.0192,
      "step": 19829
    },
    {
      "epoch": 2.7792571829011914,
      "grad_norm": 0.4989485442638397,
      "learning_rate": 2.271705333652236e-05,
      "loss": 0.0288,
      "step": 19830
    },
    {
      "epoch": 2.779397337070778,
      "grad_norm": 0.10022205114364624,
      "learning_rate": 2.27027027027027e-05,
      "loss": 0.0318,
      "step": 19831
    },
    {
      "epoch": 2.7795374912403643,
      "grad_norm": 0.10035252571105957,
      "learning_rate": 2.268835206888304e-05,
      "loss": 0.0299,
      "step": 19832
    },
    {
      "epoch": 2.7796776454099508,
      "grad_norm": 0.04994402825832367,
      "learning_rate": 2.2674001435063382e-05,
      "loss": 0.0039,
      "step": 19833
    },
    {
      "epoch": 2.7798177995795372,
      "grad_norm": 0.3298553228378296,
      "learning_rate": 2.2659650801243717e-05,
      "loss": 0.0234,
      "step": 19834
    },
    {
      "epoch": 2.779957953749124,
      "grad_norm": 0.0856083482503891,
      "learning_rate": 2.264530016742406e-05,
      "loss": 0.0078,
      "step": 19835
    },
    {
      "epoch": 2.7800981079187106,
      "grad_norm": 0.18359985947608948,
      "learning_rate": 2.2630949533604398e-05,
      "loss": 0.0206,
      "step": 19836
    },
    {
      "epoch": 2.780238262088297,
      "grad_norm": 0.5506598353385925,
      "learning_rate": 2.261659889978474e-05,
      "loss": 0.049,
      "step": 19837
    },
    {
      "epoch": 2.7803784162578835,
      "grad_norm": 0.11980695277452469,
      "learning_rate": 2.2602248265965076e-05,
      "loss": 0.0084,
      "step": 19838
    },
    {
      "epoch": 2.7805185704274704,
      "grad_norm": 0.22937439382076263,
      "learning_rate": 2.2587897632145418e-05,
      "loss": 0.0733,
      "step": 19839
    },
    {
      "epoch": 2.780658724597057,
      "grad_norm": 0.3908344805240631,
      "learning_rate": 2.2573546998325757e-05,
      "loss": 0.0376,
      "step": 19840
    },
    {
      "epoch": 2.7807988787666433,
      "grad_norm": 0.4814217984676361,
      "learning_rate": 2.25591963645061e-05,
      "loss": 0.0353,
      "step": 19841
    },
    {
      "epoch": 2.78093903293623,
      "grad_norm": 0.2078816443681717,
      "learning_rate": 2.2544845730686434e-05,
      "loss": 0.031,
      "step": 19842
    },
    {
      "epoch": 2.7810791871058163,
      "grad_norm": 0.21305575966835022,
      "learning_rate": 2.2530495096866776e-05,
      "loss": 0.052,
      "step": 19843
    },
    {
      "epoch": 2.7812193412754027,
      "grad_norm": 0.10749264806509018,
      "learning_rate": 2.2516144463047115e-05,
      "loss": 0.0193,
      "step": 19844
    },
    {
      "epoch": 2.7813594954449896,
      "grad_norm": 0.2677127718925476,
      "learning_rate": 2.2501793829227457e-05,
      "loss": 0.0435,
      "step": 19845
    },
    {
      "epoch": 2.781499649614576,
      "grad_norm": 0.24073554575443268,
      "learning_rate": 2.2487443195407792e-05,
      "loss": 0.0491,
      "step": 19846
    },
    {
      "epoch": 2.7816398037841625,
      "grad_norm": 0.3661080300807953,
      "learning_rate": 2.2473092561588134e-05,
      "loss": 0.0211,
      "step": 19847
    },
    {
      "epoch": 2.781779957953749,
      "grad_norm": 0.19566532969474792,
      "learning_rate": 2.2458741927768473e-05,
      "loss": 0.0149,
      "step": 19848
    },
    {
      "epoch": 2.781920112123336,
      "grad_norm": 0.09284968674182892,
      "learning_rate": 2.2444391293948815e-05,
      "loss": 0.0064,
      "step": 19849
    },
    {
      "epoch": 2.7820602662929224,
      "grad_norm": 0.2263614982366562,
      "learning_rate": 2.243004066012915e-05,
      "loss": 0.0399,
      "step": 19850
    },
    {
      "epoch": 2.782200420462509,
      "grad_norm": 0.040337175130844116,
      "learning_rate": 2.2415690026309493e-05,
      "loss": 0.0025,
      "step": 19851
    },
    {
      "epoch": 2.7823405746320953,
      "grad_norm": 0.2804590165615082,
      "learning_rate": 2.240133939248983e-05,
      "loss": 0.0445,
      "step": 19852
    },
    {
      "epoch": 2.7824807288016817,
      "grad_norm": 0.10034050792455673,
      "learning_rate": 2.2386988758670174e-05,
      "loss": 0.0072,
      "step": 19853
    },
    {
      "epoch": 2.782620882971268,
      "grad_norm": 0.2392459362745285,
      "learning_rate": 2.2372638124850516e-05,
      "loss": 0.0269,
      "step": 19854
    },
    {
      "epoch": 2.782761037140855,
      "grad_norm": 0.13013502955436707,
      "learning_rate": 2.235828749103085e-05,
      "loss": 0.0148,
      "step": 19855
    },
    {
      "epoch": 2.7829011913104416,
      "grad_norm": 0.14276379346847534,
      "learning_rate": 2.2343936857211193e-05,
      "loss": 0.024,
      "step": 19856
    },
    {
      "epoch": 2.783041345480028,
      "grad_norm": 0.29417479038238525,
      "learning_rate": 2.2329586223391532e-05,
      "loss": 0.0428,
      "step": 19857
    },
    {
      "epoch": 2.7831814996496145,
      "grad_norm": 0.13541923463344574,
      "learning_rate": 2.2315235589571874e-05,
      "loss": 0.0074,
      "step": 19858
    },
    {
      "epoch": 2.7833216538192014,
      "grad_norm": 0.21642860770225525,
      "learning_rate": 2.230088495575221e-05,
      "loss": 0.0329,
      "step": 19859
    },
    {
      "epoch": 2.783461807988788,
      "grad_norm": 0.10510509461164474,
      "learning_rate": 2.228653432193255e-05,
      "loss": 0.0308,
      "step": 19860
    },
    {
      "epoch": 2.7836019621583743,
      "grad_norm": 0.19402119517326355,
      "learning_rate": 2.227218368811289e-05,
      "loss": 0.008,
      "step": 19861
    },
    {
      "epoch": 2.7837421163279608,
      "grad_norm": 0.05505824834108353,
      "learning_rate": 2.2257833054293232e-05,
      "loss": 0.0022,
      "step": 19862
    },
    {
      "epoch": 2.783882270497547,
      "grad_norm": 0.1609456092119217,
      "learning_rate": 2.2243482420473568e-05,
      "loss": 0.0047,
      "step": 19863
    },
    {
      "epoch": 2.7840224246671337,
      "grad_norm": 0.37182798981666565,
      "learning_rate": 2.222913178665391e-05,
      "loss": 0.0284,
      "step": 19864
    },
    {
      "epoch": 2.78416257883672,
      "grad_norm": 0.7403817772865295,
      "learning_rate": 2.221478115283425e-05,
      "loss": 0.0107,
      "step": 19865
    },
    {
      "epoch": 2.784302733006307,
      "grad_norm": 0.015163655392825603,
      "learning_rate": 2.220043051901459e-05,
      "loss": 0.001,
      "step": 19866
    },
    {
      "epoch": 2.7844428871758935,
      "grad_norm": 0.2052098661661148,
      "learning_rate": 2.2186079885194926e-05,
      "loss": 0.0141,
      "step": 19867
    },
    {
      "epoch": 2.78458304134548,
      "grad_norm": 0.6579716205596924,
      "learning_rate": 2.2171729251375268e-05,
      "loss": 0.1095,
      "step": 19868
    },
    {
      "epoch": 2.7847231955150664,
      "grad_norm": 0.05567855387926102,
      "learning_rate": 2.2157378617555607e-05,
      "loss": 0.001,
      "step": 19869
    },
    {
      "epoch": 2.7848633496846533,
      "grad_norm": 0.7235983610153198,
      "learning_rate": 2.214302798373595e-05,
      "loss": 0.1546,
      "step": 19870
    },
    {
      "epoch": 2.78500350385424,
      "grad_norm": 0.2096051573753357,
      "learning_rate": 2.2128677349916284e-05,
      "loss": 0.0351,
      "step": 19871
    },
    {
      "epoch": 2.7851436580238262,
      "grad_norm": 0.08475883305072784,
      "learning_rate": 2.2114326716096626e-05,
      "loss": 0.0018,
      "step": 19872
    },
    {
      "epoch": 2.7852838121934127,
      "grad_norm": 0.1402767300605774,
      "learning_rate": 2.2099976082276965e-05,
      "loss": 0.0484,
      "step": 19873
    },
    {
      "epoch": 2.785423966362999,
      "grad_norm": 0.10882092267274857,
      "learning_rate": 2.2085625448457307e-05,
      "loss": 0.0062,
      "step": 19874
    },
    {
      "epoch": 2.7855641205325856,
      "grad_norm": 0.6786602735519409,
      "learning_rate": 2.2071274814637642e-05,
      "loss": 0.0318,
      "step": 19875
    },
    {
      "epoch": 2.7857042747021725,
      "grad_norm": 0.21453382074832916,
      "learning_rate": 2.2056924180817985e-05,
      "loss": 0.0408,
      "step": 19876
    },
    {
      "epoch": 2.785844428871759,
      "grad_norm": 0.1139344722032547,
      "learning_rate": 2.2042573546998323e-05,
      "loss": 0.0224,
      "step": 19877
    },
    {
      "epoch": 2.7859845830413454,
      "grad_norm": 0.12129947543144226,
      "learning_rate": 2.2028222913178665e-05,
      "loss": 0.0058,
      "step": 19878
    },
    {
      "epoch": 2.786124737210932,
      "grad_norm": 0.3911696672439575,
      "learning_rate": 2.2013872279359e-05,
      "loss": 0.0435,
      "step": 19879
    },
    {
      "epoch": 2.786264891380519,
      "grad_norm": 0.02805904485285282,
      "learning_rate": 2.1999521645539343e-05,
      "loss": 0.0018,
      "step": 19880
    },
    {
      "epoch": 2.7864050455501053,
      "grad_norm": 0.05103493481874466,
      "learning_rate": 2.198517101171968e-05,
      "loss": 0.0043,
      "step": 19881
    },
    {
      "epoch": 2.7865451997196917,
      "grad_norm": 0.3952654302120209,
      "learning_rate": 2.1970820377900024e-05,
      "loss": 0.0661,
      "step": 19882
    },
    {
      "epoch": 2.786685353889278,
      "grad_norm": 0.1119556725025177,
      "learning_rate": 2.195646974408036e-05,
      "loss": 0.0302,
      "step": 19883
    },
    {
      "epoch": 2.7868255080588646,
      "grad_norm": 0.10786882787942886,
      "learning_rate": 2.19421191102607e-05,
      "loss": 0.0136,
      "step": 19884
    },
    {
      "epoch": 2.786965662228451,
      "grad_norm": 0.3601684868335724,
      "learning_rate": 2.192776847644104e-05,
      "loss": 0.03,
      "step": 19885
    },
    {
      "epoch": 2.787105816398038,
      "grad_norm": 0.17197385430335999,
      "learning_rate": 2.1913417842621382e-05,
      "loss": 0.014,
      "step": 19886
    },
    {
      "epoch": 2.7872459705676245,
      "grad_norm": 0.10475435107946396,
      "learning_rate": 2.1899067208801717e-05,
      "loss": 0.0134,
      "step": 19887
    },
    {
      "epoch": 2.787386124737211,
      "grad_norm": 0.19809970259666443,
      "learning_rate": 2.188471657498206e-05,
      "loss": 0.0062,
      "step": 19888
    },
    {
      "epoch": 2.7875262789067974,
      "grad_norm": 0.10463893413543701,
      "learning_rate": 2.1870365941162398e-05,
      "loss": 0.0099,
      "step": 19889
    },
    {
      "epoch": 2.7876664330763843,
      "grad_norm": 0.19980274140834808,
      "learning_rate": 2.185601530734274e-05,
      "loss": 0.0326,
      "step": 19890
    },
    {
      "epoch": 2.7878065872459707,
      "grad_norm": 0.05061763897538185,
      "learning_rate": 2.1841664673523082e-05,
      "loss": 0.0028,
      "step": 19891
    },
    {
      "epoch": 2.787946741415557,
      "grad_norm": 0.17905987799167633,
      "learning_rate": 2.1827314039703418e-05,
      "loss": 0.0112,
      "step": 19892
    },
    {
      "epoch": 2.7880868955851437,
      "grad_norm": 0.24636733531951904,
      "learning_rate": 2.181296340588376e-05,
      "loss": 0.0286,
      "step": 19893
    },
    {
      "epoch": 2.78822704975473,
      "grad_norm": 0.12782379984855652,
      "learning_rate": 2.17986127720641e-05,
      "loss": 0.0089,
      "step": 19894
    },
    {
      "epoch": 2.7883672039243166,
      "grad_norm": 0.18282784521579742,
      "learning_rate": 2.178426213824444e-05,
      "loss": 0.0342,
      "step": 19895
    },
    {
      "epoch": 2.788507358093903,
      "grad_norm": 0.1590076982975006,
      "learning_rate": 2.1769911504424776e-05,
      "loss": 0.0144,
      "step": 19896
    },
    {
      "epoch": 2.78864751226349,
      "grad_norm": 0.38412681221961975,
      "learning_rate": 2.1755560870605118e-05,
      "loss": 0.0646,
      "step": 19897
    },
    {
      "epoch": 2.7887876664330764,
      "grad_norm": 0.07290639728307724,
      "learning_rate": 2.1741210236785457e-05,
      "loss": 0.0085,
      "step": 19898
    },
    {
      "epoch": 2.788927820602663,
      "grad_norm": 0.16273124516010284,
      "learning_rate": 2.17268596029658e-05,
      "loss": 0.0082,
      "step": 19899
    },
    {
      "epoch": 2.7890679747722493,
      "grad_norm": 0.36756864190101624,
      "learning_rate": 2.1712508969146134e-05,
      "loss": 0.0563,
      "step": 19900
    },
    {
      "epoch": 2.7892081289418362,
      "grad_norm": 0.047376178205013275,
      "learning_rate": 2.1698158335326476e-05,
      "loss": 0.0027,
      "step": 19901
    },
    {
      "epoch": 2.7893482831114227,
      "grad_norm": 0.07464243471622467,
      "learning_rate": 2.1683807701506815e-05,
      "loss": 0.0066,
      "step": 19902
    },
    {
      "epoch": 2.789488437281009,
      "grad_norm": 0.07505783438682556,
      "learning_rate": 2.1669457067687157e-05,
      "loss": 0.0064,
      "step": 19903
    },
    {
      "epoch": 2.7896285914505956,
      "grad_norm": 0.21771767735481262,
      "learning_rate": 2.1655106433867493e-05,
      "loss": 0.0277,
      "step": 19904
    },
    {
      "epoch": 2.789768745620182,
      "grad_norm": 0.132535919547081,
      "learning_rate": 2.1640755800047835e-05,
      "loss": 0.0235,
      "step": 19905
    },
    {
      "epoch": 2.7899088997897685,
      "grad_norm": 0.5503745675086975,
      "learning_rate": 2.1626405166228173e-05,
      "loss": 0.0523,
      "step": 19906
    },
    {
      "epoch": 2.7900490539593554,
      "grad_norm": 0.215169757604599,
      "learning_rate": 2.1612054532408515e-05,
      "loss": 0.0047,
      "step": 19907
    },
    {
      "epoch": 2.790189208128942,
      "grad_norm": 0.2809276282787323,
      "learning_rate": 2.159770389858885e-05,
      "loss": 0.0408,
      "step": 19908
    },
    {
      "epoch": 2.7903293622985283,
      "grad_norm": 0.11196155846118927,
      "learning_rate": 2.1583353264769193e-05,
      "loss": 0.0115,
      "step": 19909
    },
    {
      "epoch": 2.790469516468115,
      "grad_norm": 0.09218896925449371,
      "learning_rate": 2.156900263094953e-05,
      "loss": 0.011,
      "step": 19910
    },
    {
      "epoch": 2.7906096706377017,
      "grad_norm": 0.16444481909275055,
      "learning_rate": 2.1554651997129874e-05,
      "loss": 0.0185,
      "step": 19911
    },
    {
      "epoch": 2.790749824807288,
      "grad_norm": 0.08260773867368698,
      "learning_rate": 2.154030136331021e-05,
      "loss": 0.0046,
      "step": 19912
    },
    {
      "epoch": 2.7908899789768746,
      "grad_norm": 0.20342746376991272,
      "learning_rate": 2.152595072949055e-05,
      "loss": 0.0088,
      "step": 19913
    },
    {
      "epoch": 2.791030133146461,
      "grad_norm": 0.04194631054997444,
      "learning_rate": 2.151160009567089e-05,
      "loss": 0.0046,
      "step": 19914
    },
    {
      "epoch": 2.7911702873160475,
      "grad_norm": 0.19466149806976318,
      "learning_rate": 2.1497249461851232e-05,
      "loss": 0.0452,
      "step": 19915
    },
    {
      "epoch": 2.791310441485634,
      "grad_norm": 0.154509037733078,
      "learning_rate": 2.1482898828031567e-05,
      "loss": 0.0082,
      "step": 19916
    },
    {
      "epoch": 2.7914505956552205,
      "grad_norm": 0.7207839488983154,
      "learning_rate": 2.146854819421191e-05,
      "loss": 0.0326,
      "step": 19917
    },
    {
      "epoch": 2.7915907498248074,
      "grad_norm": 0.36176037788391113,
      "learning_rate": 2.1454197560392248e-05,
      "loss": 0.0122,
      "step": 19918
    },
    {
      "epoch": 2.791730903994394,
      "grad_norm": 0.3289072811603546,
      "learning_rate": 2.143984692657259e-05,
      "loss": 0.0084,
      "step": 19919
    },
    {
      "epoch": 2.7918710581639803,
      "grad_norm": 0.6676080822944641,
      "learning_rate": 2.1425496292752926e-05,
      "loss": 0.0375,
      "step": 19920
    },
    {
      "epoch": 2.792011212333567,
      "grad_norm": 0.14369215071201324,
      "learning_rate": 2.1411145658933268e-05,
      "loss": 0.026,
      "step": 19921
    },
    {
      "epoch": 2.7921513665031537,
      "grad_norm": 0.23929215967655182,
      "learning_rate": 2.1396795025113607e-05,
      "loss": 0.0243,
      "step": 19922
    },
    {
      "epoch": 2.79229152067274,
      "grad_norm": 0.15751765668392181,
      "learning_rate": 2.138244439129395e-05,
      "loss": 0.0197,
      "step": 19923
    },
    {
      "epoch": 2.7924316748423266,
      "grad_norm": 0.12370754033327103,
      "learning_rate": 2.1368093757474284e-05,
      "loss": 0.0347,
      "step": 19924
    },
    {
      "epoch": 2.792571829011913,
      "grad_norm": 0.2292085886001587,
      "learning_rate": 2.1353743123654626e-05,
      "loss": 0.0328,
      "step": 19925
    },
    {
      "epoch": 2.7927119831814995,
      "grad_norm": 0.2243906706571579,
      "learning_rate": 2.1339392489834965e-05,
      "loss": 0.0151,
      "step": 19926
    },
    {
      "epoch": 2.792852137351086,
      "grad_norm": 0.1460040658712387,
      "learning_rate": 2.1325041856015307e-05,
      "loss": 0.0133,
      "step": 19927
    },
    {
      "epoch": 2.792992291520673,
      "grad_norm": 0.2599943280220032,
      "learning_rate": 2.1310691222195646e-05,
      "loss": 0.0228,
      "step": 19928
    },
    {
      "epoch": 2.7931324456902593,
      "grad_norm": 0.1314019113779068,
      "learning_rate": 2.1296340588375984e-05,
      "loss": 0.0111,
      "step": 19929
    },
    {
      "epoch": 2.7932725998598458,
      "grad_norm": 0.23316337168216705,
      "learning_rate": 2.1281989954556326e-05,
      "loss": 0.0157,
      "step": 19930
    },
    {
      "epoch": 2.7934127540294322,
      "grad_norm": 0.0410420224070549,
      "learning_rate": 2.1267639320736665e-05,
      "loss": 0.003,
      "step": 19931
    },
    {
      "epoch": 2.793552908199019,
      "grad_norm": 0.020699461922049522,
      "learning_rate": 2.1253288686917004e-05,
      "loss": 0.0025,
      "step": 19932
    },
    {
      "epoch": 2.7936930623686056,
      "grad_norm": 0.2327887862920761,
      "learning_rate": 2.1238938053097343e-05,
      "loss": 0.0556,
      "step": 19933
    },
    {
      "epoch": 2.793833216538192,
      "grad_norm": 0.31376755237579346,
      "learning_rate": 2.1224587419277685e-05,
      "loss": 0.0785,
      "step": 19934
    },
    {
      "epoch": 2.7939733707077785,
      "grad_norm": 0.2757945656776428,
      "learning_rate": 2.1210236785458023e-05,
      "loss": 0.0647,
      "step": 19935
    },
    {
      "epoch": 2.794113524877365,
      "grad_norm": 0.36520570516586304,
      "learning_rate": 2.1195886151638362e-05,
      "loss": 0.0683,
      "step": 19936
    },
    {
      "epoch": 2.7942536790469514,
      "grad_norm": 0.2816203832626343,
      "learning_rate": 2.11815355178187e-05,
      "loss": 0.0387,
      "step": 19937
    },
    {
      "epoch": 2.7943938332165383,
      "grad_norm": 0.21043512225151062,
      "learning_rate": 2.1167184883999043e-05,
      "loss": 0.0586,
      "step": 19938
    },
    {
      "epoch": 2.794533987386125,
      "grad_norm": 0.05700146406888962,
      "learning_rate": 2.1152834250179382e-05,
      "loss": 0.0051,
      "step": 19939
    },
    {
      "epoch": 2.7946741415557113,
      "grad_norm": 0.051535919308662415,
      "learning_rate": 2.113848361635972e-05,
      "loss": 0.0022,
      "step": 19940
    },
    {
      "epoch": 2.7948142957252977,
      "grad_norm": 0.12905417382717133,
      "learning_rate": 2.112413298254006e-05,
      "loss": 0.0237,
      "step": 19941
    },
    {
      "epoch": 2.7949544498948846,
      "grad_norm": 0.18502721190452576,
      "learning_rate": 2.11097823487204e-05,
      "loss": 0.0272,
      "step": 19942
    },
    {
      "epoch": 2.795094604064471,
      "grad_norm": 0.35748687386512756,
      "learning_rate": 2.109543171490074e-05,
      "loss": 0.0287,
      "step": 19943
    },
    {
      "epoch": 2.7952347582340575,
      "grad_norm": 0.086206816136837,
      "learning_rate": 2.108108108108108e-05,
      "loss": 0.0086,
      "step": 19944
    },
    {
      "epoch": 2.795374912403644,
      "grad_norm": 0.12639585137367249,
      "learning_rate": 2.1066730447261418e-05,
      "loss": 0.02,
      "step": 19945
    },
    {
      "epoch": 2.7955150665732305,
      "grad_norm": 0.3880903124809265,
      "learning_rate": 2.105237981344176e-05,
      "loss": 0.0322,
      "step": 19946
    },
    {
      "epoch": 2.795655220742817,
      "grad_norm": 0.2510320544242859,
      "learning_rate": 2.10380291796221e-05,
      "loss": 0.0813,
      "step": 19947
    },
    {
      "epoch": 2.7957953749124034,
      "grad_norm": 0.03349941223859787,
      "learning_rate": 2.1023678545802437e-05,
      "loss": 0.0036,
      "step": 19948
    },
    {
      "epoch": 2.7959355290819903,
      "grad_norm": 0.2164732664823532,
      "learning_rate": 2.1009327911982776e-05,
      "loss": 0.0295,
      "step": 19949
    },
    {
      "epoch": 2.7960756832515767,
      "grad_norm": 0.19333980977535248,
      "learning_rate": 2.0994977278163118e-05,
      "loss": 0.0376,
      "step": 19950
    },
    {
      "epoch": 2.796215837421163,
      "grad_norm": 0.30675041675567627,
      "learning_rate": 2.0980626644343457e-05,
      "loss": 0.0365,
      "step": 19951
    },
    {
      "epoch": 2.79635599159075,
      "grad_norm": 0.6219843626022339,
      "learning_rate": 2.0966276010523795e-05,
      "loss": 0.0262,
      "step": 19952
    },
    {
      "epoch": 2.7964961457603366,
      "grad_norm": 0.12456759065389633,
      "learning_rate": 2.0951925376704134e-05,
      "loss": 0.007,
      "step": 19953
    },
    {
      "epoch": 2.796636299929923,
      "grad_norm": 0.5238804221153259,
      "learning_rate": 2.0937574742884476e-05,
      "loss": 0.0308,
      "step": 19954
    },
    {
      "epoch": 2.7967764540995095,
      "grad_norm": 0.29134076833724976,
      "learning_rate": 2.0923224109064815e-05,
      "loss": 0.0189,
      "step": 19955
    },
    {
      "epoch": 2.796916608269096,
      "grad_norm": 0.24165566265583038,
      "learning_rate": 2.0908873475245154e-05,
      "loss": 0.0388,
      "step": 19956
    },
    {
      "epoch": 2.7970567624386824,
      "grad_norm": 0.5876925587654114,
      "learning_rate": 2.0894522841425492e-05,
      "loss": 0.0165,
      "step": 19957
    },
    {
      "epoch": 2.797196916608269,
      "grad_norm": 0.2325659543275833,
      "learning_rate": 2.0880172207605834e-05,
      "loss": 0.0407,
      "step": 19958
    },
    {
      "epoch": 2.7973370707778558,
      "grad_norm": 0.1631787270307541,
      "learning_rate": 2.0865821573786173e-05,
      "loss": 0.0331,
      "step": 19959
    },
    {
      "epoch": 2.797477224947442,
      "grad_norm": 0.20489393174648285,
      "learning_rate": 2.0851470939966515e-05,
      "loss": 0.0177,
      "step": 19960
    },
    {
      "epoch": 2.7976173791170287,
      "grad_norm": 0.14901484549045563,
      "learning_rate": 2.083712030614685e-05,
      "loss": 0.0038,
      "step": 19961
    },
    {
      "epoch": 2.797757533286615,
      "grad_norm": 0.24769359827041626,
      "learning_rate": 2.0822769672327193e-05,
      "loss": 0.0419,
      "step": 19962
    },
    {
      "epoch": 2.797897687456202,
      "grad_norm": 0.04648546874523163,
      "learning_rate": 2.080841903850753e-05,
      "loss": 0.0039,
      "step": 19963
    },
    {
      "epoch": 2.7980378416257885,
      "grad_norm": 0.33478906750679016,
      "learning_rate": 2.0794068404687874e-05,
      "loss": 0.0472,
      "step": 19964
    },
    {
      "epoch": 2.798177995795375,
      "grad_norm": 0.24060703814029694,
      "learning_rate": 2.0779717770868212e-05,
      "loss": 0.0338,
      "step": 19965
    },
    {
      "epoch": 2.7983181499649614,
      "grad_norm": 0.8144930601119995,
      "learning_rate": 2.076536713704855e-05,
      "loss": 0.1267,
      "step": 19966
    },
    {
      "epoch": 2.798458304134548,
      "grad_norm": 0.26101940870285034,
      "learning_rate": 2.0751016503228893e-05,
      "loss": 0.0466,
      "step": 19967
    },
    {
      "epoch": 2.7985984583041343,
      "grad_norm": 1.7428154945373535,
      "learning_rate": 2.0736665869409232e-05,
      "loss": 0.0447,
      "step": 19968
    },
    {
      "epoch": 2.7987386124737212,
      "grad_norm": 1.5862023830413818,
      "learning_rate": 2.072231523558957e-05,
      "loss": 0.1341,
      "step": 19969
    },
    {
      "epoch": 2.7988787666433077,
      "grad_norm": 0.7910722494125366,
      "learning_rate": 2.070796460176991e-05,
      "loss": 0.092,
      "step": 19970
    },
    {
      "epoch": 2.799018920812894,
      "grad_norm": 0.12827037274837494,
      "learning_rate": 2.069361396795025e-05,
      "loss": 0.0378,
      "step": 19971
    },
    {
      "epoch": 2.7991590749824806,
      "grad_norm": 0.21304842829704285,
      "learning_rate": 2.067926333413059e-05,
      "loss": 0.0431,
      "step": 19972
    },
    {
      "epoch": 2.7992992291520675,
      "grad_norm": 0.16096867620944977,
      "learning_rate": 2.066491270031093e-05,
      "loss": 0.0351,
      "step": 19973
    },
    {
      "epoch": 2.799439383321654,
      "grad_norm": 0.080171138048172,
      "learning_rate": 2.0650562066491268e-05,
      "loss": 0.0115,
      "step": 19974
    },
    {
      "epoch": 2.7995795374912404,
      "grad_norm": 0.21438491344451904,
      "learning_rate": 2.063621143267161e-05,
      "loss": 0.0356,
      "step": 19975
    },
    {
      "epoch": 2.799719691660827,
      "grad_norm": 0.23420527577400208,
      "learning_rate": 2.062186079885195e-05,
      "loss": 0.0165,
      "step": 19976
    },
    {
      "epoch": 2.7998598458304134,
      "grad_norm": 0.14552585780620575,
      "learning_rate": 2.0607510165032287e-05,
      "loss": 0.0134,
      "step": 19977
    },
    {
      "epoch": 2.8,
      "grad_norm": 0.2259116917848587,
      "learning_rate": 2.0593159531212626e-05,
      "loss": 0.0156,
      "step": 19978
    },
    {
      "epoch": 2.8001401541695863,
      "grad_norm": 0.16932596266269684,
      "learning_rate": 2.0578808897392968e-05,
      "loss": 0.0184,
      "step": 19979
    },
    {
      "epoch": 2.800280308339173,
      "grad_norm": 0.019965756684541702,
      "learning_rate": 2.0564458263573307e-05,
      "loss": 0.0021,
      "step": 19980
    },
    {
      "epoch": 2.8004204625087596,
      "grad_norm": 0.07200606912374496,
      "learning_rate": 2.0550107629753645e-05,
      "loss": 0.012,
      "step": 19981
    },
    {
      "epoch": 2.800560616678346,
      "grad_norm": 0.20571266114711761,
      "learning_rate": 2.0535756995933984e-05,
      "loss": 0.0113,
      "step": 19982
    },
    {
      "epoch": 2.8007007708479326,
      "grad_norm": 0.2929847836494446,
      "learning_rate": 2.0521406362114326e-05,
      "loss": 0.025,
      "step": 19983
    },
    {
      "epoch": 2.8008409250175195,
      "grad_norm": 0.5060651302337646,
      "learning_rate": 2.0507055728294665e-05,
      "loss": 0.0457,
      "step": 19984
    },
    {
      "epoch": 2.800981079187106,
      "grad_norm": 0.33901354670524597,
      "learning_rate": 2.0492705094475004e-05,
      "loss": 0.0414,
      "step": 19985
    },
    {
      "epoch": 2.8011212333566924,
      "grad_norm": 0.07111307978630066,
      "learning_rate": 2.0478354460655342e-05,
      "loss": 0.0069,
      "step": 19986
    },
    {
      "epoch": 2.801261387526279,
      "grad_norm": 0.24877874553203583,
      "learning_rate": 2.0464003826835685e-05,
      "loss": 0.0222,
      "step": 19987
    },
    {
      "epoch": 2.8014015416958653,
      "grad_norm": 0.224320188164711,
      "learning_rate": 2.0449653193016023e-05,
      "loss": 0.0085,
      "step": 19988
    },
    {
      "epoch": 2.8015416958654518,
      "grad_norm": 0.07319874316453934,
      "learning_rate": 2.0435302559196362e-05,
      "loss": 0.0185,
      "step": 19989
    },
    {
      "epoch": 2.8016818500350387,
      "grad_norm": 0.19133345782756805,
      "learning_rate": 2.04209519253767e-05,
      "loss": 0.041,
      "step": 19990
    },
    {
      "epoch": 2.801822004204625,
      "grad_norm": 0.042898040264844894,
      "learning_rate": 2.0406601291557043e-05,
      "loss": 0.0026,
      "step": 19991
    },
    {
      "epoch": 2.8019621583742116,
      "grad_norm": 0.2284245640039444,
      "learning_rate": 2.039225065773738e-05,
      "loss": 0.0199,
      "step": 19992
    },
    {
      "epoch": 2.802102312543798,
      "grad_norm": 0.08462200313806534,
      "learning_rate": 2.037790002391772e-05,
      "loss": 0.0136,
      "step": 19993
    },
    {
      "epoch": 2.802242466713385,
      "grad_norm": 0.6263629198074341,
      "learning_rate": 2.036354939009806e-05,
      "loss": 0.0679,
      "step": 19994
    },
    {
      "epoch": 2.8023826208829714,
      "grad_norm": 0.14108793437480927,
      "learning_rate": 2.03491987562784e-05,
      "loss": 0.0088,
      "step": 19995
    },
    {
      "epoch": 2.802522775052558,
      "grad_norm": 0.3206084072589874,
      "learning_rate": 2.033484812245874e-05,
      "loss": 0.0567,
      "step": 19996
    },
    {
      "epoch": 2.8026629292221443,
      "grad_norm": 0.2152736932039261,
      "learning_rate": 2.032049748863908e-05,
      "loss": 0.0133,
      "step": 19997
    },
    {
      "epoch": 2.802803083391731,
      "grad_norm": 0.12140417844057083,
      "learning_rate": 2.0306146854819417e-05,
      "loss": 0.0194,
      "step": 19998
    },
    {
      "epoch": 2.8029432375613172,
      "grad_norm": 0.3021085262298584,
      "learning_rate": 2.029179622099976e-05,
      "loss": 0.0216,
      "step": 19999
    },
    {
      "epoch": 2.803083391730904,
      "grad_norm": 0.22126875817775726,
      "learning_rate": 2.0277445587180098e-05,
      "loss": 0.0284,
      "step": 20000
    },
    {
      "epoch": 2.8032235459004906,
      "grad_norm": 0.5474613904953003,
      "learning_rate": 2.0263094953360437e-05,
      "loss": 0.0322,
      "step": 20001
    },
    {
      "epoch": 2.803363700070077,
      "grad_norm": 0.14236284792423248,
      "learning_rate": 2.0248744319540776e-05,
      "loss": 0.006,
      "step": 20002
    },
    {
      "epoch": 2.8035038542396635,
      "grad_norm": 0.19918866455554962,
      "learning_rate": 2.0234393685721118e-05,
      "loss": 0.032,
      "step": 20003
    },
    {
      "epoch": 2.8036440084092504,
      "grad_norm": 1.1128315925598145,
      "learning_rate": 2.022004305190146e-05,
      "loss": 0.0656,
      "step": 20004
    },
    {
      "epoch": 2.803784162578837,
      "grad_norm": 0.2393268495798111,
      "learning_rate": 2.0205692418081795e-05,
      "loss": 0.0074,
      "step": 20005
    },
    {
      "epoch": 2.8039243167484234,
      "grad_norm": 0.05781044065952301,
      "learning_rate": 2.0191341784262137e-05,
      "loss": 0.0037,
      "step": 20006
    },
    {
      "epoch": 2.80406447091801,
      "grad_norm": 0.09157800674438477,
      "learning_rate": 2.0176991150442476e-05,
      "loss": 0.0149,
      "step": 20007
    },
    {
      "epoch": 2.8042046250875963,
      "grad_norm": 0.2359168380498886,
      "learning_rate": 2.0162640516622818e-05,
      "loss": 0.0171,
      "step": 20008
    },
    {
      "epoch": 2.8043447792571827,
      "grad_norm": 0.11985179036855698,
      "learning_rate": 2.0148289882803153e-05,
      "loss": 0.0087,
      "step": 20009
    },
    {
      "epoch": 2.804484933426769,
      "grad_norm": 0.02725437842309475,
      "learning_rate": 2.0133939248983496e-05,
      "loss": 0.002,
      "step": 20010
    },
    {
      "epoch": 2.804625087596356,
      "grad_norm": 0.052521172910928726,
      "learning_rate": 2.0119588615163834e-05,
      "loss": 0.0028,
      "step": 20011
    },
    {
      "epoch": 2.8047652417659426,
      "grad_norm": 0.04475374519824982,
      "learning_rate": 2.0105237981344176e-05,
      "loss": 0.0041,
      "step": 20012
    },
    {
      "epoch": 2.804905395935529,
      "grad_norm": 0.43351632356643677,
      "learning_rate": 2.0090887347524512e-05,
      "loss": 0.0854,
      "step": 20013
    },
    {
      "epoch": 2.8050455501051155,
      "grad_norm": 0.1831410676240921,
      "learning_rate": 2.0076536713704854e-05,
      "loss": 0.0191,
      "step": 20014
    },
    {
      "epoch": 2.8051857042747024,
      "grad_norm": 1.5342525243759155,
      "learning_rate": 2.0062186079885193e-05,
      "loss": 0.0371,
      "step": 20015
    },
    {
      "epoch": 2.805325858444289,
      "grad_norm": 0.011657540686428547,
      "learning_rate": 2.0047835446065535e-05,
      "loss": 0.001,
      "step": 20016
    },
    {
      "epoch": 2.8054660126138753,
      "grad_norm": 0.856203556060791,
      "learning_rate": 2.003348481224587e-05,
      "loss": 0.138,
      "step": 20017
    },
    {
      "epoch": 2.8056061667834618,
      "grad_norm": 0.23107963800430298,
      "learning_rate": 2.0019134178426212e-05,
      "loss": 0.0179,
      "step": 20018
    },
    {
      "epoch": 2.805746320953048,
      "grad_norm": 0.4495178163051605,
      "learning_rate": 2.000478354460655e-05,
      "loss": 0.0298,
      "step": 20019
    },
    {
      "epoch": 2.8058864751226347,
      "grad_norm": 0.6873546242713928,
      "learning_rate": 1.9990432910786893e-05,
      "loss": 0.0842,
      "step": 20020
    },
    {
      "epoch": 2.8060266292922216,
      "grad_norm": 0.21186579763889313,
      "learning_rate": 1.997608227696723e-05,
      "loss": 0.0243,
      "step": 20021
    },
    {
      "epoch": 2.806166783461808,
      "grad_norm": 0.2650187611579895,
      "learning_rate": 1.996173164314757e-05,
      "loss": 0.0291,
      "step": 20022
    },
    {
      "epoch": 2.8063069376313945,
      "grad_norm": 0.13266363739967346,
      "learning_rate": 1.994738100932791e-05,
      "loss": 0.0332,
      "step": 20023
    },
    {
      "epoch": 2.806447091800981,
      "grad_norm": 0.06920032948255539,
      "learning_rate": 1.993303037550825e-05,
      "loss": 0.0098,
      "step": 20024
    },
    {
      "epoch": 2.806587245970568,
      "grad_norm": 0.052378904074430466,
      "learning_rate": 1.9918679741688587e-05,
      "loss": 0.0051,
      "step": 20025
    },
    {
      "epoch": 2.8067274001401543,
      "grad_norm": 0.3211519718170166,
      "learning_rate": 1.990432910786893e-05,
      "loss": 0.0143,
      "step": 20026
    },
    {
      "epoch": 2.8068675543097408,
      "grad_norm": 0.014825846068561077,
      "learning_rate": 1.9889978474049267e-05,
      "loss": 0.0016,
      "step": 20027
    },
    {
      "epoch": 2.8070077084793272,
      "grad_norm": 0.084259532392025,
      "learning_rate": 1.987562784022961e-05,
      "loss": 0.0053,
      "step": 20028
    },
    {
      "epoch": 2.8071478626489137,
      "grad_norm": 0.02732730656862259,
      "learning_rate": 1.9861277206409945e-05,
      "loss": 0.0023,
      "step": 20029
    },
    {
      "epoch": 2.8072880168185,
      "grad_norm": 0.11814083158969879,
      "learning_rate": 1.9846926572590287e-05,
      "loss": 0.0255,
      "step": 20030
    },
    {
      "epoch": 2.807428170988087,
      "grad_norm": 0.48678386211395264,
      "learning_rate": 1.9832575938770626e-05,
      "loss": 0.0484,
      "step": 20031
    },
    {
      "epoch": 2.8075683251576735,
      "grad_norm": 0.12674832344055176,
      "learning_rate": 1.9818225304950968e-05,
      "loss": 0.0122,
      "step": 20032
    },
    {
      "epoch": 2.80770847932726,
      "grad_norm": 0.3977404534816742,
      "learning_rate": 1.9803874671131303e-05,
      "loss": 0.0182,
      "step": 20033
    },
    {
      "epoch": 2.8078486334968464,
      "grad_norm": 0.09031179547309875,
      "learning_rate": 1.9789524037311645e-05,
      "loss": 0.0213,
      "step": 20034
    },
    {
      "epoch": 2.8079887876664333,
      "grad_norm": 0.17990154027938843,
      "learning_rate": 1.9775173403491984e-05,
      "loss": 0.0285,
      "step": 20035
    },
    {
      "epoch": 2.80812894183602,
      "grad_norm": 0.10841778665781021,
      "learning_rate": 1.9760822769672326e-05,
      "loss": 0.0204,
      "step": 20036
    },
    {
      "epoch": 2.8082690960056063,
      "grad_norm": 0.0737527385354042,
      "learning_rate": 1.974647213585266e-05,
      "loss": 0.0069,
      "step": 20037
    },
    {
      "epoch": 2.8084092501751927,
      "grad_norm": 0.06787803769111633,
      "learning_rate": 1.9732121502033004e-05,
      "loss": 0.0066,
      "step": 20038
    },
    {
      "epoch": 2.808549404344779,
      "grad_norm": 0.16270709037780762,
      "learning_rate": 1.9717770868213342e-05,
      "loss": 0.015,
      "step": 20039
    },
    {
      "epoch": 2.8086895585143656,
      "grad_norm": 0.11124812811613083,
      "learning_rate": 1.9703420234393684e-05,
      "loss": 0.0145,
      "step": 20040
    },
    {
      "epoch": 2.808829712683952,
      "grad_norm": 0.23035526275634766,
      "learning_rate": 1.9689069600574027e-05,
      "loss": 0.035,
      "step": 20041
    },
    {
      "epoch": 2.808969866853539,
      "grad_norm": 0.01653190888464451,
      "learning_rate": 1.9674718966754362e-05,
      "loss": 0.0018,
      "step": 20042
    },
    {
      "epoch": 2.8091100210231255,
      "grad_norm": 0.14227880537509918,
      "learning_rate": 1.9660368332934704e-05,
      "loss": 0.0096,
      "step": 20043
    },
    {
      "epoch": 2.809250175192712,
      "grad_norm": 0.1516581028699875,
      "learning_rate": 1.9646017699115043e-05,
      "loss": 0.0062,
      "step": 20044
    },
    {
      "epoch": 2.8093903293622984,
      "grad_norm": 0.4181561768054962,
      "learning_rate": 1.9631667065295385e-05,
      "loss": 0.032,
      "step": 20045
    },
    {
      "epoch": 2.8095304835318853,
      "grad_norm": 0.17425920069217682,
      "learning_rate": 1.961731643147572e-05,
      "loss": 0.0296,
      "step": 20046
    },
    {
      "epoch": 2.8096706377014717,
      "grad_norm": 0.23781277239322662,
      "learning_rate": 1.9602965797656062e-05,
      "loss": 0.0107,
      "step": 20047
    },
    {
      "epoch": 2.809810791871058,
      "grad_norm": 0.3627985417842865,
      "learning_rate": 1.95886151638364e-05,
      "loss": 0.0181,
      "step": 20048
    },
    {
      "epoch": 2.8099509460406447,
      "grad_norm": 0.18137101829051971,
      "learning_rate": 1.9574264530016743e-05,
      "loss": 0.0219,
      "step": 20049
    },
    {
      "epoch": 2.810091100210231,
      "grad_norm": 0.260376513004303,
      "learning_rate": 1.955991389619708e-05,
      "loss": 0.048,
      "step": 20050
    },
    {
      "epoch": 2.8102312543798176,
      "grad_norm": 0.07062747329473495,
      "learning_rate": 1.954556326237742e-05,
      "loss": 0.0118,
      "step": 20051
    },
    {
      "epoch": 2.8103714085494045,
      "grad_norm": 0.360655814409256,
      "learning_rate": 1.953121262855776e-05,
      "loss": 0.0428,
      "step": 20052
    },
    {
      "epoch": 2.810511562718991,
      "grad_norm": 0.2344047725200653,
      "learning_rate": 1.95168619947381e-05,
      "loss": 0.0195,
      "step": 20053
    },
    {
      "epoch": 2.8106517168885774,
      "grad_norm": 0.06112000346183777,
      "learning_rate": 1.9502511360918437e-05,
      "loss": 0.0044,
      "step": 20054
    },
    {
      "epoch": 2.810791871058164,
      "grad_norm": 0.09430229663848877,
      "learning_rate": 1.948816072709878e-05,
      "loss": 0.0096,
      "step": 20055
    },
    {
      "epoch": 2.8109320252277508,
      "grad_norm": 0.49903953075408936,
      "learning_rate": 1.9473810093279118e-05,
      "loss": 0.0249,
      "step": 20056
    },
    {
      "epoch": 2.8110721793973372,
      "grad_norm": 0.20669111609458923,
      "learning_rate": 1.945945945945946e-05,
      "loss": 0.0104,
      "step": 20057
    },
    {
      "epoch": 2.8112123335669237,
      "grad_norm": 0.06677598506212234,
      "learning_rate": 1.9445108825639795e-05,
      "loss": 0.0034,
      "step": 20058
    },
    {
      "epoch": 2.81135248773651,
      "grad_norm": 0.20507769286632538,
      "learning_rate": 1.9430758191820137e-05,
      "loss": 0.0406,
      "step": 20059
    },
    {
      "epoch": 2.8114926419060966,
      "grad_norm": 0.20784273743629456,
      "learning_rate": 1.9416407558000476e-05,
      "loss": 0.0269,
      "step": 20060
    },
    {
      "epoch": 2.811632796075683,
      "grad_norm": 0.1675213873386383,
      "learning_rate": 1.9402056924180818e-05,
      "loss": 0.0206,
      "step": 20061
    },
    {
      "epoch": 2.8117729502452695,
      "grad_norm": 0.5399994850158691,
      "learning_rate": 1.9387706290361153e-05,
      "loss": 0.0202,
      "step": 20062
    },
    {
      "epoch": 2.8119131044148564,
      "grad_norm": 0.02934509515762329,
      "learning_rate": 1.9373355656541495e-05,
      "loss": 0.0029,
      "step": 20063
    },
    {
      "epoch": 2.812053258584443,
      "grad_norm": 0.02207348309457302,
      "learning_rate": 1.9359005022721834e-05,
      "loss": 0.0013,
      "step": 20064
    },
    {
      "epoch": 2.8121934127540293,
      "grad_norm": 0.9961301684379578,
      "learning_rate": 1.9344654388902176e-05,
      "loss": 0.0278,
      "step": 20065
    },
    {
      "epoch": 2.8123335669236162,
      "grad_norm": 0.2945929169654846,
      "learning_rate": 1.933030375508251e-05,
      "loss": 0.0422,
      "step": 20066
    },
    {
      "epoch": 2.8124737210932027,
      "grad_norm": 0.5682245492935181,
      "learning_rate": 1.9315953121262854e-05,
      "loss": 0.0893,
      "step": 20067
    },
    {
      "epoch": 2.812613875262789,
      "grad_norm": 0.3704012930393219,
      "learning_rate": 1.9301602487443192e-05,
      "loss": 0.032,
      "step": 20068
    },
    {
      "epoch": 2.8127540294323756,
      "grad_norm": 1.348705530166626,
      "learning_rate": 1.9287251853623535e-05,
      "loss": 0.0358,
      "step": 20069
    },
    {
      "epoch": 2.812894183601962,
      "grad_norm": 0.21642029285430908,
      "learning_rate": 1.927290121980387e-05,
      "loss": 0.0048,
      "step": 20070
    },
    {
      "epoch": 2.8130343377715485,
      "grad_norm": 0.5931062698364258,
      "learning_rate": 1.9258550585984212e-05,
      "loss": 0.0443,
      "step": 20071
    },
    {
      "epoch": 2.813174491941135,
      "grad_norm": 0.1410433053970337,
      "learning_rate": 1.924419995216455e-05,
      "loss": 0.0163,
      "step": 20072
    },
    {
      "epoch": 2.813314646110722,
      "grad_norm": 0.2444588541984558,
      "learning_rate": 1.9229849318344893e-05,
      "loss": 0.0616,
      "step": 20073
    },
    {
      "epoch": 2.8134548002803084,
      "grad_norm": 0.11151965707540512,
      "learning_rate": 1.9215498684525228e-05,
      "loss": 0.0221,
      "step": 20074
    },
    {
      "epoch": 2.813594954449895,
      "grad_norm": 0.2686412036418915,
      "learning_rate": 1.920114805070557e-05,
      "loss": 0.019,
      "step": 20075
    },
    {
      "epoch": 2.8137351086194813,
      "grad_norm": 0.3521868586540222,
      "learning_rate": 1.918679741688591e-05,
      "loss": 0.0322,
      "step": 20076
    },
    {
      "epoch": 2.813875262789068,
      "grad_norm": 0.010168555192649364,
      "learning_rate": 1.917244678306625e-05,
      "loss": 0.0011,
      "step": 20077
    },
    {
      "epoch": 2.8140154169586546,
      "grad_norm": 0.3403485119342804,
      "learning_rate": 1.9158096149246593e-05,
      "loss": 0.023,
      "step": 20078
    },
    {
      "epoch": 2.814155571128241,
      "grad_norm": 0.13261543214321136,
      "learning_rate": 1.914374551542693e-05,
      "loss": 0.0055,
      "step": 20079
    },
    {
      "epoch": 2.8142957252978276,
      "grad_norm": 0.23161859810352325,
      "learning_rate": 1.912939488160727e-05,
      "loss": 0.0143,
      "step": 20080
    },
    {
      "epoch": 2.814435879467414,
      "grad_norm": 0.4936937391757965,
      "learning_rate": 1.911504424778761e-05,
      "loss": 0.0411,
      "step": 20081
    },
    {
      "epoch": 2.8145760336370005,
      "grad_norm": 0.3196813464164734,
      "learning_rate": 1.910069361396795e-05,
      "loss": 0.0417,
      "step": 20082
    },
    {
      "epoch": 2.8147161878065874,
      "grad_norm": 0.29124414920806885,
      "learning_rate": 1.9086342980148287e-05,
      "loss": 0.0221,
      "step": 20083
    },
    {
      "epoch": 2.814856341976174,
      "grad_norm": 0.12162228673696518,
      "learning_rate": 1.907199234632863e-05,
      "loss": 0.013,
      "step": 20084
    },
    {
      "epoch": 2.8149964961457603,
      "grad_norm": 0.060638755559921265,
      "learning_rate": 1.9057641712508968e-05,
      "loss": 0.0076,
      "step": 20085
    },
    {
      "epoch": 2.8151366503153468,
      "grad_norm": 0.31550222635269165,
      "learning_rate": 1.904329107868931e-05,
      "loss": 0.0527,
      "step": 20086
    },
    {
      "epoch": 2.8152768044849337,
      "grad_norm": 0.08666808158159256,
      "learning_rate": 1.9028940444869645e-05,
      "loss": 0.0029,
      "step": 20087
    },
    {
      "epoch": 2.81541695865452,
      "grad_norm": 0.48483553528785706,
      "learning_rate": 1.9014589811049987e-05,
      "loss": 0.0477,
      "step": 20088
    },
    {
      "epoch": 2.8155571128241066,
      "grad_norm": 0.13772916793823242,
      "learning_rate": 1.9000239177230326e-05,
      "loss": 0.0133,
      "step": 20089
    },
    {
      "epoch": 2.815697266993693,
      "grad_norm": 0.05882766470313072,
      "learning_rate": 1.8985888543410668e-05,
      "loss": 0.0068,
      "step": 20090
    },
    {
      "epoch": 2.8158374211632795,
      "grad_norm": 0.1865001618862152,
      "learning_rate": 1.8971537909591003e-05,
      "loss": 0.0291,
      "step": 20091
    },
    {
      "epoch": 2.815977575332866,
      "grad_norm": 0.1174512431025505,
      "learning_rate": 1.8957187275771346e-05,
      "loss": 0.0425,
      "step": 20092
    },
    {
      "epoch": 2.8161177295024524,
      "grad_norm": 0.024458911269903183,
      "learning_rate": 1.8942836641951684e-05,
      "loss": 0.0012,
      "step": 20093
    },
    {
      "epoch": 2.8162578836720393,
      "grad_norm": 0.06882781535387039,
      "learning_rate": 1.8928486008132026e-05,
      "loss": 0.0081,
      "step": 20094
    },
    {
      "epoch": 2.816398037841626,
      "grad_norm": 0.10022664815187454,
      "learning_rate": 1.8914135374312362e-05,
      "loss": 0.016,
      "step": 20095
    },
    {
      "epoch": 2.8165381920112122,
      "grad_norm": 0.4673845171928406,
      "learning_rate": 1.8899784740492704e-05,
      "loss": 0.0601,
      "step": 20096
    },
    {
      "epoch": 2.816678346180799,
      "grad_norm": 0.4063095152378082,
      "learning_rate": 1.8885434106673043e-05,
      "loss": 0.045,
      "step": 20097
    },
    {
      "epoch": 2.8168185003503856,
      "grad_norm": 0.21431665122509003,
      "learning_rate": 1.8871083472853385e-05,
      "loss": 0.043,
      "step": 20098
    },
    {
      "epoch": 2.816958654519972,
      "grad_norm": 0.028018292039632797,
      "learning_rate": 1.885673283903372e-05,
      "loss": 0.0029,
      "step": 20099
    },
    {
      "epoch": 2.8170988086895585,
      "grad_norm": 0.2614838480949402,
      "learning_rate": 1.8842382205214062e-05,
      "loss": 0.0325,
      "step": 20100
    },
    {
      "epoch": 2.817238962859145,
      "grad_norm": 0.21096856892108917,
      "learning_rate": 1.88280315713944e-05,
      "loss": 0.0667,
      "step": 20101
    },
    {
      "epoch": 2.8173791170287315,
      "grad_norm": 0.04970110207796097,
      "learning_rate": 1.8813680937574743e-05,
      "loss": 0.0063,
      "step": 20102
    },
    {
      "epoch": 2.817519271198318,
      "grad_norm": 0.1813165843486786,
      "learning_rate": 1.8799330303755078e-05,
      "loss": 0.0095,
      "step": 20103
    },
    {
      "epoch": 2.817659425367905,
      "grad_norm": 0.08293382823467255,
      "learning_rate": 1.878497966993542e-05,
      "loss": 0.0068,
      "step": 20104
    },
    {
      "epoch": 2.8177995795374913,
      "grad_norm": 0.22900699079036713,
      "learning_rate": 1.877062903611576e-05,
      "loss": 0.057,
      "step": 20105
    },
    {
      "epoch": 2.8179397337070777,
      "grad_norm": 0.11686235666275024,
      "learning_rate": 1.87562784022961e-05,
      "loss": 0.0378,
      "step": 20106
    },
    {
      "epoch": 2.818079887876664,
      "grad_norm": 0.15948370099067688,
      "learning_rate": 1.874192776847644e-05,
      "loss": 0.0097,
      "step": 20107
    },
    {
      "epoch": 2.818220042046251,
      "grad_norm": 0.3438152074813843,
      "learning_rate": 1.872757713465678e-05,
      "loss": 0.0179,
      "step": 20108
    },
    {
      "epoch": 2.8183601962158376,
      "grad_norm": 0.17170532047748566,
      "learning_rate": 1.8713226500837117e-05,
      "loss": 0.005,
      "step": 20109
    },
    {
      "epoch": 2.818500350385424,
      "grad_norm": 0.16956299543380737,
      "learning_rate": 1.869887586701746e-05,
      "loss": 0.0407,
      "step": 20110
    },
    {
      "epoch": 2.8186405045550105,
      "grad_norm": 0.22396832704544067,
      "learning_rate": 1.8684525233197798e-05,
      "loss": 0.0183,
      "step": 20111
    },
    {
      "epoch": 2.818780658724597,
      "grad_norm": 0.4402769207954407,
      "learning_rate": 1.8670174599378137e-05,
      "loss": 0.0637,
      "step": 20112
    },
    {
      "epoch": 2.8189208128941834,
      "grad_norm": 0.19635671377182007,
      "learning_rate": 1.8655823965558476e-05,
      "loss": 0.0198,
      "step": 20113
    },
    {
      "epoch": 2.8190609670637703,
      "grad_norm": 0.22729657590389252,
      "learning_rate": 1.8641473331738818e-05,
      "loss": 0.0069,
      "step": 20114
    },
    {
      "epoch": 2.8192011212333568,
      "grad_norm": 0.21595925092697144,
      "learning_rate": 1.8627122697919157e-05,
      "loss": 0.0341,
      "step": 20115
    },
    {
      "epoch": 2.819341275402943,
      "grad_norm": 0.36268818378448486,
      "learning_rate": 1.8612772064099495e-05,
      "loss": 0.0249,
      "step": 20116
    },
    {
      "epoch": 2.8194814295725297,
      "grad_norm": 0.06712613254785538,
      "learning_rate": 1.8598421430279834e-05,
      "loss": 0.0021,
      "step": 20117
    },
    {
      "epoch": 2.8196215837421166,
      "grad_norm": 0.06468553096055984,
      "learning_rate": 1.8584070796460176e-05,
      "loss": 0.0018,
      "step": 20118
    },
    {
      "epoch": 2.819761737911703,
      "grad_norm": 0.5762037634849548,
      "learning_rate": 1.8569720162640515e-05,
      "loss": 0.0601,
      "step": 20119
    },
    {
      "epoch": 2.8199018920812895,
      "grad_norm": 0.1784004271030426,
      "learning_rate": 1.8555369528820854e-05,
      "loss": 0.0053,
      "step": 20120
    },
    {
      "epoch": 2.820042046250876,
      "grad_norm": 0.07527387142181396,
      "learning_rate": 1.8541018895001192e-05,
      "loss": 0.007,
      "step": 20121
    },
    {
      "epoch": 2.8201822004204624,
      "grad_norm": 0.08619034290313721,
      "learning_rate": 1.8526668261181534e-05,
      "loss": 0.0069,
      "step": 20122
    },
    {
      "epoch": 2.820322354590049,
      "grad_norm": 0.14755816757678986,
      "learning_rate": 1.8512317627361873e-05,
      "loss": 0.0356,
      "step": 20123
    },
    {
      "epoch": 2.8204625087596353,
      "grad_norm": 0.13590243458747864,
      "learning_rate": 1.8497966993542212e-05,
      "loss": 0.006,
      "step": 20124
    },
    {
      "epoch": 2.8206026629292222,
      "grad_norm": 0.17464981973171234,
      "learning_rate": 1.848361635972255e-05,
      "loss": 0.0267,
      "step": 20125
    },
    {
      "epoch": 2.8207428170988087,
      "grad_norm": 0.1709974855184555,
      "learning_rate": 1.8469265725902893e-05,
      "loss": 0.0197,
      "step": 20126
    },
    {
      "epoch": 2.820882971268395,
      "grad_norm": 0.14928732812404633,
      "learning_rate": 1.845491509208323e-05,
      "loss": 0.0173,
      "step": 20127
    },
    {
      "epoch": 2.8210231254379816,
      "grad_norm": 0.5162956118583679,
      "learning_rate": 1.844056445826357e-05,
      "loss": 0.0189,
      "step": 20128
    },
    {
      "epoch": 2.8211632796075685,
      "grad_norm": 0.3384089469909668,
      "learning_rate": 1.842621382444391e-05,
      "loss": 0.0618,
      "step": 20129
    },
    {
      "epoch": 2.821303433777155,
      "grad_norm": 0.08030930161476135,
      "learning_rate": 1.841186319062425e-05,
      "loss": 0.007,
      "step": 20130
    },
    {
      "epoch": 2.8214435879467414,
      "grad_norm": 0.2609022557735443,
      "learning_rate": 1.839751255680459e-05,
      "loss": 0.0179,
      "step": 20131
    },
    {
      "epoch": 2.821583742116328,
      "grad_norm": 0.6638238430023193,
      "learning_rate": 1.8383161922984932e-05,
      "loss": 0.0501,
      "step": 20132
    },
    {
      "epoch": 2.8217238962859144,
      "grad_norm": 0.07239685207605362,
      "learning_rate": 1.836881128916527e-05,
      "loss": 0.0203,
      "step": 20133
    },
    {
      "epoch": 2.821864050455501,
      "grad_norm": 0.3404446840286255,
      "learning_rate": 1.835446065534561e-05,
      "loss": 0.0576,
      "step": 20134
    },
    {
      "epoch": 2.8220042046250877,
      "grad_norm": 0.331024169921875,
      "learning_rate": 1.834011002152595e-05,
      "loss": 0.0086,
      "step": 20135
    },
    {
      "epoch": 2.822144358794674,
      "grad_norm": 0.3042827248573303,
      "learning_rate": 1.832575938770629e-05,
      "loss": 0.012,
      "step": 20136
    },
    {
      "epoch": 2.8222845129642606,
      "grad_norm": 0.04164764657616615,
      "learning_rate": 1.831140875388663e-05,
      "loss": 0.0034,
      "step": 20137
    },
    {
      "epoch": 2.822424667133847,
      "grad_norm": 0.0468953438103199,
      "learning_rate": 1.8297058120066968e-05,
      "loss": 0.0022,
      "step": 20138
    },
    {
      "epoch": 2.822564821303434,
      "grad_norm": 0.501755952835083,
      "learning_rate": 1.828270748624731e-05,
      "loss": 0.0602,
      "step": 20139
    },
    {
      "epoch": 2.8227049754730205,
      "grad_norm": 0.1853044480085373,
      "learning_rate": 1.826835685242765e-05,
      "loss": 0.0081,
      "step": 20140
    },
    {
      "epoch": 2.822845129642607,
      "grad_norm": 0.00757548026740551,
      "learning_rate": 1.8254006218607987e-05,
      "loss": 0.0009,
      "step": 20141
    },
    {
      "epoch": 2.8229852838121934,
      "grad_norm": 0.17804817855358124,
      "learning_rate": 1.8239655584788326e-05,
      "loss": 0.0149,
      "step": 20142
    },
    {
      "epoch": 2.82312543798178,
      "grad_norm": 0.33779144287109375,
      "learning_rate": 1.8225304950968668e-05,
      "loss": 0.0253,
      "step": 20143
    },
    {
      "epoch": 2.8232655921513663,
      "grad_norm": 0.1400834619998932,
      "learning_rate": 1.8210954317149007e-05,
      "loss": 0.029,
      "step": 20144
    },
    {
      "epoch": 2.823405746320953,
      "grad_norm": 0.16597984731197357,
      "learning_rate": 1.8196603683329345e-05,
      "loss": 0.0249,
      "step": 20145
    },
    {
      "epoch": 2.8235459004905397,
      "grad_norm": 0.08694522082805634,
      "learning_rate": 1.8182253049509684e-05,
      "loss": 0.0038,
      "step": 20146
    },
    {
      "epoch": 2.823686054660126,
      "grad_norm": 0.5231955647468567,
      "learning_rate": 1.8167902415690026e-05,
      "loss": 0.0289,
      "step": 20147
    },
    {
      "epoch": 2.8238262088297126,
      "grad_norm": 0.06470644474029541,
      "learning_rate": 1.8153551781870365e-05,
      "loss": 0.011,
      "step": 20148
    },
    {
      "epoch": 2.8239663629992995,
      "grad_norm": 0.5129496455192566,
      "learning_rate": 1.8139201148050704e-05,
      "loss": 0.0843,
      "step": 20149
    },
    {
      "epoch": 2.824106517168886,
      "grad_norm": 0.011135991662740707,
      "learning_rate": 1.8124850514231042e-05,
      "loss": 0.0011,
      "step": 20150
    },
    {
      "epoch": 2.8242466713384724,
      "grad_norm": 0.18264475464820862,
      "learning_rate": 1.8110499880411384e-05,
      "loss": 0.0461,
      "step": 20151
    },
    {
      "epoch": 2.824386825508059,
      "grad_norm": 0.08170921355485916,
      "learning_rate": 1.8096149246591723e-05,
      "loss": 0.0055,
      "step": 20152
    },
    {
      "epoch": 2.8245269796776453,
      "grad_norm": 0.2484898865222931,
      "learning_rate": 1.8081798612772062e-05,
      "loss": 0.0149,
      "step": 20153
    },
    {
      "epoch": 2.824667133847232,
      "grad_norm": 0.25872939825057983,
      "learning_rate": 1.80674479789524e-05,
      "loss": 0.0401,
      "step": 20154
    },
    {
      "epoch": 2.8248072880168182,
      "grad_norm": 0.0702553391456604,
      "learning_rate": 1.8053097345132743e-05,
      "loss": 0.0034,
      "step": 20155
    },
    {
      "epoch": 2.824947442186405,
      "grad_norm": 0.13753172755241394,
      "learning_rate": 1.803874671131308e-05,
      "loss": 0.0259,
      "step": 20156
    },
    {
      "epoch": 2.8250875963559916,
      "grad_norm": 0.31921085715293884,
      "learning_rate": 1.802439607749342e-05,
      "loss": 0.0301,
      "step": 20157
    },
    {
      "epoch": 2.825227750525578,
      "grad_norm": 0.1717853844165802,
      "learning_rate": 1.801004544367376e-05,
      "loss": 0.0084,
      "step": 20158
    },
    {
      "epoch": 2.8253679046951645,
      "grad_norm": 0.0830700471997261,
      "learning_rate": 1.79956948098541e-05,
      "loss": 0.0187,
      "step": 20159
    },
    {
      "epoch": 2.8255080588647514,
      "grad_norm": 0.17259301245212555,
      "learning_rate": 1.798134417603444e-05,
      "loss": 0.0119,
      "step": 20160
    },
    {
      "epoch": 2.825648213034338,
      "grad_norm": 0.37203341722488403,
      "learning_rate": 1.796699354221478e-05,
      "loss": 0.0252,
      "step": 20161
    },
    {
      "epoch": 2.8257883672039243,
      "grad_norm": 0.6927008628845215,
      "learning_rate": 1.7952642908395117e-05,
      "loss": 0.0461,
      "step": 20162
    },
    {
      "epoch": 2.825928521373511,
      "grad_norm": 0.1689605414867401,
      "learning_rate": 1.793829227457546e-05,
      "loss": 0.0209,
      "step": 20163
    },
    {
      "epoch": 2.8260686755430973,
      "grad_norm": 0.14327998459339142,
      "learning_rate": 1.7923941640755798e-05,
      "loss": 0.0028,
      "step": 20164
    },
    {
      "epoch": 2.8262088297126837,
      "grad_norm": 0.3046146035194397,
      "learning_rate": 1.7909591006936137e-05,
      "loss": 0.0274,
      "step": 20165
    },
    {
      "epoch": 2.8263489838822706,
      "grad_norm": 0.2714897096157074,
      "learning_rate": 1.7895240373116476e-05,
      "loss": 0.0018,
      "step": 20166
    },
    {
      "epoch": 2.826489138051857,
      "grad_norm": 0.1348365992307663,
      "learning_rate": 1.7880889739296818e-05,
      "loss": 0.0153,
      "step": 20167
    },
    {
      "epoch": 2.8266292922214435,
      "grad_norm": 0.5544025897979736,
      "learning_rate": 1.7866539105477156e-05,
      "loss": 0.0614,
      "step": 20168
    },
    {
      "epoch": 2.82676944639103,
      "grad_norm": 0.10236028581857681,
      "learning_rate": 1.7852188471657495e-05,
      "loss": 0.0066,
      "step": 20169
    },
    {
      "epoch": 2.826909600560617,
      "grad_norm": 1.294047474861145,
      "learning_rate": 1.7837837837837837e-05,
      "loss": 0.1696,
      "step": 20170
    },
    {
      "epoch": 2.8270497547302034,
      "grad_norm": 0.6213603615760803,
      "learning_rate": 1.7823487204018176e-05,
      "loss": 0.093,
      "step": 20171
    },
    {
      "epoch": 2.82718990889979,
      "grad_norm": 0.050013910979032516,
      "learning_rate": 1.7809136570198518e-05,
      "loss": 0.0049,
      "step": 20172
    },
    {
      "epoch": 2.8273300630693763,
      "grad_norm": 0.051053084433078766,
      "learning_rate": 1.7794785936378857e-05,
      "loss": 0.0059,
      "step": 20173
    },
    {
      "epoch": 2.8274702172389627,
      "grad_norm": 0.1913643628358841,
      "learning_rate": 1.7780435302559195e-05,
      "loss": 0.0266,
      "step": 20174
    },
    {
      "epoch": 2.827610371408549,
      "grad_norm": 0.15279848873615265,
      "learning_rate": 1.7766084668739534e-05,
      "loss": 0.0068,
      "step": 20175
    },
    {
      "epoch": 2.827750525578136,
      "grad_norm": 0.07097769528627396,
      "learning_rate": 1.7751734034919876e-05,
      "loss": 0.0076,
      "step": 20176
    },
    {
      "epoch": 2.8278906797477226,
      "grad_norm": 0.13913406431674957,
      "learning_rate": 1.7737383401100215e-05,
      "loss": 0.0144,
      "step": 20177
    },
    {
      "epoch": 2.828030833917309,
      "grad_norm": 0.15220694243907928,
      "learning_rate": 1.7723032767280554e-05,
      "loss": 0.0342,
      "step": 20178
    },
    {
      "epoch": 2.8281709880868955,
      "grad_norm": 0.23292815685272217,
      "learning_rate": 1.7708682133460892e-05,
      "loss": 0.029,
      "step": 20179
    },
    {
      "epoch": 2.8283111422564824,
      "grad_norm": 0.4258395731449127,
      "learning_rate": 1.7694331499641235e-05,
      "loss": 0.0283,
      "step": 20180
    },
    {
      "epoch": 2.828451296426069,
      "grad_norm": 0.8651798367500305,
      "learning_rate": 1.7679980865821573e-05,
      "loss": 0.0343,
      "step": 20181
    },
    {
      "epoch": 2.8285914505956553,
      "grad_norm": 0.09384683519601822,
      "learning_rate": 1.7665630232001912e-05,
      "loss": 0.0242,
      "step": 20182
    },
    {
      "epoch": 2.8287316047652418,
      "grad_norm": 0.22945542633533478,
      "learning_rate": 1.765127959818225e-05,
      "loss": 0.0369,
      "step": 20183
    },
    {
      "epoch": 2.8288717589348282,
      "grad_norm": 0.04409319907426834,
      "learning_rate": 1.7636928964362593e-05,
      "loss": 0.0049,
      "step": 20184
    },
    {
      "epoch": 2.8290119131044147,
      "grad_norm": 0.2397792786359787,
      "learning_rate": 1.762257833054293e-05,
      "loss": 0.0374,
      "step": 20185
    },
    {
      "epoch": 2.829152067274001,
      "grad_norm": 0.20370949804782867,
      "learning_rate": 1.760822769672327e-05,
      "loss": 0.0116,
      "step": 20186
    },
    {
      "epoch": 2.829292221443588,
      "grad_norm": 0.1491415798664093,
      "learning_rate": 1.759387706290361e-05,
      "loss": 0.0152,
      "step": 20187
    },
    {
      "epoch": 2.8294323756131745,
      "grad_norm": 0.22468191385269165,
      "learning_rate": 1.757952642908395e-05,
      "loss": 0.015,
      "step": 20188
    },
    {
      "epoch": 2.829572529782761,
      "grad_norm": 0.19595931470394135,
      "learning_rate": 1.756517579526429e-05,
      "loss": 0.0694,
      "step": 20189
    },
    {
      "epoch": 2.8297126839523474,
      "grad_norm": 0.21717269718647003,
      "learning_rate": 1.755082516144463e-05,
      "loss": 0.0135,
      "step": 20190
    },
    {
      "epoch": 2.8298528381219343,
      "grad_norm": 0.3220983147621155,
      "learning_rate": 1.7536474527624967e-05,
      "loss": 0.0242,
      "step": 20191
    },
    {
      "epoch": 2.829992992291521,
      "grad_norm": 0.08474330604076385,
      "learning_rate": 1.752212389380531e-05,
      "loss": 0.0175,
      "step": 20192
    },
    {
      "epoch": 2.8301331464611073,
      "grad_norm": 0.07432655245065689,
      "learning_rate": 1.7507773259985648e-05,
      "loss": 0.0095,
      "step": 20193
    },
    {
      "epoch": 2.8302733006306937,
      "grad_norm": 0.49580323696136475,
      "learning_rate": 1.7493422626165987e-05,
      "loss": 0.0462,
      "step": 20194
    },
    {
      "epoch": 2.83041345480028,
      "grad_norm": 0.05561661347746849,
      "learning_rate": 1.7479071992346326e-05,
      "loss": 0.0049,
      "step": 20195
    },
    {
      "epoch": 2.8305536089698666,
      "grad_norm": 0.2758733332157135,
      "learning_rate": 1.7464721358526668e-05,
      "loss": 0.0574,
      "step": 20196
    },
    {
      "epoch": 2.8306937631394535,
      "grad_norm": 0.566324770450592,
      "learning_rate": 1.7450370724707006e-05,
      "loss": 0.0299,
      "step": 20197
    },
    {
      "epoch": 2.83083391730904,
      "grad_norm": 0.24207179248332977,
      "learning_rate": 1.7436020090887345e-05,
      "loss": 0.0642,
      "step": 20198
    },
    {
      "epoch": 2.8309740714786265,
      "grad_norm": 0.3416862487792969,
      "learning_rate": 1.7421669457067684e-05,
      "loss": 0.0298,
      "step": 20199
    },
    {
      "epoch": 2.831114225648213,
      "grad_norm": 0.17715443670749664,
      "learning_rate": 1.7407318823248026e-05,
      "loss": 0.0105,
      "step": 20200
    },
    {
      "epoch": 2.8312543798178,
      "grad_norm": 0.18120242655277252,
      "learning_rate": 1.7392968189428365e-05,
      "loss": 0.011,
      "step": 20201
    },
    {
      "epoch": 2.8313945339873863,
      "grad_norm": 0.11265964061021805,
      "learning_rate": 1.7378617555608703e-05,
      "loss": 0.0447,
      "step": 20202
    },
    {
      "epoch": 2.8315346881569727,
      "grad_norm": 0.3904508650302887,
      "learning_rate": 1.7364266921789042e-05,
      "loss": 0.0408,
      "step": 20203
    },
    {
      "epoch": 2.831674842326559,
      "grad_norm": 0.025759408250451088,
      "learning_rate": 1.7349916287969384e-05,
      "loss": 0.0025,
      "step": 20204
    },
    {
      "epoch": 2.8318149964961457,
      "grad_norm": 0.05911707133054733,
      "learning_rate": 1.7335565654149723e-05,
      "loss": 0.0061,
      "step": 20205
    },
    {
      "epoch": 2.831955150665732,
      "grad_norm": 0.1870964765548706,
      "learning_rate": 1.7321215020330062e-05,
      "loss": 0.0215,
      "step": 20206
    },
    {
      "epoch": 2.832095304835319,
      "grad_norm": 0.30868837237358093,
      "learning_rate": 1.7306864386510404e-05,
      "loss": 0.0247,
      "step": 20207
    },
    {
      "epoch": 2.8322354590049055,
      "grad_norm": 0.17618998885154724,
      "learning_rate": 1.7292513752690743e-05,
      "loss": 0.007,
      "step": 20208
    },
    {
      "epoch": 2.832375613174492,
      "grad_norm": 0.3716455399990082,
      "learning_rate": 1.727816311887108e-05,
      "loss": 0.0337,
      "step": 20209
    },
    {
      "epoch": 2.8325157673440784,
      "grad_norm": 0.1306070238351822,
      "learning_rate": 1.7263812485051423e-05,
      "loss": 0.0137,
      "step": 20210
    },
    {
      "epoch": 2.8326559215136653,
      "grad_norm": 0.3776424527168274,
      "learning_rate": 1.7249461851231762e-05,
      "loss": 0.0076,
      "step": 20211
    },
    {
      "epoch": 2.8327960756832518,
      "grad_norm": 0.09000629931688309,
      "learning_rate": 1.72351112174121e-05,
      "loss": 0.0169,
      "step": 20212
    },
    {
      "epoch": 2.832936229852838,
      "grad_norm": 0.1441204845905304,
      "learning_rate": 1.722076058359244e-05,
      "loss": 0.0145,
      "step": 20213
    },
    {
      "epoch": 2.8330763840224247,
      "grad_norm": 0.08253537118434906,
      "learning_rate": 1.7206409949772782e-05,
      "loss": 0.0036,
      "step": 20214
    },
    {
      "epoch": 2.833216538192011,
      "grad_norm": 0.014261264353990555,
      "learning_rate": 1.719205931595312e-05,
      "loss": 0.0015,
      "step": 20215
    },
    {
      "epoch": 2.8333566923615976,
      "grad_norm": 0.5262055993080139,
      "learning_rate": 1.717770868213346e-05,
      "loss": 0.0391,
      "step": 20216
    },
    {
      "epoch": 2.833496846531184,
      "grad_norm": 0.025808749720454216,
      "learning_rate": 1.7163358048313798e-05,
      "loss": 0.0019,
      "step": 20217
    },
    {
      "epoch": 2.833637000700771,
      "grad_norm": 0.6952418684959412,
      "learning_rate": 1.714900741449414e-05,
      "loss": 0.072,
      "step": 20218
    },
    {
      "epoch": 2.8337771548703574,
      "grad_norm": 0.6326785683631897,
      "learning_rate": 1.713465678067448e-05,
      "loss": 0.0347,
      "step": 20219
    },
    {
      "epoch": 2.833917309039944,
      "grad_norm": 0.443508118391037,
      "learning_rate": 1.7120306146854817e-05,
      "loss": 0.0247,
      "step": 20220
    },
    {
      "epoch": 2.8340574632095303,
      "grad_norm": 0.26656726002693176,
      "learning_rate": 1.7105955513035156e-05,
      "loss": 0.0348,
      "step": 20221
    },
    {
      "epoch": 2.8341976173791172,
      "grad_norm": 0.28205418586730957,
      "learning_rate": 1.7091604879215498e-05,
      "loss": 0.019,
      "step": 20222
    },
    {
      "epoch": 2.8343377715487037,
      "grad_norm": 0.041545867919921875,
      "learning_rate": 1.7077254245395837e-05,
      "loss": 0.0043,
      "step": 20223
    },
    {
      "epoch": 2.83447792571829,
      "grad_norm": 0.3739026188850403,
      "learning_rate": 1.7062903611576176e-05,
      "loss": 0.0489,
      "step": 20224
    },
    {
      "epoch": 2.8346180798878766,
      "grad_norm": 0.09915999323129654,
      "learning_rate": 1.7048552977756514e-05,
      "loss": 0.0105,
      "step": 20225
    },
    {
      "epoch": 2.834758234057463,
      "grad_norm": 0.5638039708137512,
      "learning_rate": 1.7034202343936857e-05,
      "loss": 0.0191,
      "step": 20226
    },
    {
      "epoch": 2.8348983882270495,
      "grad_norm": 0.15827758610248566,
      "learning_rate": 1.7019851710117195e-05,
      "loss": 0.014,
      "step": 20227
    },
    {
      "epoch": 2.8350385423966364,
      "grad_norm": 0.2947181463241577,
      "learning_rate": 1.7005501076297534e-05,
      "loss": 0.0675,
      "step": 20228
    },
    {
      "epoch": 2.835178696566223,
      "grad_norm": 0.1616363823413849,
      "learning_rate": 1.6991150442477873e-05,
      "loss": 0.0326,
      "step": 20229
    },
    {
      "epoch": 2.8353188507358094,
      "grad_norm": 0.17346526682376862,
      "learning_rate": 1.6976799808658215e-05,
      "loss": 0.0333,
      "step": 20230
    },
    {
      "epoch": 2.835459004905396,
      "grad_norm": 0.25854647159576416,
      "learning_rate": 1.6962449174838554e-05,
      "loss": 0.0241,
      "step": 20231
    },
    {
      "epoch": 2.8355991590749827,
      "grad_norm": 0.23596741259098053,
      "learning_rate": 1.6948098541018892e-05,
      "loss": 0.0119,
      "step": 20232
    },
    {
      "epoch": 2.835739313244569,
      "grad_norm": 0.19777964055538177,
      "learning_rate": 1.693374790719923e-05,
      "loss": 0.0366,
      "step": 20233
    },
    {
      "epoch": 2.8358794674141556,
      "grad_norm": 0.31582239270210266,
      "learning_rate": 1.6919397273379573e-05,
      "loss": 0.0292,
      "step": 20234
    },
    {
      "epoch": 2.836019621583742,
      "grad_norm": 0.10154934227466583,
      "learning_rate": 1.6905046639559912e-05,
      "loss": 0.0106,
      "step": 20235
    },
    {
      "epoch": 2.8361597757533286,
      "grad_norm": 0.17357492446899414,
      "learning_rate": 1.689069600574025e-05,
      "loss": 0.0214,
      "step": 20236
    },
    {
      "epoch": 2.836299929922915,
      "grad_norm": 0.33176639676094055,
      "learning_rate": 1.687634537192059e-05,
      "loss": 0.0428,
      "step": 20237
    },
    {
      "epoch": 2.8364400840925015,
      "grad_norm": 0.28931349515914917,
      "learning_rate": 1.686199473810093e-05,
      "loss": 0.0191,
      "step": 20238
    },
    {
      "epoch": 2.8365802382620884,
      "grad_norm": 0.33798739314079285,
      "learning_rate": 1.684764410428127e-05,
      "loss": 0.0554,
      "step": 20239
    },
    {
      "epoch": 2.836720392431675,
      "grad_norm": 0.13870789110660553,
      "learning_rate": 1.683329347046161e-05,
      "loss": 0.0266,
      "step": 20240
    },
    {
      "epoch": 2.8368605466012613,
      "grad_norm": 0.2243368774652481,
      "learning_rate": 1.6818942836641948e-05,
      "loss": 0.032,
      "step": 20241
    },
    {
      "epoch": 2.837000700770848,
      "grad_norm": 0.0687587708234787,
      "learning_rate": 1.680459220282229e-05,
      "loss": 0.0048,
      "step": 20242
    },
    {
      "epoch": 2.8371408549404347,
      "grad_norm": 0.3018030524253845,
      "learning_rate": 1.679024156900263e-05,
      "loss": 0.0346,
      "step": 20243
    },
    {
      "epoch": 2.837281009110021,
      "grad_norm": 0.26874881982803345,
      "learning_rate": 1.677589093518297e-05,
      "loss": 0.0111,
      "step": 20244
    },
    {
      "epoch": 2.8374211632796076,
      "grad_norm": 0.14872173964977264,
      "learning_rate": 1.676154030136331e-05,
      "loss": 0.0126,
      "step": 20245
    },
    {
      "epoch": 2.837561317449194,
      "grad_norm": 0.40505680441856384,
      "learning_rate": 1.6747189667543648e-05,
      "loss": 0.0472,
      "step": 20246
    },
    {
      "epoch": 2.8377014716187805,
      "grad_norm": 0.28024744987487793,
      "learning_rate": 1.673283903372399e-05,
      "loss": 0.0357,
      "step": 20247
    },
    {
      "epoch": 2.837841625788367,
      "grad_norm": 0.14496368169784546,
      "learning_rate": 1.671848839990433e-05,
      "loss": 0.0811,
      "step": 20248
    },
    {
      "epoch": 2.837981779957954,
      "grad_norm": 0.19921255111694336,
      "learning_rate": 1.6704137766084668e-05,
      "loss": 0.027,
      "step": 20249
    },
    {
      "epoch": 2.8381219341275403,
      "grad_norm": 0.11754713952541351,
      "learning_rate": 1.6689787132265006e-05,
      "loss": 0.0085,
      "step": 20250
    },
    {
      "epoch": 2.838262088297127,
      "grad_norm": 0.44583505392074585,
      "learning_rate": 1.667543649844535e-05,
      "loss": 0.0734,
      "step": 20251
    },
    {
      "epoch": 2.8384022424667132,
      "grad_norm": 1.23448646068573,
      "learning_rate": 1.6661085864625687e-05,
      "loss": 0.0252,
      "step": 20252
    },
    {
      "epoch": 2.8385423966363,
      "grad_norm": 0.05516646429896355,
      "learning_rate": 1.6646735230806026e-05,
      "loss": 0.0061,
      "step": 20253
    },
    {
      "epoch": 2.8386825508058866,
      "grad_norm": 0.06524958461523056,
      "learning_rate": 1.6632384596986365e-05,
      "loss": 0.0208,
      "step": 20254
    },
    {
      "epoch": 2.838822704975473,
      "grad_norm": 0.25072863698005676,
      "learning_rate": 1.6618033963166707e-05,
      "loss": 0.0258,
      "step": 20255
    },
    {
      "epoch": 2.8389628591450595,
      "grad_norm": 0.08586220443248749,
      "learning_rate": 1.6603683329347045e-05,
      "loss": 0.0049,
      "step": 20256
    },
    {
      "epoch": 2.839103013314646,
      "grad_norm": 0.11943991482257843,
      "learning_rate": 1.6589332695527384e-05,
      "loss": 0.0074,
      "step": 20257
    },
    {
      "epoch": 2.8392431674842324,
      "grad_norm": 0.0650561973452568,
      "learning_rate": 1.6574982061707723e-05,
      "loss": 0.0035,
      "step": 20258
    },
    {
      "epoch": 2.8393833216538193,
      "grad_norm": 0.05371643230319023,
      "learning_rate": 1.6560631427888065e-05,
      "loss": 0.0027,
      "step": 20259
    },
    {
      "epoch": 2.839523475823406,
      "grad_norm": 0.18674317002296448,
      "learning_rate": 1.6546280794068404e-05,
      "loss": 0.0132,
      "step": 20260
    },
    {
      "epoch": 2.8396636299929923,
      "grad_norm": 0.03560641035437584,
      "learning_rate": 1.6531930160248742e-05,
      "loss": 0.0024,
      "step": 20261
    },
    {
      "epoch": 2.8398037841625787,
      "grad_norm": 0.41691285371780396,
      "learning_rate": 1.651757952642908e-05,
      "loss": 0.0165,
      "step": 20262
    },
    {
      "epoch": 2.8399439383321656,
      "grad_norm": 0.09331688284873962,
      "learning_rate": 1.6503228892609423e-05,
      "loss": 0.0099,
      "step": 20263
    },
    {
      "epoch": 2.840084092501752,
      "grad_norm": 1.6376011371612549,
      "learning_rate": 1.6488878258789762e-05,
      "loss": 0.1782,
      "step": 20264
    },
    {
      "epoch": 2.8402242466713385,
      "grad_norm": 1.7144731283187866,
      "learning_rate": 1.64745276249701e-05,
      "loss": 0.0366,
      "step": 20265
    },
    {
      "epoch": 2.840364400840925,
      "grad_norm": 0.5791053175926208,
      "learning_rate": 1.646017699115044e-05,
      "loss": 0.0161,
      "step": 20266
    },
    {
      "epoch": 2.8405045550105115,
      "grad_norm": 0.29969364404678345,
      "learning_rate": 1.644582635733078e-05,
      "loss": 0.0071,
      "step": 20267
    },
    {
      "epoch": 2.840644709180098,
      "grad_norm": 0.00609160028398037,
      "learning_rate": 1.643147572351112e-05,
      "loss": 0.0005,
      "step": 20268
    },
    {
      "epoch": 2.8407848633496844,
      "grad_norm": 0.48932984471321106,
      "learning_rate": 1.641712508969146e-05,
      "loss": 0.011,
      "step": 20269
    },
    {
      "epoch": 2.8409250175192713,
      "grad_norm": NaN,
      "learning_rate": 1.6402774455871798e-05,
      "loss": 0.1917,
      "step": 20270
    },
    {
      "epoch": 2.8410651716888577,
      "grad_norm": 0.5322683453559875,
      "learning_rate": 1.6402774455871798e-05,
      "loss": 0.0257,
      "step": 20271
    },
    {
      "epoch": 2.841205325858444,
      "grad_norm": 0.5537340641021729,
      "learning_rate": 1.638842382205214e-05,
      "loss": 0.0302,
      "step": 20272
    },
    {
      "epoch": 2.8413454800280307,
      "grad_norm": 0.12467426806688309,
      "learning_rate": 1.637407318823248e-05,
      "loss": 0.0127,
      "step": 20273
    },
    {
      "epoch": 2.8414856341976176,
      "grad_norm": 0.28427594900131226,
      "learning_rate": 1.6359722554412817e-05,
      "loss": 0.0412,
      "step": 20274
    },
    {
      "epoch": 2.841625788367204,
      "grad_norm": 0.13745933771133423,
      "learning_rate": 1.6345371920593156e-05,
      "loss": 0.013,
      "step": 20275
    },
    {
      "epoch": 2.8417659425367905,
      "grad_norm": 0.21139267086982727,
      "learning_rate": 1.6331021286773498e-05,
      "loss": 0.0323,
      "step": 20276
    },
    {
      "epoch": 2.841906096706377,
      "grad_norm": 0.15647540986537933,
      "learning_rate": 1.6316670652953837e-05,
      "loss": 0.0082,
      "step": 20277
    },
    {
      "epoch": 2.8420462508759634,
      "grad_norm": 0.07853582501411438,
      "learning_rate": 1.6302320019134176e-05,
      "loss": 0.0031,
      "step": 20278
    },
    {
      "epoch": 2.84218640504555,
      "grad_norm": 0.16068831086158752,
      "learning_rate": 1.6287969385314514e-05,
      "loss": 0.0523,
      "step": 20279
    },
    {
      "epoch": 2.8423265592151368,
      "grad_norm": 0.3388252556324005,
      "learning_rate": 1.6273618751494856e-05,
      "loss": 0.0298,
      "step": 20280
    },
    {
      "epoch": 2.8424667133847232,
      "grad_norm": 0.14630305767059326,
      "learning_rate": 1.6259268117675195e-05,
      "loss": 0.0064,
      "step": 20281
    },
    {
      "epoch": 2.8426068675543097,
      "grad_norm": 0.2632675766944885,
      "learning_rate": 1.6244917483855537e-05,
      "loss": 0.0111,
      "step": 20282
    },
    {
      "epoch": 2.842747021723896,
      "grad_norm": 0.016888601705431938,
      "learning_rate": 1.6230566850035876e-05,
      "loss": 0.0015,
      "step": 20283
    },
    {
      "epoch": 2.842887175893483,
      "grad_norm": 0.1076136976480484,
      "learning_rate": 1.6216216216216215e-05,
      "loss": 0.0117,
      "step": 20284
    },
    {
      "epoch": 2.8430273300630695,
      "grad_norm": 0.22523699700832367,
      "learning_rate": 1.6201865582396557e-05,
      "loss": 0.0301,
      "step": 20285
    },
    {
      "epoch": 2.843167484232656,
      "grad_norm": 0.09478264302015305,
      "learning_rate": 1.6187514948576896e-05,
      "loss": 0.0056,
      "step": 20286
    },
    {
      "epoch": 2.8433076384022424,
      "grad_norm": 0.1267780065536499,
      "learning_rate": 1.6173164314757234e-05,
      "loss": 0.0149,
      "step": 20287
    },
    {
      "epoch": 2.843447792571829,
      "grad_norm": 0.2338511198759079,
      "learning_rate": 1.6158813680937573e-05,
      "loss": 0.0378,
      "step": 20288
    },
    {
      "epoch": 2.8435879467414154,
      "grad_norm": 0.015144739300012589,
      "learning_rate": 1.6144463047117915e-05,
      "loss": 0.0013,
      "step": 20289
    },
    {
      "epoch": 2.8437281009110023,
      "grad_norm": 0.09922248870134354,
      "learning_rate": 1.6130112413298254e-05,
      "loss": 0.0049,
      "step": 20290
    },
    {
      "epoch": 2.8438682550805887,
      "grad_norm": 0.03603662922978401,
      "learning_rate": 1.6115761779478593e-05,
      "loss": 0.001,
      "step": 20291
    },
    {
      "epoch": 2.844008409250175,
      "grad_norm": 0.1759449988603592,
      "learning_rate": 1.610141114565893e-05,
      "loss": 0.0152,
      "step": 20292
    },
    {
      "epoch": 2.8441485634197616,
      "grad_norm": 0.08061642199754715,
      "learning_rate": 1.6087060511839273e-05,
      "loss": 0.0076,
      "step": 20293
    },
    {
      "epoch": 2.8442887175893485,
      "grad_norm": 0.07315284013748169,
      "learning_rate": 1.6072709878019612e-05,
      "loss": 0.0083,
      "step": 20294
    },
    {
      "epoch": 2.844428871758935,
      "grad_norm": 0.11606989055871964,
      "learning_rate": 1.605835924419995e-05,
      "loss": 0.0357,
      "step": 20295
    },
    {
      "epoch": 2.8445690259285215,
      "grad_norm": 0.0716494619846344,
      "learning_rate": 1.604400861038029e-05,
      "loss": 0.0024,
      "step": 20296
    },
    {
      "epoch": 2.844709180098108,
      "grad_norm": 0.24385316669940948,
      "learning_rate": 1.602965797656063e-05,
      "loss": 0.0193,
      "step": 20297
    },
    {
      "epoch": 2.8448493342676944,
      "grad_norm": 0.18004268407821655,
      "learning_rate": 1.601530734274097e-05,
      "loss": 0.0068,
      "step": 20298
    },
    {
      "epoch": 2.844989488437281,
      "grad_norm": 0.0315437950193882,
      "learning_rate": 1.600095670892131e-05,
      "loss": 0.0018,
      "step": 20299
    },
    {
      "epoch": 2.8451296426068673,
      "grad_norm": 0.21020697057247162,
      "learning_rate": 1.5986606075101648e-05,
      "loss": 0.0978,
      "step": 20300
    },
    {
      "epoch": 2.845269796776454,
      "grad_norm": 0.16509751975536346,
      "learning_rate": 1.597225544128199e-05,
      "loss": 0.009,
      "step": 20301
    },
    {
      "epoch": 2.8454099509460407,
      "grad_norm": 0.08987118303775787,
      "learning_rate": 1.595790480746233e-05,
      "loss": 0.0256,
      "step": 20302
    },
    {
      "epoch": 2.845550105115627,
      "grad_norm": 0.14443255960941315,
      "learning_rate": 1.5943554173642667e-05,
      "loss": 0.0104,
      "step": 20303
    },
    {
      "epoch": 2.8456902592852136,
      "grad_norm": 0.022094091400504112,
      "learning_rate": 1.5929203539823006e-05,
      "loss": 0.0012,
      "step": 20304
    },
    {
      "epoch": 2.8458304134548005,
      "grad_norm": 0.285263329744339,
      "learning_rate": 1.5914852906003348e-05,
      "loss": 0.0095,
      "step": 20305
    },
    {
      "epoch": 2.845970567624387,
      "grad_norm": 0.5317838788032532,
      "learning_rate": 1.5900502272183687e-05,
      "loss": 0.0181,
      "step": 20306
    },
    {
      "epoch": 2.8461107217939734,
      "grad_norm": 0.1129615530371666,
      "learning_rate": 1.5886151638364026e-05,
      "loss": 0.0082,
      "step": 20307
    },
    {
      "epoch": 2.84625087596356,
      "grad_norm": 0.27298229932785034,
      "learning_rate": 1.5871801004544364e-05,
      "loss": 0.0118,
      "step": 20308
    },
    {
      "epoch": 2.8463910301331463,
      "grad_norm": 0.045300569385290146,
      "learning_rate": 1.5857450370724707e-05,
      "loss": 0.0028,
      "step": 20309
    },
    {
      "epoch": 2.8465311843027328,
      "grad_norm": 0.5991926193237305,
      "learning_rate": 1.5843099736905045e-05,
      "loss": 0.0801,
      "step": 20310
    },
    {
      "epoch": 2.8466713384723197,
      "grad_norm": 0.05029904097318649,
      "learning_rate": 1.5828749103085384e-05,
      "loss": 0.0014,
      "step": 20311
    },
    {
      "epoch": 2.846811492641906,
      "grad_norm": 0.053701359778642654,
      "learning_rate": 1.5814398469265723e-05,
      "loss": 0.002,
      "step": 20312
    },
    {
      "epoch": 2.8469516468114926,
      "grad_norm": 0.0034739954862743616,
      "learning_rate": 1.5800047835446065e-05,
      "loss": 0.0004,
      "step": 20313
    },
    {
      "epoch": 2.847091800981079,
      "grad_norm": 0.017947666347026825,
      "learning_rate": 1.5785697201626404e-05,
      "loss": 0.0013,
      "step": 20314
    },
    {
      "epoch": 2.847231955150666,
      "grad_norm": 0.40516209602355957,
      "learning_rate": 1.5771346567806742e-05,
      "loss": 0.0321,
      "step": 20315
    },
    {
      "epoch": 2.8473721093202524,
      "grad_norm": 1.0125679969787598,
      "learning_rate": 1.575699593398708e-05,
      "loss": 0.0973,
      "step": 20316
    },
    {
      "epoch": 2.847512263489839,
      "grad_norm": 0.167917862534523,
      "learning_rate": 1.5742645300167423e-05,
      "loss": 0.0076,
      "step": 20317
    },
    {
      "epoch": 2.8476524176594253,
      "grad_norm": 0.44034916162490845,
      "learning_rate": 1.5728294666347762e-05,
      "loss": 0.0095,
      "step": 20318
    },
    {
      "epoch": 2.847792571829012,
      "grad_norm": 0.3668735921382904,
      "learning_rate": 1.5713944032528104e-05,
      "loss": 0.0706,
      "step": 20319
    },
    {
      "epoch": 2.8479327259985983,
      "grad_norm": 1.4763059616088867,
      "learning_rate": 1.5699593398708443e-05,
      "loss": 0.0599,
      "step": 20320
    },
    {
      "epoch": 2.848072880168185,
      "grad_norm": 0.3076702356338501,
      "learning_rate": 1.568524276488878e-05,
      "loss": 0.0803,
      "step": 20321
    },
    {
      "epoch": 2.8482130343377716,
      "grad_norm": 0.1260942816734314,
      "learning_rate": 1.567089213106912e-05,
      "loss": 0.0116,
      "step": 20322
    },
    {
      "epoch": 2.848353188507358,
      "grad_norm": 0.3856593072414398,
      "learning_rate": 1.5656541497249462e-05,
      "loss": 0.0464,
      "step": 20323
    },
    {
      "epoch": 2.8484933426769445,
      "grad_norm": 0.16212178766727448,
      "learning_rate": 1.56421908634298e-05,
      "loss": 0.04,
      "step": 20324
    },
    {
      "epoch": 2.8486334968465314,
      "grad_norm": 0.2588249146938324,
      "learning_rate": 1.562784022961014e-05,
      "loss": 0.0264,
      "step": 20325
    },
    {
      "epoch": 2.848773651016118,
      "grad_norm": 0.15476007759571075,
      "learning_rate": 1.561348959579048e-05,
      "loss": 0.0128,
      "step": 20326
    },
    {
      "epoch": 2.8489138051857044,
      "grad_norm": 0.07228241860866547,
      "learning_rate": 1.559913896197082e-05,
      "loss": 0.0098,
      "step": 20327
    },
    {
      "epoch": 2.849053959355291,
      "grad_norm": 0.06651583313941956,
      "learning_rate": 1.558478832815116e-05,
      "loss": 0.0059,
      "step": 20328
    },
    {
      "epoch": 2.8491941135248773,
      "grad_norm": 0.39124396443367004,
      "learning_rate": 1.5570437694331498e-05,
      "loss": 0.0606,
      "step": 20329
    },
    {
      "epoch": 2.8493342676944637,
      "grad_norm": 0.2665528357028961,
      "learning_rate": 1.5556087060511837e-05,
      "loss": 0.0193,
      "step": 20330
    },
    {
      "epoch": 2.84947442186405,
      "grad_norm": 0.1665567010641098,
      "learning_rate": 1.554173642669218e-05,
      "loss": 0.0234,
      "step": 20331
    },
    {
      "epoch": 2.849614576033637,
      "grad_norm": 0.04394285008311272,
      "learning_rate": 1.5527385792872518e-05,
      "loss": 0.0025,
      "step": 20332
    },
    {
      "epoch": 2.8497547302032236,
      "grad_norm": 0.24094484746456146,
      "learning_rate": 1.5513035159052856e-05,
      "loss": 0.0378,
      "step": 20333
    },
    {
      "epoch": 2.84989488437281,
      "grad_norm": 0.19465135037899017,
      "learning_rate": 1.5498684525233195e-05,
      "loss": 0.0337,
      "step": 20334
    },
    {
      "epoch": 2.8500350385423965,
      "grad_norm": 0.1165773943066597,
      "learning_rate": 1.5484333891413537e-05,
      "loss": 0.0075,
      "step": 20335
    },
    {
      "epoch": 2.8501751927119834,
      "grad_norm": 0.6786255836486816,
      "learning_rate": 1.5469983257593876e-05,
      "loss": 0.0334,
      "step": 20336
    },
    {
      "epoch": 2.85031534688157,
      "grad_norm": 0.011521991342306137,
      "learning_rate": 1.5455632623774215e-05,
      "loss": 0.0011,
      "step": 20337
    },
    {
      "epoch": 2.8504555010511563,
      "grad_norm": 0.0730847418308258,
      "learning_rate": 1.5441281989954553e-05,
      "loss": 0.0048,
      "step": 20338
    },
    {
      "epoch": 2.8505956552207428,
      "grad_norm": 0.22655510902404785,
      "learning_rate": 1.5426931356134895e-05,
      "loss": 0.0145,
      "step": 20339
    },
    {
      "epoch": 2.8507358093903292,
      "grad_norm": 0.15519924461841583,
      "learning_rate": 1.5412580722315234e-05,
      "loss": 0.0098,
      "step": 20340
    },
    {
      "epoch": 2.8508759635599157,
      "grad_norm": 0.05947353318333626,
      "learning_rate": 1.5398230088495573e-05,
      "loss": 0.0053,
      "step": 20341
    },
    {
      "epoch": 2.8510161177295026,
      "grad_norm": 0.16406801342964172,
      "learning_rate": 1.538387945467591e-05,
      "loss": 0.0172,
      "step": 20342
    },
    {
      "epoch": 2.851156271899089,
      "grad_norm": 0.2745133936405182,
      "learning_rate": 1.5369528820856254e-05,
      "loss": 0.0343,
      "step": 20343
    },
    {
      "epoch": 2.8512964260686755,
      "grad_norm": 0.2881617248058319,
      "learning_rate": 1.5355178187036592e-05,
      "loss": 0.0249,
      "step": 20344
    },
    {
      "epoch": 2.851436580238262,
      "grad_norm": 0.17715638875961304,
      "learning_rate": 1.534082755321693e-05,
      "loss": 0.0212,
      "step": 20345
    },
    {
      "epoch": 2.851576734407849,
      "grad_norm": 0.4022597074508667,
      "learning_rate": 1.5326476919397273e-05,
      "loss": 0.0793,
      "step": 20346
    },
    {
      "epoch": 2.8517168885774353,
      "grad_norm": 0.022772137075662613,
      "learning_rate": 1.5312126285577612e-05,
      "loss": 0.0016,
      "step": 20347
    },
    {
      "epoch": 2.851857042747022,
      "grad_norm": 0.2803971767425537,
      "learning_rate": 1.529777565175795e-05,
      "loss": 0.0491,
      "step": 20348
    },
    {
      "epoch": 2.8519971969166082,
      "grad_norm": 0.059749312698841095,
      "learning_rate": 1.528342501793829e-05,
      "loss": 0.0023,
      "step": 20349
    },
    {
      "epoch": 2.8521373510861947,
      "grad_norm": 0.010602869093418121,
      "learning_rate": 1.526907438411863e-05,
      "loss": 0.0009,
      "step": 20350
    },
    {
      "epoch": 2.852277505255781,
      "grad_norm": 0.2774984836578369,
      "learning_rate": 1.525472375029897e-05,
      "loss": 0.0139,
      "step": 20351
    },
    {
      "epoch": 2.852417659425368,
      "grad_norm": 0.01156681776046753,
      "learning_rate": 1.5240373116479309e-05,
      "loss": 0.0008,
      "step": 20352
    },
    {
      "epoch": 2.8525578135949545,
      "grad_norm": 0.2745216190814972,
      "learning_rate": 1.522602248265965e-05,
      "loss": 0.0318,
      "step": 20353
    },
    {
      "epoch": 2.852697967764541,
      "grad_norm": 0.025583239272236824,
      "learning_rate": 1.5211671848839988e-05,
      "loss": 0.0023,
      "step": 20354
    },
    {
      "epoch": 2.8528381219341274,
      "grad_norm": 0.1819443553686142,
      "learning_rate": 1.5197321215020329e-05,
      "loss": 0.0298,
      "step": 20355
    },
    {
      "epoch": 2.8529782761037144,
      "grad_norm": 0.08847843110561371,
      "learning_rate": 1.5182970581200669e-05,
      "loss": 0.0041,
      "step": 20356
    },
    {
      "epoch": 2.853118430273301,
      "grad_norm": 0.05471263825893402,
      "learning_rate": 1.516861994738101e-05,
      "loss": 0.0043,
      "step": 20357
    },
    {
      "epoch": 2.8532585844428873,
      "grad_norm": 0.2533074617385864,
      "learning_rate": 1.5154269313561348e-05,
      "loss": 0.0161,
      "step": 20358
    },
    {
      "epoch": 2.8533987386124737,
      "grad_norm": 0.22861048579216003,
      "learning_rate": 1.5139918679741688e-05,
      "loss": 0.0423,
      "step": 20359
    },
    {
      "epoch": 2.85353889278206,
      "grad_norm": 0.4860513508319855,
      "learning_rate": 1.5125568045922027e-05,
      "loss": 0.0386,
      "step": 20360
    },
    {
      "epoch": 2.8536790469516466,
      "grad_norm": 0.027445362880825996,
      "learning_rate": 1.5111217412102368e-05,
      "loss": 0.0015,
      "step": 20361
    },
    {
      "epoch": 2.853819201121233,
      "grad_norm": 0.19729383289813995,
      "learning_rate": 1.5096866778282706e-05,
      "loss": 0.01,
      "step": 20362
    },
    {
      "epoch": 2.85395935529082,
      "grad_norm": 0.45119708776474,
      "learning_rate": 1.5082516144463047e-05,
      "loss": 0.03,
      "step": 20363
    },
    {
      "epoch": 2.8540995094604065,
      "grad_norm": 0.2652711272239685,
      "learning_rate": 1.5068165510643386e-05,
      "loss": 0.0159,
      "step": 20364
    },
    {
      "epoch": 2.854239663629993,
      "grad_norm": 0.03018065355718136,
      "learning_rate": 1.5053814876823726e-05,
      "loss": 0.0011,
      "step": 20365
    },
    {
      "epoch": 2.8543798177995794,
      "grad_norm": 0.07536077499389648,
      "learning_rate": 1.5039464243004065e-05,
      "loss": 0.0049,
      "step": 20366
    },
    {
      "epoch": 2.8545199719691663,
      "grad_norm": 0.35838422179222107,
      "learning_rate": 1.5025113609184405e-05,
      "loss": 0.0267,
      "step": 20367
    },
    {
      "epoch": 2.8546601261387528,
      "grad_norm": 0.5044819116592407,
      "learning_rate": 1.5010762975364744e-05,
      "loss": 0.0675,
      "step": 20368
    },
    {
      "epoch": 2.854800280308339,
      "grad_norm": 2.760685920715332,
      "learning_rate": 1.4996412341545084e-05,
      "loss": 0.4009,
      "step": 20369
    },
    {
      "epoch": 2.8549404344779257,
      "grad_norm": 3.385244369506836,
      "learning_rate": 1.4982061707725423e-05,
      "loss": 0.1376,
      "step": 20370
    },
    {
      "epoch": 2.855080588647512,
      "grad_norm": 0.3678162097930908,
      "learning_rate": 1.4967711073905763e-05,
      "loss": 0.0242,
      "step": 20371
    },
    {
      "epoch": 2.8552207428170986,
      "grad_norm": 0.21036860346794128,
      "learning_rate": 1.4953360440086102e-05,
      "loss": 0.0291,
      "step": 20372
    },
    {
      "epoch": 2.8553608969866855,
      "grad_norm": 0.18883508443832397,
      "learning_rate": 1.4939009806266443e-05,
      "loss": 0.0221,
      "step": 20373
    },
    {
      "epoch": 2.855501051156272,
      "grad_norm": 0.2510300874710083,
      "learning_rate": 1.4924659172446781e-05,
      "loss": 0.0211,
      "step": 20374
    },
    {
      "epoch": 2.8556412053258584,
      "grad_norm": 0.12770569324493408,
      "learning_rate": 1.4910308538627122e-05,
      "loss": 0.0205,
      "step": 20375
    },
    {
      "epoch": 2.855781359495445,
      "grad_norm": 0.200319305062294,
      "learning_rate": 1.489595790480746e-05,
      "loss": 0.0134,
      "step": 20376
    },
    {
      "epoch": 2.8559215136650318,
      "grad_norm": 0.16275058686733246,
      "learning_rate": 1.48816072709878e-05,
      "loss": 0.0288,
      "step": 20377
    },
    {
      "epoch": 2.8560616678346182,
      "grad_norm": 0.19373728334903717,
      "learning_rate": 1.486725663716814e-05,
      "loss": 0.016,
      "step": 20378
    },
    {
      "epoch": 2.8562018220042047,
      "grad_norm": 0.08098205178976059,
      "learning_rate": 1.485290600334848e-05,
      "loss": 0.0058,
      "step": 20379
    },
    {
      "epoch": 2.856341976173791,
      "grad_norm": 0.20354300737380981,
      "learning_rate": 1.4838555369528819e-05,
      "loss": 0.0099,
      "step": 20380
    },
    {
      "epoch": 2.8564821303433776,
      "grad_norm": 0.4066515266895294,
      "learning_rate": 1.4824204735709159e-05,
      "loss": 0.0558,
      "step": 20381
    },
    {
      "epoch": 2.856622284512964,
      "grad_norm": 0.1347045600414276,
      "learning_rate": 1.4809854101889498e-05,
      "loss": 0.0307,
      "step": 20382
    },
    {
      "epoch": 2.8567624386825505,
      "grad_norm": 0.16052640974521637,
      "learning_rate": 1.4795503468069838e-05,
      "loss": 0.0146,
      "step": 20383
    },
    {
      "epoch": 2.8569025928521374,
      "grad_norm": 0.2812305688858032,
      "learning_rate": 1.4781152834250177e-05,
      "loss": 0.0327,
      "step": 20384
    },
    {
      "epoch": 2.857042747021724,
      "grad_norm": 0.039231885224580765,
      "learning_rate": 1.4766802200430517e-05,
      "loss": 0.004,
      "step": 20385
    },
    {
      "epoch": 2.8571829011913104,
      "grad_norm": 0.13137809932231903,
      "learning_rate": 1.4752451566610856e-05,
      "loss": 0.016,
      "step": 20386
    },
    {
      "epoch": 2.8573230553608973,
      "grad_norm": 0.2043808400630951,
      "learning_rate": 1.4738100932791197e-05,
      "loss": 0.0236,
      "step": 20387
    },
    {
      "epoch": 2.8574632095304837,
      "grad_norm": 0.34610849618911743,
      "learning_rate": 1.4723750298971535e-05,
      "loss": 0.0514,
      "step": 20388
    },
    {
      "epoch": 2.85760336370007,
      "grad_norm": 0.10060770064592361,
      "learning_rate": 1.4709399665151876e-05,
      "loss": 0.0059,
      "step": 20389
    },
    {
      "epoch": 2.8577435178696566,
      "grad_norm": 0.3698410093784332,
      "learning_rate": 1.4695049031332214e-05,
      "loss": 0.0361,
      "step": 20390
    },
    {
      "epoch": 2.857883672039243,
      "grad_norm": 0.438769668340683,
      "learning_rate": 1.4680698397512555e-05,
      "loss": 0.044,
      "step": 20391
    },
    {
      "epoch": 2.8580238262088296,
      "grad_norm": 0.37049970030784607,
      "learning_rate": 1.4666347763692894e-05,
      "loss": 0.0457,
      "step": 20392
    },
    {
      "epoch": 2.858163980378416,
      "grad_norm": 0.10531621426343918,
      "learning_rate": 1.4651997129873236e-05,
      "loss": 0.0042,
      "step": 20393
    },
    {
      "epoch": 2.858304134548003,
      "grad_norm": 0.16144564747810364,
      "learning_rate": 1.4637646496053576e-05,
      "loss": 0.0237,
      "step": 20394
    },
    {
      "epoch": 2.8584442887175894,
      "grad_norm": 0.09245503693819046,
      "learning_rate": 1.4623295862233915e-05,
      "loss": 0.0099,
      "step": 20395
    },
    {
      "epoch": 2.858584442887176,
      "grad_norm": 0.31200745701789856,
      "learning_rate": 1.4608945228414255e-05,
      "loss": 0.0167,
      "step": 20396
    },
    {
      "epoch": 2.8587245970567623,
      "grad_norm": 0.2925493121147156,
      "learning_rate": 1.4594594594594594e-05,
      "loss": 0.0769,
      "step": 20397
    },
    {
      "epoch": 2.858864751226349,
      "grad_norm": 0.09629321098327637,
      "learning_rate": 1.4580243960774934e-05,
      "loss": 0.0236,
      "step": 20398
    },
    {
      "epoch": 2.8590049053959357,
      "grad_norm": 0.09990980476140976,
      "learning_rate": 1.4565893326955273e-05,
      "loss": 0.0259,
      "step": 20399
    },
    {
      "epoch": 2.859145059565522,
      "grad_norm": 0.1156524121761322,
      "learning_rate": 1.4551542693135613e-05,
      "loss": 0.0166,
      "step": 20400
    },
    {
      "epoch": 2.8592852137351086,
      "grad_norm": 0.16261838376522064,
      "learning_rate": 1.4537192059315952e-05,
      "loss": 0.0233,
      "step": 20401
    },
    {
      "epoch": 2.859425367904695,
      "grad_norm": 0.15608060359954834,
      "learning_rate": 1.4522841425496293e-05,
      "loss": 0.0112,
      "step": 20402
    },
    {
      "epoch": 2.8595655220742815,
      "grad_norm": 0.037038516253232956,
      "learning_rate": 1.4508490791676631e-05,
      "loss": 0.0039,
      "step": 20403
    },
    {
      "epoch": 2.8597056762438684,
      "grad_norm": 0.22943833470344543,
      "learning_rate": 1.4494140157856972e-05,
      "loss": 0.0373,
      "step": 20404
    },
    {
      "epoch": 2.859845830413455,
      "grad_norm": 0.7149508595466614,
      "learning_rate": 1.447978952403731e-05,
      "loss": 0.0393,
      "step": 20405
    },
    {
      "epoch": 2.8599859845830413,
      "grad_norm": 0.3068964183330536,
      "learning_rate": 1.4465438890217651e-05,
      "loss": 0.0236,
      "step": 20406
    },
    {
      "epoch": 2.860126138752628,
      "grad_norm": 0.18925514817237854,
      "learning_rate": 1.445108825639799e-05,
      "loss": 0.0565,
      "step": 20407
    },
    {
      "epoch": 2.8602662929222147,
      "grad_norm": 0.4989233911037445,
      "learning_rate": 1.443673762257833e-05,
      "loss": 0.0271,
      "step": 20408
    },
    {
      "epoch": 2.860406447091801,
      "grad_norm": 1.1134618520736694,
      "learning_rate": 1.4422386988758669e-05,
      "loss": 0.0509,
      "step": 20409
    },
    {
      "epoch": 2.8605466012613876,
      "grad_norm": 0.3696964681148529,
      "learning_rate": 1.440803635493901e-05,
      "loss": 0.0664,
      "step": 20410
    },
    {
      "epoch": 2.860686755430974,
      "grad_norm": 0.3626372218132019,
      "learning_rate": 1.4393685721119348e-05,
      "loss": 0.0509,
      "step": 20411
    },
    {
      "epoch": 2.8608269096005605,
      "grad_norm": 0.11314673721790314,
      "learning_rate": 1.4379335087299688e-05,
      "loss": 0.0094,
      "step": 20412
    },
    {
      "epoch": 2.860967063770147,
      "grad_norm": 0.05812682956457138,
      "learning_rate": 1.4364984453480027e-05,
      "loss": 0.0044,
      "step": 20413
    },
    {
      "epoch": 2.8611072179397334,
      "grad_norm": 0.21471843123435974,
      "learning_rate": 1.4350633819660367e-05,
      "loss": 0.0163,
      "step": 20414
    },
    {
      "epoch": 2.8612473721093203,
      "grad_norm": 0.04024404287338257,
      "learning_rate": 1.4336283185840706e-05,
      "loss": 0.0027,
      "step": 20415
    },
    {
      "epoch": 2.861387526278907,
      "grad_norm": 0.3599867522716522,
      "learning_rate": 1.4321932552021047e-05,
      "loss": 0.0603,
      "step": 20416
    },
    {
      "epoch": 2.8615276804484933,
      "grad_norm": 0.6264867782592773,
      "learning_rate": 1.4307581918201385e-05,
      "loss": 0.0389,
      "step": 20417
    },
    {
      "epoch": 2.86166783461808,
      "grad_norm": 0.12961158156394958,
      "learning_rate": 1.4293231284381726e-05,
      "loss": 0.0766,
      "step": 20418
    },
    {
      "epoch": 2.8618079887876666,
      "grad_norm": 1.5094178915023804,
      "learning_rate": 1.4278880650562064e-05,
      "loss": 0.1293,
      "step": 20419
    },
    {
      "epoch": 2.861948142957253,
      "grad_norm": 1.2366629838943481,
      "learning_rate": 1.4264530016742405e-05,
      "loss": 0.1142,
      "step": 20420
    },
    {
      "epoch": 2.8620882971268395,
      "grad_norm": 0.30086979269981384,
      "learning_rate": 1.4250179382922744e-05,
      "loss": 0.0595,
      "step": 20421
    },
    {
      "epoch": 2.862228451296426,
      "grad_norm": 0.14553207159042358,
      "learning_rate": 1.4235828749103084e-05,
      "loss": 0.0174,
      "step": 20422
    },
    {
      "epoch": 2.8623686054660125,
      "grad_norm": 0.30825215578079224,
      "learning_rate": 1.4221478115283423e-05,
      "loss": 0.025,
      "step": 20423
    },
    {
      "epoch": 2.862508759635599,
      "grad_norm": 0.11741652339696884,
      "learning_rate": 1.4207127481463763e-05,
      "loss": 0.009,
      "step": 20424
    },
    {
      "epoch": 2.862648913805186,
      "grad_norm": 0.07556717097759247,
      "learning_rate": 1.4192776847644102e-05,
      "loss": 0.0168,
      "step": 20425
    },
    {
      "epoch": 2.8627890679747723,
      "grad_norm": 0.13203367590904236,
      "learning_rate": 1.4178426213824442e-05,
      "loss": 0.013,
      "step": 20426
    },
    {
      "epoch": 2.8629292221443587,
      "grad_norm": 0.35279136896133423,
      "learning_rate": 1.4164075580004781e-05,
      "loss": 0.0819,
      "step": 20427
    },
    {
      "epoch": 2.863069376313945,
      "grad_norm": 0.2570411264896393,
      "learning_rate": 1.4149724946185121e-05,
      "loss": 0.0086,
      "step": 20428
    },
    {
      "epoch": 2.863209530483532,
      "grad_norm": 0.17944258451461792,
      "learning_rate": 1.413537431236546e-05,
      "loss": 0.0151,
      "step": 20429
    },
    {
      "epoch": 2.8633496846531186,
      "grad_norm": 0.07360837608575821,
      "learning_rate": 1.41210236785458e-05,
      "loss": 0.0061,
      "step": 20430
    },
    {
      "epoch": 2.863489838822705,
      "grad_norm": 0.21010488271713257,
      "learning_rate": 1.4106673044726141e-05,
      "loss": 0.0408,
      "step": 20431
    },
    {
      "epoch": 2.8636299929922915,
      "grad_norm": 0.3781132102012634,
      "learning_rate": 1.4092322410906481e-05,
      "loss": 0.0402,
      "step": 20432
    },
    {
      "epoch": 2.863770147161878,
      "grad_norm": 0.20468628406524658,
      "learning_rate": 1.407797177708682e-05,
      "loss": 0.0262,
      "step": 20433
    },
    {
      "epoch": 2.8639103013314644,
      "grad_norm": 0.08186621963977814,
      "learning_rate": 1.406362114326716e-05,
      "loss": 0.0125,
      "step": 20434
    },
    {
      "epoch": 2.8640504555010513,
      "grad_norm": 0.41598352789878845,
      "learning_rate": 1.40492705094475e-05,
      "loss": 0.0287,
      "step": 20435
    },
    {
      "epoch": 2.8641906096706378,
      "grad_norm": 0.1666543185710907,
      "learning_rate": 1.403491987562784e-05,
      "loss": 0.0142,
      "step": 20436
    },
    {
      "epoch": 2.8643307638402242,
      "grad_norm": 0.2777056396007538,
      "learning_rate": 1.4020569241808178e-05,
      "loss": 0.0334,
      "step": 20437
    },
    {
      "epoch": 2.8644709180098107,
      "grad_norm": 0.10430144518613815,
      "learning_rate": 1.4006218607988519e-05,
      "loss": 0.0084,
      "step": 20438
    },
    {
      "epoch": 2.8646110721793976,
      "grad_norm": 0.21764230728149414,
      "learning_rate": 1.3991867974168858e-05,
      "loss": 0.0382,
      "step": 20439
    },
    {
      "epoch": 2.864751226348984,
      "grad_norm": 0.22724409401416779,
      "learning_rate": 1.3977517340349198e-05,
      "loss": 0.0174,
      "step": 20440
    },
    {
      "epoch": 2.8648913805185705,
      "grad_norm": 0.3295377492904663,
      "learning_rate": 1.3963166706529537e-05,
      "loss": 0.0164,
      "step": 20441
    },
    {
      "epoch": 2.865031534688157,
      "grad_norm": 0.3335498869419098,
      "learning_rate": 1.3948816072709877e-05,
      "loss": 0.0081,
      "step": 20442
    },
    {
      "epoch": 2.8651716888577434,
      "grad_norm": 0.10871914029121399,
      "learning_rate": 1.3934465438890218e-05,
      "loss": 0.0103,
      "step": 20443
    },
    {
      "epoch": 2.86531184302733,
      "grad_norm": 0.1450573205947876,
      "learning_rate": 1.3920114805070556e-05,
      "loss": 0.0242,
      "step": 20444
    },
    {
      "epoch": 2.8654519971969163,
      "grad_norm": 0.09006363153457642,
      "learning_rate": 1.3905764171250897e-05,
      "loss": 0.0166,
      "step": 20445
    },
    {
      "epoch": 2.8655921513665032,
      "grad_norm": 1.0678198337554932,
      "learning_rate": 1.3891413537431235e-05,
      "loss": 0.1008,
      "step": 20446
    },
    {
      "epoch": 2.8657323055360897,
      "grad_norm": 0.10796839743852615,
      "learning_rate": 1.3877062903611576e-05,
      "loss": 0.0063,
      "step": 20447
    },
    {
      "epoch": 2.865872459705676,
      "grad_norm": 0.13320298492908478,
      "learning_rate": 1.3862712269791915e-05,
      "loss": 0.0098,
      "step": 20448
    },
    {
      "epoch": 2.8660126138752626,
      "grad_norm": 0.4603837728500366,
      "learning_rate": 1.3848361635972255e-05,
      "loss": 0.0149,
      "step": 20449
    },
    {
      "epoch": 2.8661527680448495,
      "grad_norm": 0.4901426136493683,
      "learning_rate": 1.3834011002152594e-05,
      "loss": 0.0212,
      "step": 20450
    },
    {
      "epoch": 2.866292922214436,
      "grad_norm": 0.456548810005188,
      "learning_rate": 1.3819660368332934e-05,
      "loss": 0.0301,
      "step": 20451
    },
    {
      "epoch": 2.8664330763840224,
      "grad_norm": 0.09810171276330948,
      "learning_rate": 1.3805309734513273e-05,
      "loss": 0.0215,
      "step": 20452
    },
    {
      "epoch": 2.866573230553609,
      "grad_norm": 0.13737161457538605,
      "learning_rate": 1.3790959100693613e-05,
      "loss": 0.0189,
      "step": 20453
    },
    {
      "epoch": 2.8667133847231954,
      "grad_norm": 0.15599215030670166,
      "learning_rate": 1.3776608466873952e-05,
      "loss": 0.0142,
      "step": 20454
    },
    {
      "epoch": 2.866853538892782,
      "grad_norm": 0.21872660517692566,
      "learning_rate": 1.3762257833054292e-05,
      "loss": 0.0232,
      "step": 20455
    },
    {
      "epoch": 2.8669936930623687,
      "grad_norm": 0.19498136639595032,
      "learning_rate": 1.3747907199234631e-05,
      "loss": 0.0579,
      "step": 20456
    },
    {
      "epoch": 2.867133847231955,
      "grad_norm": 0.02320132963359356,
      "learning_rate": 1.3733556565414972e-05,
      "loss": 0.0026,
      "step": 20457
    },
    {
      "epoch": 2.8672740014015416,
      "grad_norm": 0.3561120331287384,
      "learning_rate": 1.371920593159531e-05,
      "loss": 0.0954,
      "step": 20458
    },
    {
      "epoch": 2.867414155571128,
      "grad_norm": 0.07305943220853806,
      "learning_rate": 1.370485529777565e-05,
      "loss": 0.0087,
      "step": 20459
    },
    {
      "epoch": 2.867554309740715,
      "grad_norm": 0.06453564018011093,
      "learning_rate": 1.369050466395599e-05,
      "loss": 0.0025,
      "step": 20460
    },
    {
      "epoch": 2.8676944639103015,
      "grad_norm": 0.2136397510766983,
      "learning_rate": 1.367615403013633e-05,
      "loss": 0.0201,
      "step": 20461
    },
    {
      "epoch": 2.867834618079888,
      "grad_norm": 0.2717055380344391,
      "learning_rate": 1.3661803396316669e-05,
      "loss": 0.02,
      "step": 20462
    },
    {
      "epoch": 2.8679747722494744,
      "grad_norm": 0.09508466720581055,
      "learning_rate": 1.3647452762497009e-05,
      "loss": 0.0115,
      "step": 20463
    },
    {
      "epoch": 2.868114926419061,
      "grad_norm": 0.09293036162853241,
      "learning_rate": 1.3633102128677348e-05,
      "loss": 0.0102,
      "step": 20464
    },
    {
      "epoch": 2.8682550805886473,
      "grad_norm": 0.12921813130378723,
      "learning_rate": 1.3618751494857688e-05,
      "loss": 0.0067,
      "step": 20465
    },
    {
      "epoch": 2.868395234758234,
      "grad_norm": 0.15859870612621307,
      "learning_rate": 1.3604400861038027e-05,
      "loss": 0.0265,
      "step": 20466
    },
    {
      "epoch": 2.8685353889278207,
      "grad_norm": 0.32528579235076904,
      "learning_rate": 1.3590050227218367e-05,
      "loss": 0.0339,
      "step": 20467
    },
    {
      "epoch": 2.868675543097407,
      "grad_norm": 0.5237740278244019,
      "learning_rate": 1.3575699593398708e-05,
      "loss": 0.0271,
      "step": 20468
    },
    {
      "epoch": 2.8688156972669936,
      "grad_norm": 1.419663906097412,
      "learning_rate": 1.3561348959579048e-05,
      "loss": 0.0669,
      "step": 20469
    },
    {
      "epoch": 2.8689558514365805,
      "grad_norm": 0.9900166392326355,
      "learning_rate": 1.3546998325759387e-05,
      "loss": 0.0935,
      "step": 20470
    },
    {
      "epoch": 2.869096005606167,
      "grad_norm": 0.05839133635163307,
      "learning_rate": 1.3532647691939727e-05,
      "loss": 0.0179,
      "step": 20471
    },
    {
      "epoch": 2.8692361597757534,
      "grad_norm": 0.09993952512741089,
      "learning_rate": 1.3518297058120066e-05,
      "loss": 0.0203,
      "step": 20472
    },
    {
      "epoch": 2.86937631394534,
      "grad_norm": 0.11818927526473999,
      "learning_rate": 1.3503946424300406e-05,
      "loss": 0.0133,
      "step": 20473
    },
    {
      "epoch": 2.8695164681149263,
      "grad_norm": 0.26086297631263733,
      "learning_rate": 1.3489595790480745e-05,
      "loss": 0.0225,
      "step": 20474
    },
    {
      "epoch": 2.869656622284513,
      "grad_norm": 0.34967029094696045,
      "learning_rate": 1.3475245156661086e-05,
      "loss": 0.0182,
      "step": 20475
    },
    {
      "epoch": 2.8697967764540993,
      "grad_norm": 0.15338881313800812,
      "learning_rate": 1.3460894522841424e-05,
      "loss": 0.0207,
      "step": 20476
    },
    {
      "epoch": 2.869936930623686,
      "grad_norm": 0.15352988243103027,
      "learning_rate": 1.3446543889021765e-05,
      "loss": 0.023,
      "step": 20477
    },
    {
      "epoch": 2.8700770847932726,
      "grad_norm": 0.15371306240558624,
      "learning_rate": 1.3432193255202103e-05,
      "loss": 0.0362,
      "step": 20478
    },
    {
      "epoch": 2.870217238962859,
      "grad_norm": 0.2695786654949188,
      "learning_rate": 1.3417842621382444e-05,
      "loss": 0.0223,
      "step": 20479
    },
    {
      "epoch": 2.8703573931324455,
      "grad_norm": 0.11328046023845673,
      "learning_rate": 1.3403491987562783e-05,
      "loss": 0.0207,
      "step": 20480
    },
    {
      "epoch": 2.8704975473020324,
      "grad_norm": 0.20594169199466705,
      "learning_rate": 1.3389141353743123e-05,
      "loss": 0.0173,
      "step": 20481
    },
    {
      "epoch": 2.870637701471619,
      "grad_norm": 0.216615691781044,
      "learning_rate": 1.3374790719923462e-05,
      "loss": 0.0193,
      "step": 20482
    },
    {
      "epoch": 2.8707778556412054,
      "grad_norm": 0.3203184902667999,
      "learning_rate": 1.3360440086103802e-05,
      "loss": 0.037,
      "step": 20483
    },
    {
      "epoch": 2.870918009810792,
      "grad_norm": 0.2253418117761612,
      "learning_rate": 1.3346089452284141e-05,
      "loss": 0.0185,
      "step": 20484
    },
    {
      "epoch": 2.8710581639803783,
      "grad_norm": 0.13091346621513367,
      "learning_rate": 1.3331738818464481e-05,
      "loss": 0.0163,
      "step": 20485
    },
    {
      "epoch": 2.8711983181499647,
      "grad_norm": 0.14826782047748566,
      "learning_rate": 1.331738818464482e-05,
      "loss": 0.0263,
      "step": 20486
    },
    {
      "epoch": 2.8713384723195516,
      "grad_norm": 0.779538094997406,
      "learning_rate": 1.330303755082516e-05,
      "loss": 0.0342,
      "step": 20487
    },
    {
      "epoch": 2.871478626489138,
      "grad_norm": 0.08050556480884552,
      "learning_rate": 1.32886869170055e-05,
      "loss": 0.0045,
      "step": 20488
    },
    {
      "epoch": 2.8716187806587246,
      "grad_norm": 0.1804610639810562,
      "learning_rate": 1.327433628318584e-05,
      "loss": 0.0264,
      "step": 20489
    },
    {
      "epoch": 2.871758934828311,
      "grad_norm": 0.6832132935523987,
      "learning_rate": 1.3259985649366178e-05,
      "loss": 0.0845,
      "step": 20490
    },
    {
      "epoch": 2.871899088997898,
      "grad_norm": 0.1840461641550064,
      "learning_rate": 1.3245635015546519e-05,
      "loss": 0.027,
      "step": 20491
    },
    {
      "epoch": 2.8720392431674844,
      "grad_norm": 0.215186208486557,
      "learning_rate": 1.3231284381726857e-05,
      "loss": 0.0587,
      "step": 20492
    },
    {
      "epoch": 2.872179397337071,
      "grad_norm": 0.10376659780740738,
      "learning_rate": 1.3216933747907198e-05,
      "loss": 0.0317,
      "step": 20493
    },
    {
      "epoch": 2.8723195515066573,
      "grad_norm": 0.19432315230369568,
      "learning_rate": 1.3202583114087537e-05,
      "loss": 0.0394,
      "step": 20494
    },
    {
      "epoch": 2.8724597056762438,
      "grad_norm": 0.07225720584392548,
      "learning_rate": 1.3188232480267877e-05,
      "loss": 0.0039,
      "step": 20495
    },
    {
      "epoch": 2.87259985984583,
      "grad_norm": 0.07259931415319443,
      "learning_rate": 1.3173881846448216e-05,
      "loss": 0.0059,
      "step": 20496
    },
    {
      "epoch": 2.872740014015417,
      "grad_norm": 0.13902798295021057,
      "learning_rate": 1.3159531212628556e-05,
      "loss": 0.0246,
      "step": 20497
    },
    {
      "epoch": 2.8728801681850036,
      "grad_norm": 0.1350734382867813,
      "learning_rate": 1.3145180578808895e-05,
      "loss": 0.0086,
      "step": 20498
    },
    {
      "epoch": 2.87302032235459,
      "grad_norm": 0.043200213462114334,
      "learning_rate": 1.3130829944989235e-05,
      "loss": 0.0025,
      "step": 20499
    },
    {
      "epoch": 2.8731604765241765,
      "grad_norm": 0.25786125659942627,
      "learning_rate": 1.3116479311169574e-05,
      "loss": 0.0165,
      "step": 20500
    },
    {
      "epoch": 2.8733006306937634,
      "grad_norm": 0.29306137561798096,
      "learning_rate": 1.3102128677349914e-05,
      "loss": 0.0224,
      "step": 20501
    },
    {
      "epoch": 2.87344078486335,
      "grad_norm": 0.30698108673095703,
      "learning_rate": 1.3087778043530253e-05,
      "loss": 0.0384,
      "step": 20502
    },
    {
      "epoch": 2.8735809390329363,
      "grad_norm": 0.2830221951007843,
      "learning_rate": 1.3073427409710594e-05,
      "loss": 0.0384,
      "step": 20503
    },
    {
      "epoch": 2.873721093202523,
      "grad_norm": 0.3073148727416992,
      "learning_rate": 1.3059076775890932e-05,
      "loss": 0.0965,
      "step": 20504
    },
    {
      "epoch": 2.8738612473721092,
      "grad_norm": 0.16444863379001617,
      "learning_rate": 1.3044726142071274e-05,
      "loss": 0.0152,
      "step": 20505
    },
    {
      "epoch": 2.8740014015416957,
      "grad_norm": 0.26114946603775024,
      "learning_rate": 1.3030375508251615e-05,
      "loss": 0.0121,
      "step": 20506
    },
    {
      "epoch": 2.874141555711282,
      "grad_norm": 1.1925033330917358,
      "learning_rate": 1.3016024874431954e-05,
      "loss": 0.106,
      "step": 20507
    },
    {
      "epoch": 2.874281709880869,
      "grad_norm": 0.09891834110021591,
      "learning_rate": 1.3001674240612294e-05,
      "loss": 0.0404,
      "step": 20508
    },
    {
      "epoch": 2.8744218640504555,
      "grad_norm": 0.3454154133796692,
      "learning_rate": 1.2987323606792633e-05,
      "loss": 0.0143,
      "step": 20509
    },
    {
      "epoch": 2.874562018220042,
      "grad_norm": 0.08711627125740051,
      "learning_rate": 1.2972972972972973e-05,
      "loss": 0.0055,
      "step": 20510
    },
    {
      "epoch": 2.8747021723896284,
      "grad_norm": 0.17824983596801758,
      "learning_rate": 1.2958622339153312e-05,
      "loss": 0.015,
      "step": 20511
    },
    {
      "epoch": 2.8748423265592153,
      "grad_norm": 0.35351672768592834,
      "learning_rate": 1.2944271705333652e-05,
      "loss": 0.0086,
      "step": 20512
    },
    {
      "epoch": 2.874982480728802,
      "grad_norm": 0.33355438709259033,
      "learning_rate": 1.2929921071513991e-05,
      "loss": 0.0071,
      "step": 20513
    },
    {
      "epoch": 2.8751226348983883,
      "grad_norm": 1.1041265726089478,
      "learning_rate": 1.2915570437694331e-05,
      "loss": 0.1779,
      "step": 20514
    },
    {
      "epoch": 2.8752627890679747,
      "grad_norm": 0.12072717398405075,
      "learning_rate": 1.290121980387467e-05,
      "loss": 0.0195,
      "step": 20515
    },
    {
      "epoch": 2.875402943237561,
      "grad_norm": 0.6618925929069519,
      "learning_rate": 1.288686917005501e-05,
      "loss": 0.0807,
      "step": 20516
    },
    {
      "epoch": 2.8755430974071476,
      "grad_norm": 0.6493064761161804,
      "learning_rate": 1.287251853623535e-05,
      "loss": 0.1032,
      "step": 20517
    },
    {
      "epoch": 2.8756832515767345,
      "grad_norm": 1.1774274110794067,
      "learning_rate": 1.285816790241569e-05,
      "loss": 0.0545,
      "step": 20518
    },
    {
      "epoch": 2.875823405746321,
      "grad_norm": 0.07859918475151062,
      "learning_rate": 1.2843817268596028e-05,
      "loss": 0.0032,
      "step": 20519
    },
    {
      "epoch": 2.8759635599159075,
      "grad_norm": 0.9119218587875366,
      "learning_rate": 1.2829466634776369e-05,
      "loss": 0.0469,
      "step": 20520
    },
    {
      "epoch": 2.876103714085494,
      "grad_norm": 0.4710049331188202,
      "learning_rate": 1.2815116000956708e-05,
      "loss": 0.0253,
      "step": 20521
    },
    {
      "epoch": 2.876243868255081,
      "grad_norm": 0.1618664711713791,
      "learning_rate": 1.2800765367137048e-05,
      "loss": 0.0172,
      "step": 20522
    },
    {
      "epoch": 2.8763840224246673,
      "grad_norm": 0.41987690329551697,
      "learning_rate": 1.2786414733317387e-05,
      "loss": 0.0327,
      "step": 20523
    },
    {
      "epoch": 2.8765241765942537,
      "grad_norm": 0.09576867520809174,
      "learning_rate": 1.2772064099497727e-05,
      "loss": 0.009,
      "step": 20524
    },
    {
      "epoch": 2.87666433076384,
      "grad_norm": 0.13913939893245697,
      "learning_rate": 1.2757713465678066e-05,
      "loss": 0.0286,
      "step": 20525
    },
    {
      "epoch": 2.8768044849334267,
      "grad_norm": 0.2558138966560364,
      "learning_rate": 1.2743362831858406e-05,
      "loss": 0.0092,
      "step": 20526
    },
    {
      "epoch": 2.876944639103013,
      "grad_norm": 0.1790776252746582,
      "learning_rate": 1.2729012198038745e-05,
      "loss": 0.0501,
      "step": 20527
    },
    {
      "epoch": 2.8770847932725996,
      "grad_norm": 0.17836473882198334,
      "learning_rate": 1.2714661564219085e-05,
      "loss": 0.0325,
      "step": 20528
    },
    {
      "epoch": 2.8772249474421865,
      "grad_norm": 0.17657721042633057,
      "learning_rate": 1.2700310930399424e-05,
      "loss": 0.051,
      "step": 20529
    },
    {
      "epoch": 2.877365101611773,
      "grad_norm": 0.01016154419630766,
      "learning_rate": 1.2685960296579765e-05,
      "loss": 0.0012,
      "step": 20530
    },
    {
      "epoch": 2.8775052557813594,
      "grad_norm": 0.2563575208187103,
      "learning_rate": 1.2671609662760103e-05,
      "loss": 0.0188,
      "step": 20531
    },
    {
      "epoch": 2.8776454099509463,
      "grad_norm": 0.13290025293827057,
      "learning_rate": 1.2657259028940444e-05,
      "loss": 0.0125,
      "step": 20532
    },
    {
      "epoch": 2.8777855641205328,
      "grad_norm": 0.17262418568134308,
      "learning_rate": 1.2642908395120782e-05,
      "loss": 0.027,
      "step": 20533
    },
    {
      "epoch": 2.8779257182901192,
      "grad_norm": 0.08365587890148163,
      "learning_rate": 1.2628557761301123e-05,
      "loss": 0.0023,
      "step": 20534
    },
    {
      "epoch": 2.8780658724597057,
      "grad_norm": 0.2181965708732605,
      "learning_rate": 1.2614207127481462e-05,
      "loss": 0.0297,
      "step": 20535
    },
    {
      "epoch": 2.878206026629292,
      "grad_norm": 0.09229299426078796,
      "learning_rate": 1.2599856493661802e-05,
      "loss": 0.0123,
      "step": 20536
    },
    {
      "epoch": 2.8783461807988786,
      "grad_norm": 0.24683672189712524,
      "learning_rate": 1.258550585984214e-05,
      "loss": 0.0305,
      "step": 20537
    },
    {
      "epoch": 2.878486334968465,
      "grad_norm": 0.24597430229187012,
      "learning_rate": 1.2571155226022481e-05,
      "loss": 0.0497,
      "step": 20538
    },
    {
      "epoch": 2.878626489138052,
      "grad_norm": 0.5246784687042236,
      "learning_rate": 1.255680459220282e-05,
      "loss": 0.034,
      "step": 20539
    },
    {
      "epoch": 2.8787666433076384,
      "grad_norm": 0.06071069836616516,
      "learning_rate": 1.254245395838316e-05,
      "loss": 0.0076,
      "step": 20540
    },
    {
      "epoch": 2.878906797477225,
      "grad_norm": 0.14612825214862823,
      "learning_rate": 1.2528103324563499e-05,
      "loss": 0.0382,
      "step": 20541
    },
    {
      "epoch": 2.8790469516468113,
      "grad_norm": 0.21174681186676025,
      "learning_rate": 1.2513752690743841e-05,
      "loss": 0.0192,
      "step": 20542
    },
    {
      "epoch": 2.8791871058163983,
      "grad_norm": 0.07328194379806519,
      "learning_rate": 1.2499402056924182e-05,
      "loss": 0.0076,
      "step": 20543
    },
    {
      "epoch": 2.8793272599859847,
      "grad_norm": 0.30224767327308655,
      "learning_rate": 1.248505142310452e-05,
      "loss": 0.0087,
      "step": 20544
    },
    {
      "epoch": 2.879467414155571,
      "grad_norm": 1.3327641487121582,
      "learning_rate": 1.247070078928486e-05,
      "loss": 0.0326,
      "step": 20545
    },
    {
      "epoch": 2.8796075683251576,
      "grad_norm": 0.07316180318593979,
      "learning_rate": 1.24563501554652e-05,
      "loss": 0.0143,
      "step": 20546
    },
    {
      "epoch": 2.879747722494744,
      "grad_norm": 0.20163120329380035,
      "learning_rate": 1.244199952164554e-05,
      "loss": 0.0321,
      "step": 20547
    },
    {
      "epoch": 2.8798878766643305,
      "grad_norm": 0.09757734835147858,
      "learning_rate": 1.2427648887825879e-05,
      "loss": 0.0064,
      "step": 20548
    },
    {
      "epoch": 2.8800280308339175,
      "grad_norm": 0.09465175122022629,
      "learning_rate": 1.2413298254006219e-05,
      "loss": 0.0041,
      "step": 20549
    },
    {
      "epoch": 2.880168185003504,
      "grad_norm": 0.22634288668632507,
      "learning_rate": 1.2398947620186558e-05,
      "loss": 0.0106,
      "step": 20550
    },
    {
      "epoch": 2.8803083391730904,
      "grad_norm": 0.1089482307434082,
      "learning_rate": 1.2384596986366898e-05,
      "loss": 0.0077,
      "step": 20551
    },
    {
      "epoch": 2.880448493342677,
      "grad_norm": 0.26292601227760315,
      "learning_rate": 1.2370246352547237e-05,
      "loss": 0.0174,
      "step": 20552
    },
    {
      "epoch": 2.8805886475122637,
      "grad_norm": 0.2562173306941986,
      "learning_rate": 1.2355895718727577e-05,
      "loss": 0.0378,
      "step": 20553
    },
    {
      "epoch": 2.88072880168185,
      "grad_norm": 0.35360920429229736,
      "learning_rate": 1.2341545084907916e-05,
      "loss": 0.0578,
      "step": 20554
    },
    {
      "epoch": 2.8808689558514367,
      "grad_norm": 0.20812025666236877,
      "learning_rate": 1.2327194451088256e-05,
      "loss": 0.0234,
      "step": 20555
    },
    {
      "epoch": 2.881009110021023,
      "grad_norm": 0.24358178675174713,
      "learning_rate": 1.2312843817268595e-05,
      "loss": 0.0328,
      "step": 20556
    },
    {
      "epoch": 2.8811492641906096,
      "grad_norm": 0.054564934223890305,
      "learning_rate": 1.2298493183448936e-05,
      "loss": 0.0016,
      "step": 20557
    },
    {
      "epoch": 2.881289418360196,
      "grad_norm": 0.6981430649757385,
      "learning_rate": 1.2284142549629274e-05,
      "loss": 0.0695,
      "step": 20558
    },
    {
      "epoch": 2.8814295725297825,
      "grad_norm": 0.04398428276181221,
      "learning_rate": 1.2269791915809615e-05,
      "loss": 0.0016,
      "step": 20559
    },
    {
      "epoch": 2.8815697266993694,
      "grad_norm": 0.2964060604572296,
      "learning_rate": 1.2255441281989953e-05,
      "loss": 0.0358,
      "step": 20560
    },
    {
      "epoch": 2.881709880868956,
      "grad_norm": 0.10761590301990509,
      "learning_rate": 1.2241090648170294e-05,
      "loss": 0.0168,
      "step": 20561
    },
    {
      "epoch": 2.8818500350385423,
      "grad_norm": 0.4069973826408386,
      "learning_rate": 1.2226740014350633e-05,
      "loss": 0.0703,
      "step": 20562
    },
    {
      "epoch": 2.881990189208129,
      "grad_norm": 0.2163611799478531,
      "learning_rate": 1.2212389380530973e-05,
      "loss": 0.0311,
      "step": 20563
    },
    {
      "epoch": 2.8821303433777157,
      "grad_norm": 0.4572719931602478,
      "learning_rate": 1.2198038746711312e-05,
      "loss": 0.0737,
      "step": 20564
    },
    {
      "epoch": 2.882270497547302,
      "grad_norm": 0.10436063259840012,
      "learning_rate": 1.2183688112891652e-05,
      "loss": 0.0085,
      "step": 20565
    },
    {
      "epoch": 2.8824106517168886,
      "grad_norm": 0.2913877069950104,
      "learning_rate": 1.216933747907199e-05,
      "loss": 0.0079,
      "step": 20566
    },
    {
      "epoch": 2.882550805886475,
      "grad_norm": 0.5824584364891052,
      "learning_rate": 1.2154986845252331e-05,
      "loss": 0.0813,
      "step": 20567
    },
    {
      "epoch": 2.8826909600560615,
      "grad_norm": 0.1155749037861824,
      "learning_rate": 1.214063621143267e-05,
      "loss": 0.0061,
      "step": 20568
    },
    {
      "epoch": 2.882831114225648,
      "grad_norm": 0.5120499134063721,
      "learning_rate": 1.212628557761301e-05,
      "loss": 0.09,
      "step": 20569
    },
    {
      "epoch": 2.882971268395235,
      "grad_norm": 0.08372246474027634,
      "learning_rate": 1.2111934943793349e-05,
      "loss": 0.007,
      "step": 20570
    },
    {
      "epoch": 2.8831114225648213,
      "grad_norm": 0.7174161672592163,
      "learning_rate": 1.209758430997369e-05,
      "loss": 0.0355,
      "step": 20571
    },
    {
      "epoch": 2.883251576734408,
      "grad_norm": 0.13389822840690613,
      "learning_rate": 1.2083233676154028e-05,
      "loss": 0.0267,
      "step": 20572
    },
    {
      "epoch": 2.8833917309039943,
      "grad_norm": 0.1598690152168274,
      "learning_rate": 1.2068883042334369e-05,
      "loss": 0.0277,
      "step": 20573
    },
    {
      "epoch": 2.883531885073581,
      "grad_norm": 0.14782583713531494,
      "learning_rate": 1.2054532408514707e-05,
      "loss": 0.0145,
      "step": 20574
    },
    {
      "epoch": 2.8836720392431676,
      "grad_norm": 0.25962793827056885,
      "learning_rate": 1.2040181774695048e-05,
      "loss": 0.0302,
      "step": 20575
    },
    {
      "epoch": 2.883812193412754,
      "grad_norm": 0.10121694207191467,
      "learning_rate": 1.2025831140875387e-05,
      "loss": 0.0047,
      "step": 20576
    },
    {
      "epoch": 2.8839523475823405,
      "grad_norm": 0.1421107053756714,
      "learning_rate": 1.2011480507055727e-05,
      "loss": 0.0064,
      "step": 20577
    },
    {
      "epoch": 2.884092501751927,
      "grad_norm": 0.1648636907339096,
      "learning_rate": 1.1997129873236066e-05,
      "loss": 0.0395,
      "step": 20578
    },
    {
      "epoch": 2.8842326559215135,
      "grad_norm": 0.06656556576490402,
      "learning_rate": 1.1982779239416408e-05,
      "loss": 0.0125,
      "step": 20579
    },
    {
      "epoch": 2.8843728100911004,
      "grad_norm": 0.13815541565418243,
      "learning_rate": 1.1968428605596747e-05,
      "loss": 0.008,
      "step": 20580
    },
    {
      "epoch": 2.884512964260687,
      "grad_norm": 0.27275902032852173,
      "learning_rate": 1.1954077971777087e-05,
      "loss": 0.0095,
      "step": 20581
    },
    {
      "epoch": 2.8846531184302733,
      "grad_norm": 0.18239572644233704,
      "learning_rate": 1.1939727337957426e-05,
      "loss": 0.0221,
      "step": 20582
    },
    {
      "epoch": 2.8847932725998597,
      "grad_norm": 0.17327091097831726,
      "learning_rate": 1.1925376704137766e-05,
      "loss": 0.0295,
      "step": 20583
    },
    {
      "epoch": 2.8849334267694466,
      "grad_norm": 0.10624104738235474,
      "learning_rate": 1.1911026070318105e-05,
      "loss": 0.0226,
      "step": 20584
    },
    {
      "epoch": 2.885073580939033,
      "grad_norm": 0.16192522644996643,
      "learning_rate": 1.1896675436498445e-05,
      "loss": 0.0089,
      "step": 20585
    },
    {
      "epoch": 2.8852137351086196,
      "grad_norm": 0.22291190922260284,
      "learning_rate": 1.1882324802678784e-05,
      "loss": 0.0274,
      "step": 20586
    },
    {
      "epoch": 2.885353889278206,
      "grad_norm": 0.24589356780052185,
      "learning_rate": 1.1867974168859124e-05,
      "loss": 0.0228,
      "step": 20587
    },
    {
      "epoch": 2.8854940434477925,
      "grad_norm": 0.5466956496238708,
      "learning_rate": 1.1853623535039463e-05,
      "loss": 0.0315,
      "step": 20588
    },
    {
      "epoch": 2.885634197617379,
      "grad_norm": 0.28918564319610596,
      "learning_rate": 1.1839272901219804e-05,
      "loss": 0.032,
      "step": 20589
    },
    {
      "epoch": 2.8857743517869654,
      "grad_norm": 0.11162767559289932,
      "learning_rate": 1.1824922267400142e-05,
      "loss": 0.0144,
      "step": 20590
    },
    {
      "epoch": 2.8859145059565523,
      "grad_norm": 0.11436593532562256,
      "learning_rate": 1.1810571633580483e-05,
      "loss": 0.0317,
      "step": 20591
    },
    {
      "epoch": 2.8860546601261388,
      "grad_norm": 0.12318064272403717,
      "learning_rate": 1.1796220999760821e-05,
      "loss": 0.0209,
      "step": 20592
    },
    {
      "epoch": 2.886194814295725,
      "grad_norm": 0.11822043359279633,
      "learning_rate": 1.1781870365941162e-05,
      "loss": 0.0262,
      "step": 20593
    },
    {
      "epoch": 2.8863349684653117,
      "grad_norm": 0.41611841320991516,
      "learning_rate": 1.17675197321215e-05,
      "loss": 0.0171,
      "step": 20594
    },
    {
      "epoch": 2.8864751226348986,
      "grad_norm": 0.269073486328125,
      "learning_rate": 1.1753169098301841e-05,
      "loss": 0.0164,
      "step": 20595
    },
    {
      "epoch": 2.886615276804485,
      "grad_norm": 0.21539461612701416,
      "learning_rate": 1.173881846448218e-05,
      "loss": 0.0295,
      "step": 20596
    },
    {
      "epoch": 2.8867554309740715,
      "grad_norm": 0.11899608373641968,
      "learning_rate": 1.172446783066252e-05,
      "loss": 0.0127,
      "step": 20597
    },
    {
      "epoch": 2.886895585143658,
      "grad_norm": 0.22435010969638824,
      "learning_rate": 1.1710117196842859e-05,
      "loss": 0.0125,
      "step": 20598
    },
    {
      "epoch": 2.8870357393132444,
      "grad_norm": 0.2176763415336609,
      "learning_rate": 1.16957665630232e-05,
      "loss": 0.054,
      "step": 20599
    },
    {
      "epoch": 2.887175893482831,
      "grad_norm": 0.0818527415394783,
      "learning_rate": 1.1681415929203538e-05,
      "loss": 0.0049,
      "step": 20600
    },
    {
      "epoch": 2.887316047652418,
      "grad_norm": 0.5863497257232666,
      "learning_rate": 1.1667065295383878e-05,
      "loss": 0.0575,
      "step": 20601
    },
    {
      "epoch": 2.8874562018220042,
      "grad_norm": 0.15718479454517365,
      "learning_rate": 1.1652714661564217e-05,
      "loss": 0.03,
      "step": 20602
    },
    {
      "epoch": 2.8875963559915907,
      "grad_norm": 0.19008733332157135,
      "learning_rate": 1.1638364027744558e-05,
      "loss": 0.0415,
      "step": 20603
    },
    {
      "epoch": 2.887736510161177,
      "grad_norm": 0.02169755846261978,
      "learning_rate": 1.1624013393924896e-05,
      "loss": 0.0016,
      "step": 20604
    },
    {
      "epoch": 2.887876664330764,
      "grad_norm": 0.044860586524009705,
      "learning_rate": 1.1609662760105237e-05,
      "loss": 0.0017,
      "step": 20605
    },
    {
      "epoch": 2.8880168185003505,
      "grad_norm": 0.37165677547454834,
      "learning_rate": 1.1595312126285575e-05,
      "loss": 0.0248,
      "step": 20606
    },
    {
      "epoch": 2.888156972669937,
      "grad_norm": 0.0824437215924263,
      "learning_rate": 1.1580961492465916e-05,
      "loss": 0.0067,
      "step": 20607
    },
    {
      "epoch": 2.8882971268395234,
      "grad_norm": 1.2491965293884277,
      "learning_rate": 1.1566610858646255e-05,
      "loss": 0.034,
      "step": 20608
    },
    {
      "epoch": 2.88843728100911,
      "grad_norm": 0.4895854890346527,
      "learning_rate": 1.1552260224826595e-05,
      "loss": 0.0993,
      "step": 20609
    },
    {
      "epoch": 2.8885774351786964,
      "grad_norm": 0.763146162033081,
      "learning_rate": 1.1537909591006934e-05,
      "loss": 0.0314,
      "step": 20610
    },
    {
      "epoch": 2.8887175893482833,
      "grad_norm": 0.1062176451086998,
      "learning_rate": 1.1523558957187274e-05,
      "loss": 0.011,
      "step": 20611
    },
    {
      "epoch": 2.8888577435178697,
      "grad_norm": 0.08441688120365143,
      "learning_rate": 1.1509208323367613e-05,
      "loss": 0.0093,
      "step": 20612
    },
    {
      "epoch": 2.888997897687456,
      "grad_norm": 0.17439578473567963,
      "learning_rate": 1.1494857689547953e-05,
      "loss": 0.0343,
      "step": 20613
    },
    {
      "epoch": 2.8891380518570426,
      "grad_norm": 0.28553247451782227,
      "learning_rate": 1.1480507055728292e-05,
      "loss": 0.0368,
      "step": 20614
    },
    {
      "epoch": 2.8892782060266295,
      "grad_norm": 0.22482764720916748,
      "learning_rate": 1.1466156421908632e-05,
      "loss": 0.012,
      "step": 20615
    },
    {
      "epoch": 2.889418360196216,
      "grad_norm": 1.044072151184082,
      "learning_rate": 1.1451805788088974e-05,
      "loss": 0.0209,
      "step": 20616
    },
    {
      "epoch": 2.8895585143658025,
      "grad_norm": 1.2441165447235107,
      "learning_rate": 1.1437455154269313e-05,
      "loss": 0.0408,
      "step": 20617
    },
    {
      "epoch": 2.889698668535389,
      "grad_norm": 0.0749635100364685,
      "learning_rate": 1.1423104520449654e-05,
      "loss": 0.0059,
      "step": 20618
    },
    {
      "epoch": 2.8898388227049754,
      "grad_norm": 0.26553890109062195,
      "learning_rate": 1.1408753886629992e-05,
      "loss": 0.1125,
      "step": 20619
    },
    {
      "epoch": 2.889978976874562,
      "grad_norm": 0.021862143650650978,
      "learning_rate": 1.1394403252810333e-05,
      "loss": 0.0014,
      "step": 20620
    },
    {
      "epoch": 2.8901191310441483,
      "grad_norm": 0.14215447008609772,
      "learning_rate": 1.1380052618990671e-05,
      "loss": 0.0048,
      "step": 20621
    },
    {
      "epoch": 2.890259285213735,
      "grad_norm": 0.4674132168292999,
      "learning_rate": 1.1365701985171012e-05,
      "loss": 0.0171,
      "step": 20622
    },
    {
      "epoch": 2.8903994393833217,
      "grad_norm": 0.1958397924900055,
      "learning_rate": 1.135135135135135e-05,
      "loss": 0.0103,
      "step": 20623
    },
    {
      "epoch": 2.890539593552908,
      "grad_norm": 0.32101282477378845,
      "learning_rate": 1.1337000717531691e-05,
      "loss": 0.071,
      "step": 20624
    },
    {
      "epoch": 2.8906797477224946,
      "grad_norm": 0.2379656285047531,
      "learning_rate": 1.132265008371203e-05,
      "loss": 0.0222,
      "step": 20625
    },
    {
      "epoch": 2.8908199018920815,
      "grad_norm": 0.1310795545578003,
      "learning_rate": 1.130829944989237e-05,
      "loss": 0.0076,
      "step": 20626
    },
    {
      "epoch": 2.890960056061668,
      "grad_norm": 0.22999387979507446,
      "learning_rate": 1.1293948816072709e-05,
      "loss": 0.0246,
      "step": 20627
    },
    {
      "epoch": 2.8911002102312544,
      "grad_norm": 0.1883697211742401,
      "learning_rate": 1.127959818225305e-05,
      "loss": 0.0147,
      "step": 20628
    },
    {
      "epoch": 2.891240364400841,
      "grad_norm": 0.0720987319946289,
      "learning_rate": 1.1265247548433388e-05,
      "loss": 0.0094,
      "step": 20629
    },
    {
      "epoch": 2.8913805185704273,
      "grad_norm": 0.3023030757904053,
      "learning_rate": 1.1250896914613728e-05,
      "loss": 0.047,
      "step": 20630
    },
    {
      "epoch": 2.891520672740014,
      "grad_norm": 0.1286933571100235,
      "learning_rate": 1.1236546280794067e-05,
      "loss": 0.0205,
      "step": 20631
    },
    {
      "epoch": 2.8916608269096007,
      "grad_norm": 0.3219941258430481,
      "learning_rate": 1.1222195646974408e-05,
      "loss": 0.0243,
      "step": 20632
    },
    {
      "epoch": 2.891800981079187,
      "grad_norm": 0.09781860560178757,
      "learning_rate": 1.1207845013154746e-05,
      "loss": 0.0053,
      "step": 20633
    },
    {
      "epoch": 2.8919411352487736,
      "grad_norm": 0.059332434087991714,
      "learning_rate": 1.1193494379335087e-05,
      "loss": 0.0053,
      "step": 20634
    },
    {
      "epoch": 2.89208128941836,
      "grad_norm": 0.1558128148317337,
      "learning_rate": 1.1179143745515425e-05,
      "loss": 0.0207,
      "step": 20635
    },
    {
      "epoch": 2.892221443587947,
      "grad_norm": 0.43020737171173096,
      "learning_rate": 1.1164793111695766e-05,
      "loss": 0.0261,
      "step": 20636
    },
    {
      "epoch": 2.8923615977575334,
      "grad_norm": 0.5143353939056396,
      "learning_rate": 1.1150442477876105e-05,
      "loss": 0.0409,
      "step": 20637
    },
    {
      "epoch": 2.89250175192712,
      "grad_norm": 0.13568823039531708,
      "learning_rate": 1.1136091844056445e-05,
      "loss": 0.0218,
      "step": 20638
    },
    {
      "epoch": 2.8926419060967064,
      "grad_norm": 0.11360125988721848,
      "learning_rate": 1.1121741210236784e-05,
      "loss": 0.012,
      "step": 20639
    },
    {
      "epoch": 2.892782060266293,
      "grad_norm": 1.168239951133728,
      "learning_rate": 1.1107390576417124e-05,
      "loss": 0.0701,
      "step": 20640
    },
    {
      "epoch": 2.8929222144358793,
      "grad_norm": 0.5349233746528625,
      "learning_rate": 1.1093039942597463e-05,
      "loss": 0.0866,
      "step": 20641
    },
    {
      "epoch": 2.893062368605466,
      "grad_norm": 0.16325749456882477,
      "learning_rate": 1.1078689308777803e-05,
      "loss": 0.0139,
      "step": 20642
    },
    {
      "epoch": 2.8932025227750526,
      "grad_norm": 0.08652399480342865,
      "learning_rate": 1.1064338674958142e-05,
      "loss": 0.0098,
      "step": 20643
    },
    {
      "epoch": 2.893342676944639,
      "grad_norm": 0.1399209052324295,
      "learning_rate": 1.1049988041138482e-05,
      "loss": 0.0135,
      "step": 20644
    },
    {
      "epoch": 2.8934828311142256,
      "grad_norm": 0.4124363958835602,
      "learning_rate": 1.1035637407318821e-05,
      "loss": 0.0576,
      "step": 20645
    },
    {
      "epoch": 2.8936229852838125,
      "grad_norm": 0.23660466074943542,
      "learning_rate": 1.1021286773499162e-05,
      "loss": 0.0382,
      "step": 20646
    },
    {
      "epoch": 2.893763139453399,
      "grad_norm": 0.40035301446914673,
      "learning_rate": 1.10069361396795e-05,
      "loss": 0.0463,
      "step": 20647
    },
    {
      "epoch": 2.8939032936229854,
      "grad_norm": 0.269803524017334,
      "learning_rate": 1.099258550585984e-05,
      "loss": 0.0317,
      "step": 20648
    },
    {
      "epoch": 2.894043447792572,
      "grad_norm": 0.1005750447511673,
      "learning_rate": 1.097823487204018e-05,
      "loss": 0.0046,
      "step": 20649
    },
    {
      "epoch": 2.8941836019621583,
      "grad_norm": 0.3053402900695801,
      "learning_rate": 1.096388423822052e-05,
      "loss": 0.0212,
      "step": 20650
    },
    {
      "epoch": 2.8943237561317448,
      "grad_norm": 0.25356629490852356,
      "learning_rate": 1.0949533604400859e-05,
      "loss": 0.0311,
      "step": 20651
    },
    {
      "epoch": 2.894463910301331,
      "grad_norm": 0.08650394529104233,
      "learning_rate": 1.0935182970581199e-05,
      "loss": 0.0043,
      "step": 20652
    },
    {
      "epoch": 2.894604064470918,
      "grad_norm": 0.25055503845214844,
      "learning_rate": 1.0920832336761541e-05,
      "loss": 0.036,
      "step": 20653
    },
    {
      "epoch": 2.8947442186405046,
      "grad_norm": 0.06276492029428482,
      "learning_rate": 1.090648170294188e-05,
      "loss": 0.0037,
      "step": 20654
    },
    {
      "epoch": 2.894884372810091,
      "grad_norm": 0.3738802373409271,
      "learning_rate": 1.089213106912222e-05,
      "loss": 0.0367,
      "step": 20655
    },
    {
      "epoch": 2.8950245269796775,
      "grad_norm": 0.14314015209674835,
      "learning_rate": 1.0877780435302559e-05,
      "loss": 0.0222,
      "step": 20656
    },
    {
      "epoch": 2.8951646811492644,
      "grad_norm": 0.06546798348426819,
      "learning_rate": 1.08634298014829e-05,
      "loss": 0.0023,
      "step": 20657
    },
    {
      "epoch": 2.895304835318851,
      "grad_norm": 0.2814999222755432,
      "learning_rate": 1.0849079167663238e-05,
      "loss": 0.0249,
      "step": 20658
    },
    {
      "epoch": 2.8954449894884373,
      "grad_norm": 0.16743066906929016,
      "learning_rate": 1.0834728533843579e-05,
      "loss": 0.0309,
      "step": 20659
    },
    {
      "epoch": 2.8955851436580238,
      "grad_norm": 0.04918849840760231,
      "learning_rate": 1.0820377900023917e-05,
      "loss": 0.0077,
      "step": 20660
    },
    {
      "epoch": 2.8957252978276102,
      "grad_norm": 0.40250658988952637,
      "learning_rate": 1.0806027266204258e-05,
      "loss": 0.0538,
      "step": 20661
    },
    {
      "epoch": 2.8958654519971967,
      "grad_norm": 0.38218533992767334,
      "learning_rate": 1.0791676632384596e-05,
      "loss": 0.0413,
      "step": 20662
    },
    {
      "epoch": 2.8960056061667836,
      "grad_norm": 0.04741606116294861,
      "learning_rate": 1.0777325998564937e-05,
      "loss": 0.0028,
      "step": 20663
    },
    {
      "epoch": 2.89614576033637,
      "grad_norm": 0.1715167760848999,
      "learning_rate": 1.0762975364745276e-05,
      "loss": 0.0258,
      "step": 20664
    },
    {
      "epoch": 2.8962859145059565,
      "grad_norm": 0.15992993116378784,
      "learning_rate": 1.0748624730925616e-05,
      "loss": 0.015,
      "step": 20665
    },
    {
      "epoch": 2.896426068675543,
      "grad_norm": 0.2722056210041046,
      "learning_rate": 1.0734274097105955e-05,
      "loss": 0.0365,
      "step": 20666
    },
    {
      "epoch": 2.89656622284513,
      "grad_norm": 0.8713580965995789,
      "learning_rate": 1.0719923463286295e-05,
      "loss": 0.1105,
      "step": 20667
    },
    {
      "epoch": 2.8967063770147163,
      "grad_norm": 0.5795403122901917,
      "learning_rate": 1.0705572829466634e-05,
      "loss": 0.0662,
      "step": 20668
    },
    {
      "epoch": 2.896846531184303,
      "grad_norm": 0.3681755065917969,
      "learning_rate": 1.0691222195646974e-05,
      "loss": 0.0075,
      "step": 20669
    },
    {
      "epoch": 2.8969866853538893,
      "grad_norm": 0.4776165783405304,
      "learning_rate": 1.0676871561827313e-05,
      "loss": 0.0787,
      "step": 20670
    },
    {
      "epoch": 2.8971268395234757,
      "grad_norm": 0.17915162444114685,
      "learning_rate": 1.0662520928007653e-05,
      "loss": 0.0254,
      "step": 20671
    },
    {
      "epoch": 2.897266993693062,
      "grad_norm": 0.7088584303855896,
      "learning_rate": 1.0648170294187992e-05,
      "loss": 0.0454,
      "step": 20672
    },
    {
      "epoch": 2.897407147862649,
      "grad_norm": 0.1979878842830658,
      "learning_rate": 1.0633819660368333e-05,
      "loss": 0.0339,
      "step": 20673
    },
    {
      "epoch": 2.8975473020322355,
      "grad_norm": 0.07501644641160965,
      "learning_rate": 1.0619469026548671e-05,
      "loss": 0.0078,
      "step": 20674
    },
    {
      "epoch": 2.897687456201822,
      "grad_norm": 0.10961319506168365,
      "learning_rate": 1.0605118392729012e-05,
      "loss": 0.0069,
      "step": 20675
    },
    {
      "epoch": 2.8978276103714085,
      "grad_norm": 0.15714536607265472,
      "learning_rate": 1.059076775890935e-05,
      "loss": 0.0084,
      "step": 20676
    },
    {
      "epoch": 2.8979677645409954,
      "grad_norm": 0.2076135277748108,
      "learning_rate": 1.0576417125089691e-05,
      "loss": 0.0081,
      "step": 20677
    },
    {
      "epoch": 2.898107918710582,
      "grad_norm": 0.098606638610363,
      "learning_rate": 1.056206649127003e-05,
      "loss": 0.0132,
      "step": 20678
    },
    {
      "epoch": 2.8982480728801683,
      "grad_norm": 0.11080455780029297,
      "learning_rate": 1.054771585745037e-05,
      "loss": 0.0218,
      "step": 20679
    },
    {
      "epoch": 2.8983882270497547,
      "grad_norm": 0.811276912689209,
      "learning_rate": 1.0533365223630709e-05,
      "loss": 0.0626,
      "step": 20680
    },
    {
      "epoch": 2.898528381219341,
      "grad_norm": 0.31210771203041077,
      "learning_rate": 1.051901458981105e-05,
      "loss": 0.0471,
      "step": 20681
    },
    {
      "epoch": 2.8986685353889277,
      "grad_norm": 0.0819486752152443,
      "learning_rate": 1.0504663955991388e-05,
      "loss": 0.0062,
      "step": 20682
    },
    {
      "epoch": 2.898808689558514,
      "grad_norm": 0.4523376524448395,
      "learning_rate": 1.0490313322171728e-05,
      "loss": 0.0461,
      "step": 20683
    },
    {
      "epoch": 2.898948843728101,
      "grad_norm": 0.15217606723308563,
      "learning_rate": 1.0475962688352067e-05,
      "loss": 0.0564,
      "step": 20684
    },
    {
      "epoch": 2.8990889978976875,
      "grad_norm": 0.0509762205183506,
      "learning_rate": 1.0461612054532407e-05,
      "loss": 0.0032,
      "step": 20685
    },
    {
      "epoch": 2.899229152067274,
      "grad_norm": 0.14570952951908112,
      "learning_rate": 1.0447261420712746e-05,
      "loss": 0.0201,
      "step": 20686
    },
    {
      "epoch": 2.8993693062368604,
      "grad_norm": 0.11255156993865967,
      "learning_rate": 1.0432910786893087e-05,
      "loss": 0.0036,
      "step": 20687
    },
    {
      "epoch": 2.8995094604064473,
      "grad_norm": 0.3249497711658478,
      "learning_rate": 1.0418560153073425e-05,
      "loss": 0.1012,
      "step": 20688
    },
    {
      "epoch": 2.8996496145760338,
      "grad_norm": 0.2085169106721878,
      "learning_rate": 1.0404209519253766e-05,
      "loss": 0.0213,
      "step": 20689
    },
    {
      "epoch": 2.89978976874562,
      "grad_norm": 0.1037660464644432,
      "learning_rate": 1.0389858885434106e-05,
      "loss": 0.0112,
      "step": 20690
    },
    {
      "epoch": 2.8999299229152067,
      "grad_norm": 0.5664847493171692,
      "learning_rate": 1.0375508251614447e-05,
      "loss": 0.0078,
      "step": 20691
    },
    {
      "epoch": 2.900070077084793,
      "grad_norm": 0.2693379819393158,
      "learning_rate": 1.0361157617794785e-05,
      "loss": 0.0182,
      "step": 20692
    },
    {
      "epoch": 2.9002102312543796,
      "grad_norm": 0.10767032951116562,
      "learning_rate": 1.0346806983975126e-05,
      "loss": 0.0043,
      "step": 20693
    },
    {
      "epoch": 2.9003503854239665,
      "grad_norm": 0.17976680397987366,
      "learning_rate": 1.0332456350155464e-05,
      "loss": 0.0079,
      "step": 20694
    },
    {
      "epoch": 2.900490539593553,
      "grad_norm": 0.06164063513278961,
      "learning_rate": 1.0318105716335805e-05,
      "loss": 0.0113,
      "step": 20695
    },
    {
      "epoch": 2.9006306937631394,
      "grad_norm": 0.18699444830417633,
      "learning_rate": 1.0303755082516144e-05,
      "loss": 0.0255,
      "step": 20696
    },
    {
      "epoch": 2.900770847932726,
      "grad_norm": 0.04256226867437363,
      "learning_rate": 1.0289404448696484e-05,
      "loss": 0.0051,
      "step": 20697
    },
    {
      "epoch": 2.900911002102313,
      "grad_norm": 0.033912040293216705,
      "learning_rate": 1.0275053814876823e-05,
      "loss": 0.0019,
      "step": 20698
    },
    {
      "epoch": 2.9010511562718992,
      "grad_norm": 0.2507621943950653,
      "learning_rate": 1.0260703181057163e-05,
      "loss": 0.0283,
      "step": 20699
    },
    {
      "epoch": 2.9011913104414857,
      "grad_norm": 0.2935486137866974,
      "learning_rate": 1.0246352547237502e-05,
      "loss": 0.0304,
      "step": 20700
    },
    {
      "epoch": 2.901331464611072,
      "grad_norm": 0.22213530540466309,
      "learning_rate": 1.0232001913417842e-05,
      "loss": 0.0118,
      "step": 20701
    },
    {
      "epoch": 2.9014716187806586,
      "grad_norm": 0.17837722599506378,
      "learning_rate": 1.0217651279598181e-05,
      "loss": 0.009,
      "step": 20702
    },
    {
      "epoch": 2.901611772950245,
      "grad_norm": 0.3788335919380188,
      "learning_rate": 1.0203300645778521e-05,
      "loss": 0.0278,
      "step": 20703
    },
    {
      "epoch": 2.9017519271198315,
      "grad_norm": 0.05208876356482506,
      "learning_rate": 1.018895001195886e-05,
      "loss": 0.0039,
      "step": 20704
    },
    {
      "epoch": 2.9018920812894184,
      "grad_norm": 0.24590559303760529,
      "learning_rate": 1.01745993781392e-05,
      "loss": 0.0241,
      "step": 20705
    },
    {
      "epoch": 2.902032235459005,
      "grad_norm": 0.22803764045238495,
      "learning_rate": 1.016024874431954e-05,
      "loss": 0.0243,
      "step": 20706
    },
    {
      "epoch": 2.9021723896285914,
      "grad_norm": 0.1533641219139099,
      "learning_rate": 1.014589811049988e-05,
      "loss": 0.0262,
      "step": 20707
    },
    {
      "epoch": 2.9023125437981783,
      "grad_norm": 0.5914942622184753,
      "learning_rate": 1.0131547476680218e-05,
      "loss": 0.0179,
      "step": 20708
    },
    {
      "epoch": 2.9024526979677647,
      "grad_norm": 0.05916742607951164,
      "learning_rate": 1.0117196842860559e-05,
      "loss": 0.0047,
      "step": 20709
    },
    {
      "epoch": 2.902592852137351,
      "grad_norm": 0.11456607282161713,
      "learning_rate": 1.0102846209040898e-05,
      "loss": 0.006,
      "step": 20710
    },
    {
      "epoch": 2.9027330063069376,
      "grad_norm": 0.2938591241836548,
      "learning_rate": 1.0088495575221238e-05,
      "loss": 0.0213,
      "step": 20711
    },
    {
      "epoch": 2.902873160476524,
      "grad_norm": 0.14763087034225464,
      "learning_rate": 1.0074144941401577e-05,
      "loss": 0.0081,
      "step": 20712
    },
    {
      "epoch": 2.9030133146461106,
      "grad_norm": 0.5123448967933655,
      "learning_rate": 1.0059794307581917e-05,
      "loss": 0.0476,
      "step": 20713
    },
    {
      "epoch": 2.903153468815697,
      "grad_norm": 0.3131152391433716,
      "learning_rate": 1.0045443673762256e-05,
      "loss": 0.0095,
      "step": 20714
    },
    {
      "epoch": 2.903293622985284,
      "grad_norm": 0.07629865407943726,
      "learning_rate": 1.0031093039942596e-05,
      "loss": 0.0311,
      "step": 20715
    },
    {
      "epoch": 2.9034337771548704,
      "grad_norm": 0.2567148506641388,
      "learning_rate": 1.0016742406122935e-05,
      "loss": 0.0085,
      "step": 20716
    },
    {
      "epoch": 2.903573931324457,
      "grad_norm": 0.04181884974241257,
      "learning_rate": 1.0002391772303275e-05,
      "loss": 0.002,
      "step": 20717
    },
    {
      "epoch": 2.9037140854940433,
      "grad_norm": 0.054413337260484695,
      "learning_rate": 9.988041138483614e-06,
      "loss": 0.0037,
      "step": 20718
    },
    {
      "epoch": 2.90385423966363,
      "grad_norm": 1.6050736904144287,
      "learning_rate": 9.973690504663955e-06,
      "loss": 0.1004,
      "step": 20719
    },
    {
      "epoch": 2.9039943938332167,
      "grad_norm": 0.1725742667913437,
      "learning_rate": 9.959339870844293e-06,
      "loss": 0.0044,
      "step": 20720
    },
    {
      "epoch": 2.904134548002803,
      "grad_norm": 0.12213186919689178,
      "learning_rate": 9.944989237024634e-06,
      "loss": 0.0208,
      "step": 20721
    },
    {
      "epoch": 2.9042747021723896,
      "grad_norm": 0.14038968086242676,
      "learning_rate": 9.930638603204972e-06,
      "loss": 0.0231,
      "step": 20722
    },
    {
      "epoch": 2.904414856341976,
      "grad_norm": 0.4169723093509674,
      "learning_rate": 9.916287969385313e-06,
      "loss": 0.037,
      "step": 20723
    },
    {
      "epoch": 2.9045550105115625,
      "grad_norm": 0.15851376950740814,
      "learning_rate": 9.901937335565652e-06,
      "loss": 0.0213,
      "step": 20724
    },
    {
      "epoch": 2.9046951646811494,
      "grad_norm": 0.7114185690879822,
      "learning_rate": 9.887586701745992e-06,
      "loss": 0.0153,
      "step": 20725
    },
    {
      "epoch": 2.904835318850736,
      "grad_norm": 0.4193439185619354,
      "learning_rate": 9.87323606792633e-06,
      "loss": 0.0261,
      "step": 20726
    },
    {
      "epoch": 2.9049754730203223,
      "grad_norm": 0.12254543602466583,
      "learning_rate": 9.858885434106671e-06,
      "loss": 0.017,
      "step": 20727
    },
    {
      "epoch": 2.905115627189909,
      "grad_norm": 0.2768273651599884,
      "learning_rate": 9.844534800287013e-06,
      "loss": 0.028,
      "step": 20728
    },
    {
      "epoch": 2.9052557813594957,
      "grad_norm": 0.29675331711769104,
      "learning_rate": 9.830184166467352e-06,
      "loss": 0.02,
      "step": 20729
    },
    {
      "epoch": 2.905395935529082,
      "grad_norm": 0.1500357985496521,
      "learning_rate": 9.815833532647692e-06,
      "loss": 0.0126,
      "step": 20730
    },
    {
      "epoch": 2.9055360896986686,
      "grad_norm": 0.3799542188644409,
      "learning_rate": 9.801482898828031e-06,
      "loss": 0.036,
      "step": 20731
    },
    {
      "epoch": 2.905676243868255,
      "grad_norm": 0.15092352032661438,
      "learning_rate": 9.787132265008372e-06,
      "loss": 0.0098,
      "step": 20732
    },
    {
      "epoch": 2.9058163980378415,
      "grad_norm": 0.20201392471790314,
      "learning_rate": 9.77278163118871e-06,
      "loss": 0.0154,
      "step": 20733
    },
    {
      "epoch": 2.905956552207428,
      "grad_norm": 0.2058466672897339,
      "learning_rate": 9.75843099736905e-06,
      "loss": 0.0416,
      "step": 20734
    },
    {
      "epoch": 2.9060967063770144,
      "grad_norm": 0.2651395797729492,
      "learning_rate": 9.74408036354939e-06,
      "loss": 0.0364,
      "step": 20735
    },
    {
      "epoch": 2.9062368605466014,
      "grad_norm": 0.25694939494132996,
      "learning_rate": 9.72972972972973e-06,
      "loss": 0.0341,
      "step": 20736
    },
    {
      "epoch": 2.906377014716188,
      "grad_norm": 0.30681174993515015,
      "learning_rate": 9.715379095910069e-06,
      "loss": 0.0414,
      "step": 20737
    },
    {
      "epoch": 2.9065171688857743,
      "grad_norm": 0.14340701699256897,
      "learning_rate": 9.701028462090409e-06,
      "loss": 0.0237,
      "step": 20738
    },
    {
      "epoch": 2.9066573230553607,
      "grad_norm": 0.40006113052368164,
      "learning_rate": 9.686677828270748e-06,
      "loss": 0.012,
      "step": 20739
    },
    {
      "epoch": 2.9067974772249476,
      "grad_norm": 0.16262710094451904,
      "learning_rate": 9.672327194451088e-06,
      "loss": 0.0178,
      "step": 20740
    },
    {
      "epoch": 2.906937631394534,
      "grad_norm": 0.15633246302604675,
      "learning_rate": 9.657976560631427e-06,
      "loss": 0.0232,
      "step": 20741
    },
    {
      "epoch": 2.9070777855641206,
      "grad_norm": 0.16302275657653809,
      "learning_rate": 9.643625926811767e-06,
      "loss": 0.0133,
      "step": 20742
    },
    {
      "epoch": 2.907217939733707,
      "grad_norm": 0.2189866006374359,
      "learning_rate": 9.629275292992106e-06,
      "loss": 0.0526,
      "step": 20743
    },
    {
      "epoch": 2.9073580939032935,
      "grad_norm": 0.32441413402557373,
      "learning_rate": 9.614924659172446e-06,
      "loss": 0.0227,
      "step": 20744
    },
    {
      "epoch": 2.90749824807288,
      "grad_norm": 0.19663912057876587,
      "learning_rate": 9.600574025352785e-06,
      "loss": 0.0194,
      "step": 20745
    },
    {
      "epoch": 2.907638402242467,
      "grad_norm": 0.1286417692899704,
      "learning_rate": 9.586223391533126e-06,
      "loss": 0.0137,
      "step": 20746
    },
    {
      "epoch": 2.9077785564120533,
      "grad_norm": 0.38042134046554565,
      "learning_rate": 9.571872757713464e-06,
      "loss": 0.0213,
      "step": 20747
    },
    {
      "epoch": 2.9079187105816398,
      "grad_norm": 0.3720579445362091,
      "learning_rate": 9.557522123893805e-06,
      "loss": 0.0215,
      "step": 20748
    },
    {
      "epoch": 2.908058864751226,
      "grad_norm": 2.1758949756622314,
      "learning_rate": 9.543171490074143e-06,
      "loss": 0.0153,
      "step": 20749
    },
    {
      "epoch": 2.908199018920813,
      "grad_norm": 0.26301172375679016,
      "learning_rate": 9.528820856254484e-06,
      "loss": 0.0649,
      "step": 20750
    },
    {
      "epoch": 2.9083391730903996,
      "grad_norm": 0.12222176045179367,
      "learning_rate": 9.514470222434823e-06,
      "loss": 0.0325,
      "step": 20751
    },
    {
      "epoch": 2.908479327259986,
      "grad_norm": 0.10018991678953171,
      "learning_rate": 9.500119588615163e-06,
      "loss": 0.0105,
      "step": 20752
    },
    {
      "epoch": 2.9086194814295725,
      "grad_norm": 0.27538055181503296,
      "learning_rate": 9.485768954795502e-06,
      "loss": 0.0629,
      "step": 20753
    },
    {
      "epoch": 2.908759635599159,
      "grad_norm": 0.09087176620960236,
      "learning_rate": 9.471418320975842e-06,
      "loss": 0.0058,
      "step": 20754
    },
    {
      "epoch": 2.9088997897687454,
      "grad_norm": 0.35563525557518005,
      "learning_rate": 9.457067687156181e-06,
      "loss": 0.0486,
      "step": 20755
    },
    {
      "epoch": 2.9090399439383323,
      "grad_norm": 0.6977300643920898,
      "learning_rate": 9.442717053336521e-06,
      "loss": 0.0529,
      "step": 20756
    },
    {
      "epoch": 2.9091800981079188,
      "grad_norm": 0.03876902535557747,
      "learning_rate": 9.42836641951686e-06,
      "loss": 0.0036,
      "step": 20757
    },
    {
      "epoch": 2.9093202522775052,
      "grad_norm": 0.2578149437904358,
      "learning_rate": 9.4140157856972e-06,
      "loss": 0.01,
      "step": 20758
    },
    {
      "epoch": 2.9094604064470917,
      "grad_norm": 0.34983158111572266,
      "learning_rate": 9.399665151877539e-06,
      "loss": 0.0328,
      "step": 20759
    },
    {
      "epoch": 2.9096005606166786,
      "grad_norm": 0.11553102731704712,
      "learning_rate": 9.38531451805788e-06,
      "loss": 0.0096,
      "step": 20760
    },
    {
      "epoch": 2.909740714786265,
      "grad_norm": 0.11713887006044388,
      "learning_rate": 9.37096388423822e-06,
      "loss": 0.011,
      "step": 20761
    },
    {
      "epoch": 2.9098808689558515,
      "grad_norm": 0.03882862627506256,
      "learning_rate": 9.356613250418559e-06,
      "loss": 0.0037,
      "step": 20762
    },
    {
      "epoch": 2.910021023125438,
      "grad_norm": 0.1979157030582428,
      "learning_rate": 9.342262616598899e-06,
      "loss": 0.017,
      "step": 20763
    },
    {
      "epoch": 2.9101611772950244,
      "grad_norm": 0.1617109775543213,
      "learning_rate": 9.327911982779238e-06,
      "loss": 0.0091,
      "step": 20764
    },
    {
      "epoch": 2.910301331464611,
      "grad_norm": 0.11878762394189835,
      "learning_rate": 9.313561348959578e-06,
      "loss": 0.007,
      "step": 20765
    },
    {
      "epoch": 2.9104414856341974,
      "grad_norm": 0.4726565480232239,
      "learning_rate": 9.299210715139917e-06,
      "loss": 0.0657,
      "step": 20766
    },
    {
      "epoch": 2.9105816398037843,
      "grad_norm": 0.35598182678222656,
      "learning_rate": 9.284860081320257e-06,
      "loss": 0.026,
      "step": 20767
    },
    {
      "epoch": 2.9107217939733707,
      "grad_norm": 0.8986560106277466,
      "learning_rate": 9.270509447500596e-06,
      "loss": 0.176,
      "step": 20768
    },
    {
      "epoch": 2.910861948142957,
      "grad_norm": 0.4295255243778229,
      "learning_rate": 9.256158813680937e-06,
      "loss": 0.0282,
      "step": 20769
    },
    {
      "epoch": 2.9110021023125436,
      "grad_norm": 2.182746410369873,
      "learning_rate": 9.241808179861275e-06,
      "loss": 0.1048,
      "step": 20770
    },
    {
      "epoch": 2.9111422564821305,
      "grad_norm": 0.13311715424060822,
      "learning_rate": 9.227457546041616e-06,
      "loss": 0.0087,
      "step": 20771
    },
    {
      "epoch": 2.911282410651717,
      "grad_norm": 0.17250411212444305,
      "learning_rate": 9.213106912221954e-06,
      "loss": 0.0265,
      "step": 20772
    },
    {
      "epoch": 2.9114225648213035,
      "grad_norm": 0.2883392870426178,
      "learning_rate": 9.198756278402295e-06,
      "loss": 0.0361,
      "step": 20773
    },
    {
      "epoch": 2.91156271899089,
      "grad_norm": 0.1712668240070343,
      "learning_rate": 9.184405644582635e-06,
      "loss": 0.0166,
      "step": 20774
    },
    {
      "epoch": 2.9117028731604764,
      "grad_norm": 0.1162927895784378,
      "learning_rate": 9.170055010762976e-06,
      "loss": 0.0445,
      "step": 20775
    },
    {
      "epoch": 2.911843027330063,
      "grad_norm": 0.2495385706424713,
      "learning_rate": 9.155704376943314e-06,
      "loss": 0.025,
      "step": 20776
    },
    {
      "epoch": 2.9119831814996497,
      "grad_norm": 0.15533806383609772,
      "learning_rate": 9.141353743123655e-06,
      "loss": 0.0122,
      "step": 20777
    },
    {
      "epoch": 2.912123335669236,
      "grad_norm": 0.1594160497188568,
      "learning_rate": 9.127003109303994e-06,
      "loss": 0.0545,
      "step": 20778
    },
    {
      "epoch": 2.9122634898388227,
      "grad_norm": 0.23043681681156158,
      "learning_rate": 9.112652475484334e-06,
      "loss": 0.0167,
      "step": 20779
    },
    {
      "epoch": 2.912403644008409,
      "grad_norm": 0.08006884902715683,
      "learning_rate": 9.098301841664673e-06,
      "loss": 0.0043,
      "step": 20780
    },
    {
      "epoch": 2.912543798177996,
      "grad_norm": 1.5369068384170532,
      "learning_rate": 9.083951207845013e-06,
      "loss": 0.0375,
      "step": 20781
    },
    {
      "epoch": 2.9126839523475825,
      "grad_norm": 0.21777760982513428,
      "learning_rate": 9.069600574025352e-06,
      "loss": 0.0207,
      "step": 20782
    },
    {
      "epoch": 2.912824106517169,
      "grad_norm": 0.3543589115142822,
      "learning_rate": 9.055249940205692e-06,
      "loss": 0.0362,
      "step": 20783
    },
    {
      "epoch": 2.9129642606867554,
      "grad_norm": 0.21634580194950104,
      "learning_rate": 9.040899306386031e-06,
      "loss": 0.0152,
      "step": 20784
    },
    {
      "epoch": 2.913104414856342,
      "grad_norm": 0.11264508962631226,
      "learning_rate": 9.026548672566371e-06,
      "loss": 0.0177,
      "step": 20785
    },
    {
      "epoch": 2.9132445690259283,
      "grad_norm": 0.4805937111377716,
      "learning_rate": 9.01219803874671e-06,
      "loss": 0.0404,
      "step": 20786
    },
    {
      "epoch": 2.9133847231955152,
      "grad_norm": 0.04068087786436081,
      "learning_rate": 8.99784740492705e-06,
      "loss": 0.0019,
      "step": 20787
    },
    {
      "epoch": 2.9135248773651017,
      "grad_norm": 0.3326500356197357,
      "learning_rate": 8.98349677110739e-06,
      "loss": 0.0094,
      "step": 20788
    },
    {
      "epoch": 2.913665031534688,
      "grad_norm": 0.4000054597854614,
      "learning_rate": 8.96914613728773e-06,
      "loss": 0.0356,
      "step": 20789
    },
    {
      "epoch": 2.9138051857042746,
      "grad_norm": 0.20616628229618073,
      "learning_rate": 8.954795503468068e-06,
      "loss": 0.0224,
      "step": 20790
    },
    {
      "epoch": 2.9139453398738615,
      "grad_norm": 0.127966970205307,
      "learning_rate": 8.940444869648409e-06,
      "loss": 0.0112,
      "step": 20791
    },
    {
      "epoch": 2.914085494043448,
      "grad_norm": 0.6540483236312866,
      "learning_rate": 8.926094235828748e-06,
      "loss": 0.0584,
      "step": 20792
    },
    {
      "epoch": 2.9142256482130344,
      "grad_norm": 0.7131612300872803,
      "learning_rate": 8.911743602009088e-06,
      "loss": 0.0341,
      "step": 20793
    },
    {
      "epoch": 2.914365802382621,
      "grad_norm": 0.03764643520116806,
      "learning_rate": 8.897392968189428e-06,
      "loss": 0.0029,
      "step": 20794
    },
    {
      "epoch": 2.9145059565522073,
      "grad_norm": 0.0637982189655304,
      "learning_rate": 8.883042334369767e-06,
      "loss": 0.0061,
      "step": 20795
    },
    {
      "epoch": 2.914646110721794,
      "grad_norm": 0.22289808094501495,
      "learning_rate": 8.868691700550108e-06,
      "loss": 0.0315,
      "step": 20796
    },
    {
      "epoch": 2.9147862648913803,
      "grad_norm": 0.2513267695903778,
      "learning_rate": 8.854341066730446e-06,
      "loss": 0.0389,
      "step": 20797
    },
    {
      "epoch": 2.914926419060967,
      "grad_norm": 0.18218311667442322,
      "learning_rate": 8.839990432910787e-06,
      "loss": 0.0108,
      "step": 20798
    },
    {
      "epoch": 2.9150665732305536,
      "grad_norm": 0.11901721358299255,
      "learning_rate": 8.825639799091125e-06,
      "loss": 0.01,
      "step": 20799
    },
    {
      "epoch": 2.91520672740014,
      "grad_norm": 0.43263545632362366,
      "learning_rate": 8.811289165271466e-06,
      "loss": 0.0094,
      "step": 20800
    },
    {
      "epoch": 2.9153468815697265,
      "grad_norm": 0.20341414213180542,
      "learning_rate": 8.796938531451805e-06,
      "loss": 0.0226,
      "step": 20801
    },
    {
      "epoch": 2.9154870357393134,
      "grad_norm": 0.34665700793266296,
      "learning_rate": 8.782587897632145e-06,
      "loss": 0.0462,
      "step": 20802
    },
    {
      "epoch": 2.9156271899089,
      "grad_norm": 0.36099669337272644,
      "learning_rate": 8.768237263812484e-06,
      "loss": 0.0528,
      "step": 20803
    },
    {
      "epoch": 2.9157673440784864,
      "grad_norm": 0.033484939485788345,
      "learning_rate": 8.753886629992824e-06,
      "loss": 0.0014,
      "step": 20804
    },
    {
      "epoch": 2.915907498248073,
      "grad_norm": 0.06174330785870552,
      "learning_rate": 8.739535996173163e-06,
      "loss": 0.0045,
      "step": 20805
    },
    {
      "epoch": 2.9160476524176593,
      "grad_norm": 0.44212549924850464,
      "learning_rate": 8.725185362353503e-06,
      "loss": 0.0132,
      "step": 20806
    },
    {
      "epoch": 2.9161878065872457,
      "grad_norm": 0.1343972235918045,
      "learning_rate": 8.710834728533842e-06,
      "loss": 0.0257,
      "step": 20807
    },
    {
      "epoch": 2.9163279607568326,
      "grad_norm": 0.19658316671848297,
      "learning_rate": 8.696484094714182e-06,
      "loss": 0.0124,
      "step": 20808
    },
    {
      "epoch": 2.916468114926419,
      "grad_norm": 0.43633759021759033,
      "learning_rate": 8.682133460894521e-06,
      "loss": 0.0335,
      "step": 20809
    },
    {
      "epoch": 2.9166082690960056,
      "grad_norm": 0.13331104815006256,
      "learning_rate": 8.667782827074862e-06,
      "loss": 0.0165,
      "step": 20810
    },
    {
      "epoch": 2.916748423265592,
      "grad_norm": 0.2501240074634552,
      "learning_rate": 8.653432193255202e-06,
      "loss": 0.0246,
      "step": 20811
    },
    {
      "epoch": 2.916888577435179,
      "grad_norm": 0.08909827470779419,
      "learning_rate": 8.63908155943554e-06,
      "loss": 0.0112,
      "step": 20812
    },
    {
      "epoch": 2.9170287316047654,
      "grad_norm": 0.18700209259986877,
      "learning_rate": 8.624730925615881e-06,
      "loss": 0.0457,
      "step": 20813
    },
    {
      "epoch": 2.917168885774352,
      "grad_norm": 0.3018716275691986,
      "learning_rate": 8.61038029179622e-06,
      "loss": 0.0132,
      "step": 20814
    },
    {
      "epoch": 2.9173090399439383,
      "grad_norm": 0.5658467411994934,
      "learning_rate": 8.59602965797656e-06,
      "loss": 0.0736,
      "step": 20815
    },
    {
      "epoch": 2.9174491941135248,
      "grad_norm": 0.17314061522483826,
      "learning_rate": 8.581679024156899e-06,
      "loss": 0.0037,
      "step": 20816
    },
    {
      "epoch": 2.9175893482831112,
      "grad_norm": 1.6219761371612549,
      "learning_rate": 8.56732839033724e-06,
      "loss": 0.1375,
      "step": 20817
    },
    {
      "epoch": 2.917729502452698,
      "grad_norm": 1.7230589389801025,
      "learning_rate": 8.552977756517578e-06,
      "loss": 0.2051,
      "step": 20818
    },
    {
      "epoch": 2.9178696566222846,
      "grad_norm": 4.344144344329834,
      "learning_rate": 8.538627122697919e-06,
      "loss": 0.1788,
      "step": 20819
    },
    {
      "epoch": 2.918009810791871,
      "grad_norm": 0.03655634820461273,
      "learning_rate": 8.524276488878257e-06,
      "loss": 0.0027,
      "step": 20820
    },
    {
      "epoch": 2.9181499649614575,
      "grad_norm": 0.25102561712265015,
      "learning_rate": 8.509925855058598e-06,
      "loss": 0.0456,
      "step": 20821
    },
    {
      "epoch": 2.9182901191310444,
      "grad_norm": 0.07444274425506592,
      "learning_rate": 8.495575221238936e-06,
      "loss": 0.0038,
      "step": 20822
    },
    {
      "epoch": 2.918430273300631,
      "grad_norm": 0.11049601435661316,
      "learning_rate": 8.481224587419277e-06,
      "loss": 0.062,
      "step": 20823
    },
    {
      "epoch": 2.9185704274702173,
      "grad_norm": 0.060240164399147034,
      "learning_rate": 8.466873953599616e-06,
      "loss": 0.0063,
      "step": 20824
    },
    {
      "epoch": 2.918710581639804,
      "grad_norm": 0.4039372503757477,
      "learning_rate": 8.452523319779956e-06,
      "loss": 0.0247,
      "step": 20825
    },
    {
      "epoch": 2.9188507358093903,
      "grad_norm": 0.2745022177696228,
      "learning_rate": 8.438172685960295e-06,
      "loss": 0.0472,
      "step": 20826
    },
    {
      "epoch": 2.9189908899789767,
      "grad_norm": 0.2167574018239975,
      "learning_rate": 8.423822052140635e-06,
      "loss": 0.0187,
      "step": 20827
    },
    {
      "epoch": 2.919131044148563,
      "grad_norm": 0.4564080238342285,
      "learning_rate": 8.409471418320974e-06,
      "loss": 0.035,
      "step": 20828
    },
    {
      "epoch": 2.91927119831815,
      "grad_norm": 0.04281604662537575,
      "learning_rate": 8.395120784501314e-06,
      "loss": 0.0025,
      "step": 20829
    },
    {
      "epoch": 2.9194113524877365,
      "grad_norm": 0.16506488621234894,
      "learning_rate": 8.380770150681655e-06,
      "loss": 0.0524,
      "step": 20830
    },
    {
      "epoch": 2.919551506657323,
      "grad_norm": 0.06750055402517319,
      "learning_rate": 8.366419516861995e-06,
      "loss": 0.0036,
      "step": 20831
    },
    {
      "epoch": 2.9196916608269095,
      "grad_norm": 0.16076305508613586,
      "learning_rate": 8.352068883042334e-06,
      "loss": 0.0074,
      "step": 20832
    },
    {
      "epoch": 2.9198318149964964,
      "grad_norm": 0.07818535715341568,
      "learning_rate": 8.337718249222674e-06,
      "loss": 0.0052,
      "step": 20833
    },
    {
      "epoch": 2.919971969166083,
      "grad_norm": 0.19906705617904663,
      "learning_rate": 8.323367615403013e-06,
      "loss": 0.0392,
      "step": 20834
    },
    {
      "epoch": 2.9201121233356693,
      "grad_norm": 0.15549708902835846,
      "learning_rate": 8.309016981583353e-06,
      "loss": 0.0334,
      "step": 20835
    },
    {
      "epoch": 2.9202522775052557,
      "grad_norm": 0.08082341402769089,
      "learning_rate": 8.294666347763692e-06,
      "loss": 0.0093,
      "step": 20836
    },
    {
      "epoch": 2.920392431674842,
      "grad_norm": 0.14476197957992554,
      "learning_rate": 8.280315713944032e-06,
      "loss": 0.0607,
      "step": 20837
    },
    {
      "epoch": 2.9205325858444287,
      "grad_norm": 0.31879347562789917,
      "learning_rate": 8.265965080124371e-06,
      "loss": 0.0756,
      "step": 20838
    },
    {
      "epoch": 2.9206727400140156,
      "grad_norm": 0.460928738117218,
      "learning_rate": 8.251614446304712e-06,
      "loss": 0.032,
      "step": 20839
    },
    {
      "epoch": 2.920812894183602,
      "grad_norm": 0.24567623436450958,
      "learning_rate": 8.23726381248505e-06,
      "loss": 0.0311,
      "step": 20840
    },
    {
      "epoch": 2.9209530483531885,
      "grad_norm": 0.1459200233221054,
      "learning_rate": 8.22291317866539e-06,
      "loss": 0.0203,
      "step": 20841
    },
    {
      "epoch": 2.921093202522775,
      "grad_norm": 0.12736356258392334,
      "learning_rate": 8.20856254484573e-06,
      "loss": 0.026,
      "step": 20842
    },
    {
      "epoch": 2.921233356692362,
      "grad_norm": 0.15769840776920319,
      "learning_rate": 8.19421191102607e-06,
      "loss": 0.0296,
      "step": 20843
    },
    {
      "epoch": 2.9213735108619483,
      "grad_norm": 0.10037697106599808,
      "learning_rate": 8.179861277206409e-06,
      "loss": 0.026,
      "step": 20844
    },
    {
      "epoch": 2.9215136650315348,
      "grad_norm": 0.13630695641040802,
      "learning_rate": 8.165510643386749e-06,
      "loss": 0.0161,
      "step": 20845
    },
    {
      "epoch": 2.921653819201121,
      "grad_norm": 0.2527487277984619,
      "learning_rate": 8.151160009567088e-06,
      "loss": 0.0628,
      "step": 20846
    },
    {
      "epoch": 2.9217939733707077,
      "grad_norm": 0.17540931701660156,
      "learning_rate": 8.136809375747428e-06,
      "loss": 0.021,
      "step": 20847
    },
    {
      "epoch": 2.921934127540294,
      "grad_norm": 0.08566952496767044,
      "learning_rate": 8.122458741927769e-06,
      "loss": 0.0046,
      "step": 20848
    },
    {
      "epoch": 2.9220742817098806,
      "grad_norm": 0.19337183237075806,
      "learning_rate": 8.108108108108107e-06,
      "loss": 0.0197,
      "step": 20849
    },
    {
      "epoch": 2.9222144358794675,
      "grad_norm": 0.562846839427948,
      "learning_rate": 8.093757474288448e-06,
      "loss": 0.0345,
      "step": 20850
    },
    {
      "epoch": 2.922354590049054,
      "grad_norm": 0.20644716918468475,
      "learning_rate": 8.079406840468786e-06,
      "loss": 0.0393,
      "step": 20851
    },
    {
      "epoch": 2.9224947442186404,
      "grad_norm": 0.03776136413216591,
      "learning_rate": 8.065056206649127e-06,
      "loss": 0.0022,
      "step": 20852
    },
    {
      "epoch": 2.9226348983882273,
      "grad_norm": 0.06299964338541031,
      "learning_rate": 8.050705572829466e-06,
      "loss": 0.0055,
      "step": 20853
    },
    {
      "epoch": 2.922775052557814,
      "grad_norm": 0.24422027170658112,
      "learning_rate": 8.036354939009806e-06,
      "loss": 0.0333,
      "step": 20854
    },
    {
      "epoch": 2.9229152067274002,
      "grad_norm": 0.1088394895195961,
      "learning_rate": 8.022004305190145e-06,
      "loss": 0.0103,
      "step": 20855
    },
    {
      "epoch": 2.9230553608969867,
      "grad_norm": 0.35398152470588684,
      "learning_rate": 8.007653671370485e-06,
      "loss": 0.0715,
      "step": 20856
    },
    {
      "epoch": 2.923195515066573,
      "grad_norm": 0.1836041361093521,
      "learning_rate": 7.993303037550824e-06,
      "loss": 0.0601,
      "step": 20857
    },
    {
      "epoch": 2.9233356692361596,
      "grad_norm": 0.18405325710773468,
      "learning_rate": 7.978952403731164e-06,
      "loss": 0.0216,
      "step": 20858
    },
    {
      "epoch": 2.923475823405746,
      "grad_norm": 0.12270653247833252,
      "learning_rate": 7.964601769911503e-06,
      "loss": 0.0066,
      "step": 20859
    },
    {
      "epoch": 2.923615977575333,
      "grad_norm": 0.12997110188007355,
      "learning_rate": 7.950251136091843e-06,
      "loss": 0.011,
      "step": 20860
    },
    {
      "epoch": 2.9237561317449194,
      "grad_norm": 0.3330586552619934,
      "learning_rate": 7.935900502272182e-06,
      "loss": 0.0414,
      "step": 20861
    },
    {
      "epoch": 2.923896285914506,
      "grad_norm": 0.10077826678752899,
      "learning_rate": 7.921549868452523e-06,
      "loss": 0.004,
      "step": 20862
    },
    {
      "epoch": 2.9240364400840924,
      "grad_norm": 0.4784476161003113,
      "learning_rate": 7.907199234632861e-06,
      "loss": 0.0728,
      "step": 20863
    },
    {
      "epoch": 2.9241765942536793,
      "grad_norm": 0.0951777920126915,
      "learning_rate": 7.892848600813202e-06,
      "loss": 0.0057,
      "step": 20864
    },
    {
      "epoch": 2.9243167484232657,
      "grad_norm": 0.03978104516863823,
      "learning_rate": 7.87849796699354e-06,
      "loss": 0.0017,
      "step": 20865
    },
    {
      "epoch": 2.924456902592852,
      "grad_norm": 0.21975819766521454,
      "learning_rate": 7.864147333173881e-06,
      "loss": 0.0144,
      "step": 20866
    },
    {
      "epoch": 2.9245970567624386,
      "grad_norm": 0.31612688302993774,
      "learning_rate": 7.849796699354221e-06,
      "loss": 0.032,
      "step": 20867
    },
    {
      "epoch": 2.924737210932025,
      "grad_norm": 0.6473194360733032,
      "learning_rate": 7.83544606553456e-06,
      "loss": 0.0366,
      "step": 20868
    },
    {
      "epoch": 2.9248773651016116,
      "grad_norm": 0.9579076170921326,
      "learning_rate": 7.8210954317149e-06,
      "loss": 0.0867,
      "step": 20869
    },
    {
      "epoch": 2.9250175192711985,
      "grad_norm": 1.9025181531906128,
      "learning_rate": 7.80674479789524e-06,
      "loss": 0.4027,
      "step": 20870
    },
    {
      "epoch": 2.925157673440785,
      "grad_norm": 0.14700014889240265,
      "learning_rate": 7.79239416407558e-06,
      "loss": 0.0133,
      "step": 20871
    },
    {
      "epoch": 2.9252978276103714,
      "grad_norm": 0.36744391918182373,
      "learning_rate": 7.778043530255918e-06,
      "loss": 0.0489,
      "step": 20872
    },
    {
      "epoch": 2.925437981779958,
      "grad_norm": 0.11922232806682587,
      "learning_rate": 7.763692896436259e-06,
      "loss": 0.0072,
      "step": 20873
    },
    {
      "epoch": 2.9255781359495447,
      "grad_norm": 0.22037041187286377,
      "learning_rate": 7.749342262616597e-06,
      "loss": 0.0154,
      "step": 20874
    },
    {
      "epoch": 2.925718290119131,
      "grad_norm": 0.34085753560066223,
      "learning_rate": 7.734991628796938e-06,
      "loss": 0.0306,
      "step": 20875
    },
    {
      "epoch": 2.9258584442887177,
      "grad_norm": 0.19489246606826782,
      "learning_rate": 7.720640994977277e-06,
      "loss": 0.0471,
      "step": 20876
    },
    {
      "epoch": 2.925998598458304,
      "grad_norm": 0.32996195554733276,
      "learning_rate": 7.706290361157617e-06,
      "loss": 0.0216,
      "step": 20877
    },
    {
      "epoch": 2.9261387526278906,
      "grad_norm": 0.2208753526210785,
      "learning_rate": 7.691939727337956e-06,
      "loss": 0.018,
      "step": 20878
    },
    {
      "epoch": 2.926278906797477,
      "grad_norm": 0.3136223256587982,
      "learning_rate": 7.677589093518296e-06,
      "loss": 0.0189,
      "step": 20879
    },
    {
      "epoch": 2.9264190609670635,
      "grad_norm": 0.12015900760889053,
      "learning_rate": 7.663238459698637e-06,
      "loss": 0.0114,
      "step": 20880
    },
    {
      "epoch": 2.9265592151366504,
      "grad_norm": 0.16330033540725708,
      "learning_rate": 7.648887825878975e-06,
      "loss": 0.038,
      "step": 20881
    },
    {
      "epoch": 2.926699369306237,
      "grad_norm": 0.3236497640609741,
      "learning_rate": 7.634537192059316e-06,
      "loss": 0.0714,
      "step": 20882
    },
    {
      "epoch": 2.9268395234758233,
      "grad_norm": 0.252256840467453,
      "learning_rate": 7.6201865582396545e-06,
      "loss": 0.0673,
      "step": 20883
    },
    {
      "epoch": 2.9269796776454102,
      "grad_norm": 0.28608256578445435,
      "learning_rate": 7.605835924419994e-06,
      "loss": 0.036,
      "step": 20884
    },
    {
      "epoch": 2.9271198318149967,
      "grad_norm": 0.5605271458625793,
      "learning_rate": 7.5914852906003345e-06,
      "loss": 0.0568,
      "step": 20885
    },
    {
      "epoch": 2.927259985984583,
      "grad_norm": 0.28941836953163147,
      "learning_rate": 7.577134656780674e-06,
      "loss": 0.0377,
      "step": 20886
    },
    {
      "epoch": 2.9274001401541696,
      "grad_norm": 0.09259998053312302,
      "learning_rate": 7.562784022961014e-06,
      "loss": 0.0032,
      "step": 20887
    },
    {
      "epoch": 2.927540294323756,
      "grad_norm": 0.07618865370750427,
      "learning_rate": 7.548433389141353e-06,
      "loss": 0.0194,
      "step": 20888
    },
    {
      "epoch": 2.9276804484933425,
      "grad_norm": 0.08588466048240662,
      "learning_rate": 7.534082755321693e-06,
      "loss": 0.021,
      "step": 20889
    },
    {
      "epoch": 2.927820602662929,
      "grad_norm": 0.16338613629341125,
      "learning_rate": 7.519732121502032e-06,
      "loss": 0.0282,
      "step": 20890
    },
    {
      "epoch": 2.927960756832516,
      "grad_norm": 0.111556276679039,
      "learning_rate": 7.505381487682372e-06,
      "loss": 0.0292,
      "step": 20891
    },
    {
      "epoch": 2.9281009110021023,
      "grad_norm": 0.302435040473938,
      "learning_rate": 7.4910308538627115e-06,
      "loss": 0.073,
      "step": 20892
    },
    {
      "epoch": 2.928241065171689,
      "grad_norm": 0.13511331379413605,
      "learning_rate": 7.476680220043051e-06,
      "loss": 0.0241,
      "step": 20893
    },
    {
      "epoch": 2.9283812193412753,
      "grad_norm": 0.36202773451805115,
      "learning_rate": 7.462329586223391e-06,
      "loss": 0.0216,
      "step": 20894
    },
    {
      "epoch": 2.928521373510862,
      "grad_norm": 0.08806430548429489,
      "learning_rate": 7.44797895240373e-06,
      "loss": 0.0046,
      "step": 20895
    },
    {
      "epoch": 2.9286615276804486,
      "grad_norm": 0.08994189649820328,
      "learning_rate": 7.43362831858407e-06,
      "loss": 0.03,
      "step": 20896
    },
    {
      "epoch": 2.928801681850035,
      "grad_norm": 0.1963607668876648,
      "learning_rate": 7.419277684764409e-06,
      "loss": 0.024,
      "step": 20897
    },
    {
      "epoch": 2.9289418360196215,
      "grad_norm": 0.18027660250663757,
      "learning_rate": 7.404927050944749e-06,
      "loss": 0.0222,
      "step": 20898
    },
    {
      "epoch": 2.929081990189208,
      "grad_norm": 0.11307781934738159,
      "learning_rate": 7.3905764171250885e-06,
      "loss": 0.0209,
      "step": 20899
    },
    {
      "epoch": 2.9292221443587945,
      "grad_norm": 0.04108583927154541,
      "learning_rate": 7.376225783305428e-06,
      "loss": 0.0028,
      "step": 20900
    },
    {
      "epoch": 2.9293622985283814,
      "grad_norm": 0.11263587325811386,
      "learning_rate": 7.361875149485768e-06,
      "loss": 0.0132,
      "step": 20901
    },
    {
      "epoch": 2.929502452697968,
      "grad_norm": 0.11941960453987122,
      "learning_rate": 7.347524515666107e-06,
      "loss": 0.0087,
      "step": 20902
    },
    {
      "epoch": 2.9296426068675543,
      "grad_norm": 0.19769345223903656,
      "learning_rate": 7.333173881846447e-06,
      "loss": 0.0143,
      "step": 20903
    },
    {
      "epoch": 2.9297827610371407,
      "grad_norm": 0.6844443082809448,
      "learning_rate": 7.318823248026788e-06,
      "loss": 0.0443,
      "step": 20904
    },
    {
      "epoch": 2.9299229152067277,
      "grad_norm": 0.21895532310009003,
      "learning_rate": 7.304472614207128e-06,
      "loss": 0.0176,
      "step": 20905
    },
    {
      "epoch": 2.930063069376314,
      "grad_norm": 0.09981156140565872,
      "learning_rate": 7.290121980387467e-06,
      "loss": 0.0153,
      "step": 20906
    },
    {
      "epoch": 2.9302032235459006,
      "grad_norm": 0.5449497699737549,
      "learning_rate": 7.275771346567807e-06,
      "loss": 0.03,
      "step": 20907
    },
    {
      "epoch": 2.930343377715487,
      "grad_norm": 0.26879242062568665,
      "learning_rate": 7.261420712748146e-06,
      "loss": 0.0128,
      "step": 20908
    },
    {
      "epoch": 2.9304835318850735,
      "grad_norm": 0.4231356680393219,
      "learning_rate": 7.247070078928486e-06,
      "loss": 0.0168,
      "step": 20909
    },
    {
      "epoch": 2.93062368605466,
      "grad_norm": 0.1815986931324005,
      "learning_rate": 7.2327194451088255e-06,
      "loss": 0.0261,
      "step": 20910
    },
    {
      "epoch": 2.9307638402242464,
      "grad_norm": 0.14714835584163666,
      "learning_rate": 7.218368811289165e-06,
      "loss": 0.0175,
      "step": 20911
    },
    {
      "epoch": 2.9309039943938333,
      "grad_norm": 0.17421674728393555,
      "learning_rate": 7.204018177469505e-06,
      "loss": 0.0279,
      "step": 20912
    },
    {
      "epoch": 2.9310441485634198,
      "grad_norm": 0.2407115399837494,
      "learning_rate": 7.189667543649844e-06,
      "loss": 0.045,
      "step": 20913
    },
    {
      "epoch": 2.9311843027330062,
      "grad_norm": 0.06666436046361923,
      "learning_rate": 7.175316909830184e-06,
      "loss": 0.0027,
      "step": 20914
    },
    {
      "epoch": 2.9313244569025927,
      "grad_norm": 0.5216501951217651,
      "learning_rate": 7.160966276010523e-06,
      "loss": 0.0203,
      "step": 20915
    },
    {
      "epoch": 2.9314646110721796,
      "grad_norm": 0.8742838501930237,
      "learning_rate": 7.146615642190863e-06,
      "loss": 0.201,
      "step": 20916
    },
    {
      "epoch": 2.931604765241766,
      "grad_norm": 0.307233065366745,
      "learning_rate": 7.1322650083712025e-06,
      "loss": 0.0371,
      "step": 20917
    },
    {
      "epoch": 2.9317449194113525,
      "grad_norm": 0.40731897950172424,
      "learning_rate": 7.117914374551542e-06,
      "loss": 0.0065,
      "step": 20918
    },
    {
      "epoch": 2.931885073580939,
      "grad_norm": 2.596923828125,
      "learning_rate": 7.103563740731882e-06,
      "loss": 0.1723,
      "step": 20919
    },
    {
      "epoch": 2.9320252277505254,
      "grad_norm": 1.0663001537322998,
      "learning_rate": 7.089213106912221e-06,
      "loss": 0.0273,
      "step": 20920
    },
    {
      "epoch": 2.932165381920112,
      "grad_norm": 0.32411348819732666,
      "learning_rate": 7.074862473092561e-06,
      "loss": 0.0346,
      "step": 20921
    },
    {
      "epoch": 2.932305536089699,
      "grad_norm": 0.03323211148381233,
      "learning_rate": 7.0605118392729e-06,
      "loss": 0.0026,
      "step": 20922
    },
    {
      "epoch": 2.9324456902592853,
      "grad_norm": 0.43752187490463257,
      "learning_rate": 7.046161205453241e-06,
      "loss": 0.0448,
      "step": 20923
    },
    {
      "epoch": 2.9325858444288717,
      "grad_norm": 0.0739763006567955,
      "learning_rate": 7.03181057163358e-06,
      "loss": 0.0099,
      "step": 20924
    },
    {
      "epoch": 2.932725998598458,
      "grad_norm": 0.19363577663898468,
      "learning_rate": 7.01745993781392e-06,
      "loss": 0.0458,
      "step": 20925
    },
    {
      "epoch": 2.932866152768045,
      "grad_norm": 0.17047443985939026,
      "learning_rate": 7.0031093039942594e-06,
      "loss": 0.0215,
      "step": 20926
    },
    {
      "epoch": 2.9330063069376315,
      "grad_norm": 0.2737269401550293,
      "learning_rate": 6.988758670174599e-06,
      "loss": 0.0096,
      "step": 20927
    },
    {
      "epoch": 2.933146461107218,
      "grad_norm": 0.16186510026454926,
      "learning_rate": 6.974408036354939e-06,
      "loss": 0.0608,
      "step": 20928
    },
    {
      "epoch": 2.9332866152768045,
      "grad_norm": 0.34598278999328613,
      "learning_rate": 6.960057402535278e-06,
      "loss": 0.0354,
      "step": 20929
    },
    {
      "epoch": 2.933426769446391,
      "grad_norm": 0.2241280972957611,
      "learning_rate": 6.945706768715618e-06,
      "loss": 0.0195,
      "step": 20930
    },
    {
      "epoch": 2.9335669236159774,
      "grad_norm": 0.06546413153409958,
      "learning_rate": 6.931356134895957e-06,
      "loss": 0.0055,
      "step": 20931
    },
    {
      "epoch": 2.9337070777855643,
      "grad_norm": 0.12368803471326828,
      "learning_rate": 6.917005501076297e-06,
      "loss": 0.0106,
      "step": 20932
    },
    {
      "epoch": 2.9338472319551507,
      "grad_norm": 0.2745272219181061,
      "learning_rate": 6.9026548672566364e-06,
      "loss": 0.018,
      "step": 20933
    },
    {
      "epoch": 2.933987386124737,
      "grad_norm": 0.16309945285320282,
      "learning_rate": 6.888304233436976e-06,
      "loss": 0.0103,
      "step": 20934
    },
    {
      "epoch": 2.9341275402943237,
      "grad_norm": 0.355155348777771,
      "learning_rate": 6.873953599617316e-06,
      "loss": 0.0625,
      "step": 20935
    },
    {
      "epoch": 2.9342676944639106,
      "grad_norm": 0.144060418009758,
      "learning_rate": 6.859602965797655e-06,
      "loss": 0.0081,
      "step": 20936
    },
    {
      "epoch": 2.934407848633497,
      "grad_norm": 0.10298395156860352,
      "learning_rate": 6.845252331977995e-06,
      "loss": 0.0266,
      "step": 20937
    },
    {
      "epoch": 2.9345480028030835,
      "grad_norm": 0.13036443293094635,
      "learning_rate": 6.830901698158334e-06,
      "loss": 0.0335,
      "step": 20938
    },
    {
      "epoch": 2.93468815697267,
      "grad_norm": 0.21388401091098785,
      "learning_rate": 6.816551064338674e-06,
      "loss": 0.0282,
      "step": 20939
    },
    {
      "epoch": 2.9348283111422564,
      "grad_norm": 0.12957242131233215,
      "learning_rate": 6.8022004305190134e-06,
      "loss": 0.0272,
      "step": 20940
    },
    {
      "epoch": 2.934968465311843,
      "grad_norm": 0.03360304608941078,
      "learning_rate": 6.787849796699354e-06,
      "loss": 0.0102,
      "step": 20941
    },
    {
      "epoch": 2.9351086194814293,
      "grad_norm": 0.0971352681517601,
      "learning_rate": 6.7734991628796934e-06,
      "loss": 0.0137,
      "step": 20942
    },
    {
      "epoch": 2.935248773651016,
      "grad_norm": 0.04756896197795868,
      "learning_rate": 6.759148529060033e-06,
      "loss": 0.0038,
      "step": 20943
    },
    {
      "epoch": 2.9353889278206027,
      "grad_norm": 0.4916985332965851,
      "learning_rate": 6.744797895240373e-06,
      "loss": 0.0213,
      "step": 20944
    },
    {
      "epoch": 2.935529081990189,
      "grad_norm": 0.0612313486635685,
      "learning_rate": 6.730447261420712e-06,
      "loss": 0.0055,
      "step": 20945
    },
    {
      "epoch": 2.9356692361597756,
      "grad_norm": 0.1277768462896347,
      "learning_rate": 6.716096627601052e-06,
      "loss": 0.0269,
      "step": 20946
    },
    {
      "epoch": 2.9358093903293625,
      "grad_norm": 0.313719242811203,
      "learning_rate": 6.701745993781391e-06,
      "loss": 0.0131,
      "step": 20947
    },
    {
      "epoch": 2.935949544498949,
      "grad_norm": 0.12861113250255585,
      "learning_rate": 6.687395359961731e-06,
      "loss": 0.0144,
      "step": 20948
    },
    {
      "epoch": 2.9360896986685354,
      "grad_norm": 0.14126648008823395,
      "learning_rate": 6.6730447261420704e-06,
      "loss": 0.0075,
      "step": 20949
    },
    {
      "epoch": 2.936229852838122,
      "grad_norm": 0.2856072187423706,
      "learning_rate": 6.65869409232241e-06,
      "loss": 0.0198,
      "step": 20950
    },
    {
      "epoch": 2.9363700070077083,
      "grad_norm": 0.4148334860801697,
      "learning_rate": 6.64434345850275e-06,
      "loss": 0.0272,
      "step": 20951
    },
    {
      "epoch": 2.936510161177295,
      "grad_norm": 0.20596837997436523,
      "learning_rate": 6.629992824683089e-06,
      "loss": 0.0086,
      "step": 20952
    },
    {
      "epoch": 2.9366503153468817,
      "grad_norm": 0.1942443996667862,
      "learning_rate": 6.615642190863429e-06,
      "loss": 0.0118,
      "step": 20953
    },
    {
      "epoch": 2.936790469516468,
      "grad_norm": 0.11931493878364563,
      "learning_rate": 6.601291557043768e-06,
      "loss": 0.0175,
      "step": 20954
    },
    {
      "epoch": 2.9369306236860546,
      "grad_norm": 0.1311112493276596,
      "learning_rate": 6.586940923224108e-06,
      "loss": 0.0084,
      "step": 20955
    },
    {
      "epoch": 2.937070777855641,
      "grad_norm": 0.3379358947277069,
      "learning_rate": 6.5725902894044474e-06,
      "loss": 0.0734,
      "step": 20956
    },
    {
      "epoch": 2.937210932025228,
      "grad_norm": 0.6628038287162781,
      "learning_rate": 6.558239655584787e-06,
      "loss": 0.0354,
      "step": 20957
    },
    {
      "epoch": 2.9373510861948144,
      "grad_norm": 0.18850652873516083,
      "learning_rate": 6.543889021765127e-06,
      "loss": 0.0132,
      "step": 20958
    },
    {
      "epoch": 2.937491240364401,
      "grad_norm": 0.2557535171508789,
      "learning_rate": 6.529538387945466e-06,
      "loss": 0.0361,
      "step": 20959
    },
    {
      "epoch": 2.9376313945339874,
      "grad_norm": 0.12206394970417023,
      "learning_rate": 6.515187754125807e-06,
      "loss": 0.0113,
      "step": 20960
    },
    {
      "epoch": 2.937771548703574,
      "grad_norm": 0.13461731374263763,
      "learning_rate": 6.500837120306147e-06,
      "loss": 0.0254,
      "step": 20961
    },
    {
      "epoch": 2.9379117028731603,
      "grad_norm": 0.0747813731431961,
      "learning_rate": 6.4864864864864866e-06,
      "loss": 0.004,
      "step": 20962
    },
    {
      "epoch": 2.938051857042747,
      "grad_norm": 0.2458028346300125,
      "learning_rate": 6.472135852666826e-06,
      "loss": 0.0224,
      "step": 20963
    },
    {
      "epoch": 2.9381920112123336,
      "grad_norm": 0.2038590908050537,
      "learning_rate": 6.457785218847166e-06,
      "loss": 0.0077,
      "step": 20964
    },
    {
      "epoch": 2.93833216538192,
      "grad_norm": 0.20349450409412384,
      "learning_rate": 6.443434585027505e-06,
      "loss": 0.0297,
      "step": 20965
    },
    {
      "epoch": 2.9384723195515066,
      "grad_norm": 0.4349125325679779,
      "learning_rate": 6.429083951207845e-06,
      "loss": 0.0349,
      "step": 20966
    },
    {
      "epoch": 2.9386124737210935,
      "grad_norm": 0.07646788656711578,
      "learning_rate": 6.414733317388184e-06,
      "loss": 0.0032,
      "step": 20967
    },
    {
      "epoch": 2.93875262789068,
      "grad_norm": 0.05840045213699341,
      "learning_rate": 6.400382683568524e-06,
      "loss": 0.0022,
      "step": 20968
    },
    {
      "epoch": 2.9388927820602664,
      "grad_norm": 1.8922349214553833,
      "learning_rate": 6.3860320497488636e-06,
      "loss": 0.1689,
      "step": 20969
    },
    {
      "epoch": 2.939032936229853,
      "grad_norm": 0.00875465851277113,
      "learning_rate": 6.371681415929203e-06,
      "loss": 0.0007,
      "step": 20970
    },
    {
      "epoch": 2.9391730903994393,
      "grad_norm": 0.19148333370685577,
      "learning_rate": 6.357330782109543e-06,
      "loss": 0.0096,
      "step": 20971
    },
    {
      "epoch": 2.9393132445690258,
      "grad_norm": 0.23200567066669464,
      "learning_rate": 6.342980148289882e-06,
      "loss": 0.049,
      "step": 20972
    },
    {
      "epoch": 2.939453398738612,
      "grad_norm": 0.19728942215442657,
      "learning_rate": 6.328629514470222e-06,
      "loss": 0.0168,
      "step": 20973
    },
    {
      "epoch": 2.939593552908199,
      "grad_norm": 0.21060748398303986,
      "learning_rate": 6.314278880650561e-06,
      "loss": 0.0074,
      "step": 20974
    },
    {
      "epoch": 2.9397337070777856,
      "grad_norm": 0.20466677844524384,
      "learning_rate": 6.299928246830901e-06,
      "loss": 0.0289,
      "step": 20975
    },
    {
      "epoch": 2.939873861247372,
      "grad_norm": 0.30256277322769165,
      "learning_rate": 6.2855776130112406e-06,
      "loss": 0.0273,
      "step": 20976
    },
    {
      "epoch": 2.9400140154169585,
      "grad_norm": 0.9201028943061829,
      "learning_rate": 6.27122697919158e-06,
      "loss": 0.0444,
      "step": 20977
    },
    {
      "epoch": 2.9401541695865454,
      "grad_norm": 0.2992725372314453,
      "learning_rate": 6.2568763453719206e-06,
      "loss": 0.0507,
      "step": 20978
    },
    {
      "epoch": 2.940294323756132,
      "grad_norm": 0.18159158527851105,
      "learning_rate": 6.24252571155226e-06,
      "loss": 0.0093,
      "step": 20979
    },
    {
      "epoch": 2.9404344779257183,
      "grad_norm": 0.23910143971443176,
      "learning_rate": 6.2281750777326e-06,
      "loss": 0.0218,
      "step": 20980
    },
    {
      "epoch": 2.940574632095305,
      "grad_norm": 0.36086729168891907,
      "learning_rate": 6.213824443912939e-06,
      "loss": 0.0342,
      "step": 20981
    },
    {
      "epoch": 2.9407147862648912,
      "grad_norm": 0.1406344622373581,
      "learning_rate": 6.199473810093279e-06,
      "loss": 0.0211,
      "step": 20982
    },
    {
      "epoch": 2.9408549404344777,
      "grad_norm": 0.2438364028930664,
      "learning_rate": 6.185123176273618e-06,
      "loss": 0.0077,
      "step": 20983
    },
    {
      "epoch": 2.9409950946040646,
      "grad_norm": 0.23444446921348572,
      "learning_rate": 6.170772542453958e-06,
      "loss": 0.0204,
      "step": 20984
    },
    {
      "epoch": 2.941135248773651,
      "grad_norm": 0.160509392619133,
      "learning_rate": 6.1564219086342976e-06,
      "loss": 0.018,
      "step": 20985
    },
    {
      "epoch": 2.9412754029432375,
      "grad_norm": 0.25123390555381775,
      "learning_rate": 6.142071274814637e-06,
      "loss": 0.0131,
      "step": 20986
    },
    {
      "epoch": 2.941415557112824,
      "grad_norm": 0.16712206602096558,
      "learning_rate": 6.127720640994977e-06,
      "loss": 0.0285,
      "step": 20987
    },
    {
      "epoch": 2.941555711282411,
      "grad_norm": 0.14667803049087524,
      "learning_rate": 6.113370007175316e-06,
      "loss": 0.0247,
      "step": 20988
    },
    {
      "epoch": 2.9416958654519973,
      "grad_norm": 0.12914063036441803,
      "learning_rate": 6.099019373355656e-06,
      "loss": 0.0035,
      "step": 20989
    },
    {
      "epoch": 2.941836019621584,
      "grad_norm": 0.010141808539628983,
      "learning_rate": 6.084668739535995e-06,
      "loss": 0.0004,
      "step": 20990
    },
    {
      "epoch": 2.9419761737911703,
      "grad_norm": 0.2180391550064087,
      "learning_rate": 6.070318105716335e-06,
      "loss": 0.0074,
      "step": 20991
    },
    {
      "epoch": 2.9421163279607567,
      "grad_norm": 0.38126474618911743,
      "learning_rate": 6.0559674718966746e-06,
      "loss": 0.0357,
      "step": 20992
    },
    {
      "epoch": 2.942256482130343,
      "grad_norm": 0.17474308609962463,
      "learning_rate": 6.041616838077014e-06,
      "loss": 0.035,
      "step": 20993
    },
    {
      "epoch": 2.9423966362999296,
      "grad_norm": 0.13245925307273865,
      "learning_rate": 6.027266204257354e-06,
      "loss": 0.0416,
      "step": 20994
    },
    {
      "epoch": 2.9425367904695165,
      "grad_norm": 0.6099827289581299,
      "learning_rate": 6.012915570437693e-06,
      "loss": 0.0387,
      "step": 20995
    },
    {
      "epoch": 2.942676944639103,
      "grad_norm": 0.04228270426392555,
      "learning_rate": 5.998564936618033e-06,
      "loss": 0.0031,
      "step": 20996
    },
    {
      "epoch": 2.9428170988086895,
      "grad_norm": 0.12102396786212921,
      "learning_rate": 5.984214302798373e-06,
      "loss": 0.0143,
      "step": 20997
    },
    {
      "epoch": 2.9429572529782764,
      "grad_norm": 0.33325010538101196,
      "learning_rate": 5.969863668978713e-06,
      "loss": 0.015,
      "step": 20998
    },
    {
      "epoch": 2.943097407147863,
      "grad_norm": 0.2509569823741913,
      "learning_rate": 5.955513035159052e-06,
      "loss": 0.0295,
      "step": 20999
    },
    {
      "epoch": 2.9432375613174493,
      "grad_norm": 0.2118096649646759,
      "learning_rate": 5.941162401339392e-06,
      "loss": 0.0169,
      "step": 21000
    },
    {
      "epoch": 2.9433777154870358,
      "grad_norm": 0.14205168187618256,
      "learning_rate": 5.9268117675197315e-06,
      "loss": 0.0124,
      "step": 21001
    },
    {
      "epoch": 2.943517869656622,
      "grad_norm": 0.15084637701511383,
      "learning_rate": 5.912461133700071e-06,
      "loss": 0.0179,
      "step": 21002
    },
    {
      "epoch": 2.9436580238262087,
      "grad_norm": 0.18869581818580627,
      "learning_rate": 5.898110499880411e-06,
      "loss": 0.0116,
      "step": 21003
    },
    {
      "epoch": 2.943798177995795,
      "grad_norm": 0.3209415078163147,
      "learning_rate": 5.88375986606075e-06,
      "loss": 0.0609,
      "step": 21004
    },
    {
      "epoch": 2.943938332165382,
      "grad_norm": 0.3442994952201843,
      "learning_rate": 5.86940923224109e-06,
      "loss": 0.0214,
      "step": 21005
    },
    {
      "epoch": 2.9440784863349685,
      "grad_norm": 0.12832537293434143,
      "learning_rate": 5.855058598421429e-06,
      "loss": 0.0085,
      "step": 21006
    },
    {
      "epoch": 2.944218640504555,
      "grad_norm": 0.6553977131843567,
      "learning_rate": 5.840707964601769e-06,
      "loss": 0.0234,
      "step": 21007
    },
    {
      "epoch": 2.9443587946741414,
      "grad_norm": 0.01101483590900898,
      "learning_rate": 5.8263573307821085e-06,
      "loss": 0.0007,
      "step": 21008
    },
    {
      "epoch": 2.9444989488437283,
      "grad_norm": 0.11822237819433212,
      "learning_rate": 5.812006696962448e-06,
      "loss": 0.0114,
      "step": 21009
    },
    {
      "epoch": 2.9446391030133148,
      "grad_norm": 0.10611727088689804,
      "learning_rate": 5.797656063142788e-06,
      "loss": 0.0063,
      "step": 21010
    },
    {
      "epoch": 2.9447792571829012,
      "grad_norm": 0.6857806444168091,
      "learning_rate": 5.783305429323127e-06,
      "loss": 0.0143,
      "step": 21011
    },
    {
      "epoch": 2.9449194113524877,
      "grad_norm": 0.3767915368080139,
      "learning_rate": 5.768954795503467e-06,
      "loss": 0.0351,
      "step": 21012
    },
    {
      "epoch": 2.945059565522074,
      "grad_norm": 0.21097201108932495,
      "learning_rate": 5.754604161683806e-06,
      "loss": 0.0335,
      "step": 21013
    },
    {
      "epoch": 2.9451997196916606,
      "grad_norm": 0.12750670313835144,
      "learning_rate": 5.740253527864146e-06,
      "loss": 0.0163,
      "step": 21014
    },
    {
      "epoch": 2.9453398738612475,
      "grad_norm": 0.21914798021316528,
      "learning_rate": 5.725902894044487e-06,
      "loss": 0.0111,
      "step": 21015
    },
    {
      "epoch": 2.945480028030834,
      "grad_norm": 0.19769099354743958,
      "learning_rate": 5.711552260224827e-06,
      "loss": 0.0098,
      "step": 21016
    },
    {
      "epoch": 2.9456201822004204,
      "grad_norm": 0.3958126902580261,
      "learning_rate": 5.697201626405166e-06,
      "loss": 0.0127,
      "step": 21017
    },
    {
      "epoch": 2.945760336370007,
      "grad_norm": 0.5439279675483704,
      "learning_rate": 5.682850992585506e-06,
      "loss": 0.0253,
      "step": 21018
    },
    {
      "epoch": 2.945900490539594,
      "grad_norm": 0.5001281499862671,
      "learning_rate": 5.6685003587658455e-06,
      "loss": 0.035,
      "step": 21019
    },
    {
      "epoch": 2.9460406447091803,
      "grad_norm": 0.44221031665802,
      "learning_rate": 5.654149724946185e-06,
      "loss": 0.0478,
      "step": 21020
    },
    {
      "epoch": 2.9461807988787667,
      "grad_norm": 0.08216775953769684,
      "learning_rate": 5.639799091126525e-06,
      "loss": 0.0056,
      "step": 21021
    },
    {
      "epoch": 2.946320953048353,
      "grad_norm": 0.07792092114686966,
      "learning_rate": 5.625448457306864e-06,
      "loss": 0.0026,
      "step": 21022
    },
    {
      "epoch": 2.9464611072179396,
      "grad_norm": 0.38634616136550903,
      "learning_rate": 5.611097823487204e-06,
      "loss": 0.0954,
      "step": 21023
    },
    {
      "epoch": 2.946601261387526,
      "grad_norm": 0.19332477450370789,
      "learning_rate": 5.596747189667543e-06,
      "loss": 0.0284,
      "step": 21024
    },
    {
      "epoch": 2.9467414155571126,
      "grad_norm": 0.19380851089954376,
      "learning_rate": 5.582396555847883e-06,
      "loss": 0.015,
      "step": 21025
    },
    {
      "epoch": 2.9468815697266995,
      "grad_norm": 0.20197142660617828,
      "learning_rate": 5.5680459220282225e-06,
      "loss": 0.0073,
      "step": 21026
    },
    {
      "epoch": 2.947021723896286,
      "grad_norm": 1.0442070960998535,
      "learning_rate": 5.553695288208562e-06,
      "loss": 0.1178,
      "step": 21027
    },
    {
      "epoch": 2.9471618780658724,
      "grad_norm": 0.08172272890806198,
      "learning_rate": 5.539344654388902e-06,
      "loss": 0.0118,
      "step": 21028
    },
    {
      "epoch": 2.9473020322354593,
      "grad_norm": 0.1193513423204422,
      "learning_rate": 5.524994020569241e-06,
      "loss": 0.0063,
      "step": 21029
    },
    {
      "epoch": 2.9474421864050457,
      "grad_norm": 0.12481007725000381,
      "learning_rate": 5.510643386749581e-06,
      "loss": 0.0062,
      "step": 21030
    },
    {
      "epoch": 2.947582340574632,
      "grad_norm": 0.24042418599128723,
      "learning_rate": 5.49629275292992e-06,
      "loss": 0.0182,
      "step": 21031
    },
    {
      "epoch": 2.9477224947442187,
      "grad_norm": 0.14098285138607025,
      "learning_rate": 5.48194211911026e-06,
      "loss": 0.0231,
      "step": 21032
    },
    {
      "epoch": 2.947862648913805,
      "grad_norm": 0.18717357516288757,
      "learning_rate": 5.4675914852905995e-06,
      "loss": 0.0125,
      "step": 21033
    },
    {
      "epoch": 2.9480028030833916,
      "grad_norm": 0.16105882823467255,
      "learning_rate": 5.45324085147094e-06,
      "loss": 0.0056,
      "step": 21034
    },
    {
      "epoch": 2.948142957252978,
      "grad_norm": 0.10270168632268906,
      "learning_rate": 5.4388902176512795e-06,
      "loss": 0.0148,
      "step": 21035
    },
    {
      "epoch": 2.948283111422565,
      "grad_norm": 0.08530911803245544,
      "learning_rate": 5.424539583831619e-06,
      "loss": 0.0086,
      "step": 21036
    },
    {
      "epoch": 2.9484232655921514,
      "grad_norm": 0.21199356019496918,
      "learning_rate": 5.410188950011959e-06,
      "loss": 0.0268,
      "step": 21037
    },
    {
      "epoch": 2.948563419761738,
      "grad_norm": 0.20915906131267548,
      "learning_rate": 5.395838316192298e-06,
      "loss": 0.0375,
      "step": 21038
    },
    {
      "epoch": 2.9487035739313243,
      "grad_norm": 0.167807936668396,
      "learning_rate": 5.381487682372638e-06,
      "loss": 0.0631,
      "step": 21039
    },
    {
      "epoch": 2.948843728100911,
      "grad_norm": 0.1882878690958023,
      "learning_rate": 5.367137048552977e-06,
      "loss": 0.0257,
      "step": 21040
    },
    {
      "epoch": 2.9489838822704977,
      "grad_norm": 0.262844055891037,
      "learning_rate": 5.352786414733317e-06,
      "loss": 0.0155,
      "step": 21041
    },
    {
      "epoch": 2.949124036440084,
      "grad_norm": 0.04019749164581299,
      "learning_rate": 5.3384357809136565e-06,
      "loss": 0.0028,
      "step": 21042
    },
    {
      "epoch": 2.9492641906096706,
      "grad_norm": 0.3905564844608307,
      "learning_rate": 5.324085147093996e-06,
      "loss": 0.0444,
      "step": 21043
    },
    {
      "epoch": 2.949404344779257,
      "grad_norm": 0.04536590352654457,
      "learning_rate": 5.309734513274336e-06,
      "loss": 0.0096,
      "step": 21044
    },
    {
      "epoch": 2.9495444989488435,
      "grad_norm": 0.20009024441242218,
      "learning_rate": 5.295383879454675e-06,
      "loss": 0.0506,
      "step": 21045
    },
    {
      "epoch": 2.9496846531184304,
      "grad_norm": 0.3012358844280243,
      "learning_rate": 5.281033245635015e-06,
      "loss": 0.0456,
      "step": 21046
    },
    {
      "epoch": 2.949824807288017,
      "grad_norm": 0.2205495983362198,
      "learning_rate": 5.266682611815354e-06,
      "loss": 0.0215,
      "step": 21047
    },
    {
      "epoch": 2.9499649614576033,
      "grad_norm": 0.21878769993782043,
      "learning_rate": 5.252331977995694e-06,
      "loss": 0.0315,
      "step": 21048
    },
    {
      "epoch": 2.95010511562719,
      "grad_norm": 0.1739622801542282,
      "learning_rate": 5.2379813441760335e-06,
      "loss": 0.0106,
      "step": 21049
    },
    {
      "epoch": 2.9502452697967767,
      "grad_norm": 0.08818540722131729,
      "learning_rate": 5.223630710356373e-06,
      "loss": 0.0109,
      "step": 21050
    },
    {
      "epoch": 2.950385423966363,
      "grad_norm": 0.08541856706142426,
      "learning_rate": 5.209280076536713e-06,
      "loss": 0.014,
      "step": 21051
    },
    {
      "epoch": 2.9505255781359496,
      "grad_norm": 0.20288494229316711,
      "learning_rate": 5.194929442717053e-06,
      "loss": 0.0134,
      "step": 21052
    },
    {
      "epoch": 2.950665732305536,
      "grad_norm": 0.3345883786678314,
      "learning_rate": 5.180578808897393e-06,
      "loss": 0.0544,
      "step": 21053
    },
    {
      "epoch": 2.9508058864751225,
      "grad_norm": 0.18844570219516754,
      "learning_rate": 5.166228175077732e-06,
      "loss": 0.0161,
      "step": 21054
    },
    {
      "epoch": 2.950946040644709,
      "grad_norm": 0.06456686556339264,
      "learning_rate": 5.151877541258072e-06,
      "loss": 0.0041,
      "step": 21055
    },
    {
      "epoch": 2.9510861948142955,
      "grad_norm": 0.14287172257900238,
      "learning_rate": 5.137526907438411e-06,
      "loss": 0.0277,
      "step": 21056
    },
    {
      "epoch": 2.9512263489838824,
      "grad_norm": 0.29339340329170227,
      "learning_rate": 5.123176273618751e-06,
      "loss": 0.009,
      "step": 21057
    },
    {
      "epoch": 2.951366503153469,
      "grad_norm": 0.036791812628507614,
      "learning_rate": 5.1088256397990905e-06,
      "loss": 0.0022,
      "step": 21058
    },
    {
      "epoch": 2.9515066573230553,
      "grad_norm": 0.09941355139017105,
      "learning_rate": 5.09447500597943e-06,
      "loss": 0.0189,
      "step": 21059
    },
    {
      "epoch": 2.9516468114926417,
      "grad_norm": 0.193936288356781,
      "learning_rate": 5.08012437215977e-06,
      "loss": 0.0222,
      "step": 21060
    },
    {
      "epoch": 2.9517869656622286,
      "grad_norm": 0.24540333449840546,
      "learning_rate": 5.065773738340109e-06,
      "loss": 0.0199,
      "step": 21061
    },
    {
      "epoch": 2.951927119831815,
      "grad_norm": 0.18111878633499146,
      "learning_rate": 5.051423104520449e-06,
      "loss": 0.021,
      "step": 21062
    },
    {
      "epoch": 2.9520672740014016,
      "grad_norm": 0.20715396106243134,
      "learning_rate": 5.037072470700788e-06,
      "loss": 0.0663,
      "step": 21063
    },
    {
      "epoch": 2.952207428170988,
      "grad_norm": 0.1730603128671646,
      "learning_rate": 5.022721836881128e-06,
      "loss": 0.0097,
      "step": 21064
    },
    {
      "epoch": 2.9523475823405745,
      "grad_norm": 0.5647817254066467,
      "learning_rate": 5.0083712030614675e-06,
      "loss": 0.0167,
      "step": 21065
    },
    {
      "epoch": 2.952487736510161,
      "grad_norm": 0.45230671763420105,
      "learning_rate": 4.994020569241807e-06,
      "loss": 0.0523,
      "step": 21066
    },
    {
      "epoch": 2.952627890679748,
      "grad_norm": 1.4154618978500366,
      "learning_rate": 4.979669935422147e-06,
      "loss": 0.049,
      "step": 21067
    },
    {
      "epoch": 2.9527680448493343,
      "grad_norm": 0.04920990765094757,
      "learning_rate": 4.965319301602486e-06,
      "loss": 0.0029,
      "step": 21068
    },
    {
      "epoch": 2.9529081990189208,
      "grad_norm": 0.29503774642944336,
      "learning_rate": 4.950968667782826e-06,
      "loss": 0.0183,
      "step": 21069
    },
    {
      "epoch": 2.9530483531885072,
      "grad_norm": 0.24438603222370148,
      "learning_rate": 4.936618033963165e-06,
      "loss": 0.0072,
      "step": 21070
    },
    {
      "epoch": 2.953188507358094,
      "grad_norm": 0.3728465735912323,
      "learning_rate": 4.922267400143507e-06,
      "loss": 0.037,
      "step": 21071
    },
    {
      "epoch": 2.9533286615276806,
      "grad_norm": 0.057553790509700775,
      "learning_rate": 4.907916766323846e-06,
      "loss": 0.0034,
      "step": 21072
    },
    {
      "epoch": 2.953468815697267,
      "grad_norm": 0.11790420114994049,
      "learning_rate": 4.893566132504186e-06,
      "loss": 0.0171,
      "step": 21073
    },
    {
      "epoch": 2.9536089698668535,
      "grad_norm": 0.09027604758739471,
      "learning_rate": 4.879215498684525e-06,
      "loss": 0.0051,
      "step": 21074
    },
    {
      "epoch": 2.95374912403644,
      "grad_norm": 0.19138166308403015,
      "learning_rate": 4.864864864864865e-06,
      "loss": 0.0184,
      "step": 21075
    },
    {
      "epoch": 2.9538892782060264,
      "grad_norm": 0.2814255654811859,
      "learning_rate": 4.8505142310452045e-06,
      "loss": 0.0476,
      "step": 21076
    },
    {
      "epoch": 2.9540294323756133,
      "grad_norm": 0.08900079876184464,
      "learning_rate": 4.836163597225544e-06,
      "loss": 0.0165,
      "step": 21077
    },
    {
      "epoch": 2.9541695865452,
      "grad_norm": 0.12783344089984894,
      "learning_rate": 4.821812963405884e-06,
      "loss": 0.011,
      "step": 21078
    },
    {
      "epoch": 2.9543097407147862,
      "grad_norm": 0.07159767299890518,
      "learning_rate": 4.807462329586223e-06,
      "loss": 0.0079,
      "step": 21079
    },
    {
      "epoch": 2.9544498948843727,
      "grad_norm": 0.09055927395820618,
      "learning_rate": 4.793111695766563e-06,
      "loss": 0.0042,
      "step": 21080
    },
    {
      "epoch": 2.9545900490539596,
      "grad_norm": 0.21537242829799652,
      "learning_rate": 4.778761061946902e-06,
      "loss": 0.0115,
      "step": 21081
    },
    {
      "epoch": 2.954730203223546,
      "grad_norm": 0.10648280382156372,
      "learning_rate": 4.764410428127242e-06,
      "loss": 0.0293,
      "step": 21082
    },
    {
      "epoch": 2.9548703573931325,
      "grad_norm": 0.149746835231781,
      "learning_rate": 4.7500597943075815e-06,
      "loss": 0.0133,
      "step": 21083
    },
    {
      "epoch": 2.955010511562719,
      "grad_norm": 0.1368332803249359,
      "learning_rate": 4.735709160487921e-06,
      "loss": 0.011,
      "step": 21084
    },
    {
      "epoch": 2.9551506657323054,
      "grad_norm": 1.235548973083496,
      "learning_rate": 4.721358526668261e-06,
      "loss": 0.101,
      "step": 21085
    },
    {
      "epoch": 2.955290819901892,
      "grad_norm": 0.23924636840820312,
      "learning_rate": 4.7070078928486e-06,
      "loss": 0.0611,
      "step": 21086
    },
    {
      "epoch": 2.9554309740714784,
      "grad_norm": 0.07855143398046494,
      "learning_rate": 4.69265725902894e-06,
      "loss": 0.0086,
      "step": 21087
    },
    {
      "epoch": 2.9555711282410653,
      "grad_norm": 0.23724548518657684,
      "learning_rate": 4.678306625209279e-06,
      "loss": 0.0327,
      "step": 21088
    },
    {
      "epoch": 2.9557112824106517,
      "grad_norm": 0.16771820187568665,
      "learning_rate": 4.663955991389619e-06,
      "loss": 0.0384,
      "step": 21089
    },
    {
      "epoch": 2.955851436580238,
      "grad_norm": 0.2794395685195923,
      "learning_rate": 4.6496053575699585e-06,
      "loss": 0.0205,
      "step": 21090
    },
    {
      "epoch": 2.9559915907498246,
      "grad_norm": 0.09262455254793167,
      "learning_rate": 4.635254723750298e-06,
      "loss": 0.0026,
      "step": 21091
    },
    {
      "epoch": 2.9561317449194116,
      "grad_norm": 0.2059822827577591,
      "learning_rate": 4.620904089930638e-06,
      "loss": 0.0166,
      "step": 21092
    },
    {
      "epoch": 2.956271899088998,
      "grad_norm": 0.19926834106445312,
      "learning_rate": 4.606553456110977e-06,
      "loss": 0.0222,
      "step": 21093
    },
    {
      "epoch": 2.9564120532585845,
      "grad_norm": 0.4881141483783722,
      "learning_rate": 4.592202822291318e-06,
      "loss": 0.0277,
      "step": 21094
    },
    {
      "epoch": 2.956552207428171,
      "grad_norm": 0.03848414495587349,
      "learning_rate": 4.577852188471657e-06,
      "loss": 0.003,
      "step": 21095
    },
    {
      "epoch": 2.9566923615977574,
      "grad_norm": 0.21560364961624146,
      "learning_rate": 4.563501554651997e-06,
      "loss": 0.0371,
      "step": 21096
    },
    {
      "epoch": 2.956832515767344,
      "grad_norm": 0.4213297963142395,
      "learning_rate": 4.549150920832336e-06,
      "loss": 0.0545,
      "step": 21097
    },
    {
      "epoch": 2.9569726699369308,
      "grad_norm": 0.06178172677755356,
      "learning_rate": 4.534800287012676e-06,
      "loss": 0.0107,
      "step": 21098
    },
    {
      "epoch": 2.957112824106517,
      "grad_norm": 0.09656988084316254,
      "learning_rate": 4.5204496531930155e-06,
      "loss": 0.0055,
      "step": 21099
    },
    {
      "epoch": 2.9572529782761037,
      "grad_norm": 0.23314829170703888,
      "learning_rate": 4.506099019373355e-06,
      "loss": 0.0149,
      "step": 21100
    },
    {
      "epoch": 2.95739313244569,
      "grad_norm": 0.16913571953773499,
      "learning_rate": 4.491748385553695e-06,
      "loss": 0.0411,
      "step": 21101
    },
    {
      "epoch": 2.957533286615277,
      "grad_norm": 0.08227922022342682,
      "learning_rate": 4.477397751734034e-06,
      "loss": 0.0119,
      "step": 21102
    },
    {
      "epoch": 2.9576734407848635,
      "grad_norm": 0.14493155479431152,
      "learning_rate": 4.463047117914374e-06,
      "loss": 0.0038,
      "step": 21103
    },
    {
      "epoch": 2.95781359495445,
      "grad_norm": 0.18913684785366058,
      "learning_rate": 4.448696484094714e-06,
      "loss": 0.0423,
      "step": 21104
    },
    {
      "epoch": 2.9579537491240364,
      "grad_norm": 0.2183585911989212,
      "learning_rate": 4.434345850275054e-06,
      "loss": 0.0207,
      "step": 21105
    },
    {
      "epoch": 2.958093903293623,
      "grad_norm": 0.24108831584453583,
      "learning_rate": 4.419995216455393e-06,
      "loss": 0.0143,
      "step": 21106
    },
    {
      "epoch": 2.9582340574632093,
      "grad_norm": 0.5341862440109253,
      "learning_rate": 4.405644582635733e-06,
      "loss": 0.0259,
      "step": 21107
    },
    {
      "epoch": 2.9583742116327962,
      "grad_norm": 0.10886256396770477,
      "learning_rate": 4.3912939488160725e-06,
      "loss": 0.0101,
      "step": 21108
    },
    {
      "epoch": 2.9585143658023827,
      "grad_norm": 0.09370721131563187,
      "learning_rate": 4.376943314996412e-06,
      "loss": 0.0157,
      "step": 21109
    },
    {
      "epoch": 2.958654519971969,
      "grad_norm": 0.5097808241844177,
      "learning_rate": 4.362592681176752e-06,
      "loss": 0.0176,
      "step": 21110
    },
    {
      "epoch": 2.9587946741415556,
      "grad_norm": 0.18530426919460297,
      "learning_rate": 4.348242047357091e-06,
      "loss": 0.0092,
      "step": 21111
    },
    {
      "epoch": 2.9589348283111425,
      "grad_norm": 0.23256361484527588,
      "learning_rate": 4.333891413537431e-06,
      "loss": 0.022,
      "step": 21112
    },
    {
      "epoch": 2.959074982480729,
      "grad_norm": 0.223111093044281,
      "learning_rate": 4.31954077971777e-06,
      "loss": 0.0198,
      "step": 21113
    },
    {
      "epoch": 2.9592151366503154,
      "grad_norm": 0.11564124375581741,
      "learning_rate": 4.30519014589811e-06,
      "loss": 0.0093,
      "step": 21114
    },
    {
      "epoch": 2.959355290819902,
      "grad_norm": 0.34722933173179626,
      "learning_rate": 4.2908395120784495e-06,
      "loss": 0.0111,
      "step": 21115
    },
    {
      "epoch": 2.9594954449894884,
      "grad_norm": 0.17747914791107178,
      "learning_rate": 4.276488878258789e-06,
      "loss": 0.0322,
      "step": 21116
    },
    {
      "epoch": 2.959635599159075,
      "grad_norm": 0.0960494801402092,
      "learning_rate": 4.262138244439129e-06,
      "loss": 0.0105,
      "step": 21117
    },
    {
      "epoch": 2.9597757533286613,
      "grad_norm": 0.870903730392456,
      "learning_rate": 4.247787610619468e-06,
      "loss": 0.1474,
      "step": 21118
    },
    {
      "epoch": 2.959915907498248,
      "grad_norm": 2.2370543479919434,
      "learning_rate": 4.233436976799808e-06,
      "loss": 0.4678,
      "step": 21119
    },
    {
      "epoch": 2.9600560616678346,
      "grad_norm": 0.023159483447670937,
      "learning_rate": 4.219086342980147e-06,
      "loss": 0.0016,
      "step": 21120
    },
    {
      "epoch": 2.960196215837421,
      "grad_norm": 0.12340908497571945,
      "learning_rate": 4.204735709160487e-06,
      "loss": 0.0101,
      "step": 21121
    },
    {
      "epoch": 2.9603363700070076,
      "grad_norm": 0.463870644569397,
      "learning_rate": 4.190385075340827e-06,
      "loss": 0.0266,
      "step": 21122
    },
    {
      "epoch": 2.9604765241765945,
      "grad_norm": 0.34553825855255127,
      "learning_rate": 4.176034441521167e-06,
      "loss": 0.0283,
      "step": 21123
    },
    {
      "epoch": 2.960616678346181,
      "grad_norm": 0.10228119790554047,
      "learning_rate": 4.1616838077015065e-06,
      "loss": 0.0027,
      "step": 21124
    },
    {
      "epoch": 2.9607568325157674,
      "grad_norm": 0.07837118953466415,
      "learning_rate": 4.147333173881846e-06,
      "loss": 0.0062,
      "step": 21125
    },
    {
      "epoch": 2.960896986685354,
      "grad_norm": 0.3279400169849396,
      "learning_rate": 4.132982540062186e-06,
      "loss": 0.0451,
      "step": 21126
    },
    {
      "epoch": 2.9610371408549403,
      "grad_norm": 0.25230130553245544,
      "learning_rate": 4.118631906242525e-06,
      "loss": 0.0199,
      "step": 21127
    },
    {
      "epoch": 2.9611772950245268,
      "grad_norm": 0.10527370870113373,
      "learning_rate": 4.104281272422865e-06,
      "loss": 0.0244,
      "step": 21128
    },
    {
      "epoch": 2.9613174491941137,
      "grad_norm": 0.2723374664783478,
      "learning_rate": 4.089930638603204e-06,
      "loss": 0.0619,
      "step": 21129
    },
    {
      "epoch": 2.9614576033637,
      "grad_norm": 0.15058882534503937,
      "learning_rate": 4.075580004783544e-06,
      "loss": 0.0183,
      "step": 21130
    },
    {
      "epoch": 2.9615977575332866,
      "grad_norm": 0.15751156210899353,
      "learning_rate": 4.061229370963884e-06,
      "loss": 0.0055,
      "step": 21131
    },
    {
      "epoch": 2.961737911702873,
      "grad_norm": 0.12466356158256531,
      "learning_rate": 4.046878737144224e-06,
      "loss": 0.0231,
      "step": 21132
    },
    {
      "epoch": 2.96187806587246,
      "grad_norm": 0.15521079301834106,
      "learning_rate": 4.0325281033245635e-06,
      "loss": 0.0354,
      "step": 21133
    },
    {
      "epoch": 2.9620182200420464,
      "grad_norm": 0.05635342374444008,
      "learning_rate": 4.018177469504903e-06,
      "loss": 0.0033,
      "step": 21134
    },
    {
      "epoch": 2.962158374211633,
      "grad_norm": 0.2846575677394867,
      "learning_rate": 4.003826835685243e-06,
      "loss": 0.0201,
      "step": 21135
    },
    {
      "epoch": 2.9622985283812193,
      "grad_norm": 0.14820873737335205,
      "learning_rate": 3.989476201865582e-06,
      "loss": 0.0329,
      "step": 21136
    },
    {
      "epoch": 2.962438682550806,
      "grad_norm": 0.01651100255548954,
      "learning_rate": 3.975125568045922e-06,
      "loss": 0.0014,
      "step": 21137
    },
    {
      "epoch": 2.9625788367203922,
      "grad_norm": 0.08997229486703873,
      "learning_rate": 3.960774934226261e-06,
      "loss": 0.0097,
      "step": 21138
    },
    {
      "epoch": 2.962718990889979,
      "grad_norm": 0.0956442579627037,
      "learning_rate": 3.946424300406601e-06,
      "loss": 0.012,
      "step": 21139
    },
    {
      "epoch": 2.9628591450595656,
      "grad_norm": 0.3473750650882721,
      "learning_rate": 3.9320736665869405e-06,
      "loss": 0.0194,
      "step": 21140
    },
    {
      "epoch": 2.962999299229152,
      "grad_norm": 0.347017765045166,
      "learning_rate": 3.91772303276728e-06,
      "loss": 0.0183,
      "step": 21141
    },
    {
      "epoch": 2.9631394533987385,
      "grad_norm": 0.21526440978050232,
      "learning_rate": 3.90337239894762e-06,
      "loss": 0.019,
      "step": 21142
    },
    {
      "epoch": 2.9632796075683254,
      "grad_norm": 0.08299297094345093,
      "learning_rate": 3.889021765127959e-06,
      "loss": 0.0106,
      "step": 21143
    },
    {
      "epoch": 2.963419761737912,
      "grad_norm": 0.35823145508766174,
      "learning_rate": 3.874671131308299e-06,
      "loss": 0.09,
      "step": 21144
    },
    {
      "epoch": 2.9635599159074983,
      "grad_norm": 0.3212105333805084,
      "learning_rate": 3.860320497488638e-06,
      "loss": 0.0308,
      "step": 21145
    },
    {
      "epoch": 2.963700070077085,
      "grad_norm": 0.13644059002399445,
      "learning_rate": 3.845969863668978e-06,
      "loss": 0.0044,
      "step": 21146
    },
    {
      "epoch": 2.9638402242466713,
      "grad_norm": 0.09371822327375412,
      "learning_rate": 3.831619229849318e-06,
      "loss": 0.0174,
      "step": 21147
    },
    {
      "epoch": 2.9639803784162577,
      "grad_norm": 0.1725771129131317,
      "learning_rate": 3.817268596029658e-06,
      "loss": 0.0202,
      "step": 21148
    },
    {
      "epoch": 2.964120532585844,
      "grad_norm": 0.5339385867118835,
      "learning_rate": 3.802917962209997e-06,
      "loss": 0.0257,
      "step": 21149
    },
    {
      "epoch": 2.964260686755431,
      "grad_norm": 0.10800144821405411,
      "learning_rate": 3.788567328390337e-06,
      "loss": 0.0304,
      "step": 21150
    },
    {
      "epoch": 2.9644008409250175,
      "grad_norm": 0.054744333028793335,
      "learning_rate": 3.7742166945706766e-06,
      "loss": 0.0031,
      "step": 21151
    },
    {
      "epoch": 2.964540995094604,
      "grad_norm": 0.0457860492169857,
      "learning_rate": 3.759866060751016e-06,
      "loss": 0.0048,
      "step": 21152
    },
    {
      "epoch": 2.9646811492641905,
      "grad_norm": 0.0574570968747139,
      "learning_rate": 3.7455154269313557e-06,
      "loss": 0.0041,
      "step": 21153
    },
    {
      "epoch": 2.9648213034337774,
      "grad_norm": 0.14818532764911652,
      "learning_rate": 3.7311647931116953e-06,
      "loss": 0.0069,
      "step": 21154
    },
    {
      "epoch": 2.964961457603364,
      "grad_norm": 0.08829306066036224,
      "learning_rate": 3.716814159292035e-06,
      "loss": 0.0142,
      "step": 21155
    },
    {
      "epoch": 2.9651016117729503,
      "grad_norm": 0.07578791677951813,
      "learning_rate": 3.7024635254723744e-06,
      "loss": 0.006,
      "step": 21156
    },
    {
      "epoch": 2.9652417659425367,
      "grad_norm": 0.04134010896086693,
      "learning_rate": 3.688112891652714e-06,
      "loss": 0.0022,
      "step": 21157
    },
    {
      "epoch": 2.965381920112123,
      "grad_norm": 0.08086366951465607,
      "learning_rate": 3.6737622578330536e-06,
      "loss": 0.0142,
      "step": 21158
    },
    {
      "epoch": 2.9655220742817097,
      "grad_norm": 0.05936136841773987,
      "learning_rate": 3.659411624013394e-06,
      "loss": 0.007,
      "step": 21159
    },
    {
      "epoch": 2.9656622284512966,
      "grad_norm": 0.2730533182621002,
      "learning_rate": 3.6450609901937336e-06,
      "loss": 0.0184,
      "step": 21160
    },
    {
      "epoch": 2.965802382620883,
      "grad_norm": 0.2608994245529175,
      "learning_rate": 3.630710356374073e-06,
      "loss": 0.0182,
      "step": 21161
    },
    {
      "epoch": 2.9659425367904695,
      "grad_norm": 0.16505511105060577,
      "learning_rate": 3.6163597225544127e-06,
      "loss": 0.005,
      "step": 21162
    },
    {
      "epoch": 2.966082690960056,
      "grad_norm": 0.00855859275907278,
      "learning_rate": 3.6020090887347523e-06,
      "loss": 0.0007,
      "step": 21163
    },
    {
      "epoch": 2.966222845129643,
      "grad_norm": 0.5090882778167725,
      "learning_rate": 3.587658454915092e-06,
      "loss": 0.0544,
      "step": 21164
    },
    {
      "epoch": 2.9663629992992293,
      "grad_norm": 0.7754547595977783,
      "learning_rate": 3.5733078210954314e-06,
      "loss": 0.0114,
      "step": 21165
    },
    {
      "epoch": 2.9665031534688158,
      "grad_norm": 0.2170303910970688,
      "learning_rate": 3.558957187275771e-06,
      "loss": 0.0614,
      "step": 21166
    },
    {
      "epoch": 2.9666433076384022,
      "grad_norm": 0.669796347618103,
      "learning_rate": 3.5446065534561106e-06,
      "loss": 0.0094,
      "step": 21167
    },
    {
      "epoch": 2.9667834618079887,
      "grad_norm": 0.6994901299476624,
      "learning_rate": 3.53025591963645e-06,
      "loss": 0.0144,
      "step": 21168
    },
    {
      "epoch": 2.966923615977575,
      "grad_norm": 0.37669000029563904,
      "learning_rate": 3.51590528581679e-06,
      "loss": 0.0921,
      "step": 21169
    },
    {
      "epoch": 2.9670637701471616,
      "grad_norm": 0.6713768243789673,
      "learning_rate": 3.5015546519971297e-06,
      "loss": 0.1055,
      "step": 21170
    },
    {
      "epoch": 2.9672039243167485,
      "grad_norm": 0.20481841266155243,
      "learning_rate": 3.4872040181774693e-06,
      "loss": 0.0346,
      "step": 21171
    },
    {
      "epoch": 2.967344078486335,
      "grad_norm": 0.20191913843154907,
      "learning_rate": 3.472853384357809e-06,
      "loss": 0.0474,
      "step": 21172
    },
    {
      "epoch": 2.9674842326559214,
      "grad_norm": 0.12157867103815079,
      "learning_rate": 3.4585027505381484e-06,
      "loss": 0.0121,
      "step": 21173
    },
    {
      "epoch": 2.9676243868255083,
      "grad_norm": 0.01615927927196026,
      "learning_rate": 3.444152116718488e-06,
      "loss": 0.0011,
      "step": 21174
    },
    {
      "epoch": 2.967764540995095,
      "grad_norm": 0.12683695554733276,
      "learning_rate": 3.4298014828988276e-06,
      "loss": 0.0139,
      "step": 21175
    },
    {
      "epoch": 2.9679046951646813,
      "grad_norm": 0.2118932157754898,
      "learning_rate": 3.415450849079167e-06,
      "loss": 0.0132,
      "step": 21176
    },
    {
      "epoch": 2.9680448493342677,
      "grad_norm": 0.17958365380764008,
      "learning_rate": 3.4011002152595067e-06,
      "loss": 0.0442,
      "step": 21177
    },
    {
      "epoch": 2.968185003503854,
      "grad_norm": 0.20443286001682281,
      "learning_rate": 3.3867495814398467e-06,
      "loss": 0.0278,
      "step": 21178
    },
    {
      "epoch": 2.9683251576734406,
      "grad_norm": 0.1908264011144638,
      "learning_rate": 3.3723989476201863e-06,
      "loss": 0.0257,
      "step": 21179
    },
    {
      "epoch": 2.968465311843027,
      "grad_norm": 0.09006085246801376,
      "learning_rate": 3.358048313800526e-06,
      "loss": 0.0269,
      "step": 21180
    },
    {
      "epoch": 2.968605466012614,
      "grad_norm": 0.3181600570678711,
      "learning_rate": 3.3436976799808654e-06,
      "loss": 0.0253,
      "step": 21181
    },
    {
      "epoch": 2.9687456201822005,
      "grad_norm": 0.16416595876216888,
      "learning_rate": 3.329347046161205e-06,
      "loss": 0.0102,
      "step": 21182
    },
    {
      "epoch": 2.968885774351787,
      "grad_norm": 0.10876847058534622,
      "learning_rate": 3.3149964123415446e-06,
      "loss": 0.0197,
      "step": 21183
    },
    {
      "epoch": 2.9690259285213734,
      "grad_norm": 0.17660491168498993,
      "learning_rate": 3.300645778521884e-06,
      "loss": 0.0163,
      "step": 21184
    },
    {
      "epoch": 2.9691660826909603,
      "grad_norm": 0.08246561139822006,
      "learning_rate": 3.2862951447022237e-06,
      "loss": 0.0086,
      "step": 21185
    },
    {
      "epoch": 2.9693062368605467,
      "grad_norm": 0.18474839627742767,
      "learning_rate": 3.2719445108825633e-06,
      "loss": 0.0479,
      "step": 21186
    },
    {
      "epoch": 2.969446391030133,
      "grad_norm": 0.31320396065711975,
      "learning_rate": 3.2575938770629037e-06,
      "loss": 0.0675,
      "step": 21187
    },
    {
      "epoch": 2.9695865451997197,
      "grad_norm": 0.11197633296251297,
      "learning_rate": 3.2432432432432433e-06,
      "loss": 0.0124,
      "step": 21188
    },
    {
      "epoch": 2.969726699369306,
      "grad_norm": 0.05305375158786774,
      "learning_rate": 3.228892609423583e-06,
      "loss": 0.0065,
      "step": 21189
    },
    {
      "epoch": 2.9698668535388926,
      "grad_norm": 0.31757596135139465,
      "learning_rate": 3.2145419756039224e-06,
      "loss": 0.0538,
      "step": 21190
    },
    {
      "epoch": 2.9700070077084795,
      "grad_norm": 0.1560099720954895,
      "learning_rate": 3.200191341784262e-06,
      "loss": 0.0303,
      "step": 21191
    },
    {
      "epoch": 2.970147161878066,
      "grad_norm": 0.2689273953437805,
      "learning_rate": 3.1858407079646016e-06,
      "loss": 0.0148,
      "step": 21192
    },
    {
      "epoch": 2.9702873160476524,
      "grad_norm": 0.04956325888633728,
      "learning_rate": 3.171490074144941e-06,
      "loss": 0.0029,
      "step": 21193
    },
    {
      "epoch": 2.970427470217239,
      "grad_norm": 0.1273495852947235,
      "learning_rate": 3.1571394403252807e-06,
      "loss": 0.0151,
      "step": 21194
    },
    {
      "epoch": 2.9705676243868258,
      "grad_norm": 0.2906442880630493,
      "learning_rate": 3.1427888065056203e-06,
      "loss": 0.0348,
      "step": 21195
    },
    {
      "epoch": 2.970707778556412,
      "grad_norm": 0.11216489970684052,
      "learning_rate": 3.1284381726859603e-06,
      "loss": 0.0067,
      "step": 21196
    },
    {
      "epoch": 2.9708479327259987,
      "grad_norm": 0.05651843920350075,
      "learning_rate": 3.1140875388663e-06,
      "loss": 0.0057,
      "step": 21197
    },
    {
      "epoch": 2.970988086895585,
      "grad_norm": 0.26865845918655396,
      "learning_rate": 3.0997369050466394e-06,
      "loss": 0.0145,
      "step": 21198
    },
    {
      "epoch": 2.9711282410651716,
      "grad_norm": 0.10654281079769135,
      "learning_rate": 3.085386271226979e-06,
      "loss": 0.0064,
      "step": 21199
    },
    {
      "epoch": 2.971268395234758,
      "grad_norm": 0.18553903698921204,
      "learning_rate": 3.0710356374073186e-06,
      "loss": 0.0255,
      "step": 21200
    },
    {
      "epoch": 2.9714085494043445,
      "grad_norm": 0.2749457061290741,
      "learning_rate": 3.056685003587658e-06,
      "loss": 0.0313,
      "step": 21201
    },
    {
      "epoch": 2.9715487035739314,
      "grad_norm": 0.2068312168121338,
      "learning_rate": 3.0423343697679977e-06,
      "loss": 0.0119,
      "step": 21202
    },
    {
      "epoch": 2.971688857743518,
      "grad_norm": 0.42408275604248047,
      "learning_rate": 3.0279837359483373e-06,
      "loss": 0.0464,
      "step": 21203
    },
    {
      "epoch": 2.9718290119131043,
      "grad_norm": 0.2625313997268677,
      "learning_rate": 3.013633102128677e-06,
      "loss": 0.049,
      "step": 21204
    },
    {
      "epoch": 2.971969166082691,
      "grad_norm": 0.15315940976142883,
      "learning_rate": 2.9992824683090164e-06,
      "loss": 0.0158,
      "step": 21205
    },
    {
      "epoch": 2.9721093202522777,
      "grad_norm": 0.23790588974952698,
      "learning_rate": 2.9849318344893564e-06,
      "loss": 0.0157,
      "step": 21206
    },
    {
      "epoch": 2.972249474421864,
      "grad_norm": 0.18927103281021118,
      "learning_rate": 2.970581200669696e-06,
      "loss": 0.0529,
      "step": 21207
    },
    {
      "epoch": 2.9723896285914506,
      "grad_norm": 0.16035114228725433,
      "learning_rate": 2.9562305668500356e-06,
      "loss": 0.0169,
      "step": 21208
    },
    {
      "epoch": 2.972529782761037,
      "grad_norm": 0.11114153265953064,
      "learning_rate": 2.941879933030375e-06,
      "loss": 0.0174,
      "step": 21209
    },
    {
      "epoch": 2.9726699369306235,
      "grad_norm": 0.3049420118331909,
      "learning_rate": 2.9275292992107147e-06,
      "loss": 0.0421,
      "step": 21210
    },
    {
      "epoch": 2.97281009110021,
      "grad_norm": 0.006898100487887859,
      "learning_rate": 2.9131786653910543e-06,
      "loss": 0.0007,
      "step": 21211
    },
    {
      "epoch": 2.972950245269797,
      "grad_norm": 0.4988664984703064,
      "learning_rate": 2.898828031571394e-06,
      "loss": 0.079,
      "step": 21212
    },
    {
      "epoch": 2.9730903994393834,
      "grad_norm": 0.5038439035415649,
      "learning_rate": 2.8844773977517334e-06,
      "loss": 0.0535,
      "step": 21213
    },
    {
      "epoch": 2.97323055360897,
      "grad_norm": 0.31087058782577515,
      "learning_rate": 2.870126763932073e-06,
      "loss": 0.0182,
      "step": 21214
    },
    {
      "epoch": 2.9733707077785563,
      "grad_norm": 0.1631285548210144,
      "learning_rate": 2.8557761301124134e-06,
      "loss": 0.0085,
      "step": 21215
    },
    {
      "epoch": 2.973510861948143,
      "grad_norm": 0.05021699145436287,
      "learning_rate": 2.841425496292753e-06,
      "loss": 0.0045,
      "step": 21216
    },
    {
      "epoch": 2.9736510161177296,
      "grad_norm": 0.9280708432197571,
      "learning_rate": 2.8270748624730925e-06,
      "loss": 0.0167,
      "step": 21217
    },
    {
      "epoch": 2.973791170287316,
      "grad_norm": 0.6436378955841064,
      "learning_rate": 2.812724228653432e-06,
      "loss": 0.1539,
      "step": 21218
    },
    {
      "epoch": 2.9739313244569026,
      "grad_norm": 0.02286190167069435,
      "learning_rate": 2.7983735948337717e-06,
      "loss": 0.0015,
      "step": 21219
    },
    {
      "epoch": 2.974071478626489,
      "grad_norm": 0.34060630202293396,
      "learning_rate": 2.7840229610141113e-06,
      "loss": 0.0462,
      "step": 21220
    },
    {
      "epoch": 2.9742116327960755,
      "grad_norm": 0.15698304772377014,
      "learning_rate": 2.769672327194451e-06,
      "loss": 0.0312,
      "step": 21221
    },
    {
      "epoch": 2.9743517869656624,
      "grad_norm": 0.18912239372730255,
      "learning_rate": 2.7553216933747904e-06,
      "loss": 0.0247,
      "step": 21222
    },
    {
      "epoch": 2.974491941135249,
      "grad_norm": 0.18822844326496124,
      "learning_rate": 2.74097105955513e-06,
      "loss": 0.0383,
      "step": 21223
    },
    {
      "epoch": 2.9746320953048353,
      "grad_norm": 0.12350863218307495,
      "learning_rate": 2.72662042573547e-06,
      "loss": 0.0126,
      "step": 21224
    },
    {
      "epoch": 2.9747722494744218,
      "grad_norm": 0.1239120289683342,
      "learning_rate": 2.7122697919158095e-06,
      "loss": 0.0069,
      "step": 21225
    },
    {
      "epoch": 2.9749124036440087,
      "grad_norm": 0.03043368086218834,
      "learning_rate": 2.697919158096149e-06,
      "loss": 0.0055,
      "step": 21226
    },
    {
      "epoch": 2.975052557813595,
      "grad_norm": 0.14320045709609985,
      "learning_rate": 2.6835685242764887e-06,
      "loss": 0.0126,
      "step": 21227
    },
    {
      "epoch": 2.9751927119831816,
      "grad_norm": 0.1972735971212387,
      "learning_rate": 2.6692178904568283e-06,
      "loss": 0.0165,
      "step": 21228
    },
    {
      "epoch": 2.975332866152768,
      "grad_norm": 0.4378927946090698,
      "learning_rate": 2.654867256637168e-06,
      "loss": 0.0173,
      "step": 21229
    },
    {
      "epoch": 2.9754730203223545,
      "grad_norm": 0.17313575744628906,
      "learning_rate": 2.6405166228175074e-06,
      "loss": 0.0119,
      "step": 21230
    },
    {
      "epoch": 2.975613174491941,
      "grad_norm": 0.3695804178714752,
      "learning_rate": 2.626165988997847e-06,
      "loss": 0.0159,
      "step": 21231
    },
    {
      "epoch": 2.9757533286615274,
      "grad_norm": 0.10442319512367249,
      "learning_rate": 2.6118153551781865e-06,
      "loss": 0.0276,
      "step": 21232
    },
    {
      "epoch": 2.9758934828311143,
      "grad_norm": 0.08686045557260513,
      "learning_rate": 2.5974647213585265e-06,
      "loss": 0.0076,
      "step": 21233
    },
    {
      "epoch": 2.976033637000701,
      "grad_norm": 0.15557065606117249,
      "learning_rate": 2.583114087538866e-06,
      "loss": 0.0437,
      "step": 21234
    },
    {
      "epoch": 2.9761737911702872,
      "grad_norm": 0.09110182523727417,
      "learning_rate": 2.5687634537192057e-06,
      "loss": 0.0123,
      "step": 21235
    },
    {
      "epoch": 2.9763139453398737,
      "grad_norm": 0.13279235363006592,
      "learning_rate": 2.5544128198995453e-06,
      "loss": 0.009,
      "step": 21236
    },
    {
      "epoch": 2.9764540995094606,
      "grad_norm": 0.1045272946357727,
      "learning_rate": 2.540062186079885e-06,
      "loss": 0.0132,
      "step": 21237
    },
    {
      "epoch": 2.976594253679047,
      "grad_norm": 0.4879589378833771,
      "learning_rate": 2.5257115522602244e-06,
      "loss": 0.0385,
      "step": 21238
    },
    {
      "epoch": 2.9767344078486335,
      "grad_norm": 0.4535438120365143,
      "learning_rate": 2.511360918440564e-06,
      "loss": 0.0265,
      "step": 21239
    },
    {
      "epoch": 2.97687456201822,
      "grad_norm": 0.1670013666152954,
      "learning_rate": 2.4970102846209035e-06,
      "loss": 0.0199,
      "step": 21240
    },
    {
      "epoch": 2.9770147161878064,
      "grad_norm": 0.08886120468378067,
      "learning_rate": 2.482659650801243e-06,
      "loss": 0.0071,
      "step": 21241
    },
    {
      "epoch": 2.977154870357393,
      "grad_norm": 0.12215469032526016,
      "learning_rate": 2.4683090169815827e-06,
      "loss": 0.0065,
      "step": 21242
    },
    {
      "epoch": 2.97729502452698,
      "grad_norm": 0.0543903149664402,
      "learning_rate": 2.453958383161923e-06,
      "loss": 0.0078,
      "step": 21243
    },
    {
      "epoch": 2.9774351786965663,
      "grad_norm": 0.1725052148103714,
      "learning_rate": 2.4396077493422627e-06,
      "loss": 0.0447,
      "step": 21244
    },
    {
      "epoch": 2.9775753328661527,
      "grad_norm": 0.1441146433353424,
      "learning_rate": 2.4252571155226022e-06,
      "loss": 0.0101,
      "step": 21245
    },
    {
      "epoch": 2.977715487035739,
      "grad_norm": 0.3440361022949219,
      "learning_rate": 2.410906481702942e-06,
      "loss": 0.0211,
      "step": 21246
    },
    {
      "epoch": 2.977855641205326,
      "grad_norm": 0.3911580741405487,
      "learning_rate": 2.3965558478832814e-06,
      "loss": 0.0714,
      "step": 21247
    },
    {
      "epoch": 2.9779957953749125,
      "grad_norm": 0.2645511031150818,
      "learning_rate": 2.382205214063621e-06,
      "loss": 0.0086,
      "step": 21248
    },
    {
      "epoch": 2.978135949544499,
      "grad_norm": 0.04280121996998787,
      "learning_rate": 2.3678545802439605e-06,
      "loss": 0.0034,
      "step": 21249
    },
    {
      "epoch": 2.9782761037140855,
      "grad_norm": 0.15920425951480865,
      "learning_rate": 2.3535039464243e-06,
      "loss": 0.0224,
      "step": 21250
    },
    {
      "epoch": 2.978416257883672,
      "grad_norm": 0.4336516261100769,
      "learning_rate": 2.3391533126046397e-06,
      "loss": 0.0482,
      "step": 21251
    },
    {
      "epoch": 2.9785564120532584,
      "grad_norm": 0.039143770933151245,
      "learning_rate": 2.3248026787849792e-06,
      "loss": 0.0046,
      "step": 21252
    },
    {
      "epoch": 2.9786965662228453,
      "grad_norm": 0.6950653195381165,
      "learning_rate": 2.310452044965319e-06,
      "loss": 0.0808,
      "step": 21253
    },
    {
      "epoch": 2.9788367203924317,
      "grad_norm": 0.03702020272612572,
      "learning_rate": 2.296101411145659e-06,
      "loss": 0.0016,
      "step": 21254
    },
    {
      "epoch": 2.978976874562018,
      "grad_norm": 0.11000578850507736,
      "learning_rate": 2.2817507773259984e-06,
      "loss": 0.0097,
      "step": 21255
    },
    {
      "epoch": 2.9791170287316047,
      "grad_norm": 0.08814690262079239,
      "learning_rate": 2.267400143506338e-06,
      "loss": 0.0066,
      "step": 21256
    },
    {
      "epoch": 2.9792571829011916,
      "grad_norm": 0.1446048766374588,
      "learning_rate": 2.2530495096866775e-06,
      "loss": 0.0113,
      "step": 21257
    },
    {
      "epoch": 2.979397337070778,
      "grad_norm": 0.028340548276901245,
      "learning_rate": 2.238698875867017e-06,
      "loss": 0.001,
      "step": 21258
    },
    {
      "epoch": 2.9795374912403645,
      "grad_norm": 0.07785959541797638,
      "learning_rate": 2.224348242047357e-06,
      "loss": 0.0071,
      "step": 21259
    },
    {
      "epoch": 2.979677645409951,
      "grad_norm": 0.24910004436969757,
      "learning_rate": 2.2099976082276967e-06,
      "loss": 0.0154,
      "step": 21260
    },
    {
      "epoch": 2.9798177995795374,
      "grad_norm": 1.5766299962997437,
      "learning_rate": 2.1956469744080362e-06,
      "loss": 0.1005,
      "step": 21261
    },
    {
      "epoch": 2.979957953749124,
      "grad_norm": 0.3562525510787964,
      "learning_rate": 2.181296340588376e-06,
      "loss": 0.0675,
      "step": 21262
    },
    {
      "epoch": 2.9800981079187103,
      "grad_norm": 0.22078558802604675,
      "learning_rate": 2.1669457067687154e-06,
      "loss": 0.0323,
      "step": 21263
    },
    {
      "epoch": 2.9802382620882972,
      "grad_norm": 0.26065829396247864,
      "learning_rate": 2.152595072949055e-06,
      "loss": 0.0148,
      "step": 21264
    },
    {
      "epoch": 2.9803784162578837,
      "grad_norm": 0.18954956531524658,
      "learning_rate": 2.1382444391293945e-06,
      "loss": 0.0076,
      "step": 21265
    },
    {
      "epoch": 2.98051857042747,
      "grad_norm": 0.704615592956543,
      "learning_rate": 2.123893805309734e-06,
      "loss": 0.0202,
      "step": 21266
    },
    {
      "epoch": 2.9806587245970566,
      "grad_norm": 0.4070110321044922,
      "learning_rate": 2.1095431714900737e-06,
      "loss": 0.0789,
      "step": 21267
    },
    {
      "epoch": 2.9807988787666435,
      "grad_norm": 1.0493751764297485,
      "learning_rate": 2.0951925376704137e-06,
      "loss": 0.0793,
      "step": 21268
    },
    {
      "epoch": 2.98093903293623,
      "grad_norm": 0.14510242640972137,
      "learning_rate": 2.0808419038507532e-06,
      "loss": 0.0108,
      "step": 21269
    },
    {
      "epoch": 2.9810791871058164,
      "grad_norm": 1.8625648021697998,
      "learning_rate": 2.066491270031093e-06,
      "loss": 0.0589,
      "step": 21270
    },
    {
      "epoch": 2.981219341275403,
      "grad_norm": 0.09420888870954514,
      "learning_rate": 2.0521406362114324e-06,
      "loss": 0.0089,
      "step": 21271
    },
    {
      "epoch": 2.9813594954449893,
      "grad_norm": 0.15646785497665405,
      "learning_rate": 2.037790002391772e-06,
      "loss": 0.0227,
      "step": 21272
    },
    {
      "epoch": 2.981499649614576,
      "grad_norm": 0.2971125841140747,
      "learning_rate": 2.023439368572112e-06,
      "loss": 0.0716,
      "step": 21273
    },
    {
      "epoch": 2.9816398037841627,
      "grad_norm": 0.8802768588066101,
      "learning_rate": 2.0090887347524515e-06,
      "loss": 0.0774,
      "step": 21274
    },
    {
      "epoch": 2.981779957953749,
      "grad_norm": 0.120146244764328,
      "learning_rate": 1.994738100932791e-06,
      "loss": 0.0082,
      "step": 21275
    },
    {
      "epoch": 2.9819201121233356,
      "grad_norm": 0.2393527626991272,
      "learning_rate": 1.9803874671131307e-06,
      "loss": 0.0176,
      "step": 21276
    },
    {
      "epoch": 2.982060266292922,
      "grad_norm": 0.18858370184898376,
      "learning_rate": 1.9660368332934702e-06,
      "loss": 0.0153,
      "step": 21277
    },
    {
      "epoch": 2.982200420462509,
      "grad_norm": 0.2839561402797699,
      "learning_rate": 1.95168619947381e-06,
      "loss": 0.0292,
      "step": 21278
    },
    {
      "epoch": 2.9823405746320955,
      "grad_norm": 0.2131810337305069,
      "learning_rate": 1.9373355656541494e-06,
      "loss": 0.0108,
      "step": 21279
    },
    {
      "epoch": 2.982480728801682,
      "grad_norm": 0.04391753301024437,
      "learning_rate": 1.922984931834489e-06,
      "loss": 0.0029,
      "step": 21280
    },
    {
      "epoch": 2.9826208829712684,
      "grad_norm": 0.33264344930648804,
      "learning_rate": 1.908634298014829e-06,
      "loss": 0.0608,
      "step": 21281
    },
    {
      "epoch": 2.982761037140855,
      "grad_norm": 0.13141390681266785,
      "learning_rate": 1.8942836641951685e-06,
      "loss": 0.0228,
      "step": 21282
    },
    {
      "epoch": 2.9829011913104413,
      "grad_norm": 0.1918085515499115,
      "learning_rate": 1.879933030375508e-06,
      "loss": 0.0096,
      "step": 21283
    },
    {
      "epoch": 2.983041345480028,
      "grad_norm": 0.152059406042099,
      "learning_rate": 1.8655823965558477e-06,
      "loss": 0.0314,
      "step": 21284
    },
    {
      "epoch": 2.9831814996496147,
      "grad_norm": 0.17291635274887085,
      "learning_rate": 1.8512317627361872e-06,
      "loss": 0.0386,
      "step": 21285
    },
    {
      "epoch": 2.983321653819201,
      "grad_norm": 0.05877182260155678,
      "learning_rate": 1.8368811289165268e-06,
      "loss": 0.0035,
      "step": 21286
    },
    {
      "epoch": 2.9834618079887876,
      "grad_norm": 0.13336123526096344,
      "learning_rate": 1.8225304950968668e-06,
      "loss": 0.0142,
      "step": 21287
    },
    {
      "epoch": 2.9836019621583745,
      "grad_norm": 0.2553093731403351,
      "learning_rate": 1.8081798612772064e-06,
      "loss": 0.0566,
      "step": 21288
    },
    {
      "epoch": 2.983742116327961,
      "grad_norm": 0.15615487098693848,
      "learning_rate": 1.793829227457546e-06,
      "loss": 0.0256,
      "step": 21289
    },
    {
      "epoch": 2.9838822704975474,
      "grad_norm": 0.19886676967144012,
      "learning_rate": 1.7794785936378855e-06,
      "loss": 0.0665,
      "step": 21290
    },
    {
      "epoch": 2.984022424667134,
      "grad_norm": 0.03692477568984032,
      "learning_rate": 1.765127959818225e-06,
      "loss": 0.0052,
      "step": 21291
    },
    {
      "epoch": 2.9841625788367203,
      "grad_norm": 0.2963419556617737,
      "learning_rate": 1.7507773259985649e-06,
      "loss": 0.0457,
      "step": 21292
    },
    {
      "epoch": 2.9843027330063068,
      "grad_norm": 0.2805601954460144,
      "learning_rate": 1.7364266921789044e-06,
      "loss": 0.0425,
      "step": 21293
    },
    {
      "epoch": 2.9844428871758932,
      "grad_norm": 0.6528148651123047,
      "learning_rate": 1.722076058359244e-06,
      "loss": 0.0406,
      "step": 21294
    },
    {
      "epoch": 2.98458304134548,
      "grad_norm": 0.2736854553222656,
      "learning_rate": 1.7077254245395836e-06,
      "loss": 0.0068,
      "step": 21295
    },
    {
      "epoch": 2.9847231955150666,
      "grad_norm": 0.06643366068601608,
      "learning_rate": 1.6933747907199234e-06,
      "loss": 0.0037,
      "step": 21296
    },
    {
      "epoch": 2.984863349684653,
      "grad_norm": 0.710904061794281,
      "learning_rate": 1.679024156900263e-06,
      "loss": 0.0413,
      "step": 21297
    },
    {
      "epoch": 2.9850035038542395,
      "grad_norm": 0.231459379196167,
      "learning_rate": 1.6646735230806025e-06,
      "loss": 0.0232,
      "step": 21298
    },
    {
      "epoch": 2.9851436580238264,
      "grad_norm": 0.051056526601314545,
      "learning_rate": 1.650322889260942e-06,
      "loss": 0.0029,
      "step": 21299
    },
    {
      "epoch": 2.985283812193413,
      "grad_norm": 0.03607397899031639,
      "learning_rate": 1.6359722554412816e-06,
      "loss": 0.002,
      "step": 21300
    },
    {
      "epoch": 2.9854239663629993,
      "grad_norm": 0.14366395771503448,
      "learning_rate": 1.6216216216216216e-06,
      "loss": 0.0094,
      "step": 21301
    },
    {
      "epoch": 2.985564120532586,
      "grad_norm": 0.22030656039714813,
      "learning_rate": 1.6072709878019612e-06,
      "loss": 0.0211,
      "step": 21302
    },
    {
      "epoch": 2.9857042747021723,
      "grad_norm": 0.5090020895004272,
      "learning_rate": 1.5929203539823008e-06,
      "loss": 0.024,
      "step": 21303
    },
    {
      "epoch": 2.9858444288717587,
      "grad_norm": 0.1953844577074051,
      "learning_rate": 1.5785697201626404e-06,
      "loss": 0.0234,
      "step": 21304
    },
    {
      "epoch": 2.9859845830413456,
      "grad_norm": 0.12596209347248077,
      "learning_rate": 1.5642190863429801e-06,
      "loss": 0.0037,
      "step": 21305
    },
    {
      "epoch": 2.986124737210932,
      "grad_norm": 0.22437943518161774,
      "learning_rate": 1.5498684525233197e-06,
      "loss": 0.0233,
      "step": 21306
    },
    {
      "epoch": 2.9862648913805185,
      "grad_norm": 0.09167762845754623,
      "learning_rate": 1.5355178187036593e-06,
      "loss": 0.0065,
      "step": 21307
    },
    {
      "epoch": 2.986405045550105,
      "grad_norm": 0.05860983580350876,
      "learning_rate": 1.5211671848839989e-06,
      "loss": 0.0028,
      "step": 21308
    },
    {
      "epoch": 2.986545199719692,
      "grad_norm": 0.11762921512126923,
      "learning_rate": 1.5068165510643384e-06,
      "loss": 0.0271,
      "step": 21309
    },
    {
      "epoch": 2.9866853538892784,
      "grad_norm": 0.022226884961128235,
      "learning_rate": 1.4924659172446782e-06,
      "loss": 0.0022,
      "step": 21310
    },
    {
      "epoch": 2.986825508058865,
      "grad_norm": 0.03915843367576599,
      "learning_rate": 1.4781152834250178e-06,
      "loss": 0.0021,
      "step": 21311
    },
    {
      "epoch": 2.9869656622284513,
      "grad_norm": 0.09635857492685318,
      "learning_rate": 1.4637646496053574e-06,
      "loss": 0.0047,
      "step": 21312
    },
    {
      "epoch": 2.9871058163980377,
      "grad_norm": 0.051178429275751114,
      "learning_rate": 1.449414015785697e-06,
      "loss": 0.0037,
      "step": 21313
    },
    {
      "epoch": 2.987245970567624,
      "grad_norm": 0.19735737144947052,
      "learning_rate": 1.4350633819660365e-06,
      "loss": 0.0386,
      "step": 21314
    },
    {
      "epoch": 2.9873861247372107,
      "grad_norm": 0.14851321280002594,
      "learning_rate": 1.4207127481463765e-06,
      "loss": 0.0079,
      "step": 21315
    },
    {
      "epoch": 2.9875262789067976,
      "grad_norm": 0.516215443611145,
      "learning_rate": 1.406362114326716e-06,
      "loss": 0.0962,
      "step": 21316
    },
    {
      "epoch": 2.987666433076384,
      "grad_norm": 0.7529941201210022,
      "learning_rate": 1.3920114805070556e-06,
      "loss": 0.0861,
      "step": 21317
    },
    {
      "epoch": 2.9878065872459705,
      "grad_norm": 0.7254121899604797,
      "learning_rate": 1.3776608466873952e-06,
      "loss": 0.1756,
      "step": 21318
    },
    {
      "epoch": 2.9879467414155574,
      "grad_norm": 2.3045730590820312,
      "learning_rate": 1.363310212867735e-06,
      "loss": 0.122,
      "step": 21319
    },
    {
      "epoch": 2.988086895585144,
      "grad_norm": 4.316004753112793,
      "learning_rate": 1.3489595790480746e-06,
      "loss": 0.14,
      "step": 21320
    },
    {
      "epoch": 2.9882270497547303,
      "grad_norm": 0.09581112861633301,
      "learning_rate": 1.3346089452284141e-06,
      "loss": 0.0219,
      "step": 21321
    },
    {
      "epoch": 2.9883672039243168,
      "grad_norm": 0.2235007882118225,
      "learning_rate": 1.3202583114087537e-06,
      "loss": 0.0186,
      "step": 21322
    },
    {
      "epoch": 2.988507358093903,
      "grad_norm": 0.010522022843360901,
      "learning_rate": 1.3059076775890933e-06,
      "loss": 0.0012,
      "step": 21323
    },
    {
      "epoch": 2.9886475122634897,
      "grad_norm": 0.2181089222431183,
      "learning_rate": 1.291557043769433e-06,
      "loss": 0.0139,
      "step": 21324
    },
    {
      "epoch": 2.988787666433076,
      "grad_norm": 0.16764657199382782,
      "learning_rate": 1.2772064099497726e-06,
      "loss": 0.014,
      "step": 21325
    },
    {
      "epoch": 2.988927820602663,
      "grad_norm": 1.1151686906814575,
      "learning_rate": 1.2628557761301122e-06,
      "loss": 0.0912,
      "step": 21326
    },
    {
      "epoch": 2.9890679747722495,
      "grad_norm": 0.20999130606651306,
      "learning_rate": 1.2485051423104518e-06,
      "loss": 0.0318,
      "step": 21327
    },
    {
      "epoch": 2.989208128941836,
      "grad_norm": 0.04904772341251373,
      "learning_rate": 1.2341545084907913e-06,
      "loss": 0.0052,
      "step": 21328
    },
    {
      "epoch": 2.9893482831114224,
      "grad_norm": 0.2710166573524475,
      "learning_rate": 1.2198038746711313e-06,
      "loss": 0.0395,
      "step": 21329
    },
    {
      "epoch": 2.9894884372810093,
      "grad_norm": 0.1824885606765747,
      "learning_rate": 1.205453240851471e-06,
      "loss": 0.0268,
      "step": 21330
    },
    {
      "epoch": 2.989628591450596,
      "grad_norm": 0.1584746390581131,
      "learning_rate": 1.1911026070318105e-06,
      "loss": 0.0279,
      "step": 21331
    },
    {
      "epoch": 2.9897687456201822,
      "grad_norm": 0.1529160737991333,
      "learning_rate": 1.17675197321215e-06,
      "loss": 0.0258,
      "step": 21332
    },
    {
      "epoch": 2.9899088997897687,
      "grad_norm": 0.08004710078239441,
      "learning_rate": 1.1624013393924896e-06,
      "loss": 0.0055,
      "step": 21333
    },
    {
      "epoch": 2.990049053959355,
      "grad_norm": 0.10253626108169556,
      "learning_rate": 1.1480507055728294e-06,
      "loss": 0.0083,
      "step": 21334
    },
    {
      "epoch": 2.9901892081289416,
      "grad_norm": 0.2089439183473587,
      "learning_rate": 1.133700071753169e-06,
      "loss": 0.0068,
      "step": 21335
    },
    {
      "epoch": 2.9903293622985285,
      "grad_norm": 1.4994295835494995,
      "learning_rate": 1.1193494379335085e-06,
      "loss": 0.0809,
      "step": 21336
    },
    {
      "epoch": 2.990469516468115,
      "grad_norm": 0.08568329364061356,
      "learning_rate": 1.1049988041138483e-06,
      "loss": 0.0151,
      "step": 21337
    },
    {
      "epoch": 2.9906096706377014,
      "grad_norm": 0.052259791642427444,
      "learning_rate": 1.090648170294188e-06,
      "loss": 0.0049,
      "step": 21338
    },
    {
      "epoch": 2.990749824807288,
      "grad_norm": 0.2366602122783661,
      "learning_rate": 1.0762975364745275e-06,
      "loss": 0.0199,
      "step": 21339
    },
    {
      "epoch": 2.990889978976875,
      "grad_norm": 0.20673762261867523,
      "learning_rate": 1.061946902654867e-06,
      "loss": 0.0403,
      "step": 21340
    },
    {
      "epoch": 2.9910301331464613,
      "grad_norm": 0.05970551073551178,
      "learning_rate": 1.0475962688352068e-06,
      "loss": 0.0031,
      "step": 21341
    },
    {
      "epoch": 2.9911702873160477,
      "grad_norm": 0.09055374562740326,
      "learning_rate": 1.0332456350155464e-06,
      "loss": 0.012,
      "step": 21342
    },
    {
      "epoch": 2.991310441485634,
      "grad_norm": 0.05353236943483353,
      "learning_rate": 1.018895001195886e-06,
      "loss": 0.0009,
      "step": 21343
    },
    {
      "epoch": 2.9914505956552206,
      "grad_norm": 0.20226335525512695,
      "learning_rate": 1.0045443673762258e-06,
      "loss": 0.0185,
      "step": 21344
    },
    {
      "epoch": 2.991590749824807,
      "grad_norm": 0.14523659646511078,
      "learning_rate": 9.901937335565653e-07,
      "loss": 0.0277,
      "step": 21345
    },
    {
      "epoch": 2.9917309039943936,
      "grad_norm": 0.08523275703191757,
      "learning_rate": 9.75843099736905e-07,
      "loss": 0.0064,
      "step": 21346
    },
    {
      "epoch": 2.9918710581639805,
      "grad_norm": 0.16134516894817352,
      "learning_rate": 9.614924659172445e-07,
      "loss": 0.0076,
      "step": 21347
    },
    {
      "epoch": 2.992011212333567,
      "grad_norm": 0.081428661942482,
      "learning_rate": 9.471418320975843e-07,
      "loss": 0.007,
      "step": 21348
    },
    {
      "epoch": 2.9921513665031534,
      "grad_norm": 0.1241818368434906,
      "learning_rate": 9.327911982779238e-07,
      "loss": 0.025,
      "step": 21349
    },
    {
      "epoch": 2.9922915206727403,
      "grad_norm": 0.14690424501895905,
      "learning_rate": 9.184405644582634e-07,
      "loss": 0.0093,
      "step": 21350
    },
    {
      "epoch": 2.9924316748423267,
      "grad_norm": 1.0814913511276245,
      "learning_rate": 9.040899306386032e-07,
      "loss": 0.0431,
      "step": 21351
    },
    {
      "epoch": 2.992571829011913,
      "grad_norm": 0.10704734176397324,
      "learning_rate": 8.897392968189428e-07,
      "loss": 0.0089,
      "step": 21352
    },
    {
      "epoch": 2.9927119831814997,
      "grad_norm": 0.35222530364990234,
      "learning_rate": 8.753886629992824e-07,
      "loss": 0.0514,
      "step": 21353
    },
    {
      "epoch": 2.992852137351086,
      "grad_norm": 0.1267102211713791,
      "learning_rate": 8.61038029179622e-07,
      "loss": 0.0112,
      "step": 21354
    },
    {
      "epoch": 2.9929922915206726,
      "grad_norm": 0.10336574167013168,
      "learning_rate": 8.466873953599617e-07,
      "loss": 0.0115,
      "step": 21355
    },
    {
      "epoch": 2.993132445690259,
      "grad_norm": 0.16795602440834045,
      "learning_rate": 8.323367615403013e-07,
      "loss": 0.018,
      "step": 21356
    },
    {
      "epoch": 2.993272599859846,
      "grad_norm": 0.10170432180166245,
      "learning_rate": 8.179861277206408e-07,
      "loss": 0.005,
      "step": 21357
    },
    {
      "epoch": 2.9934127540294324,
      "grad_norm": 0.1272394359111786,
      "learning_rate": 8.036354939009806e-07,
      "loss": 0.0203,
      "step": 21358
    },
    {
      "epoch": 2.993552908199019,
      "grad_norm": 0.5077134966850281,
      "learning_rate": 7.892848600813202e-07,
      "loss": 0.0343,
      "step": 21359
    },
    {
      "epoch": 2.9936930623686053,
      "grad_norm": 0.2041243463754654,
      "learning_rate": 7.749342262616599e-07,
      "loss": 0.0199,
      "step": 21360
    },
    {
      "epoch": 2.9938332165381922,
      "grad_norm": 0.2168726772069931,
      "learning_rate": 7.605835924419994e-07,
      "loss": 0.0078,
      "step": 21361
    },
    {
      "epoch": 2.9939733707077787,
      "grad_norm": 0.33751949667930603,
      "learning_rate": 7.462329586223391e-07,
      "loss": 0.022,
      "step": 21362
    },
    {
      "epoch": 2.994113524877365,
      "grad_norm": 0.3545292913913727,
      "learning_rate": 7.318823248026787e-07,
      "loss": 0.1091,
      "step": 21363
    },
    {
      "epoch": 2.9942536790469516,
      "grad_norm": 0.3890790343284607,
      "learning_rate": 7.175316909830182e-07,
      "loss": 0.0303,
      "step": 21364
    },
    {
      "epoch": 2.994393833216538,
      "grad_norm": 0.26840487122535706,
      "learning_rate": 7.03181057163358e-07,
      "loss": 0.0202,
      "step": 21365
    },
    {
      "epoch": 2.9945339873861245,
      "grad_norm": 0.39443060755729675,
      "learning_rate": 6.888304233436976e-07,
      "loss": 0.0757,
      "step": 21366
    },
    {
      "epoch": 2.9946741415557114,
      "grad_norm": 0.2796522080898285,
      "learning_rate": 6.744797895240373e-07,
      "loss": 0.0254,
      "step": 21367
    },
    {
      "epoch": 2.994814295725298,
      "grad_norm": 0.12309366464614868,
      "learning_rate": 6.601291557043769e-07,
      "loss": 0.009,
      "step": 21368
    },
    {
      "epoch": 2.9949544498948844,
      "grad_norm": 0.027093442156910896,
      "learning_rate": 6.457785218847165e-07,
      "loss": 0.0015,
      "step": 21369
    },
    {
      "epoch": 2.995094604064471,
      "grad_norm": 1.459133267402649,
      "learning_rate": 6.314278880650561e-07,
      "loss": 0.0662,
      "step": 21370
    },
    {
      "epoch": 2.9952347582340577,
      "grad_norm": 0.19384761154651642,
      "learning_rate": 6.170772542453957e-07,
      "loss": 0.0319,
      "step": 21371
    },
    {
      "epoch": 2.995374912403644,
      "grad_norm": 0.3839840888977051,
      "learning_rate": 6.027266204257355e-07,
      "loss": 0.0403,
      "step": 21372
    },
    {
      "epoch": 2.9955150665732306,
      "grad_norm": 0.19356566667556763,
      "learning_rate": 5.88375986606075e-07,
      "loss": 0.0227,
      "step": 21373
    },
    {
      "epoch": 2.995655220742817,
      "grad_norm": 0.07258197665214539,
      "learning_rate": 5.740253527864147e-07,
      "loss": 0.0063,
      "step": 21374
    },
    {
      "epoch": 2.9957953749124036,
      "grad_norm": 0.13831639289855957,
      "learning_rate": 5.596747189667543e-07,
      "loss": 0.0475,
      "step": 21375
    },
    {
      "epoch": 2.99593552908199,
      "grad_norm": 0.2333473265171051,
      "learning_rate": 5.45324085147094e-07,
      "loss": 0.0209,
      "step": 21376
    },
    {
      "epoch": 2.9960756832515765,
      "grad_norm": 0.2990017831325531,
      "learning_rate": 5.309734513274335e-07,
      "loss": 0.0344,
      "step": 21377
    },
    {
      "epoch": 2.9962158374211634,
      "grad_norm": 0.6658338308334351,
      "learning_rate": 5.166228175077732e-07,
      "loss": 0.0465,
      "step": 21378
    },
    {
      "epoch": 2.99635599159075,
      "grad_norm": 0.355704128742218,
      "learning_rate": 5.022721836881129e-07,
      "loss": 0.0393,
      "step": 21379
    },
    {
      "epoch": 2.9964961457603363,
      "grad_norm": 0.07343513518571854,
      "learning_rate": 4.879215498684525e-07,
      "loss": 0.0151,
      "step": 21380
    },
    {
      "epoch": 2.9966362999299228,
      "grad_norm": 0.3340516686439514,
      "learning_rate": 4.7357091604879213e-07,
      "loss": 0.024,
      "step": 21381
    },
    {
      "epoch": 2.9967764540995097,
      "grad_norm": 0.4459291696548462,
      "learning_rate": 4.592202822291317e-07,
      "loss": 0.0484,
      "step": 21382
    },
    {
      "epoch": 2.996916608269096,
      "grad_norm": 0.4163734018802643,
      "learning_rate": 4.448696484094714e-07,
      "loss": 0.0599,
      "step": 21383
    },
    {
      "epoch": 2.9970567624386826,
      "grad_norm": 0.06866970658302307,
      "learning_rate": 4.30519014589811e-07,
      "loss": 0.0068,
      "step": 21384
    },
    {
      "epoch": 2.997196916608269,
      "grad_norm": 0.5593795776367188,
      "learning_rate": 4.161683807701506e-07,
      "loss": 0.0562,
      "step": 21385
    },
    {
      "epoch": 2.9973370707778555,
      "grad_norm": 0.2872020900249481,
      "learning_rate": 4.018177469504903e-07,
      "loss": 0.0248,
      "step": 21386
    },
    {
      "epoch": 2.997477224947442,
      "grad_norm": 0.14189307391643524,
      "learning_rate": 3.8746711313082993e-07,
      "loss": 0.0235,
      "step": 21387
    },
    {
      "epoch": 2.997617379117029,
      "grad_norm": 0.21212796866893768,
      "learning_rate": 3.7311647931116955e-07,
      "loss": 0.0597,
      "step": 21388
    },
    {
      "epoch": 2.9977575332866153,
      "grad_norm": 0.09847323596477509,
      "learning_rate": 3.587658454915091e-07,
      "loss": 0.0061,
      "step": 21389
    },
    {
      "epoch": 2.9978976874562018,
      "grad_norm": 0.6569809913635254,
      "learning_rate": 3.444152116718488e-07,
      "loss": 0.098,
      "step": 21390
    },
    {
      "epoch": 2.9980378416257882,
      "grad_norm": 0.15362878143787384,
      "learning_rate": 3.300645778521884e-07,
      "loss": 0.0213,
      "step": 21391
    },
    {
      "epoch": 2.998177995795375,
      "grad_norm": 0.557316243648529,
      "learning_rate": 3.1571394403252805e-07,
      "loss": 0.029,
      "step": 21392
    },
    {
      "epoch": 2.9983181499649616,
      "grad_norm": 0.17047187685966492,
      "learning_rate": 3.0136331021286773e-07,
      "loss": 0.0096,
      "step": 21393
    },
    {
      "epoch": 2.998458304134548,
      "grad_norm": 0.09355489164590836,
      "learning_rate": 2.8701267639320735e-07,
      "loss": 0.0142,
      "step": 21394
    },
    {
      "epoch": 2.9985984583041345,
      "grad_norm": 0.3209400475025177,
      "learning_rate": 2.72662042573547e-07,
      "loss": 0.0285,
      "step": 21395
    },
    {
      "epoch": 2.998738612473721,
      "grad_norm": 0.024161234498023987,
      "learning_rate": 2.583114087538866e-07,
      "loss": 0.0014,
      "step": 21396
    },
    {
      "epoch": 2.9988787666433074,
      "grad_norm": 0.1335512101650238,
      "learning_rate": 2.439607749342262e-07,
      "loss": 0.0057,
      "step": 21397
    },
    {
      "epoch": 2.9990189208128943,
      "grad_norm": 0.16399157047271729,
      "learning_rate": 2.2961014111456585e-07,
      "loss": 0.0073,
      "step": 21398
    },
    {
      "epoch": 2.999159074982481,
      "grad_norm": 0.2587020993232727,
      "learning_rate": 2.152595072949055e-07,
      "loss": 0.0263,
      "step": 21399
    },
    {
      "epoch": 2.9992992291520673,
      "grad_norm": 0.20869041979312897,
      "learning_rate": 2.0090887347524515e-07,
      "loss": 0.0302,
      "step": 21400
    },
    {
      "epoch": 2.9994393833216537,
      "grad_norm": 0.11802941560745239,
      "learning_rate": 1.8655823965558478e-07,
      "loss": 0.0109,
      "step": 21401
    },
    {
      "epoch": 2.9995795374912406,
      "grad_norm": 0.1735084056854248,
      "learning_rate": 1.722076058359244e-07,
      "loss": 0.005,
      "step": 21402
    },
    {
      "epoch": 2.999719691660827,
      "grad_norm": 0.20583558082580566,
      "learning_rate": 1.5785697201626402e-07,
      "loss": 0.007,
      "step": 21403
    },
    {
      "epoch": 2.9998598458304135,
      "grad_norm": 0.97768235206604,
      "learning_rate": 1.4350633819660368e-07,
      "loss": 0.0669,
      "step": 21404
    },
    {
      "epoch": 3.0,
      "grad_norm": 3.6988003253936768,
      "learning_rate": 1.291557043769433e-07,
      "loss": 0.2459,
      "step": 21405
    },
    {
      "epoch": 3.0,
      "step": 21405,
      "total_flos": 3.3190826326923264e+19,
      "train_loss": 0.19606071691714266,
      "train_runtime": 14592.9757,
      "train_samples_per_second": 5.867,
      "train_steps_per_second": 1.467
    }
  ],
  "logging_steps": 1.0,
  "max_steps": 21405,
  "num_input_tokens_seen": 0,
  "num_train_epochs": 3,
  "save_steps": 400,
  "stateful_callbacks": {
    "TrainerControl": {
      "args": {
        "should_epoch_stop": false,
        "should_evaluate": false,
        "should_log": false,
        "should_save": true,
        "should_training_stop": true
      },
      "attributes": {}
    }
  },
  "total_flos": 3.3190826326923264e+19,
  "train_batch_size": 4,
  "trial_name": null,
  "trial_params": null
}