{
  "best_metric": null,
  "best_model_checkpoint": null,
  "epoch": 0.41981528127623846,
  "eval_steps": 500,
  "global_step": 2000,
  "is_hyper_param_search": false,
  "is_local_process_zero": true,
  "is_world_process_zero": true,
  "log_history": [
    {
      "epoch": 0.00020990764063811922,
      "grad_norm": 0.8211116790771484,
      "learning_rate": 2.09643605870021e-07,
      "loss": 0.9775,
      "step": 1
    },
    {
      "epoch": 0.00041981528127623844,
      "grad_norm": 0.7017123103141785,
      "learning_rate": 4.19287211740042e-07,
      "loss": 0.868,
      "step": 2
    },
    {
      "epoch": 0.0006297229219143577,
      "grad_norm": 0.6900752782821655,
      "learning_rate": 6.28930817610063e-07,
      "loss": 0.8862,
      "step": 3
    },
    {
      "epoch": 0.0008396305625524769,
      "grad_norm": 0.7597951292991638,
      "learning_rate": 8.38574423480084e-07,
      "loss": 0.923,
      "step": 4
    },
    {
      "epoch": 0.0010495382031905961,
      "grad_norm": 0.8734421730041504,
      "learning_rate": 1.048218029350105e-06,
      "loss": 1.0745,
      "step": 5
    },
    {
      "epoch": 0.0012594458438287153,
      "grad_norm": 0.8313158750534058,
      "learning_rate": 1.257861635220126e-06,
      "loss": 0.9817,
      "step": 6
    },
    {
      "epoch": 0.0014693534844668346,
      "grad_norm": 0.656913697719574,
      "learning_rate": 1.467505241090147e-06,
      "loss": 0.8544,
      "step": 7
    },
    {
      "epoch": 0.0016792611251049538,
      "grad_norm": 0.7147523164749146,
      "learning_rate": 1.677148846960168e-06,
      "loss": 0.8955,
      "step": 8
    },
    {
      "epoch": 0.001889168765743073,
      "grad_norm": 0.753795325756073,
      "learning_rate": 1.8867924528301887e-06,
      "loss": 0.881,
      "step": 9
    },
    {
      "epoch": 0.0020990764063811922,
      "grad_norm": 0.7141973972320557,
      "learning_rate": 2.09643605870021e-06,
      "loss": 0.9229,
      "step": 10
    },
    {
      "epoch": 0.0023089840470193117,
      "grad_norm": 0.8047632575035095,
      "learning_rate": 2.306079664570231e-06,
      "loss": 0.9438,
      "step": 11
    },
    {
      "epoch": 0.0025188916876574307,
      "grad_norm": 0.8492773175239563,
      "learning_rate": 2.515723270440252e-06,
      "loss": 0.9977,
      "step": 12
    },
    {
      "epoch": 0.00272879932829555,
      "grad_norm": 0.744103729724884,
      "learning_rate": 2.7253668763102727e-06,
      "loss": 0.9049,
      "step": 13
    },
    {
      "epoch": 0.002938706968933669,
      "grad_norm": 0.7073290944099426,
      "learning_rate": 2.935010482180294e-06,
      "loss": 0.8824,
      "step": 14
    },
    {
      "epoch": 0.0031486146095717885,
      "grad_norm": 0.6808626651763916,
      "learning_rate": 3.1446540880503146e-06,
      "loss": 0.8854,
      "step": 15
    },
    {
      "epoch": 0.0033585222502099076,
      "grad_norm": 0.6526097655296326,
      "learning_rate": 3.354297693920336e-06,
      "loss": 0.8219,
      "step": 16
    },
    {
      "epoch": 0.003568429890848027,
      "grad_norm": 0.8246333003044128,
      "learning_rate": 3.563941299790356e-06,
      "loss": 0.9948,
      "step": 17
    },
    {
      "epoch": 0.003778337531486146,
      "grad_norm": 0.8494473099708557,
      "learning_rate": 3.7735849056603773e-06,
      "loss": 1.0187,
      "step": 18
    },
    {
      "epoch": 0.003988245172124265,
      "grad_norm": 0.9259452223777771,
      "learning_rate": 3.9832285115303985e-06,
      "loss": 1.0864,
      "step": 19
    },
    {
      "epoch": 0.0041981528127623844,
      "grad_norm": 0.8235483765602112,
      "learning_rate": 4.19287211740042e-06,
      "loss": 1.0255,
      "step": 20
    },
    {
      "epoch": 0.004408060453400504,
      "grad_norm": 0.8101536631584167,
      "learning_rate": 4.40251572327044e-06,
      "loss": 0.9892,
      "step": 21
    },
    {
      "epoch": 0.004617968094038623,
      "grad_norm": 0.7406070232391357,
      "learning_rate": 4.612159329140462e-06,
      "loss": 0.9266,
      "step": 22
    },
    {
      "epoch": 0.004827875734676742,
      "grad_norm": 0.7151230573654175,
      "learning_rate": 4.821802935010482e-06,
      "loss": 0.902,
      "step": 23
    },
    {
      "epoch": 0.005037783375314861,
      "grad_norm": 0.6921148896217346,
      "learning_rate": 5.031446540880504e-06,
      "loss": 0.8621,
      "step": 24
    },
    {
      "epoch": 0.005247691015952981,
      "grad_norm": 0.6169761419296265,
      "learning_rate": 5.241090146750524e-06,
      "loss": 0.8296,
      "step": 25
    },
    {
      "epoch": 0.0054575986565911,
      "grad_norm": 0.679093599319458,
      "learning_rate": 5.4507337526205454e-06,
      "loss": 0.8593,
      "step": 26
    },
    {
      "epoch": 0.005667506297229219,
      "grad_norm": 0.8058671951293945,
      "learning_rate": 5.660377358490566e-06,
      "loss": 0.9504,
      "step": 27
    },
    {
      "epoch": 0.005877413937867338,
      "grad_norm": 0.8705667853355408,
      "learning_rate": 5.870020964360588e-06,
      "loss": 1.0245,
      "step": 28
    },
    {
      "epoch": 0.006087321578505458,
      "grad_norm": 0.7334048748016357,
      "learning_rate": 6.079664570230608e-06,
      "loss": 0.9172,
      "step": 29
    },
    {
      "epoch": 0.006297229219143577,
      "grad_norm": 0.7490406632423401,
      "learning_rate": 6.289308176100629e-06,
      "loss": 0.9329,
      "step": 30
    },
    {
      "epoch": 0.006507136859781696,
      "grad_norm": 0.8739404678344727,
      "learning_rate": 6.49895178197065e-06,
      "loss": 0.9584,
      "step": 31
    },
    {
      "epoch": 0.006717044500419815,
      "grad_norm": 0.7686614990234375,
      "learning_rate": 6.708595387840672e-06,
      "loss": 0.9493,
      "step": 32
    },
    {
      "epoch": 0.0069269521410579345,
      "grad_norm": 0.9184572100639343,
      "learning_rate": 6.918238993710692e-06,
      "loss": 1.0678,
      "step": 33
    },
    {
      "epoch": 0.007136859781696054,
      "grad_norm": 0.8303453326225281,
      "learning_rate": 7.127882599580712e-06,
      "loss": 0.9834,
      "step": 34
    },
    {
      "epoch": 0.0073467674223341725,
      "grad_norm": 0.6670796871185303,
      "learning_rate": 7.337526205450735e-06,
      "loss": 0.8434,
      "step": 35
    },
    {
      "epoch": 0.007556675062972292,
      "grad_norm": 0.6518784761428833,
      "learning_rate": 7.547169811320755e-06,
      "loss": 0.8445,
      "step": 36
    },
    {
      "epoch": 0.007766582703610411,
      "grad_norm": 0.8396414518356323,
      "learning_rate": 7.756813417190776e-06,
      "loss": 0.9229,
      "step": 37
    },
    {
      "epoch": 0.00797649034424853,
      "grad_norm": 0.9019722938537598,
      "learning_rate": 7.966457023060797e-06,
      "loss": 0.9545,
      "step": 38
    },
    {
      "epoch": 0.00818639798488665,
      "grad_norm": 0.6634522676467896,
      "learning_rate": 8.176100628930818e-06,
      "loss": 0.8279,
      "step": 39
    },
    {
      "epoch": 0.008396305625524769,
      "grad_norm": 0.86155104637146,
      "learning_rate": 8.38574423480084e-06,
      "loss": 0.9326,
      "step": 40
    },
    {
      "epoch": 0.008606213266162888,
      "grad_norm": 0.6448208689689636,
      "learning_rate": 8.59538784067086e-06,
      "loss": 0.8261,
      "step": 41
    },
    {
      "epoch": 0.008816120906801008,
      "grad_norm": 0.7442598938941956,
      "learning_rate": 8.80503144654088e-06,
      "loss": 0.9352,
      "step": 42
    },
    {
      "epoch": 0.009026028547439127,
      "grad_norm": 0.8854546546936035,
      "learning_rate": 9.014675052410902e-06,
      "loss": 1.017,
      "step": 43
    },
    {
      "epoch": 0.009235936188077247,
      "grad_norm": 0.66485196352005,
      "learning_rate": 9.224318658280923e-06,
      "loss": 0.8096,
      "step": 44
    },
    {
      "epoch": 0.009445843828715366,
      "grad_norm": 0.5998132824897766,
      "learning_rate": 9.433962264150944e-06,
      "loss": 0.7932,
      "step": 45
    },
    {
      "epoch": 0.009655751469353484,
      "grad_norm": 0.6402536034584045,
      "learning_rate": 9.643605870020965e-06,
      "loss": 0.8146,
      "step": 46
    },
    {
      "epoch": 0.009865659109991603,
      "grad_norm": 0.6665769219398499,
      "learning_rate": 9.853249475890985e-06,
      "loss": 0.8435,
      "step": 47
    },
    {
      "epoch": 0.010075566750629723,
      "grad_norm": 0.6947203874588013,
      "learning_rate": 1.0062893081761008e-05,
      "loss": 0.8712,
      "step": 48
    },
    {
      "epoch": 0.010285474391267842,
      "grad_norm": 0.6707759499549866,
      "learning_rate": 1.0272536687631027e-05,
      "loss": 0.8382,
      "step": 49
    },
    {
      "epoch": 0.010495382031905962,
      "grad_norm": 0.6716253161430359,
      "learning_rate": 1.0482180293501048e-05,
      "loss": 0.8887,
      "step": 50
    },
    {
      "epoch": 0.010705289672544081,
      "grad_norm": 0.6954374313354492,
      "learning_rate": 1.069182389937107e-05,
      "loss": 0.895,
      "step": 51
    },
    {
      "epoch": 0.0109151973131822,
      "grad_norm": 0.5633372068405151,
      "learning_rate": 1.0901467505241091e-05,
      "loss": 0.7971,
      "step": 52
    },
    {
      "epoch": 0.01112510495382032,
      "grad_norm": 0.5513401031494141,
      "learning_rate": 1.1111111111111112e-05,
      "loss": 0.7848,
      "step": 53
    },
    {
      "epoch": 0.011335012594458438,
      "grad_norm": 0.6735419034957886,
      "learning_rate": 1.1320754716981132e-05,
      "loss": 0.8558,
      "step": 54
    },
    {
      "epoch": 0.011544920235096557,
      "grad_norm": 0.5273690223693848,
      "learning_rate": 1.1530398322851153e-05,
      "loss": 0.7531,
      "step": 55
    },
    {
      "epoch": 0.011754827875734676,
      "grad_norm": 0.612610936164856,
      "learning_rate": 1.1740041928721176e-05,
      "loss": 0.8009,
      "step": 56
    },
    {
      "epoch": 0.011964735516372796,
      "grad_norm": 0.5250133275985718,
      "learning_rate": 1.1949685534591196e-05,
      "loss": 0.7411,
      "step": 57
    },
    {
      "epoch": 0.012174643157010915,
      "grad_norm": 0.6210602521896362,
      "learning_rate": 1.2159329140461215e-05,
      "loss": 0.814,
      "step": 58
    },
    {
      "epoch": 0.012384550797649035,
      "grad_norm": 0.5767892003059387,
      "learning_rate": 1.2368972746331238e-05,
      "loss": 0.7512,
      "step": 59
    },
    {
      "epoch": 0.012594458438287154,
      "grad_norm": 0.5112836360931396,
      "learning_rate": 1.2578616352201259e-05,
      "loss": 0.7405,
      "step": 60
    },
    {
      "epoch": 0.012804366078925274,
      "grad_norm": 0.5214811563491821,
      "learning_rate": 1.2788259958071281e-05,
      "loss": 0.7152,
      "step": 61
    },
    {
      "epoch": 0.013014273719563391,
      "grad_norm": 0.5820367932319641,
      "learning_rate": 1.29979035639413e-05,
      "loss": 0.7359,
      "step": 62
    },
    {
      "epoch": 0.01322418136020151,
      "grad_norm": 0.5728296637535095,
      "learning_rate": 1.320754716981132e-05,
      "loss": 0.7096,
      "step": 63
    },
    {
      "epoch": 0.01343408900083963,
      "grad_norm": 0.514997124671936,
      "learning_rate": 1.3417190775681343e-05,
      "loss": 0.6945,
      "step": 64
    },
    {
      "epoch": 0.01364399664147775,
      "grad_norm": 0.5707572102546692,
      "learning_rate": 1.3626834381551362e-05,
      "loss": 0.7065,
      "step": 65
    },
    {
      "epoch": 0.013853904282115869,
      "grad_norm": 0.5674712657928467,
      "learning_rate": 1.3836477987421385e-05,
      "loss": 0.6759,
      "step": 66
    },
    {
      "epoch": 0.014063811922753989,
      "grad_norm": 0.5445975661277771,
      "learning_rate": 1.4046121593291406e-05,
      "loss": 0.6487,
      "step": 67
    },
    {
      "epoch": 0.014273719563392108,
      "grad_norm": 0.5629355311393738,
      "learning_rate": 1.4255765199161425e-05,
      "loss": 0.6635,
      "step": 68
    },
    {
      "epoch": 0.014483627204030227,
      "grad_norm": 0.47151610255241394,
      "learning_rate": 1.4465408805031447e-05,
      "loss": 0.5976,
      "step": 69
    },
    {
      "epoch": 0.014693534844668345,
      "grad_norm": 0.44633767008781433,
      "learning_rate": 1.467505241090147e-05,
      "loss": 0.5738,
      "step": 70
    },
    {
      "epoch": 0.014903442485306465,
      "grad_norm": 0.48507657647132874,
      "learning_rate": 1.488469601677149e-05,
      "loss": 0.5859,
      "step": 71
    },
    {
      "epoch": 0.015113350125944584,
      "grad_norm": 0.4147733151912689,
      "learning_rate": 1.509433962264151e-05,
      "loss": 0.5386,
      "step": 72
    },
    {
      "epoch": 0.015323257766582703,
      "grad_norm": 0.635608434677124,
      "learning_rate": 1.530398322851153e-05,
      "loss": 0.6529,
      "step": 73
    },
    {
      "epoch": 0.015533165407220823,
      "grad_norm": 0.5556919574737549,
      "learning_rate": 1.5513626834381552e-05,
      "loss": 0.5935,
      "step": 74
    },
    {
      "epoch": 0.015743073047858942,
      "grad_norm": 0.5627433657646179,
      "learning_rate": 1.572327044025157e-05,
      "loss": 0.5677,
      "step": 75
    },
    {
      "epoch": 0.01595298068849706,
      "grad_norm": 0.5727344155311584,
      "learning_rate": 1.5932914046121594e-05,
      "loss": 0.5701,
      "step": 76
    },
    {
      "epoch": 0.01616288832913518,
      "grad_norm": 0.5192092657089233,
      "learning_rate": 1.6142557651991616e-05,
      "loss": 0.537,
      "step": 77
    },
    {
      "epoch": 0.0163727959697733,
      "grad_norm": 0.6583610773086548,
      "learning_rate": 1.6352201257861635e-05,
      "loss": 0.5718,
      "step": 78
    },
    {
      "epoch": 0.01658270361041142,
      "grad_norm": 0.4762994050979614,
      "learning_rate": 1.6561844863731658e-05,
      "loss": 0.5016,
      "step": 79
    },
    {
      "epoch": 0.016792611251049538,
      "grad_norm": 0.7903013825416565,
      "learning_rate": 1.677148846960168e-05,
      "loss": 0.6123,
      "step": 80
    },
    {
      "epoch": 0.01700251889168766,
      "grad_norm": 0.6027877330780029,
      "learning_rate": 1.69811320754717e-05,
      "loss": 0.5085,
      "step": 81
    },
    {
      "epoch": 0.017212426532325777,
      "grad_norm": 0.6400225162506104,
      "learning_rate": 1.719077568134172e-05,
      "loss": 0.5325,
      "step": 82
    },
    {
      "epoch": 0.017422334172963894,
      "grad_norm": 0.5193424224853516,
      "learning_rate": 1.740041928721174e-05,
      "loss": 0.476,
      "step": 83
    },
    {
      "epoch": 0.017632241813602016,
      "grad_norm": 0.5318325757980347,
      "learning_rate": 1.761006289308176e-05,
      "loss": 0.4574,
      "step": 84
    },
    {
      "epoch": 0.017842149454240133,
      "grad_norm": 0.5530166029930115,
      "learning_rate": 1.7819706498951782e-05,
      "loss": 0.4331,
      "step": 85
    },
    {
      "epoch": 0.018052057094878254,
      "grad_norm": 0.5483909845352173,
      "learning_rate": 1.8029350104821805e-05,
      "loss": 0.4254,
      "step": 86
    },
    {
      "epoch": 0.018261964735516372,
      "grad_norm": 0.4599871039390564,
      "learning_rate": 1.8238993710691824e-05,
      "loss": 0.3828,
      "step": 87
    },
    {
      "epoch": 0.018471872376154493,
      "grad_norm": 0.47489672899246216,
      "learning_rate": 1.8448637316561846e-05,
      "loss": 0.3989,
      "step": 88
    },
    {
      "epoch": 0.01868178001679261,
      "grad_norm": 0.6532279253005981,
      "learning_rate": 1.865828092243187e-05,
      "loss": 0.3968,
      "step": 89
    },
    {
      "epoch": 0.018891687657430732,
      "grad_norm": 0.707245945930481,
      "learning_rate": 1.8867924528301888e-05,
      "loss": 0.4099,
      "step": 90
    },
    {
      "epoch": 0.01910159529806885,
      "grad_norm": 0.4290582835674286,
      "learning_rate": 1.9077568134171907e-05,
      "loss": 0.3546,
      "step": 91
    },
    {
      "epoch": 0.019311502938706968,
      "grad_norm": 0.4370197057723999,
      "learning_rate": 1.928721174004193e-05,
      "loss": 0.3519,
      "step": 92
    },
    {
      "epoch": 0.01952141057934509,
      "grad_norm": 0.4862878918647766,
      "learning_rate": 1.9496855345911952e-05,
      "loss": 0.3494,
      "step": 93
    },
    {
      "epoch": 0.019731318219983206,
      "grad_norm": 0.514576256275177,
      "learning_rate": 1.970649895178197e-05,
      "loss": 0.326,
      "step": 94
    },
    {
      "epoch": 0.019941225860621328,
      "grad_norm": 0.3785364031791687,
      "learning_rate": 1.9916142557651993e-05,
      "loss": 0.3196,
      "step": 95
    },
    {
      "epoch": 0.020151133501259445,
      "grad_norm": 0.4459572732448578,
      "learning_rate": 2.0125786163522016e-05,
      "loss": 0.3245,
      "step": 96
    },
    {
      "epoch": 0.020361041141897566,
      "grad_norm": 0.3634876310825348,
      "learning_rate": 2.0335429769392035e-05,
      "loss": 0.2999,
      "step": 97
    },
    {
      "epoch": 0.020570948782535684,
      "grad_norm": 0.39789989590644836,
      "learning_rate": 2.0545073375262054e-05,
      "loss": 0.2908,
      "step": 98
    },
    {
      "epoch": 0.020780856423173802,
      "grad_norm": 0.3628767430782318,
      "learning_rate": 2.0754716981132076e-05,
      "loss": 0.278,
      "step": 99
    },
    {
      "epoch": 0.020990764063811923,
      "grad_norm": 0.3945654332637787,
      "learning_rate": 2.0964360587002095e-05,
      "loss": 0.2902,
      "step": 100
    },
    {
      "epoch": 0.02120067170445004,
      "grad_norm": 0.2995467185974121,
      "learning_rate": 2.1174004192872118e-05,
      "loss": 0.2748,
      "step": 101
    },
    {
      "epoch": 0.021410579345088162,
      "grad_norm": 0.2776371240615845,
      "learning_rate": 2.138364779874214e-05,
      "loss": 0.2772,
      "step": 102
    },
    {
      "epoch": 0.02162048698572628,
      "grad_norm": 0.292316734790802,
      "learning_rate": 2.159329140461216e-05,
      "loss": 0.2732,
      "step": 103
    },
    {
      "epoch": 0.0218303946263644,
      "grad_norm": 0.26235565543174744,
      "learning_rate": 2.1802935010482182e-05,
      "loss": 0.2592,
      "step": 104
    },
    {
      "epoch": 0.02204030226700252,
      "grad_norm": 0.2782291769981384,
      "learning_rate": 2.2012578616352204e-05,
      "loss": 0.2586,
      "step": 105
    },
    {
      "epoch": 0.02225020990764064,
      "grad_norm": 0.24781855940818787,
      "learning_rate": 2.2222222222222223e-05,
      "loss": 0.2571,
      "step": 106
    },
    {
      "epoch": 0.022460117548278757,
      "grad_norm": 0.3645837604999542,
      "learning_rate": 2.2431865828092242e-05,
      "loss": 0.247,
      "step": 107
    },
    {
      "epoch": 0.022670025188916875,
      "grad_norm": 0.4096992313861847,
      "learning_rate": 2.2641509433962265e-05,
      "loss": 0.2648,
      "step": 108
    },
    {
      "epoch": 0.022879932829554996,
      "grad_norm": 0.36801740527153015,
      "learning_rate": 2.2851153039832284e-05,
      "loss": 0.2677,
      "step": 109
    },
    {
      "epoch": 0.023089840470193114,
      "grad_norm": 0.221563920378685,
      "learning_rate": 2.3060796645702306e-05,
      "loss": 0.2544,
      "step": 110
    },
    {
      "epoch": 0.023299748110831235,
      "grad_norm": 0.17734010517597198,
      "learning_rate": 2.327044025157233e-05,
      "loss": 0.2458,
      "step": 111
    },
    {
      "epoch": 0.023509655751469353,
      "grad_norm": 0.23815220594406128,
      "learning_rate": 2.348008385744235e-05,
      "loss": 0.2249,
      "step": 112
    },
    {
      "epoch": 0.023719563392107474,
      "grad_norm": 0.24534103274345398,
      "learning_rate": 2.368972746331237e-05,
      "loss": 0.2283,
      "step": 113
    },
    {
      "epoch": 0.02392947103274559,
      "grad_norm": 0.17044654488563538,
      "learning_rate": 2.3899371069182393e-05,
      "loss": 0.2543,
      "step": 114
    },
    {
      "epoch": 0.02413937867338371,
      "grad_norm": 0.22411420941352844,
      "learning_rate": 2.4109014675052412e-05,
      "loss": 0.2537,
      "step": 115
    },
    {
      "epoch": 0.02434928631402183,
      "grad_norm": 0.17107880115509033,
      "learning_rate": 2.431865828092243e-05,
      "loss": 0.2477,
      "step": 116
    },
    {
      "epoch": 0.02455919395465995,
      "grad_norm": 0.19663883745670319,
      "learning_rate": 2.4528301886792453e-05,
      "loss": 0.2503,
      "step": 117
    },
    {
      "epoch": 0.02476910159529807,
      "grad_norm": 0.2755718231201172,
      "learning_rate": 2.4737945492662476e-05,
      "loss": 0.201,
      "step": 118
    },
    {
      "epoch": 0.024979009235936187,
      "grad_norm": 0.16535572707653046,
      "learning_rate": 2.4947589098532495e-05,
      "loss": 0.2364,
      "step": 119
    },
    {
      "epoch": 0.02518891687657431,
      "grad_norm": 0.15798090398311615,
      "learning_rate": 2.5157232704402517e-05,
      "loss": 0.214,
      "step": 120
    },
    {
      "epoch": 0.025398824517212426,
      "grad_norm": 0.1900860220193863,
      "learning_rate": 2.5366876310272536e-05,
      "loss": 0.2182,
      "step": 121
    },
    {
      "epoch": 0.025608732157850547,
      "grad_norm": 0.18855144083499908,
      "learning_rate": 2.5576519916142562e-05,
      "loss": 0.2123,
      "step": 122
    },
    {
      "epoch": 0.025818639798488665,
      "grad_norm": 0.1780049353837967,
      "learning_rate": 2.578616352201258e-05,
      "loss": 0.2197,
      "step": 123
    },
    {
      "epoch": 0.026028547439126783,
      "grad_norm": 0.27040061354637146,
      "learning_rate": 2.59958071278826e-05,
      "loss": 0.2395,
      "step": 124
    },
    {
      "epoch": 0.026238455079764904,
      "grad_norm": 0.1983417570590973,
      "learning_rate": 2.6205450733752623e-05,
      "loss": 0.2197,
      "step": 125
    },
    {
      "epoch": 0.02644836272040302,
      "grad_norm": 0.1723383665084839,
      "learning_rate": 2.641509433962264e-05,
      "loss": 0.2183,
      "step": 126
    },
    {
      "epoch": 0.026658270361041143,
      "grad_norm": 0.24477605521678925,
      "learning_rate": 2.662473794549266e-05,
      "loss": 0.2293,
      "step": 127
    },
    {
      "epoch": 0.02686817800167926,
      "grad_norm": 0.15110079944133759,
      "learning_rate": 2.6834381551362687e-05,
      "loss": 0.2245,
      "step": 128
    },
    {
      "epoch": 0.02707808564231738,
      "grad_norm": 0.22366492450237274,
      "learning_rate": 2.7044025157232706e-05,
      "loss": 0.2031,
      "step": 129
    },
    {
      "epoch": 0.0272879932829555,
      "grad_norm": 0.32891571521759033,
      "learning_rate": 2.7253668763102725e-05,
      "loss": 0.1821,
      "step": 130
    },
    {
      "epoch": 0.02749790092359362,
      "grad_norm": 0.20027081668376923,
      "learning_rate": 2.746331236897275e-05,
      "loss": 0.223,
      "step": 131
    },
    {
      "epoch": 0.027707808564231738,
      "grad_norm": 0.2269366830587387,
      "learning_rate": 2.767295597484277e-05,
      "loss": 0.2246,
      "step": 132
    },
    {
      "epoch": 0.027917716204869856,
      "grad_norm": 0.1355280727148056,
      "learning_rate": 2.788259958071279e-05,
      "loss": 0.2183,
      "step": 133
    },
    {
      "epoch": 0.028127623845507977,
      "grad_norm": 0.29291626811027527,
      "learning_rate": 2.809224318658281e-05,
      "loss": 0.2049,
      "step": 134
    },
    {
      "epoch": 0.028337531486146095,
      "grad_norm": 0.1778186410665512,
      "learning_rate": 2.830188679245283e-05,
      "loss": 0.2095,
      "step": 135
    },
    {
      "epoch": 0.028547439126784216,
      "grad_norm": 0.23263931274414062,
      "learning_rate": 2.851153039832285e-05,
      "loss": 0.2372,
      "step": 136
    },
    {
      "epoch": 0.028757346767422334,
      "grad_norm": 0.2121749222278595,
      "learning_rate": 2.8721174004192875e-05,
      "loss": 0.2107,
      "step": 137
    },
    {
      "epoch": 0.028967254408060455,
      "grad_norm": 0.19954991340637207,
      "learning_rate": 2.8930817610062894e-05,
      "loss": 0.2181,
      "step": 138
    },
    {
      "epoch": 0.029177162048698572,
      "grad_norm": 0.15431921184062958,
      "learning_rate": 2.9140461215932913e-05,
      "loss": 0.2198,
      "step": 139
    },
    {
      "epoch": 0.02938706968933669,
      "grad_norm": 0.17603729665279388,
      "learning_rate": 2.935010482180294e-05,
      "loss": 0.2339,
      "step": 140
    },
    {
      "epoch": 0.02959697732997481,
      "grad_norm": 0.1604471355676651,
      "learning_rate": 2.9559748427672958e-05,
      "loss": 0.2064,
      "step": 141
    },
    {
      "epoch": 0.02980688497061293,
      "grad_norm": 0.17169184982776642,
      "learning_rate": 2.976939203354298e-05,
      "loss": 0.1987,
      "step": 142
    },
    {
      "epoch": 0.03001679261125105,
      "grad_norm": 0.1285230815410614,
      "learning_rate": 2.9979035639413e-05,
      "loss": 0.2342,
      "step": 143
    },
    {
      "epoch": 0.030226700251889168,
      "grad_norm": 0.1755395084619522,
      "learning_rate": 3.018867924528302e-05,
      "loss": 0.222,
      "step": 144
    },
    {
      "epoch": 0.03043660789252729,
      "grad_norm": 0.15474987030029297,
      "learning_rate": 3.0398322851153044e-05,
      "loss": 0.2159,
      "step": 145
    },
    {
      "epoch": 0.030646515533165407,
      "grad_norm": 0.12986472249031067,
      "learning_rate": 3.060796645702306e-05,
      "loss": 0.2124,
      "step": 146
    },
    {
      "epoch": 0.030856423173803528,
      "grad_norm": 0.1458188146352768,
      "learning_rate": 3.081761006289308e-05,
      "loss": 0.214,
      "step": 147
    },
    {
      "epoch": 0.031066330814441646,
      "grad_norm": 0.1323792040348053,
      "learning_rate": 3.1027253668763105e-05,
      "loss": 0.2153,
      "step": 148
    },
    {
      "epoch": 0.03127623845507976,
      "grad_norm": 0.16542711853981018,
      "learning_rate": 3.1236897274633124e-05,
      "loss": 0.2154,
      "step": 149
    },
    {
      "epoch": 0.031486146095717885,
      "grad_norm": 0.17730407416820526,
      "learning_rate": 3.144654088050314e-05,
      "loss": 0.2202,
      "step": 150
    },
    {
      "epoch": 0.031696053736356006,
      "grad_norm": 0.15039502084255219,
      "learning_rate": 3.165618448637317e-05,
      "loss": 0.2056,
      "step": 151
    },
    {
      "epoch": 0.03190596137699412,
      "grad_norm": 0.20309150218963623,
      "learning_rate": 3.186582809224319e-05,
      "loss": 0.2175,
      "step": 152
    },
    {
      "epoch": 0.03211586901763224,
      "grad_norm": 0.16652604937553406,
      "learning_rate": 3.207547169811321e-05,
      "loss": 0.2076,
      "step": 153
    },
    {
      "epoch": 0.03232577665827036,
      "grad_norm": 0.14530467987060547,
      "learning_rate": 3.228511530398323e-05,
      "loss": 0.2167,
      "step": 154
    },
    {
      "epoch": 0.032535684298908484,
      "grad_norm": 0.13003528118133545,
      "learning_rate": 3.249475890985325e-05,
      "loss": 0.2089,
      "step": 155
    },
    {
      "epoch": 0.0327455919395466,
      "grad_norm": 0.16985855996608734,
      "learning_rate": 3.270440251572327e-05,
      "loss": 0.1994,
      "step": 156
    },
    {
      "epoch": 0.03295549958018472,
      "grad_norm": 0.18479777872562408,
      "learning_rate": 3.29140461215933e-05,
      "loss": 0.2127,
      "step": 157
    },
    {
      "epoch": 0.03316540722082284,
      "grad_norm": 0.1541491150856018,
      "learning_rate": 3.3123689727463316e-05,
      "loss": 0.2208,
      "step": 158
    },
    {
      "epoch": 0.033375314861460954,
      "grad_norm": 0.13511165976524353,
      "learning_rate": 3.3333333333333335e-05,
      "loss": 0.218,
      "step": 159
    },
    {
      "epoch": 0.033585222502099076,
      "grad_norm": 0.1392865628004074,
      "learning_rate": 3.354297693920336e-05,
      "loss": 0.226,
      "step": 160
    },
    {
      "epoch": 0.0337951301427372,
      "grad_norm": 0.1614847183227539,
      "learning_rate": 3.375262054507338e-05,
      "loss": 0.2072,
      "step": 161
    },
    {
      "epoch": 0.03400503778337532,
      "grad_norm": 0.12186679244041443,
      "learning_rate": 3.39622641509434e-05,
      "loss": 0.2193,
      "step": 162
    },
    {
      "epoch": 0.03421494542401343,
      "grad_norm": 0.1777278333902359,
      "learning_rate": 3.417190775681342e-05,
      "loss": 0.202,
      "step": 163
    },
    {
      "epoch": 0.03442485306465155,
      "grad_norm": 0.13323499262332916,
      "learning_rate": 3.438155136268344e-05,
      "loss": 0.1983,
      "step": 164
    },
    {
      "epoch": 0.034634760705289674,
      "grad_norm": 0.22301942110061646,
      "learning_rate": 3.4591194968553456e-05,
      "loss": 0.2252,
      "step": 165
    },
    {
      "epoch": 0.03484466834592779,
      "grad_norm": 0.1538372039794922,
      "learning_rate": 3.480083857442348e-05,
      "loss": 0.2071,
      "step": 166
    },
    {
      "epoch": 0.03505457598656591,
      "grad_norm": 0.128251314163208,
      "learning_rate": 3.50104821802935e-05,
      "loss": 0.214,
      "step": 167
    },
    {
      "epoch": 0.03526448362720403,
      "grad_norm": 0.26556313037872314,
      "learning_rate": 3.522012578616352e-05,
      "loss": 0.1863,
      "step": 168
    },
    {
      "epoch": 0.03547439126784215,
      "grad_norm": 0.14938302338123322,
      "learning_rate": 3.5429769392033546e-05,
      "loss": 0.2152,
      "step": 169
    },
    {
      "epoch": 0.035684298908480266,
      "grad_norm": 0.19645771384239197,
      "learning_rate": 3.5639412997903565e-05,
      "loss": 0.1977,
      "step": 170
    },
    {
      "epoch": 0.03589420654911839,
      "grad_norm": 0.1725340038537979,
      "learning_rate": 3.5849056603773584e-05,
      "loss": 0.1981,
      "step": 171
    },
    {
      "epoch": 0.03610411418975651,
      "grad_norm": 0.16048069298267365,
      "learning_rate": 3.605870020964361e-05,
      "loss": 0.1983,
      "step": 172
    },
    {
      "epoch": 0.03631402183039462,
      "grad_norm": 0.21855585277080536,
      "learning_rate": 3.626834381551363e-05,
      "loss": 0.2321,
      "step": 173
    },
    {
      "epoch": 0.036523929471032744,
      "grad_norm": 0.1136787012219429,
      "learning_rate": 3.647798742138365e-05,
      "loss": 0.1973,
      "step": 174
    },
    {
      "epoch": 0.036733837111670865,
      "grad_norm": 0.15145623683929443,
      "learning_rate": 3.6687631027253674e-05,
      "loss": 0.1947,
      "step": 175
    },
    {
      "epoch": 0.036943744752308987,
      "grad_norm": 0.21631890535354614,
      "learning_rate": 3.689727463312369e-05,
      "loss": 0.2151,
      "step": 176
    },
    {
      "epoch": 0.0371536523929471,
      "grad_norm": 0.2623152434825897,
      "learning_rate": 3.710691823899371e-05,
      "loss": 0.1814,
      "step": 177
    },
    {
      "epoch": 0.03736356003358522,
      "grad_norm": 0.1753605455160141,
      "learning_rate": 3.731656184486374e-05,
      "loss": 0.203,
      "step": 178
    },
    {
      "epoch": 0.03757346767422334,
      "grad_norm": 0.10878176242113113,
      "learning_rate": 3.752620545073376e-05,
      "loss": 0.2052,
      "step": 179
    },
    {
      "epoch": 0.037783375314861464,
      "grad_norm": 0.13699688017368317,
      "learning_rate": 3.7735849056603776e-05,
      "loss": 0.2004,
      "step": 180
    },
    {
      "epoch": 0.03799328295549958,
      "grad_norm": 0.14288806915283203,
      "learning_rate": 3.7945492662473795e-05,
      "loss": 0.1745,
      "step": 181
    },
    {
      "epoch": 0.0382031905961377,
      "grad_norm": 0.12457548081874847,
      "learning_rate": 3.8155136268343814e-05,
      "loss": 0.194,
      "step": 182
    },
    {
      "epoch": 0.03841309823677582,
      "grad_norm": 0.167145237326622,
      "learning_rate": 3.836477987421384e-05,
      "loss": 0.2027,
      "step": 183
    },
    {
      "epoch": 0.038623005877413935,
      "grad_norm": 0.12857979536056519,
      "learning_rate": 3.857442348008386e-05,
      "loss": 0.2126,
      "step": 184
    },
    {
      "epoch": 0.038832913518052056,
      "grad_norm": 0.16190126538276672,
      "learning_rate": 3.878406708595388e-05,
      "loss": 0.2037,
      "step": 185
    },
    {
      "epoch": 0.03904282115869018,
      "grad_norm": 0.168744757771492,
      "learning_rate": 3.8993710691823904e-05,
      "loss": 0.2108,
      "step": 186
    },
    {
      "epoch": 0.0392527287993283,
      "grad_norm": 0.1676539033651352,
      "learning_rate": 3.920335429769392e-05,
      "loss": 0.216,
      "step": 187
    },
    {
      "epoch": 0.03946263643996641,
      "grad_norm": 0.13556820154190063,
      "learning_rate": 3.941299790356394e-05,
      "loss": 0.2085,
      "step": 188
    },
    {
      "epoch": 0.039672544080604534,
      "grad_norm": 0.1797979772090912,
      "learning_rate": 3.962264150943397e-05,
      "loss": 0.1933,
      "step": 189
    },
    {
      "epoch": 0.039882451721242655,
      "grad_norm": 0.20826327800750732,
      "learning_rate": 3.983228511530399e-05,
      "loss": 0.2214,
      "step": 190
    },
    {
      "epoch": 0.04009235936188077,
      "grad_norm": 0.19972363114356995,
      "learning_rate": 4.0041928721174006e-05,
      "loss": 0.1941,
      "step": 191
    },
    {
      "epoch": 0.04030226700251889,
      "grad_norm": 0.149556502699852,
      "learning_rate": 4.025157232704403e-05,
      "loss": 0.1956,
      "step": 192
    },
    {
      "epoch": 0.04051217464315701,
      "grad_norm": 0.22496013343334198,
      "learning_rate": 4.046121593291405e-05,
      "loss": 0.1973,
      "step": 193
    },
    {
      "epoch": 0.04072208228379513,
      "grad_norm": 0.16132576763629913,
      "learning_rate": 4.067085953878407e-05,
      "loss": 0.203,
      "step": 194
    },
    {
      "epoch": 0.04093198992443325,
      "grad_norm": 0.17156128585338593,
      "learning_rate": 4.088050314465409e-05,
      "loss": 0.1942,
      "step": 195
    },
    {
      "epoch": 0.04114189756507137,
      "grad_norm": 0.14846180379390717,
      "learning_rate": 4.109014675052411e-05,
      "loss": 0.2005,
      "step": 196
    },
    {
      "epoch": 0.04135180520570949,
      "grad_norm": 0.20252752304077148,
      "learning_rate": 4.129979035639413e-05,
      "loss": 0.186,
      "step": 197
    },
    {
      "epoch": 0.041561712846347604,
      "grad_norm": 0.16286462545394897,
      "learning_rate": 4.150943396226415e-05,
      "loss": 0.1997,
      "step": 198
    },
    {
      "epoch": 0.041771620486985725,
      "grad_norm": 0.13008786737918854,
      "learning_rate": 4.171907756813417e-05,
      "loss": 0.2052,
      "step": 199
    },
    {
      "epoch": 0.041981528127623846,
      "grad_norm": 0.13853180408477783,
      "learning_rate": 4.192872117400419e-05,
      "loss": 0.1907,
      "step": 200
    },
    {
      "epoch": 0.04219143576826197,
      "grad_norm": 0.20382314920425415,
      "learning_rate": 4.213836477987422e-05,
      "loss": 0.1931,
      "step": 201
    },
    {
      "epoch": 0.04240134340890008,
      "grad_norm": 0.23206844925880432,
      "learning_rate": 4.2348008385744236e-05,
      "loss": 0.1849,
      "step": 202
    },
    {
      "epoch": 0.0426112510495382,
      "grad_norm": 0.2456827312707901,
      "learning_rate": 4.2557651991614255e-05,
      "loss": 0.2088,
      "step": 203
    },
    {
      "epoch": 0.042821158690176324,
      "grad_norm": 0.15247821807861328,
      "learning_rate": 4.276729559748428e-05,
      "loss": 0.1944,
      "step": 204
    },
    {
      "epoch": 0.043031066330814445,
      "grad_norm": 0.174981027841568,
      "learning_rate": 4.29769392033543e-05,
      "loss": 0.2161,
      "step": 205
    },
    {
      "epoch": 0.04324097397145256,
      "grad_norm": 0.20193031430244446,
      "learning_rate": 4.318658280922432e-05,
      "loss": 0.2084,
      "step": 206
    },
    {
      "epoch": 0.04345088161209068,
      "grad_norm": 0.20125791430473328,
      "learning_rate": 4.3396226415094345e-05,
      "loss": 0.189,
      "step": 207
    },
    {
      "epoch": 0.0436607892527288,
      "grad_norm": 0.16958673298358917,
      "learning_rate": 4.3605870020964364e-05,
      "loss": 0.1952,
      "step": 208
    },
    {
      "epoch": 0.043870696893366916,
      "grad_norm": 0.20714177191257477,
      "learning_rate": 4.381551362683438e-05,
      "loss": 0.1832,
      "step": 209
    },
    {
      "epoch": 0.04408060453400504,
      "grad_norm": 0.14229562878608704,
      "learning_rate": 4.402515723270441e-05,
      "loss": 0.1806,
      "step": 210
    },
    {
      "epoch": 0.04429051217464316,
      "grad_norm": 0.1985626220703125,
      "learning_rate": 4.423480083857443e-05,
      "loss": 0.1973,
      "step": 211
    },
    {
      "epoch": 0.04450041981528128,
      "grad_norm": 0.13714846968650818,
      "learning_rate": 4.4444444444444447e-05,
      "loss": 0.2018,
      "step": 212
    },
    {
      "epoch": 0.044710327455919394,
      "grad_norm": 0.25591611862182617,
      "learning_rate": 4.4654088050314466e-05,
      "loss": 0.2001,
      "step": 213
    },
    {
      "epoch": 0.044920235096557515,
      "grad_norm": 0.29017260670661926,
      "learning_rate": 4.4863731656184485e-05,
      "loss": 0.2278,
      "step": 214
    },
    {
      "epoch": 0.045130142737195636,
      "grad_norm": 0.12594066560268402,
      "learning_rate": 4.5073375262054504e-05,
      "loss": 0.2023,
      "step": 215
    },
    {
      "epoch": 0.04534005037783375,
      "grad_norm": 0.1602821797132492,
      "learning_rate": 4.528301886792453e-05,
      "loss": 0.2101,
      "step": 216
    },
    {
      "epoch": 0.04554995801847187,
      "grad_norm": 0.36664336919784546,
      "learning_rate": 4.549266247379455e-05,
      "loss": 0.1665,
      "step": 217
    },
    {
      "epoch": 0.04575986565910999,
      "grad_norm": 0.20665952563285828,
      "learning_rate": 4.570230607966457e-05,
      "loss": 0.1965,
      "step": 218
    },
    {
      "epoch": 0.045969773299748114,
      "grad_norm": 0.15664128959178925,
      "learning_rate": 4.5911949685534594e-05,
      "loss": 0.2104,
      "step": 219
    },
    {
      "epoch": 0.04617968094038623,
      "grad_norm": 0.14733830094337463,
      "learning_rate": 4.612159329140461e-05,
      "loss": 0.1955,
      "step": 220
    },
    {
      "epoch": 0.04638958858102435,
      "grad_norm": 0.19135522842407227,
      "learning_rate": 4.633123689727464e-05,
      "loss": 0.1947,
      "step": 221
    },
    {
      "epoch": 0.04659949622166247,
      "grad_norm": 0.24508413672447205,
      "learning_rate": 4.654088050314466e-05,
      "loss": 0.2005,
      "step": 222
    },
    {
      "epoch": 0.046809403862300585,
      "grad_norm": 0.16794531047344208,
      "learning_rate": 4.6750524109014677e-05,
      "loss": 0.1958,
      "step": 223
    },
    {
      "epoch": 0.047019311502938706,
      "grad_norm": 0.1870536506175995,
      "learning_rate": 4.69601677148847e-05,
      "loss": 0.1868,
      "step": 224
    },
    {
      "epoch": 0.04722921914357683,
      "grad_norm": 0.16214439272880554,
      "learning_rate": 4.716981132075472e-05,
      "loss": 0.2003,
      "step": 225
    },
    {
      "epoch": 0.04743912678421495,
      "grad_norm": 0.24978841841220856,
      "learning_rate": 4.737945492662474e-05,
      "loss": 0.1909,
      "step": 226
    },
    {
      "epoch": 0.04764903442485306,
      "grad_norm": 0.163265198469162,
      "learning_rate": 4.7589098532494766e-05,
      "loss": 0.2056,
      "step": 227
    },
    {
      "epoch": 0.04785894206549118,
      "grad_norm": 0.24885287880897522,
      "learning_rate": 4.7798742138364785e-05,
      "loss": 0.2035,
      "step": 228
    },
    {
      "epoch": 0.048068849706129305,
      "grad_norm": 0.1393681764602661,
      "learning_rate": 4.8008385744234804e-05,
      "loss": 0.1976,
      "step": 229
    },
    {
      "epoch": 0.04827875734676742,
      "grad_norm": 0.17042241990566254,
      "learning_rate": 4.8218029350104823e-05,
      "loss": 0.1824,
      "step": 230
    },
    {
      "epoch": 0.04848866498740554,
      "grad_norm": 0.1625502109527588,
      "learning_rate": 4.842767295597484e-05,
      "loss": 0.1904,
      "step": 231
    },
    {
      "epoch": 0.04869857262804366,
      "grad_norm": 0.14994169771671295,
      "learning_rate": 4.863731656184486e-05,
      "loss": 0.1926,
      "step": 232
    },
    {
      "epoch": 0.04890848026868178,
      "grad_norm": 0.15602821111679077,
      "learning_rate": 4.884696016771489e-05,
      "loss": 0.1828,
      "step": 233
    },
    {
      "epoch": 0.0491183879093199,
      "grad_norm": 0.13405688107013702,
      "learning_rate": 4.9056603773584906e-05,
      "loss": 0.1989,
      "step": 234
    },
    {
      "epoch": 0.04932829554995802,
      "grad_norm": 0.18559689819812775,
      "learning_rate": 4.9266247379454926e-05,
      "loss": 0.2131,
      "step": 235
    },
    {
      "epoch": 0.04953820319059614,
      "grad_norm": 0.1557319462299347,
      "learning_rate": 4.947589098532495e-05,
      "loss": 0.19,
      "step": 236
    },
    {
      "epoch": 0.04974811083123426,
      "grad_norm": 0.2162303477525711,
      "learning_rate": 4.968553459119497e-05,
      "loss": 0.196,
      "step": 237
    },
    {
      "epoch": 0.049958018471872374,
      "grad_norm": 0.17403477430343628,
      "learning_rate": 4.989517819706499e-05,
      "loss": 0.1997,
      "step": 238
    },
    {
      "epoch": 0.050167926112510496,
      "grad_norm": 0.11738390475511551,
      "learning_rate": 5.010482180293501e-05,
      "loss": 0.1955,
      "step": 239
    },
    {
      "epoch": 0.05037783375314862,
      "grad_norm": 0.15942999720573425,
      "learning_rate": 5.0314465408805034e-05,
      "loss": 0.1899,
      "step": 240
    },
    {
      "epoch": 0.05058774139378673,
      "grad_norm": 0.14695511758327484,
      "learning_rate": 5.052410901467506e-05,
      "loss": 0.2032,
      "step": 241
    },
    {
      "epoch": 0.05079764903442485,
      "grad_norm": 0.16291062533855438,
      "learning_rate": 5.073375262054507e-05,
      "loss": 0.1929,
      "step": 242
    },
    {
      "epoch": 0.05100755667506297,
      "grad_norm": 0.18514905869960785,
      "learning_rate": 5.09433962264151e-05,
      "loss": 0.1899,
      "step": 243
    },
    {
      "epoch": 0.051217464315701094,
      "grad_norm": 0.2196233868598938,
      "learning_rate": 5.1153039832285124e-05,
      "loss": 0.1741,
      "step": 244
    },
    {
      "epoch": 0.05142737195633921,
      "grad_norm": 0.19183433055877686,
      "learning_rate": 5.1362683438155136e-05,
      "loss": 0.1985,
      "step": 245
    },
    {
      "epoch": 0.05163727959697733,
      "grad_norm": 0.1604142189025879,
      "learning_rate": 5.157232704402516e-05,
      "loss": 0.1865,
      "step": 246
    },
    {
      "epoch": 0.05184718723761545,
      "grad_norm": 0.18311725556850433,
      "learning_rate": 5.178197064989518e-05,
      "loss": 0.2005,
      "step": 247
    },
    {
      "epoch": 0.052057094878253565,
      "grad_norm": 0.21732251346111298,
      "learning_rate": 5.19916142557652e-05,
      "loss": 0.2032,
      "step": 248
    },
    {
      "epoch": 0.052267002518891686,
      "grad_norm": 0.2600694000720978,
      "learning_rate": 5.220125786163522e-05,
      "loss": 0.1838,
      "step": 249
    },
    {
      "epoch": 0.05247691015952981,
      "grad_norm": 0.18634290993213654,
      "learning_rate": 5.2410901467505245e-05,
      "loss": 0.1809,
      "step": 250
    },
    {
      "epoch": 0.05268681780016793,
      "grad_norm": 0.14735780656337738,
      "learning_rate": 5.262054507337526e-05,
      "loss": 0.1981,
      "step": 251
    },
    {
      "epoch": 0.05289672544080604,
      "grad_norm": 0.16118381917476654,
      "learning_rate": 5.283018867924528e-05,
      "loss": 0.1836,
      "step": 252
    },
    {
      "epoch": 0.053106633081444164,
      "grad_norm": 0.17707999050617218,
      "learning_rate": 5.303983228511531e-05,
      "loss": 0.2044,
      "step": 253
    },
    {
      "epoch": 0.053316540722082285,
      "grad_norm": 0.25897523760795593,
      "learning_rate": 5.324947589098532e-05,
      "loss": 0.1753,
      "step": 254
    },
    {
      "epoch": 0.0535264483627204,
      "grad_norm": 0.1371389776468277,
      "learning_rate": 5.345911949685535e-05,
      "loss": 0.1839,
      "step": 255
    },
    {
      "epoch": 0.05373635600335852,
      "grad_norm": 0.21849682927131653,
      "learning_rate": 5.366876310272537e-05,
      "loss": 0.1938,
      "step": 256
    },
    {
      "epoch": 0.05394626364399664,
      "grad_norm": 0.16861748695373535,
      "learning_rate": 5.3878406708595385e-05,
      "loss": 0.1751,
      "step": 257
    },
    {
      "epoch": 0.05415617128463476,
      "grad_norm": 0.19400931894779205,
      "learning_rate": 5.408805031446541e-05,
      "loss": 0.2117,
      "step": 258
    },
    {
      "epoch": 0.05436607892527288,
      "grad_norm": 0.18074113130569458,
      "learning_rate": 5.429769392033544e-05,
      "loss": 0.1949,
      "step": 259
    },
    {
      "epoch": 0.054575986565911,
      "grad_norm": 0.17707990109920502,
      "learning_rate": 5.450733752620545e-05,
      "loss": 0.199,
      "step": 260
    },
    {
      "epoch": 0.05478589420654912,
      "grad_norm": 0.16568966209888458,
      "learning_rate": 5.4716981132075475e-05,
      "loss": 0.2027,
      "step": 261
    },
    {
      "epoch": 0.05499580184718724,
      "grad_norm": 0.24486149847507477,
      "learning_rate": 5.49266247379455e-05,
      "loss": 0.2042,
      "step": 262
    },
    {
      "epoch": 0.055205709487825355,
      "grad_norm": 0.20431551337242126,
      "learning_rate": 5.513626834381551e-05,
      "loss": 0.2013,
      "step": 263
    },
    {
      "epoch": 0.055415617128463476,
      "grad_norm": 0.21347559988498688,
      "learning_rate": 5.534591194968554e-05,
      "loss": 0.1687,
      "step": 264
    },
    {
      "epoch": 0.0556255247691016,
      "grad_norm": 0.22354485094547272,
      "learning_rate": 5.555555555555556e-05,
      "loss": 0.1819,
      "step": 265
    },
    {
      "epoch": 0.05583543240973971,
      "grad_norm": 0.2891826033592224,
      "learning_rate": 5.576519916142558e-05,
      "loss": 0.1927,
      "step": 266
    },
    {
      "epoch": 0.05604534005037783,
      "grad_norm": 0.2465565800666809,
      "learning_rate": 5.5974842767295596e-05,
      "loss": 0.1852,
      "step": 267
    },
    {
      "epoch": 0.056255247691015954,
      "grad_norm": 0.1743604689836502,
      "learning_rate": 5.618448637316562e-05,
      "loss": 0.1722,
      "step": 268
    },
    {
      "epoch": 0.056465155331654075,
      "grad_norm": 0.2461417019367218,
      "learning_rate": 5.6394129979035634e-05,
      "loss": 0.1816,
      "step": 269
    },
    {
      "epoch": 0.05667506297229219,
      "grad_norm": 0.21032604575157166,
      "learning_rate": 5.660377358490566e-05,
      "loss": 0.182,
      "step": 270
    },
    {
      "epoch": 0.05688497061293031,
      "grad_norm": 0.2683754861354828,
      "learning_rate": 5.6813417190775686e-05,
      "loss": 0.1718,
      "step": 271
    },
    {
      "epoch": 0.05709487825356843,
      "grad_norm": 0.18455228209495544,
      "learning_rate": 5.70230607966457e-05,
      "loss": 0.1954,
      "step": 272
    },
    {
      "epoch": 0.057304785894206546,
      "grad_norm": 0.22255247831344604,
      "learning_rate": 5.7232704402515724e-05,
      "loss": 0.1922,
      "step": 273
    },
    {
      "epoch": 0.05751469353484467,
      "grad_norm": 0.22789119184017181,
      "learning_rate": 5.744234800838575e-05,
      "loss": 0.1882,
      "step": 274
    },
    {
      "epoch": 0.05772460117548279,
      "grad_norm": 0.20874802768230438,
      "learning_rate": 5.765199161425576e-05,
      "loss": 0.1859,
      "step": 275
    },
    {
      "epoch": 0.05793450881612091,
      "grad_norm": 0.17554089426994324,
      "learning_rate": 5.786163522012579e-05,
      "loss": 0.1793,
      "step": 276
    },
    {
      "epoch": 0.058144416456759024,
      "grad_norm": 0.2011173665523529,
      "learning_rate": 5.8071278825995814e-05,
      "loss": 0.2009,
      "step": 277
    },
    {
      "epoch": 0.058354324097397145,
      "grad_norm": 0.28261420130729675,
      "learning_rate": 5.8280922431865826e-05,
      "loss": 0.2002,
      "step": 278
    },
    {
      "epoch": 0.058564231738035266,
      "grad_norm": 0.2356766015291214,
      "learning_rate": 5.849056603773585e-05,
      "loss": 0.2095,
      "step": 279
    },
    {
      "epoch": 0.05877413937867338,
      "grad_norm": 0.15072722733020782,
      "learning_rate": 5.870020964360588e-05,
      "loss": 0.1781,
      "step": 280
    },
    {
      "epoch": 0.0589840470193115,
      "grad_norm": 0.1446981132030487,
      "learning_rate": 5.89098532494759e-05,
      "loss": 0.1813,
      "step": 281
    },
    {
      "epoch": 0.05919395465994962,
      "grad_norm": 0.16516901552677155,
      "learning_rate": 5.9119496855345916e-05,
      "loss": 0.1946,
      "step": 282
    },
    {
      "epoch": 0.059403862300587744,
      "grad_norm": 0.16732774674892426,
      "learning_rate": 5.9329140461215935e-05,
      "loss": 0.1813,
      "step": 283
    },
    {
      "epoch": 0.05961376994122586,
      "grad_norm": 0.2000836730003357,
      "learning_rate": 5.953878406708596e-05,
      "loss": 0.1918,
      "step": 284
    },
    {
      "epoch": 0.05982367758186398,
      "grad_norm": 0.15576116740703583,
      "learning_rate": 5.974842767295597e-05,
      "loss": 0.1814,
      "step": 285
    },
    {
      "epoch": 0.0600335852225021,
      "grad_norm": 0.18421867489814758,
      "learning_rate": 5.9958071278826e-05,
      "loss": 0.1745,
      "step": 286
    },
    {
      "epoch": 0.06024349286314022,
      "grad_norm": 0.2663988471031189,
      "learning_rate": 6.0167714884696025e-05,
      "loss": 0.1889,
      "step": 287
    },
    {
      "epoch": 0.060453400503778336,
      "grad_norm": 0.18096649646759033,
      "learning_rate": 6.037735849056604e-05,
      "loss": 0.191,
      "step": 288
    },
    {
      "epoch": 0.06066330814441646,
      "grad_norm": 0.24025796353816986,
      "learning_rate": 6.058700209643606e-05,
      "loss": 0.1907,
      "step": 289
    },
    {
      "epoch": 0.06087321578505458,
      "grad_norm": 0.1682557910680771,
      "learning_rate": 6.079664570230609e-05,
      "loss": 0.1747,
      "step": 290
    },
    {
      "epoch": 0.06108312342569269,
      "grad_norm": 0.2801767587661743,
      "learning_rate": 6.10062893081761e-05,
      "loss": 0.1813,
      "step": 291
    },
    {
      "epoch": 0.061293031066330814,
      "grad_norm": 0.2288123071193695,
      "learning_rate": 6.121593291404612e-05,
      "loss": 0.1681,
      "step": 292
    },
    {
      "epoch": 0.061502938706968935,
      "grad_norm": 0.14664186537265778,
      "learning_rate": 6.142557651991615e-05,
      "loss": 0.1723,
      "step": 293
    },
    {
      "epoch": 0.061712846347607056,
      "grad_norm": 0.15858514606952667,
      "learning_rate": 6.163522012578616e-05,
      "loss": 0.198,
      "step": 294
    },
    {
      "epoch": 0.06192275398824517,
      "grad_norm": 0.24689258635044098,
      "learning_rate": 6.184486373165618e-05,
      "loss": 0.1865,
      "step": 295
    },
    {
      "epoch": 0.06213266162888329,
      "grad_norm": 0.19141900539398193,
      "learning_rate": 6.205450733752621e-05,
      "loss": 0.2103,
      "step": 296
    },
    {
      "epoch": 0.06234256926952141,
      "grad_norm": 0.22445173561573029,
      "learning_rate": 6.226415094339622e-05,
      "loss": 0.1778,
      "step": 297
    },
    {
      "epoch": 0.06255247691015953,
      "grad_norm": 0.1805533766746521,
      "learning_rate": 6.247379454926625e-05,
      "loss": 0.1752,
      "step": 298
    },
    {
      "epoch": 0.06276238455079765,
      "grad_norm": 0.17849349975585938,
      "learning_rate": 6.268343815513627e-05,
      "loss": 0.1735,
      "step": 299
    },
    {
      "epoch": 0.06297229219143577,
      "grad_norm": 0.2601464092731476,
      "learning_rate": 6.289308176100629e-05,
      "loss": 0.1868,
      "step": 300
    },
    {
      "epoch": 0.06318219983207389,
      "grad_norm": 0.2865089178085327,
      "learning_rate": 6.310272536687631e-05,
      "loss": 0.2058,
      "step": 301
    },
    {
      "epoch": 0.06339210747271201,
      "grad_norm": 0.1764407902956009,
      "learning_rate": 6.331236897274634e-05,
      "loss": 0.1828,
      "step": 302
    },
    {
      "epoch": 0.06360201511335013,
      "grad_norm": 0.136027991771698,
      "learning_rate": 6.352201257861635e-05,
      "loss": 0.1841,
      "step": 303
    },
    {
      "epoch": 0.06381192275398824,
      "grad_norm": 0.26960527896881104,
      "learning_rate": 6.373165618448638e-05,
      "loss": 0.1806,
      "step": 304
    },
    {
      "epoch": 0.06402183039462636,
      "grad_norm": 0.2371356189250946,
      "learning_rate": 6.39412997903564e-05,
      "loss": 0.1755,
      "step": 305
    },
    {
      "epoch": 0.06423173803526448,
      "grad_norm": 0.16067345440387726,
      "learning_rate": 6.415094339622641e-05,
      "loss": 0.1899,
      "step": 306
    },
    {
      "epoch": 0.0644416456759026,
      "grad_norm": 0.1733190417289734,
      "learning_rate": 6.436058700209644e-05,
      "loss": 0.1829,
      "step": 307
    },
    {
      "epoch": 0.06465155331654072,
      "grad_norm": 0.19170600175857544,
      "learning_rate": 6.457023060796647e-05,
      "loss": 0.1777,
      "step": 308
    },
    {
      "epoch": 0.06486146095717885,
      "grad_norm": 0.17290905117988586,
      "learning_rate": 6.477987421383648e-05,
      "loss": 0.187,
      "step": 309
    },
    {
      "epoch": 0.06507136859781697,
      "grad_norm": 0.3149113059043884,
      "learning_rate": 6.49895178197065e-05,
      "loss": 0.2037,
      "step": 310
    },
    {
      "epoch": 0.06528127623845507,
      "grad_norm": 0.15404744446277618,
      "learning_rate": 6.519916142557653e-05,
      "loss": 0.1909,
      "step": 311
    },
    {
      "epoch": 0.0654911838790932,
      "grad_norm": 0.157347172498703,
      "learning_rate": 6.540880503144654e-05,
      "loss": 0.191,
      "step": 312
    },
    {
      "epoch": 0.06570109151973132,
      "grad_norm": 0.23342733085155487,
      "learning_rate": 6.561844863731657e-05,
      "loss": 0.1867,
      "step": 313
    },
    {
      "epoch": 0.06591099916036944,
      "grad_norm": 0.2597595155239105,
      "learning_rate": 6.58280922431866e-05,
      "loss": 0.1949,
      "step": 314
    },
    {
      "epoch": 0.06612090680100756,
      "grad_norm": 0.23665842413902283,
      "learning_rate": 6.60377358490566e-05,
      "loss": 0.1898,
      "step": 315
    },
    {
      "epoch": 0.06633081444164568,
      "grad_norm": 0.18460237979888916,
      "learning_rate": 6.624737945492663e-05,
      "loss": 0.1824,
      "step": 316
    },
    {
      "epoch": 0.0665407220822838,
      "grad_norm": 0.1677280068397522,
      "learning_rate": 6.645702306079666e-05,
      "loss": 0.1871,
      "step": 317
    },
    {
      "epoch": 0.06675062972292191,
      "grad_norm": 0.1769377440214157,
      "learning_rate": 6.666666666666667e-05,
      "loss": 0.1779,
      "step": 318
    },
    {
      "epoch": 0.06696053736356003,
      "grad_norm": 0.1881011724472046,
      "learning_rate": 6.68763102725367e-05,
      "loss": 0.1711,
      "step": 319
    },
    {
      "epoch": 0.06717044500419815,
      "grad_norm": 0.19110549986362457,
      "learning_rate": 6.708595387840672e-05,
      "loss": 0.1823,
      "step": 320
    },
    {
      "epoch": 0.06738035264483627,
      "grad_norm": 0.26796162128448486,
      "learning_rate": 6.729559748427673e-05,
      "loss": 0.1863,
      "step": 321
    },
    {
      "epoch": 0.0675902602854744,
      "grad_norm": 0.17290090024471283,
      "learning_rate": 6.750524109014676e-05,
      "loss": 0.1816,
      "step": 322
    },
    {
      "epoch": 0.06780016792611251,
      "grad_norm": 0.2324109524488449,
      "learning_rate": 6.771488469601677e-05,
      "loss": 0.1944,
      "step": 323
    },
    {
      "epoch": 0.06801007556675064,
      "grad_norm": 0.24944299459457397,
      "learning_rate": 6.79245283018868e-05,
      "loss": 0.1965,
      "step": 324
    },
    {
      "epoch": 0.06821998320738874,
      "grad_norm": 0.2102229744195938,
      "learning_rate": 6.813417190775681e-05,
      "loss": 0.221,
      "step": 325
    },
    {
      "epoch": 0.06842989084802686,
      "grad_norm": 0.22497773170471191,
      "learning_rate": 6.834381551362684e-05,
      "loss": 0.1818,
      "step": 326
    },
    {
      "epoch": 0.06863979848866499,
      "grad_norm": 0.19047041237354279,
      "learning_rate": 6.855345911949685e-05,
      "loss": 0.1817,
      "step": 327
    },
    {
      "epoch": 0.0688497061293031,
      "grad_norm": 0.19890040159225464,
      "learning_rate": 6.876310272536687e-05,
      "loss": 0.2054,
      "step": 328
    },
    {
      "epoch": 0.06905961376994123,
      "grad_norm": 0.18274420499801636,
      "learning_rate": 6.89727463312369e-05,
      "loss": 0.1732,
      "step": 329
    },
    {
      "epoch": 0.06926952141057935,
      "grad_norm": 0.20556879043579102,
      "learning_rate": 6.918238993710691e-05,
      "loss": 0.1773,
      "step": 330
    },
    {
      "epoch": 0.06947942905121747,
      "grad_norm": 0.19632075726985931,
      "learning_rate": 6.939203354297694e-05,
      "loss": 0.1841,
      "step": 331
    },
    {
      "epoch": 0.06968933669185558,
      "grad_norm": 0.25757917761802673,
      "learning_rate": 6.960167714884696e-05,
      "loss": 0.1861,
      "step": 332
    },
    {
      "epoch": 0.0698992443324937,
      "grad_norm": 0.1654757410287857,
      "learning_rate": 6.981132075471698e-05,
      "loss": 0.1797,
      "step": 333
    },
    {
      "epoch": 0.07010915197313182,
      "grad_norm": 0.1424175500869751,
      "learning_rate": 7.0020964360587e-05,
      "loss": 0.1659,
      "step": 334
    },
    {
      "epoch": 0.07031905961376994,
      "grad_norm": 0.21559248864650726,
      "learning_rate": 7.023060796645703e-05,
      "loss": 0.175,
      "step": 335
    },
    {
      "epoch": 0.07052896725440806,
      "grad_norm": 0.3100188076496124,
      "learning_rate": 7.044025157232704e-05,
      "loss": 0.1644,
      "step": 336
    },
    {
      "epoch": 0.07073887489504618,
      "grad_norm": 0.2530849277973175,
      "learning_rate": 7.064989517819707e-05,
      "loss": 0.1697,
      "step": 337
    },
    {
      "epoch": 0.0709487825356843,
      "grad_norm": 0.2111438810825348,
      "learning_rate": 7.085953878406709e-05,
      "loss": 0.202,
      "step": 338
    },
    {
      "epoch": 0.07115869017632241,
      "grad_norm": 0.21798165142536163,
      "learning_rate": 7.10691823899371e-05,
      "loss": 0.1832,
      "step": 339
    },
    {
      "epoch": 0.07136859781696053,
      "grad_norm": 0.24779516458511353,
      "learning_rate": 7.127882599580713e-05,
      "loss": 0.1934,
      "step": 340
    },
    {
      "epoch": 0.07157850545759865,
      "grad_norm": 0.21718356013298035,
      "learning_rate": 7.148846960167716e-05,
      "loss": 0.1951,
      "step": 341
    },
    {
      "epoch": 0.07178841309823678,
      "grad_norm": 0.22320568561553955,
      "learning_rate": 7.169811320754717e-05,
      "loss": 0.1619,
      "step": 342
    },
    {
      "epoch": 0.0719983207388749,
      "grad_norm": 0.19393590092658997,
      "learning_rate": 7.19077568134172e-05,
      "loss": 0.201,
      "step": 343
    },
    {
      "epoch": 0.07220822837951302,
      "grad_norm": 0.1626208871603012,
      "learning_rate": 7.211740041928722e-05,
      "loss": 0.1848,
      "step": 344
    },
    {
      "epoch": 0.07241813602015114,
      "grad_norm": 0.2256711721420288,
      "learning_rate": 7.232704402515723e-05,
      "loss": 0.192,
      "step": 345
    },
    {
      "epoch": 0.07262804366078925,
      "grad_norm": 0.2225414514541626,
      "learning_rate": 7.253668763102726e-05,
      "loss": 0.1907,
      "step": 346
    },
    {
      "epoch": 0.07283795130142737,
      "grad_norm": 0.1786690205335617,
      "learning_rate": 7.274633123689728e-05,
      "loss": 0.1756,
      "step": 347
    },
    {
      "epoch": 0.07304785894206549,
      "grad_norm": 0.2155577391386032,
      "learning_rate": 7.29559748427673e-05,
      "loss": 0.1739,
      "step": 348
    },
    {
      "epoch": 0.07325776658270361,
      "grad_norm": 0.32533329725265503,
      "learning_rate": 7.316561844863732e-05,
      "loss": 0.1988,
      "step": 349
    },
    {
      "epoch": 0.07346767422334173,
      "grad_norm": 0.1870083510875702,
      "learning_rate": 7.337526205450735e-05,
      "loss": 0.1615,
      "step": 350
    },
    {
      "epoch": 0.07367758186397985,
      "grad_norm": 0.2160840779542923,
      "learning_rate": 7.358490566037736e-05,
      "loss": 0.1901,
      "step": 351
    },
    {
      "epoch": 0.07388748950461797,
      "grad_norm": 0.19049416482448578,
      "learning_rate": 7.379454926624739e-05,
      "loss": 0.1763,
      "step": 352
    },
    {
      "epoch": 0.0740973971452561,
      "grad_norm": 0.15733250975608826,
      "learning_rate": 7.400419287211741e-05,
      "loss": 0.1812,
      "step": 353
    },
    {
      "epoch": 0.0743073047858942,
      "grad_norm": 0.22470858693122864,
      "learning_rate": 7.421383647798742e-05,
      "loss": 0.194,
      "step": 354
    },
    {
      "epoch": 0.07451721242653232,
      "grad_norm": 0.3271860182285309,
      "learning_rate": 7.442348008385745e-05,
      "loss": 0.1708,
      "step": 355
    },
    {
      "epoch": 0.07472712006717044,
      "grad_norm": 0.17839424312114716,
      "learning_rate": 7.463312368972748e-05,
      "loss": 0.182,
      "step": 356
    },
    {
      "epoch": 0.07493702770780857,
      "grad_norm": 0.1907908171415329,
      "learning_rate": 7.484276729559749e-05,
      "loss": 0.163,
      "step": 357
    },
    {
      "epoch": 0.07514693534844669,
      "grad_norm": 0.20342503488063812,
      "learning_rate": 7.505241090146751e-05,
      "loss": 0.2029,
      "step": 358
    },
    {
      "epoch": 0.07535684298908481,
      "grad_norm": 0.21872438490390778,
      "learning_rate": 7.526205450733753e-05,
      "loss": 0.192,
      "step": 359
    },
    {
      "epoch": 0.07556675062972293,
      "grad_norm": 0.22313977777957916,
      "learning_rate": 7.547169811320755e-05,
      "loss": 0.1823,
      "step": 360
    },
    {
      "epoch": 0.07577665827036104,
      "grad_norm": 0.1931924819946289,
      "learning_rate": 7.568134171907756e-05,
      "loss": 0.1992,
      "step": 361
    },
    {
      "epoch": 0.07598656591099916,
      "grad_norm": 0.2859954535961151,
      "learning_rate": 7.589098532494759e-05,
      "loss": 0.1684,
      "step": 362
    },
    {
      "epoch": 0.07619647355163728,
      "grad_norm": 0.18601499497890472,
      "learning_rate": 7.610062893081762e-05,
      "loss": 0.1871,
      "step": 363
    },
    {
      "epoch": 0.0764063811922754,
      "grad_norm": 0.26345667243003845,
      "learning_rate": 7.631027253668763e-05,
      "loss": 0.1883,
      "step": 364
    },
    {
      "epoch": 0.07661628883291352,
      "grad_norm": 0.24455974996089935,
      "learning_rate": 7.651991614255765e-05,
      "loss": 0.1991,
      "step": 365
    },
    {
      "epoch": 0.07682619647355164,
      "grad_norm": 0.1787412464618683,
      "learning_rate": 7.672955974842768e-05,
      "loss": 0.1895,
      "step": 366
    },
    {
      "epoch": 0.07703610411418976,
      "grad_norm": 0.2711624205112457,
      "learning_rate": 7.693920335429769e-05,
      "loss": 0.1889,
      "step": 367
    },
    {
      "epoch": 0.07724601175482787,
      "grad_norm": 0.2764052152633667,
      "learning_rate": 7.714884696016772e-05,
      "loss": 0.1895,
      "step": 368
    },
    {
      "epoch": 0.07745591939546599,
      "grad_norm": 0.15490169823169708,
      "learning_rate": 7.735849056603774e-05,
      "loss": 0.1812,
      "step": 369
    },
    {
      "epoch": 0.07766582703610411,
      "grad_norm": 0.18659183382987976,
      "learning_rate": 7.756813417190776e-05,
      "loss": 0.1871,
      "step": 370
    },
    {
      "epoch": 0.07787573467674223,
      "grad_norm": 0.19188903272151947,
      "learning_rate": 7.777777777777778e-05,
      "loss": 0.1716,
      "step": 371
    },
    {
      "epoch": 0.07808564231738035,
      "grad_norm": 0.22174161672592163,
      "learning_rate": 7.798742138364781e-05,
      "loss": 0.1795,
      "step": 372
    },
    {
      "epoch": 0.07829554995801848,
      "grad_norm": 0.1882723569869995,
      "learning_rate": 7.819706498951782e-05,
      "loss": 0.1874,
      "step": 373
    },
    {
      "epoch": 0.0785054575986566,
      "grad_norm": 0.1762145459651947,
      "learning_rate": 7.840670859538785e-05,
      "loss": 0.1852,
      "step": 374
    },
    {
      "epoch": 0.0787153652392947,
      "grad_norm": 0.17931701242923737,
      "learning_rate": 7.861635220125787e-05,
      "loss": 0.1817,
      "step": 375
    },
    {
      "epoch": 0.07892527287993283,
      "grad_norm": 0.1833990216255188,
      "learning_rate": 7.882599580712788e-05,
      "loss": 0.1852,
      "step": 376
    },
    {
      "epoch": 0.07913518052057095,
      "grad_norm": 0.2758026421070099,
      "learning_rate": 7.903563941299791e-05,
      "loss": 0.1868,
      "step": 377
    },
    {
      "epoch": 0.07934508816120907,
      "grad_norm": 0.24222204089164734,
      "learning_rate": 7.924528301886794e-05,
      "loss": 0.186,
      "step": 378
    },
    {
      "epoch": 0.07955499580184719,
      "grad_norm": 0.17609156668186188,
      "learning_rate": 7.945492662473795e-05,
      "loss": 0.1807,
      "step": 379
    },
    {
      "epoch": 0.07976490344248531,
      "grad_norm": 0.23695167899131775,
      "learning_rate": 7.966457023060797e-05,
      "loss": 0.1799,
      "step": 380
    },
    {
      "epoch": 0.07997481108312343,
      "grad_norm": 0.25356245040893555,
      "learning_rate": 7.9874213836478e-05,
      "loss": 0.1899,
      "step": 381
    },
    {
      "epoch": 0.08018471872376154,
      "grad_norm": 0.23144365847110748,
      "learning_rate": 8.008385744234801e-05,
      "loss": 0.192,
      "step": 382
    },
    {
      "epoch": 0.08039462636439966,
      "grad_norm": 0.1521812379360199,
      "learning_rate": 8.029350104821804e-05,
      "loss": 0.1864,
      "step": 383
    },
    {
      "epoch": 0.08060453400503778,
      "grad_norm": 0.16725748777389526,
      "learning_rate": 8.050314465408806e-05,
      "loss": 0.202,
      "step": 384
    },
    {
      "epoch": 0.0808144416456759,
      "grad_norm": 0.21173058450222015,
      "learning_rate": 8.071278825995808e-05,
      "loss": 0.1751,
      "step": 385
    },
    {
      "epoch": 0.08102434928631402,
      "grad_norm": 0.15676653385162354,
      "learning_rate": 8.09224318658281e-05,
      "loss": 0.1946,
      "step": 386
    },
    {
      "epoch": 0.08123425692695214,
      "grad_norm": 0.21838362514972687,
      "learning_rate": 8.113207547169813e-05,
      "loss": 0.1889,
      "step": 387
    },
    {
      "epoch": 0.08144416456759027,
      "grad_norm": 0.19586238265037537,
      "learning_rate": 8.134171907756814e-05,
      "loss": 0.1884,
      "step": 388
    },
    {
      "epoch": 0.08165407220822837,
      "grad_norm": 0.21012739837169647,
      "learning_rate": 8.155136268343817e-05,
      "loss": 0.1822,
      "step": 389
    },
    {
      "epoch": 0.0818639798488665,
      "grad_norm": 0.2092917114496231,
      "learning_rate": 8.176100628930818e-05,
      "loss": 0.1783,
      "step": 390
    },
    {
      "epoch": 0.08207388748950462,
      "grad_norm": 0.3745954930782318,
      "learning_rate": 8.19706498951782e-05,
      "loss": 0.1976,
      "step": 391
    },
    {
      "epoch": 0.08228379513014274,
      "grad_norm": 0.2579379081726074,
      "learning_rate": 8.218029350104822e-05,
      "loss": 0.197,
      "step": 392
    },
    {
      "epoch": 0.08249370277078086,
      "grad_norm": 0.18806852400302887,
      "learning_rate": 8.238993710691824e-05,
      "loss": 0.1873,
      "step": 393
    },
    {
      "epoch": 0.08270361041141898,
      "grad_norm": 0.24592849612236023,
      "learning_rate": 8.259958071278825e-05,
      "loss": 0.2083,
      "step": 394
    },
    {
      "epoch": 0.0829135180520571,
      "grad_norm": 0.2678208649158478,
      "learning_rate": 8.280922431865828e-05,
      "loss": 0.1878,
      "step": 395
    },
    {
      "epoch": 0.08312342569269521,
      "grad_norm": 0.2023075520992279,
      "learning_rate": 8.30188679245283e-05,
      "loss": 0.1861,
      "step": 396
    },
    {
      "epoch": 0.08333333333333333,
      "grad_norm": 0.2390558272600174,
      "learning_rate": 8.322851153039832e-05,
      "loss": 0.1802,
      "step": 397
    },
    {
      "epoch": 0.08354324097397145,
      "grad_norm": 0.175222247838974,
      "learning_rate": 8.343815513626834e-05,
      "loss": 0.1674,
      "step": 398
    },
    {
      "epoch": 0.08375314861460957,
      "grad_norm": 0.2340380698442459,
      "learning_rate": 8.364779874213837e-05,
      "loss": 0.17,
      "step": 399
    },
    {
      "epoch": 0.08396305625524769,
      "grad_norm": 0.20814655721187592,
      "learning_rate": 8.385744234800838e-05,
      "loss": 0.1513,
      "step": 400
    },
    {
      "epoch": 0.08417296389588581,
      "grad_norm": 0.34867948293685913,
      "learning_rate": 8.406708595387841e-05,
      "loss": 0.1644,
      "step": 401
    },
    {
      "epoch": 0.08438287153652393,
      "grad_norm": 0.21573619544506073,
      "learning_rate": 8.427672955974843e-05,
      "loss": 0.1785,
      "step": 402
    },
    {
      "epoch": 0.08459277917716204,
      "grad_norm": 0.21437713503837585,
      "learning_rate": 8.448637316561845e-05,
      "loss": 0.1523,
      "step": 403
    },
    {
      "epoch": 0.08480268681780016,
      "grad_norm": 0.2250152826309204,
      "learning_rate": 8.469601677148847e-05,
      "loss": 0.1728,
      "step": 404
    },
    {
      "epoch": 0.08501259445843828,
      "grad_norm": 0.2514733672142029,
      "learning_rate": 8.49056603773585e-05,
      "loss": 0.1755,
      "step": 405
    },
    {
      "epoch": 0.0852225020990764,
      "grad_norm": 0.16003377735614777,
      "learning_rate": 8.511530398322851e-05,
      "loss": 0.1935,
      "step": 406
    },
    {
      "epoch": 0.08543240973971453,
      "grad_norm": 0.1792365163564682,
      "learning_rate": 8.532494758909854e-05,
      "loss": 0.1991,
      "step": 407
    },
    {
      "epoch": 0.08564231738035265,
      "grad_norm": 0.16854703426361084,
      "learning_rate": 8.553459119496856e-05,
      "loss": 0.1904,
      "step": 408
    },
    {
      "epoch": 0.08585222502099077,
      "grad_norm": 0.19401603937149048,
      "learning_rate": 8.574423480083857e-05,
      "loss": 0.1681,
      "step": 409
    },
    {
      "epoch": 0.08606213266162889,
      "grad_norm": 0.15639828145503998,
      "learning_rate": 8.59538784067086e-05,
      "loss": 0.1628,
      "step": 410
    },
    {
      "epoch": 0.086272040302267,
      "grad_norm": 0.15357258915901184,
      "learning_rate": 8.616352201257863e-05,
      "loss": 0.1905,
      "step": 411
    },
    {
      "epoch": 0.08648194794290512,
      "grad_norm": 0.256944477558136,
      "learning_rate": 8.637316561844864e-05,
      "loss": 0.2015,
      "step": 412
    },
    {
      "epoch": 0.08669185558354324,
      "grad_norm": 0.13482192158699036,
      "learning_rate": 8.658280922431866e-05,
      "loss": 0.1718,
      "step": 413
    },
    {
      "epoch": 0.08690176322418136,
      "grad_norm": 0.16525831818580627,
      "learning_rate": 8.679245283018869e-05,
      "loss": 0.1778,
      "step": 414
    },
    {
      "epoch": 0.08711167086481948,
      "grad_norm": 0.2145531326532364,
      "learning_rate": 8.70020964360587e-05,
      "loss": 0.1714,
      "step": 415
    },
    {
      "epoch": 0.0873215785054576,
      "grad_norm": 0.20481255650520325,
      "learning_rate": 8.721174004192873e-05,
      "loss": 0.1832,
      "step": 416
    },
    {
      "epoch": 0.08753148614609572,
      "grad_norm": 0.14865756034851074,
      "learning_rate": 8.742138364779875e-05,
      "loss": 0.1659,
      "step": 417
    },
    {
      "epoch": 0.08774139378673383,
      "grad_norm": 0.14713706076145172,
      "learning_rate": 8.763102725366877e-05,
      "loss": 0.1674,
      "step": 418
    },
    {
      "epoch": 0.08795130142737195,
      "grad_norm": 0.1728464812040329,
      "learning_rate": 8.784067085953879e-05,
      "loss": 0.1727,
      "step": 419
    },
    {
      "epoch": 0.08816120906801007,
      "grad_norm": 0.2646033465862274,
      "learning_rate": 8.805031446540882e-05,
      "loss": 0.1973,
      "step": 420
    },
    {
      "epoch": 0.0883711167086482,
      "grad_norm": 0.2262433022260666,
      "learning_rate": 8.825995807127883e-05,
      "loss": 0.1817,
      "step": 421
    },
    {
      "epoch": 0.08858102434928632,
      "grad_norm": 0.16398945450782776,
      "learning_rate": 8.846960167714886e-05,
      "loss": 0.1628,
      "step": 422
    },
    {
      "epoch": 0.08879093198992444,
      "grad_norm": 0.15976634621620178,
      "learning_rate": 8.867924528301888e-05,
      "loss": 0.1678,
      "step": 423
    },
    {
      "epoch": 0.08900083963056256,
      "grad_norm": 0.19314904510974884,
      "learning_rate": 8.888888888888889e-05,
      "loss": 0.1753,
      "step": 424
    },
    {
      "epoch": 0.08921074727120067,
      "grad_norm": 0.21701818704605103,
      "learning_rate": 8.909853249475892e-05,
      "loss": 0.19,
      "step": 425
    },
    {
      "epoch": 0.08942065491183879,
      "grad_norm": 0.22768795490264893,
      "learning_rate": 8.930817610062893e-05,
      "loss": 0.1613,
      "step": 426
    },
    {
      "epoch": 0.08963056255247691,
      "grad_norm": 0.13172288239002228,
      "learning_rate": 8.951781970649896e-05,
      "loss": 0.1749,
      "step": 427
    },
    {
      "epoch": 0.08984047019311503,
      "grad_norm": 0.2015480250120163,
      "learning_rate": 8.972746331236897e-05,
      "loss": 0.1857,
      "step": 428
    },
    {
      "epoch": 0.09005037783375315,
      "grad_norm": 0.20239531993865967,
      "learning_rate": 8.9937106918239e-05,
      "loss": 0.1804,
      "step": 429
    },
    {
      "epoch": 0.09026028547439127,
      "grad_norm": 0.17528317868709564,
      "learning_rate": 9.014675052410901e-05,
      "loss": 0.1729,
      "step": 430
    },
    {
      "epoch": 0.0904701931150294,
      "grad_norm": 0.16663801670074463,
      "learning_rate": 9.035639412997903e-05,
      "loss": 0.1896,
      "step": 431
    },
    {
      "epoch": 0.0906801007556675,
      "grad_norm": 0.18777558207511902,
      "learning_rate": 9.056603773584906e-05,
      "loss": 0.1801,
      "step": 432
    },
    {
      "epoch": 0.09089000839630562,
      "grad_norm": 0.1440989226102829,
      "learning_rate": 9.077568134171907e-05,
      "loss": 0.179,
      "step": 433
    },
    {
      "epoch": 0.09109991603694374,
      "grad_norm": 0.1786854863166809,
      "learning_rate": 9.09853249475891e-05,
      "loss": 0.1799,
      "step": 434
    },
    {
      "epoch": 0.09130982367758186,
      "grad_norm": 0.20794442296028137,
      "learning_rate": 9.119496855345912e-05,
      "loss": 0.1731,
      "step": 435
    },
    {
      "epoch": 0.09151973131821999,
      "grad_norm": 0.16377133131027222,
      "learning_rate": 9.140461215932914e-05,
      "loss": 0.1815,
      "step": 436
    },
    {
      "epoch": 0.0917296389588581,
      "grad_norm": 0.174666628241539,
      "learning_rate": 9.161425576519916e-05,
      "loss": 0.1798,
      "step": 437
    },
    {
      "epoch": 0.09193954659949623,
      "grad_norm": 0.2127188891172409,
      "learning_rate": 9.182389937106919e-05,
      "loss": 0.1805,
      "step": 438
    },
    {
      "epoch": 0.09214945424013433,
      "grad_norm": 0.1936446726322174,
      "learning_rate": 9.203354297693921e-05,
      "loss": 0.1723,
      "step": 439
    },
    {
      "epoch": 0.09235936188077246,
      "grad_norm": 0.18736332654953003,
      "learning_rate": 9.224318658280923e-05,
      "loss": 0.1849,
      "step": 440
    },
    {
      "epoch": 0.09256926952141058,
      "grad_norm": 0.16817238926887512,
      "learning_rate": 9.245283018867925e-05,
      "loss": 0.1746,
      "step": 441
    },
    {
      "epoch": 0.0927791771620487,
      "grad_norm": 0.18249107897281647,
      "learning_rate": 9.266247379454928e-05,
      "loss": 0.1628,
      "step": 442
    },
    {
      "epoch": 0.09298908480268682,
      "grad_norm": 0.1728898137807846,
      "learning_rate": 9.287211740041929e-05,
      "loss": 0.203,
      "step": 443
    },
    {
      "epoch": 0.09319899244332494,
      "grad_norm": 0.16144797205924988,
      "learning_rate": 9.308176100628931e-05,
      "loss": 0.1843,
      "step": 444
    },
    {
      "epoch": 0.09340890008396306,
      "grad_norm": 0.19680747389793396,
      "learning_rate": 9.329140461215934e-05,
      "loss": 0.168,
      "step": 445
    },
    {
      "epoch": 0.09361880772460117,
      "grad_norm": 0.16198395192623138,
      "learning_rate": 9.350104821802935e-05,
      "loss": 0.1598,
      "step": 446
    },
    {
      "epoch": 0.09382871536523929,
      "grad_norm": 0.17398878931999207,
      "learning_rate": 9.371069182389938e-05,
      "loss": 0.17,
      "step": 447
    },
    {
      "epoch": 0.09403862300587741,
      "grad_norm": 0.18602675199508667,
      "learning_rate": 9.39203354297694e-05,
      "loss": 0.1716,
      "step": 448
    },
    {
      "epoch": 0.09424853064651553,
      "grad_norm": 0.18403322994709015,
      "learning_rate": 9.412997903563942e-05,
      "loss": 0.1948,
      "step": 449
    },
    {
      "epoch": 0.09445843828715365,
      "grad_norm": 0.18783587217330933,
      "learning_rate": 9.433962264150944e-05,
      "loss": 0.1812,
      "step": 450
    },
    {
      "epoch": 0.09466834592779177,
      "grad_norm": 0.20252300798892975,
      "learning_rate": 9.454926624737947e-05,
      "loss": 0.1666,
      "step": 451
    },
    {
      "epoch": 0.0948782535684299,
      "grad_norm": 0.20129899680614471,
      "learning_rate": 9.475890985324948e-05,
      "loss": 0.1695,
      "step": 452
    },
    {
      "epoch": 0.095088161209068,
      "grad_norm": 0.17035968601703644,
      "learning_rate": 9.496855345911951e-05,
      "loss": 0.1678,
      "step": 453
    },
    {
      "epoch": 0.09529806884970612,
      "grad_norm": 0.20403030514717102,
      "learning_rate": 9.517819706498953e-05,
      "loss": 0.1843,
      "step": 454
    },
    {
      "epoch": 0.09550797649034425,
      "grad_norm": 0.17489562928676605,
      "learning_rate": 9.538784067085954e-05,
      "loss": 0.1657,
      "step": 455
    },
    {
      "epoch": 0.09571788413098237,
      "grad_norm": 0.19699983298778534,
      "learning_rate": 9.559748427672957e-05,
      "loss": 0.1634,
      "step": 456
    },
    {
      "epoch": 0.09592779177162049,
      "grad_norm": 0.17113354802131653,
      "learning_rate": 9.58071278825996e-05,
      "loss": 0.1879,
      "step": 457
    },
    {
      "epoch": 0.09613769941225861,
      "grad_norm": 0.2290397435426712,
      "learning_rate": 9.601677148846961e-05,
      "loss": 0.1784,
      "step": 458
    },
    {
      "epoch": 0.09634760705289673,
      "grad_norm": 0.2173147350549698,
      "learning_rate": 9.622641509433963e-05,
      "loss": 0.1673,
      "step": 459
    },
    {
      "epoch": 0.09655751469353484,
      "grad_norm": 0.18280835449695587,
      "learning_rate": 9.643605870020965e-05,
      "loss": 0.1806,
      "step": 460
    },
    {
      "epoch": 0.09676742233417296,
      "grad_norm": 0.14638672769069672,
      "learning_rate": 9.664570230607967e-05,
      "loss": 0.1833,
      "step": 461
    },
    {
      "epoch": 0.09697732997481108,
      "grad_norm": 0.16228064894676208,
      "learning_rate": 9.685534591194969e-05,
      "loss": 0.1762,
      "step": 462
    },
    {
      "epoch": 0.0971872376154492,
      "grad_norm": 0.1836690902709961,
      "learning_rate": 9.706498951781971e-05,
      "loss": 0.1782,
      "step": 463
    },
    {
      "epoch": 0.09739714525608732,
      "grad_norm": 0.19470515847206116,
      "learning_rate": 9.727463312368972e-05,
      "loss": 0.185,
      "step": 464
    },
    {
      "epoch": 0.09760705289672544,
      "grad_norm": 0.1833791732788086,
      "learning_rate": 9.748427672955975e-05,
      "loss": 0.1753,
      "step": 465
    },
    {
      "epoch": 0.09781696053736356,
      "grad_norm": 0.22608265280723572,
      "learning_rate": 9.769392033542977e-05,
      "loss": 0.1792,
      "step": 466
    },
    {
      "epoch": 0.09802686817800169,
      "grad_norm": 0.16552825272083282,
      "learning_rate": 9.790356394129979e-05,
      "loss": 0.1808,
      "step": 467
    },
    {
      "epoch": 0.0982367758186398,
      "grad_norm": 0.2294851690530777,
      "learning_rate": 9.811320754716981e-05,
      "loss": 0.1935,
      "step": 468
    },
    {
      "epoch": 0.09844668345927791,
      "grad_norm": 0.26589101552963257,
      "learning_rate": 9.832285115303984e-05,
      "loss": 0.1676,
      "step": 469
    },
    {
      "epoch": 0.09865659109991604,
      "grad_norm": 0.22315791249275208,
      "learning_rate": 9.853249475890985e-05,
      "loss": 0.1806,
      "step": 470
    },
    {
      "epoch": 0.09886649874055416,
      "grad_norm": 0.16855137050151825,
      "learning_rate": 9.874213836477988e-05,
      "loss": 0.1816,
      "step": 471
    },
    {
      "epoch": 0.09907640638119228,
      "grad_norm": 0.19197392463684082,
      "learning_rate": 9.89517819706499e-05,
      "loss": 0.1923,
      "step": 472
    },
    {
      "epoch": 0.0992863140218304,
      "grad_norm": 0.18722014129161835,
      "learning_rate": 9.916142557651992e-05,
      "loss": 0.1823,
      "step": 473
    },
    {
      "epoch": 0.09949622166246852,
      "grad_norm": 0.15668706595897675,
      "learning_rate": 9.937106918238994e-05,
      "loss": 0.1896,
      "step": 474
    },
    {
      "epoch": 0.09970612930310663,
      "grad_norm": 0.17297013103961945,
      "learning_rate": 9.958071278825997e-05,
      "loss": 0.1907,
      "step": 475
    },
    {
      "epoch": 0.09991603694374475,
      "grad_norm": 0.23546694219112396,
      "learning_rate": 9.979035639412998e-05,
      "loss": 0.1647,
      "step": 476
    },
    {
      "epoch": 0.10012594458438287,
      "grad_norm": 0.1627054661512375,
      "learning_rate": 0.0001,
      "loss": 0.1686,
      "step": 477
    },
    {
      "epoch": 0.10033585222502099,
      "grad_norm": 0.21043647825717926,
      "learning_rate": 9.999998657442895e-05,
      "loss": 0.1865,
      "step": 478
    },
    {
      "epoch": 0.10054575986565911,
      "grad_norm": 0.19615764915943146,
      "learning_rate": 9.999994629772298e-05,
      "loss": 0.1683,
      "step": 479
    },
    {
      "epoch": 0.10075566750629723,
      "grad_norm": 0.2884671092033386,
      "learning_rate": 9.999987916990372e-05,
      "loss": 0.1858,
      "step": 480
    },
    {
      "epoch": 0.10096557514693535,
      "grad_norm": 0.2381323128938675,
      "learning_rate": 9.999978519100723e-05,
      "loss": 0.1879,
      "step": 481
    },
    {
      "epoch": 0.10117548278757346,
      "grad_norm": 0.19187557697296143,
      "learning_rate": 9.999966436108398e-05,
      "loss": 0.1808,
      "step": 482
    },
    {
      "epoch": 0.10138539042821158,
      "grad_norm": 0.19443491101264954,
      "learning_rate": 9.999951668019887e-05,
      "loss": 0.1659,
      "step": 483
    },
    {
      "epoch": 0.1015952980688497,
      "grad_norm": 0.20151716470718384,
      "learning_rate": 9.999934214843116e-05,
      "loss": 0.1679,
      "step": 484
    },
    {
      "epoch": 0.10180520570948783,
      "grad_norm": 0.2896507978439331,
      "learning_rate": 9.999914076587464e-05,
      "loss": 0.1734,
      "step": 485
    },
    {
      "epoch": 0.10201511335012595,
      "grad_norm": 0.21598441898822784,
      "learning_rate": 9.999891253263741e-05,
      "loss": 0.1779,
      "step": 486
    },
    {
      "epoch": 0.10222502099076407,
      "grad_norm": 0.196011021733284,
      "learning_rate": 9.999865744884207e-05,
      "loss": 0.1815,
      "step": 487
    },
    {
      "epoch": 0.10243492863140219,
      "grad_norm": 0.20962318778038025,
      "learning_rate": 9.999837551462558e-05,
      "loss": 0.1727,
      "step": 488
    },
    {
      "epoch": 0.1026448362720403,
      "grad_norm": 0.18339572846889496,
      "learning_rate": 9.999806673013935e-05,
      "loss": 0.1689,
      "step": 489
    },
    {
      "epoch": 0.10285474391267842,
      "grad_norm": 0.20994813740253448,
      "learning_rate": 9.999773109554922e-05,
      "loss": 0.1768,
      "step": 490
    },
    {
      "epoch": 0.10306465155331654,
      "grad_norm": 0.163935124874115,
      "learning_rate": 9.999736861103541e-05,
      "loss": 0.1777,
      "step": 491
    },
    {
      "epoch": 0.10327455919395466,
      "grad_norm": 0.1465967446565628,
      "learning_rate": 9.99969792767926e-05,
      "loss": 0.1846,
      "step": 492
    },
    {
      "epoch": 0.10348446683459278,
      "grad_norm": 0.21443922817707062,
      "learning_rate": 9.999656309302987e-05,
      "loss": 0.1807,
      "step": 493
    },
    {
      "epoch": 0.1036943744752309,
      "grad_norm": 0.18504248559474945,
      "learning_rate": 9.999612005997071e-05,
      "loss": 0.1548,
      "step": 494
    },
    {
      "epoch": 0.10390428211586902,
      "grad_norm": 0.15490441024303436,
      "learning_rate": 9.999565017785305e-05,
      "loss": 0.1696,
      "step": 495
    },
    {
      "epoch": 0.10411418975650713,
      "grad_norm": 0.1881389319896698,
      "learning_rate": 9.999515344692923e-05,
      "loss": 0.1852,
      "step": 496
    },
    {
      "epoch": 0.10432409739714525,
      "grad_norm": 0.16337451338768005,
      "learning_rate": 9.999462986746598e-05,
      "loss": 0.1834,
      "step": 497
    },
    {
      "epoch": 0.10453400503778337,
      "grad_norm": 0.16641898453235626,
      "learning_rate": 9.99940794397445e-05,
      "loss": 0.187,
      "step": 498
    },
    {
      "epoch": 0.1047439126784215,
      "grad_norm": 0.15948446094989777,
      "learning_rate": 9.999350216406038e-05,
      "loss": 0.1835,
      "step": 499
    },
    {
      "epoch": 0.10495382031905962,
      "grad_norm": 0.1550200879573822,
      "learning_rate": 9.999289804072363e-05,
      "loss": 0.1706,
      "step": 500
    },
    {
      "epoch": 0.10516372795969774,
      "grad_norm": 0.16597698628902435,
      "learning_rate": 9.999226707005867e-05,
      "loss": 0.1811,
      "step": 501
    },
    {
      "epoch": 0.10537363560033586,
      "grad_norm": 0.17551501095294952,
      "learning_rate": 9.999160925240434e-05,
      "loss": 0.1677,
      "step": 502
    },
    {
      "epoch": 0.10558354324097396,
      "grad_norm": 0.15515847504138947,
      "learning_rate": 9.999092458811393e-05,
      "loss": 0.1789,
      "step": 503
    },
    {
      "epoch": 0.10579345088161209,
      "grad_norm": 0.18121638894081116,
      "learning_rate": 9.99902130775551e-05,
      "loss": 0.1684,
      "step": 504
    },
    {
      "epoch": 0.10600335852225021,
      "grad_norm": 0.1853945255279541,
      "learning_rate": 9.998947472110994e-05,
      "loss": 0.1907,
      "step": 505
    },
    {
      "epoch": 0.10621326616288833,
      "grad_norm": 0.21303139626979828,
      "learning_rate": 9.998870951917496e-05,
      "loss": 0.1712,
      "step": 506
    },
    {
      "epoch": 0.10642317380352645,
      "grad_norm": 0.16773764789104462,
      "learning_rate": 9.998791747216113e-05,
      "loss": 0.1756,
      "step": 507
    },
    {
      "epoch": 0.10663308144416457,
      "grad_norm": 0.18033501505851746,
      "learning_rate": 9.998709858049376e-05,
      "loss": 0.1654,
      "step": 508
    },
    {
      "epoch": 0.10684298908480269,
      "grad_norm": 0.14199328422546387,
      "learning_rate": 9.998625284461263e-05,
      "loss": 0.1587,
      "step": 509
    },
    {
      "epoch": 0.1070528967254408,
      "grad_norm": 0.19968685507774353,
      "learning_rate": 9.998538026497192e-05,
      "loss": 0.1796,
      "step": 510
    },
    {
      "epoch": 0.10726280436607892,
      "grad_norm": 0.17311611771583557,
      "learning_rate": 9.998448084204021e-05,
      "loss": 0.1864,
      "step": 511
    },
    {
      "epoch": 0.10747271200671704,
      "grad_norm": 0.20124119520187378,
      "learning_rate": 9.998355457630053e-05,
      "loss": 0.1829,
      "step": 512
    },
    {
      "epoch": 0.10768261964735516,
      "grad_norm": 0.12473297864198685,
      "learning_rate": 9.998260146825029e-05,
      "loss": 0.175,
      "step": 513
    },
    {
      "epoch": 0.10789252728799328,
      "grad_norm": 0.1696644425392151,
      "learning_rate": 9.998162151840135e-05,
      "loss": 0.1762,
      "step": 514
    },
    {
      "epoch": 0.1081024349286314,
      "grad_norm": 0.1781477928161621,
      "learning_rate": 9.998061472727996e-05,
      "loss": 0.1679,
      "step": 515
    },
    {
      "epoch": 0.10831234256926953,
      "grad_norm": 0.19112960994243622,
      "learning_rate": 9.997958109542675e-05,
      "loss": 0.1553,
      "step": 516
    },
    {
      "epoch": 0.10852225020990765,
      "grad_norm": 0.1417030394077301,
      "learning_rate": 9.997852062339685e-05,
      "loss": 0.1737,
      "step": 517
    },
    {
      "epoch": 0.10873215785054575,
      "grad_norm": 0.15080858767032623,
      "learning_rate": 9.997743331175976e-05,
      "loss": 0.1595,
      "step": 518
    },
    {
      "epoch": 0.10894206549118388,
      "grad_norm": 0.2046668380498886,
      "learning_rate": 9.997631916109937e-05,
      "loss": 0.1839,
      "step": 519
    },
    {
      "epoch": 0.109151973131822,
      "grad_norm": 0.19941595196723938,
      "learning_rate": 9.997517817201401e-05,
      "loss": 0.1718,
      "step": 520
    },
    {
      "epoch": 0.10936188077246012,
      "grad_norm": 0.15989692509174347,
      "learning_rate": 9.997401034511642e-05,
      "loss": 0.1613,
      "step": 521
    },
    {
      "epoch": 0.10957178841309824,
      "grad_norm": 0.1697997897863388,
      "learning_rate": 9.997281568103374e-05,
      "loss": 0.1603,
      "step": 522
    },
    {
      "epoch": 0.10978169605373636,
      "grad_norm": 0.1840822696685791,
      "learning_rate": 9.997159418040754e-05,
      "loss": 0.1735,
      "step": 523
    },
    {
      "epoch": 0.10999160369437448,
      "grad_norm": 0.20991730690002441,
      "learning_rate": 9.99703458438938e-05,
      "loss": 0.154,
      "step": 524
    },
    {
      "epoch": 0.11020151133501259,
      "grad_norm": 0.16802968084812164,
      "learning_rate": 9.99690706721629e-05,
      "loss": 0.1761,
      "step": 525
    },
    {
      "epoch": 0.11041141897565071,
      "grad_norm": 0.18329255282878876,
      "learning_rate": 9.996776866589962e-05,
      "loss": 0.1609,
      "step": 526
    },
    {
      "epoch": 0.11062132661628883,
      "grad_norm": 0.18645748496055603,
      "learning_rate": 9.996643982580318e-05,
      "loss": 0.1793,
      "step": 527
    },
    {
      "epoch": 0.11083123425692695,
      "grad_norm": 0.1966720074415207,
      "learning_rate": 9.996508415258722e-05,
      "loss": 0.1714,
      "step": 528
    },
    {
      "epoch": 0.11104114189756507,
      "grad_norm": 0.18155452609062195,
      "learning_rate": 9.996370164697974e-05,
      "loss": 0.1673,
      "step": 529
    },
    {
      "epoch": 0.1112510495382032,
      "grad_norm": 0.2004195600748062,
      "learning_rate": 9.996229230972317e-05,
      "loss": 0.1865,
      "step": 530
    },
    {
      "epoch": 0.11146095717884132,
      "grad_norm": 0.15521694719791412,
      "learning_rate": 9.996085614157438e-05,
      "loss": 0.1757,
      "step": 531
    },
    {
      "epoch": 0.11167086481947942,
      "grad_norm": 0.1686578243970871,
      "learning_rate": 9.995939314330462e-05,
      "loss": 0.1768,
      "step": 532
    },
    {
      "epoch": 0.11188077246011754,
      "grad_norm": 0.20034368336200714,
      "learning_rate": 9.995790331569954e-05,
      "loss": 0.1823,
      "step": 533
    },
    {
      "epoch": 0.11209068010075567,
      "grad_norm": 0.1494702696800232,
      "learning_rate": 9.995638665955922e-05,
      "loss": 0.175,
      "step": 534
    },
    {
      "epoch": 0.11230058774139379,
      "grad_norm": 0.16365233063697815,
      "learning_rate": 9.995484317569814e-05,
      "loss": 0.1716,
      "step": 535
    },
    {
      "epoch": 0.11251049538203191,
      "grad_norm": 0.19227434694766998,
      "learning_rate": 9.995327286494521e-05,
      "loss": 0.1605,
      "step": 536
    },
    {
      "epoch": 0.11272040302267003,
      "grad_norm": 0.21946166455745697,
      "learning_rate": 9.995167572814365e-05,
      "loss": 0.182,
      "step": 537
    },
    {
      "epoch": 0.11293031066330815,
      "grad_norm": 0.2211793065071106,
      "learning_rate": 9.995005176615124e-05,
      "loss": 0.1783,
      "step": 538
    },
    {
      "epoch": 0.11314021830394626,
      "grad_norm": 0.2154102325439453,
      "learning_rate": 9.994840097984006e-05,
      "loss": 0.1888,
      "step": 539
    },
    {
      "epoch": 0.11335012594458438,
      "grad_norm": 0.20600587129592896,
      "learning_rate": 9.994672337009658e-05,
      "loss": 0.1871,
      "step": 540
    },
    {
      "epoch": 0.1135600335852225,
      "grad_norm": 0.22028079628944397,
      "learning_rate": 9.994501893782176e-05,
      "loss": 0.1855,
      "step": 541
    },
    {
      "epoch": 0.11376994122586062,
      "grad_norm": 0.23957398533821106,
      "learning_rate": 9.99432876839309e-05,
      "loss": 0.1616,
      "step": 542
    },
    {
      "epoch": 0.11397984886649874,
      "grad_norm": 0.14516577124595642,
      "learning_rate": 9.994152960935375e-05,
      "loss": 0.1864,
      "step": 543
    },
    {
      "epoch": 0.11418975650713686,
      "grad_norm": 0.14327426254749298,
      "learning_rate": 9.99397447150344e-05,
      "loss": 0.1596,
      "step": 544
    },
    {
      "epoch": 0.11439966414777498,
      "grad_norm": 0.15387804806232452,
      "learning_rate": 9.99379330019314e-05,
      "loss": 0.1638,
      "step": 545
    },
    {
      "epoch": 0.11460957178841309,
      "grad_norm": 0.14998720586299896,
      "learning_rate": 9.993609447101767e-05,
      "loss": 0.1456,
      "step": 546
    },
    {
      "epoch": 0.11481947942905121,
      "grad_norm": 0.21649526059627533,
      "learning_rate": 9.993422912328054e-05,
      "loss": 0.1656,
      "step": 547
    },
    {
      "epoch": 0.11502938706968933,
      "grad_norm": 0.19370022416114807,
      "learning_rate": 9.993233695972175e-05,
      "loss": 0.173,
      "step": 548
    },
    {
      "epoch": 0.11523929471032746,
      "grad_norm": 0.20101432502269745,
      "learning_rate": 9.993041798135745e-05,
      "loss": 0.1914,
      "step": 549
    },
    {
      "epoch": 0.11544920235096558,
      "grad_norm": 0.18047718703746796,
      "learning_rate": 9.992847218921816e-05,
      "loss": 0.2084,
      "step": 550
    },
    {
      "epoch": 0.1156591099916037,
      "grad_norm": 0.18755191564559937,
      "learning_rate": 9.99264995843488e-05,
      "loss": 0.1646,
      "step": 551
    },
    {
      "epoch": 0.11586901763224182,
      "grad_norm": 0.2605830430984497,
      "learning_rate": 9.992450016780876e-05,
      "loss": 0.1767,
      "step": 552
    },
    {
      "epoch": 0.11607892527287993,
      "grad_norm": 0.1701487898826599,
      "learning_rate": 9.99224739406717e-05,
      "loss": 0.173,
      "step": 553
    },
    {
      "epoch": 0.11628883291351805,
      "grad_norm": 0.19497598707675934,
      "learning_rate": 9.99204209040258e-05,
      "loss": 0.1857,
      "step": 554
    },
    {
      "epoch": 0.11649874055415617,
      "grad_norm": 0.16379636526107788,
      "learning_rate": 9.991834105897356e-05,
      "loss": 0.154,
      "step": 555
    },
    {
      "epoch": 0.11670864819479429,
      "grad_norm": 0.18630164861679077,
      "learning_rate": 9.991623440663192e-05,
      "loss": 0.1871,
      "step": 556
    },
    {
      "epoch": 0.11691855583543241,
      "grad_norm": 0.219542995095253,
      "learning_rate": 9.991410094813221e-05,
      "loss": 0.1877,
      "step": 557
    },
    {
      "epoch": 0.11712846347607053,
      "grad_norm": 0.1855912059545517,
      "learning_rate": 9.991194068462011e-05,
      "loss": 0.1771,
      "step": 558
    },
    {
      "epoch": 0.11733837111670865,
      "grad_norm": 0.15919265151023865,
      "learning_rate": 9.990975361725577e-05,
      "loss": 0.1684,
      "step": 559
    },
    {
      "epoch": 0.11754827875734676,
      "grad_norm": 0.14570386707782745,
      "learning_rate": 9.990753974721366e-05,
      "loss": 0.1656,
      "step": 560
    },
    {
      "epoch": 0.11775818639798488,
      "grad_norm": 0.14153516292572021,
      "learning_rate": 9.990529907568272e-05,
      "loss": 0.1642,
      "step": 561
    },
    {
      "epoch": 0.117968094038623,
      "grad_norm": 0.14015786349773407,
      "learning_rate": 9.99030316038662e-05,
      "loss": 0.1704,
      "step": 562
    },
    {
      "epoch": 0.11817800167926112,
      "grad_norm": 0.2615254819393158,
      "learning_rate": 9.99007373329818e-05,
      "loss": 0.1601,
      "step": 563
    },
    {
      "epoch": 0.11838790931989925,
      "grad_norm": 0.2077956199645996,
      "learning_rate": 9.989841626426162e-05,
      "loss": 0.1715,
      "step": 564
    },
    {
      "epoch": 0.11859781696053737,
      "grad_norm": 0.1743435561656952,
      "learning_rate": 9.989606839895208e-05,
      "loss": 0.1725,
      "step": 565
    },
    {
      "epoch": 0.11880772460117549,
      "grad_norm": 0.14333437383174896,
      "learning_rate": 9.989369373831407e-05,
      "loss": 0.1731,
      "step": 566
    },
    {
      "epoch": 0.1190176322418136,
      "grad_norm": 0.16472546756267548,
      "learning_rate": 9.989129228362284e-05,
      "loss": 0.1641,
      "step": 567
    },
    {
      "epoch": 0.11922753988245172,
      "grad_norm": 0.19595298171043396,
      "learning_rate": 9.988886403616802e-05,
      "loss": 0.1789,
      "step": 568
    },
    {
      "epoch": 0.11943744752308984,
      "grad_norm": 0.15337203443050385,
      "learning_rate": 9.988640899725361e-05,
      "loss": 0.166,
      "step": 569
    },
    {
      "epoch": 0.11964735516372796,
      "grad_norm": 0.13052469491958618,
      "learning_rate": 9.988392716819806e-05,
      "loss": 0.1762,
      "step": 570
    },
    {
      "epoch": 0.11985726280436608,
      "grad_norm": 0.17160564661026,
      "learning_rate": 9.988141855033415e-05,
      "loss": 0.1742,
      "step": 571
    },
    {
      "epoch": 0.1200671704450042,
      "grad_norm": 0.15600250661373138,
      "learning_rate": 9.987888314500906e-05,
      "loss": 0.1819,
      "step": 572
    },
    {
      "epoch": 0.12027707808564232,
      "grad_norm": 0.17436926066875458,
      "learning_rate": 9.987632095358437e-05,
      "loss": 0.154,
      "step": 573
    },
    {
      "epoch": 0.12048698572628044,
      "grad_norm": 0.1583249568939209,
      "learning_rate": 9.987373197743603e-05,
      "loss": 0.1812,
      "step": 574
    },
    {
      "epoch": 0.12069689336691855,
      "grad_norm": 0.14877773821353912,
      "learning_rate": 9.987111621795437e-05,
      "loss": 0.1591,
      "step": 575
    },
    {
      "epoch": 0.12090680100755667,
      "grad_norm": 0.13348308205604553,
      "learning_rate": 9.986847367654414e-05,
      "loss": 0.164,
      "step": 576
    },
    {
      "epoch": 0.12111670864819479,
      "grad_norm": 0.16104738414287567,
      "learning_rate": 9.986580435462443e-05,
      "loss": 0.1568,
      "step": 577
    },
    {
      "epoch": 0.12132661628883291,
      "grad_norm": 0.1393367350101471,
      "learning_rate": 9.98631082536287e-05,
      "loss": 0.1504,
      "step": 578
    },
    {
      "epoch": 0.12153652392947104,
      "grad_norm": 0.17166094481945038,
      "learning_rate": 9.986038537500488e-05,
      "loss": 0.1746,
      "step": 579
    },
    {
      "epoch": 0.12174643157010916,
      "grad_norm": 0.19518902897834778,
      "learning_rate": 9.985763572021516e-05,
      "loss": 0.1961,
      "step": 580
    },
    {
      "epoch": 0.12195633921074728,
      "grad_norm": 0.17867891490459442,
      "learning_rate": 9.985485929073619e-05,
      "loss": 0.1836,
      "step": 581
    },
    {
      "epoch": 0.12216624685138538,
      "grad_norm": 0.2270008772611618,
      "learning_rate": 9.9852056088059e-05,
      "loss": 0.1847,
      "step": 582
    },
    {
      "epoch": 0.1223761544920235,
      "grad_norm": 0.16426697373390198,
      "learning_rate": 9.984922611368892e-05,
      "loss": 0.1587,
      "step": 583
    },
    {
      "epoch": 0.12258606213266163,
      "grad_norm": 0.16188107430934906,
      "learning_rate": 9.984636936914575e-05,
      "loss": 0.1783,
      "step": 584
    },
    {
      "epoch": 0.12279596977329975,
      "grad_norm": 0.16157647967338562,
      "learning_rate": 9.984348585596361e-05,
      "loss": 0.1612,
      "step": 585
    },
    {
      "epoch": 0.12300587741393787,
      "grad_norm": 0.17590083181858063,
      "learning_rate": 9.984057557569104e-05,
      "loss": 0.1852,
      "step": 586
    },
    {
      "epoch": 0.12321578505457599,
      "grad_norm": 0.13868169486522675,
      "learning_rate": 9.983763852989088e-05,
      "loss": 0.1915,
      "step": 587
    },
    {
      "epoch": 0.12342569269521411,
      "grad_norm": 0.15027277171611786,
      "learning_rate": 9.983467472014043e-05,
      "loss": 0.1924,
      "step": 588
    },
    {
      "epoch": 0.12363560033585222,
      "grad_norm": 0.16492144763469696,
      "learning_rate": 9.983168414803132e-05,
      "loss": 0.1768,
      "step": 589
    },
    {
      "epoch": 0.12384550797649034,
      "grad_norm": 0.16169002652168274,
      "learning_rate": 9.982866681516954e-05,
      "loss": 0.1704,
      "step": 590
    },
    {
      "epoch": 0.12405541561712846,
      "grad_norm": 0.16292813420295715,
      "learning_rate": 9.982562272317546e-05,
      "loss": 0.1687,
      "step": 591
    },
    {
      "epoch": 0.12426532325776658,
      "grad_norm": 0.16432535648345947,
      "learning_rate": 9.982255187368386e-05,
      "loss": 0.1702,
      "step": 592
    },
    {
      "epoch": 0.1244752308984047,
      "grad_norm": 0.13967812061309814,
      "learning_rate": 9.981945426834382e-05,
      "loss": 0.1858,
      "step": 593
    },
    {
      "epoch": 0.12468513853904283,
      "grad_norm": 0.16663575172424316,
      "learning_rate": 9.981632990881885e-05,
      "loss": 0.158,
      "step": 594
    },
    {
      "epoch": 0.12489504617968095,
      "grad_norm": 0.17746829986572266,
      "learning_rate": 9.981317879678679e-05,
      "loss": 0.1679,
      "step": 595
    },
    {
      "epoch": 0.12510495382031905,
      "grad_norm": 0.15006081759929657,
      "learning_rate": 9.981000093393986e-05,
      "loss": 0.1659,
      "step": 596
    },
    {
      "epoch": 0.1253148614609572,
      "grad_norm": 0.15521980822086334,
      "learning_rate": 9.980679632198466e-05,
      "loss": 0.1587,
      "step": 597
    },
    {
      "epoch": 0.1255247691015953,
      "grad_norm": 0.17882917821407318,
      "learning_rate": 9.980356496264212e-05,
      "loss": 0.1571,
      "step": 598
    },
    {
      "epoch": 0.12573467674223343,
      "grad_norm": 0.17295823991298676,
      "learning_rate": 9.980030685764754e-05,
      "loss": 0.1936,
      "step": 599
    },
    {
      "epoch": 0.12594458438287154,
      "grad_norm": 0.1699322760105133,
      "learning_rate": 9.979702200875065e-05,
      "loss": 0.1555,
      "step": 600
    },
    {
      "epoch": 0.12615449202350965,
      "grad_norm": 0.15549197793006897,
      "learning_rate": 9.979371041771543e-05,
      "loss": 0.1569,
      "step": 601
    },
    {
      "epoch": 0.12636439966414778,
      "grad_norm": 0.16207100450992584,
      "learning_rate": 9.979037208632034e-05,
      "loss": 0.1609,
      "step": 602
    },
    {
      "epoch": 0.1265743073047859,
      "grad_norm": 0.156686931848526,
      "learning_rate": 9.978700701635807e-05,
      "loss": 0.165,
      "step": 603
    },
    {
      "epoch": 0.12678421494542402,
      "grad_norm": 0.16895289719104767,
      "learning_rate": 9.97836152096358e-05,
      "loss": 0.1623,
      "step": 604
    },
    {
      "epoch": 0.12699412258606213,
      "grad_norm": 0.14077980816364288,
      "learning_rate": 9.978019666797498e-05,
      "loss": 0.1573,
      "step": 605
    },
    {
      "epoch": 0.12720403022670027,
      "grad_norm": 0.14448733627796173,
      "learning_rate": 9.977675139321146e-05,
      "loss": 0.1671,
      "step": 606
    },
    {
      "epoch": 0.12741393786733837,
      "grad_norm": 0.1879337728023529,
      "learning_rate": 9.977327938719541e-05,
      "loss": 0.1644,
      "step": 607
    },
    {
      "epoch": 0.12762384550797648,
      "grad_norm": 0.22603319585323334,
      "learning_rate": 9.976978065179138e-05,
      "loss": 0.1817,
      "step": 608
    },
    {
      "epoch": 0.12783375314861462,
      "grad_norm": 0.19939488172531128,
      "learning_rate": 9.976625518887828e-05,
      "loss": 0.1925,
      "step": 609
    },
    {
      "epoch": 0.12804366078925272,
      "grad_norm": 0.17629611492156982,
      "learning_rate": 9.976270300034936e-05,
      "loss": 0.1603,
      "step": 610
    },
    {
      "epoch": 0.12825356842989086,
      "grad_norm": 0.17778918147087097,
      "learning_rate": 9.975912408811223e-05,
      "loss": 0.1849,
      "step": 611
    },
    {
      "epoch": 0.12846347607052896,
      "grad_norm": 0.2002590447664261,
      "learning_rate": 9.975551845408886e-05,
      "loss": 0.166,
      "step": 612
    },
    {
      "epoch": 0.1286733837111671,
      "grad_norm": 0.20324808359146118,
      "learning_rate": 9.975188610021553e-05,
      "loss": 0.1673,
      "step": 613
    },
    {
      "epoch": 0.1288832913518052,
      "grad_norm": 0.20010024309158325,
      "learning_rate": 9.974822702844291e-05,
      "loss": 0.1529,
      "step": 614
    },
    {
      "epoch": 0.12909319899244331,
      "grad_norm": 0.18944194912910461,
      "learning_rate": 9.974454124073603e-05,
      "loss": 0.1652,
      "step": 615
    },
    {
      "epoch": 0.12930310663308145,
      "grad_norm": 0.16956883668899536,
      "learning_rate": 9.974082873907418e-05,
      "loss": 0.1812,
      "step": 616
    },
    {
      "epoch": 0.12951301427371956,
      "grad_norm": 0.21157757937908173,
      "learning_rate": 9.973708952545111e-05,
      "loss": 0.1862,
      "step": 617
    },
    {
      "epoch": 0.1297229219143577,
      "grad_norm": 0.20554983615875244,
      "learning_rate": 9.973332360187486e-05,
      "loss": 0.1705,
      "step": 618
    },
    {
      "epoch": 0.1299328295549958,
      "grad_norm": 0.21964187920093536,
      "learning_rate": 9.97295309703678e-05,
      "loss": 0.1807,
      "step": 619
    },
    {
      "epoch": 0.13014273719563393,
      "grad_norm": 0.15438182651996613,
      "learning_rate": 9.972571163296666e-05,
      "loss": 0.1951,
      "step": 620
    },
    {
      "epoch": 0.13035264483627204,
      "grad_norm": 0.17805755138397217,
      "learning_rate": 9.972186559172253e-05,
      "loss": 0.1543,
      "step": 621
    },
    {
      "epoch": 0.13056255247691015,
      "grad_norm": 0.1914507895708084,
      "learning_rate": 9.97179928487008e-05,
      "loss": 0.1815,
      "step": 622
    },
    {
      "epoch": 0.13077246011754828,
      "grad_norm": 0.18952740728855133,
      "learning_rate": 9.971409340598123e-05,
      "loss": 0.1601,
      "step": 623
    },
    {
      "epoch": 0.1309823677581864,
      "grad_norm": 0.1550108641386032,
      "learning_rate": 9.971016726565791e-05,
      "loss": 0.1726,
      "step": 624
    },
    {
      "epoch": 0.13119227539882453,
      "grad_norm": 0.15205375850200653,
      "learning_rate": 9.970621442983929e-05,
      "loss": 0.1536,
      "step": 625
    },
    {
      "epoch": 0.13140218303946263,
      "grad_norm": 0.15574775636196136,
      "learning_rate": 9.970223490064809e-05,
      "loss": 0.167,
      "step": 626
    },
    {
      "epoch": 0.13161209068010077,
      "grad_norm": 0.18050767481327057,
      "learning_rate": 9.969822868022143e-05,
      "loss": 0.1752,
      "step": 627
    },
    {
      "epoch": 0.13182199832073888,
      "grad_norm": 0.1507614701986313,
      "learning_rate": 9.969419577071076e-05,
      "loss": 0.1712,
      "step": 628
    },
    {
      "epoch": 0.13203190596137698,
      "grad_norm": 0.20433206856250763,
      "learning_rate": 9.96901361742818e-05,
      "loss": 0.1782,
      "step": 629
    },
    {
      "epoch": 0.13224181360201512,
      "grad_norm": 0.14705337584018707,
      "learning_rate": 9.968604989311467e-05,
      "loss": 0.1554,
      "step": 630
    },
    {
      "epoch": 0.13245172124265323,
      "grad_norm": 0.1573822945356369,
      "learning_rate": 9.968193692940382e-05,
      "loss": 0.1762,
      "step": 631
    },
    {
      "epoch": 0.13266162888329136,
      "grad_norm": 0.17816203832626343,
      "learning_rate": 9.967779728535797e-05,
      "loss": 0.1783,
      "step": 632
    },
    {
      "epoch": 0.13287153652392947,
      "grad_norm": 0.1705409288406372,
      "learning_rate": 9.967363096320022e-05,
      "loss": 0.1734,
      "step": 633
    },
    {
      "epoch": 0.1330814441645676,
      "grad_norm": 0.14574241638183594,
      "learning_rate": 9.966943796516798e-05,
      "loss": 0.1599,
      "step": 634
    },
    {
      "epoch": 0.1332913518052057,
      "grad_norm": 0.1341760754585266,
      "learning_rate": 9.966521829351297e-05,
      "loss": 0.1833,
      "step": 635
    },
    {
      "epoch": 0.13350125944584382,
      "grad_norm": 0.1680106520652771,
      "learning_rate": 9.966097195050128e-05,
      "loss": 0.1701,
      "step": 636
    },
    {
      "epoch": 0.13371116708648195,
      "grad_norm": 0.1687798947095871,
      "learning_rate": 9.965669893841326e-05,
      "loss": 0.1669,
      "step": 637
    },
    {
      "epoch": 0.13392107472712006,
      "grad_norm": 0.1792893260717392,
      "learning_rate": 9.965239925954364e-05,
      "loss": 0.1705,
      "step": 638
    },
    {
      "epoch": 0.1341309823677582,
      "grad_norm": 0.15732963383197784,
      "learning_rate": 9.964807291620144e-05,
      "loss": 0.176,
      "step": 639
    },
    {
      "epoch": 0.1343408900083963,
      "grad_norm": 0.14176061749458313,
      "learning_rate": 9.964371991070999e-05,
      "loss": 0.1755,
      "step": 640
    },
    {
      "epoch": 0.13455079764903444,
      "grad_norm": 0.1526860147714615,
      "learning_rate": 9.963934024540698e-05,
      "loss": 0.1589,
      "step": 641
    },
    {
      "epoch": 0.13476070528967254,
      "grad_norm": 0.15608391165733337,
      "learning_rate": 9.963493392264435e-05,
      "loss": 0.1652,
      "step": 642
    },
    {
      "epoch": 0.13497061293031065,
      "grad_norm": 0.14473848044872284,
      "learning_rate": 9.963050094478845e-05,
      "loss": 0.1742,
      "step": 643
    },
    {
      "epoch": 0.1351805205709488,
      "grad_norm": 0.1382717341184616,
      "learning_rate": 9.962604131421984e-05,
      "loss": 0.1713,
      "step": 644
    },
    {
      "epoch": 0.1353904282115869,
      "grad_norm": 0.14345118403434753,
      "learning_rate": 9.962155503333348e-05,
      "loss": 0.1648,
      "step": 645
    },
    {
      "epoch": 0.13560033585222503,
      "grad_norm": 0.1398243010044098,
      "learning_rate": 9.96170421045386e-05,
      "loss": 0.1889,
      "step": 646
    },
    {
      "epoch": 0.13581024349286314,
      "grad_norm": 0.14956693351268768,
      "learning_rate": 9.96125025302587e-05,
      "loss": 0.1744,
      "step": 647
    },
    {
      "epoch": 0.13602015113350127,
      "grad_norm": 0.15064063668251038,
      "learning_rate": 9.96079363129317e-05,
      "loss": 0.1625,
      "step": 648
    },
    {
      "epoch": 0.13623005877413938,
      "grad_norm": 0.15413698554039001,
      "learning_rate": 9.960334345500974e-05,
      "loss": 0.1745,
      "step": 649
    },
    {
      "epoch": 0.13643996641477749,
      "grad_norm": 0.18636789917945862,
      "learning_rate": 9.959872395895929e-05,
      "loss": 0.1639,
      "step": 650
    },
    {
      "epoch": 0.13664987405541562,
      "grad_norm": 0.2444021999835968,
      "learning_rate": 9.959407782726108e-05,
      "loss": 0.1826,
      "step": 651
    },
    {
      "epoch": 0.13685978169605373,
      "grad_norm": 0.16845685243606567,
      "learning_rate": 9.958940506241026e-05,
      "loss": 0.1793,
      "step": 652
    },
    {
      "epoch": 0.13706968933669186,
      "grad_norm": 0.16233941912651062,
      "learning_rate": 9.958470566691618e-05,
      "loss": 0.18,
      "step": 653
    },
    {
      "epoch": 0.13727959697732997,
      "grad_norm": 0.18665330111980438,
      "learning_rate": 9.95799796433025e-05,
      "loss": 0.1662,
      "step": 654
    },
    {
      "epoch": 0.1374895046179681,
      "grad_norm": 0.18732531368732452,
      "learning_rate": 9.957522699410723e-05,
      "loss": 0.1646,
      "step": 655
    },
    {
      "epoch": 0.1376994122586062,
      "grad_norm": 0.12580807507038116,
      "learning_rate": 9.957044772188266e-05,
      "loss": 0.1616,
      "step": 656
    },
    {
      "epoch": 0.13790931989924432,
      "grad_norm": 0.17410576343536377,
      "learning_rate": 9.956564182919535e-05,
      "loss": 0.1628,
      "step": 657
    },
    {
      "epoch": 0.13811922753988246,
      "grad_norm": 0.1753510683774948,
      "learning_rate": 9.95608093186262e-05,
      "loss": 0.1867,
      "step": 658
    },
    {
      "epoch": 0.13832913518052056,
      "grad_norm": 0.17664988338947296,
      "learning_rate": 9.955595019277032e-05,
      "loss": 0.1819,
      "step": 659
    },
    {
      "epoch": 0.1385390428211587,
      "grad_norm": 0.1787986010313034,
      "learning_rate": 9.955106445423722e-05,
      "loss": 0.1668,
      "step": 660
    },
    {
      "epoch": 0.1387489504617968,
      "grad_norm": 0.14035004377365112,
      "learning_rate": 9.954615210565065e-05,
      "loss": 0.1651,
      "step": 661
    },
    {
      "epoch": 0.13895885810243494,
      "grad_norm": 0.20008955895900726,
      "learning_rate": 9.954121314964864e-05,
      "loss": 0.179,
      "step": 662
    },
    {
      "epoch": 0.13916876574307305,
      "grad_norm": 0.22378800809383392,
      "learning_rate": 9.953624758888352e-05,
      "loss": 0.1729,
      "step": 663
    },
    {
      "epoch": 0.13937867338371115,
      "grad_norm": 0.18687045574188232,
      "learning_rate": 9.953125542602193e-05,
      "loss": 0.1756,
      "step": 664
    },
    {
      "epoch": 0.1395885810243493,
      "grad_norm": 0.1506877988576889,
      "learning_rate": 9.952623666374475e-05,
      "loss": 0.1616,
      "step": 665
    },
    {
      "epoch": 0.1397984886649874,
      "grad_norm": 0.20918136835098267,
      "learning_rate": 9.95211913047472e-05,
      "loss": 0.1888,
      "step": 666
    },
    {
      "epoch": 0.14000839630562553,
      "grad_norm": 0.17521612346172333,
      "learning_rate": 9.951611935173872e-05,
      "loss": 0.1886,
      "step": 667
    },
    {
      "epoch": 0.14021830394626364,
      "grad_norm": 0.14788690209388733,
      "learning_rate": 9.951102080744308e-05,
      "loss": 0.162,
      "step": 668
    },
    {
      "epoch": 0.14042821158690177,
      "grad_norm": 0.18560314178466797,
      "learning_rate": 9.950589567459832e-05,
      "loss": 0.1573,
      "step": 669
    },
    {
      "epoch": 0.14063811922753988,
      "grad_norm": 0.16913674771785736,
      "learning_rate": 9.950074395595675e-05,
      "loss": 0.1713,
      "step": 670
    },
    {
      "epoch": 0.140848026868178,
      "grad_norm": 0.13768184185028076,
      "learning_rate": 9.949556565428496e-05,
      "loss": 0.1733,
      "step": 671
    },
    {
      "epoch": 0.14105793450881612,
      "grad_norm": 0.19472239911556244,
      "learning_rate": 9.949036077236382e-05,
      "loss": 0.1638,
      "step": 672
    },
    {
      "epoch": 0.14126784214945423,
      "grad_norm": 0.17684867978096008,
      "learning_rate": 9.948512931298846e-05,
      "loss": 0.1686,
      "step": 673
    },
    {
      "epoch": 0.14147774979009237,
      "grad_norm": 0.20061515271663666,
      "learning_rate": 9.94798712789683e-05,
      "loss": 0.1921,
      "step": 674
    },
    {
      "epoch": 0.14168765743073047,
      "grad_norm": 0.17213481664657593,
      "learning_rate": 9.9474586673127e-05,
      "loss": 0.1636,
      "step": 675
    },
    {
      "epoch": 0.1418975650713686,
      "grad_norm": 0.16144217550754547,
      "learning_rate": 9.946927549830258e-05,
      "loss": 0.1594,
      "step": 676
    },
    {
      "epoch": 0.14210747271200672,
      "grad_norm": 0.16045495867729187,
      "learning_rate": 9.946393775734719e-05,
      "loss": 0.1585,
      "step": 677
    },
    {
      "epoch": 0.14231738035264482,
      "grad_norm": 0.168419748544693,
      "learning_rate": 9.945857345312735e-05,
      "loss": 0.1618,
      "step": 678
    },
    {
      "epoch": 0.14252728799328296,
      "grad_norm": 0.16631141304969788,
      "learning_rate": 9.945318258852383e-05,
      "loss": 0.1648,
      "step": 679
    },
    {
      "epoch": 0.14273719563392107,
      "grad_norm": 0.17133933305740356,
      "learning_rate": 9.944776516643161e-05,
      "loss": 0.1902,
      "step": 680
    },
    {
      "epoch": 0.1429471032745592,
      "grad_norm": 0.144994854927063,
      "learning_rate": 9.944232118976e-05,
      "loss": 0.1645,
      "step": 681
    },
    {
      "epoch": 0.1431570109151973,
      "grad_norm": 0.13521502912044525,
      "learning_rate": 9.943685066143252e-05,
      "loss": 0.1679,
      "step": 682
    },
    {
      "epoch": 0.14336691855583544,
      "grad_norm": 0.1505574733018875,
      "learning_rate": 9.943135358438698e-05,
      "loss": 0.1497,
      "step": 683
    },
    {
      "epoch": 0.14357682619647355,
      "grad_norm": 0.1701841652393341,
      "learning_rate": 9.942582996157544e-05,
      "loss": 0.141,
      "step": 684
    },
    {
      "epoch": 0.14378673383711166,
      "grad_norm": 0.16892337799072266,
      "learning_rate": 9.94202797959642e-05,
      "loss": 0.1845,
      "step": 685
    },
    {
      "epoch": 0.1439966414777498,
      "grad_norm": 0.1322741061449051,
      "learning_rate": 9.941470309053384e-05,
      "loss": 0.1635,
      "step": 686
    },
    {
      "epoch": 0.1442065491183879,
      "grad_norm": 0.18180270493030548,
      "learning_rate": 9.940909984827915e-05,
      "loss": 0.1521,
      "step": 687
    },
    {
      "epoch": 0.14441645675902604,
      "grad_norm": 0.17136745154857635,
      "learning_rate": 9.940347007220924e-05,
      "loss": 0.1694,
      "step": 688
    },
    {
      "epoch": 0.14462636439966414,
      "grad_norm": 0.16122983396053314,
      "learning_rate": 9.93978137653474e-05,
      "loss": 0.1726,
      "step": 689
    },
    {
      "epoch": 0.14483627204030228,
      "grad_norm": 0.15024663507938385,
      "learning_rate": 9.939213093073118e-05,
      "loss": 0.1703,
      "step": 690
    },
    {
      "epoch": 0.14504617968094038,
      "grad_norm": 0.14193399250507355,
      "learning_rate": 9.938642157141245e-05,
      "loss": 0.1837,
      "step": 691
    },
    {
      "epoch": 0.1452560873215785,
      "grad_norm": 0.15650463104248047,
      "learning_rate": 9.938068569045721e-05,
      "loss": 0.1665,
      "step": 692
    },
    {
      "epoch": 0.14546599496221663,
      "grad_norm": 0.14254000782966614,
      "learning_rate": 9.937492329094577e-05,
      "loss": 0.175,
      "step": 693
    },
    {
      "epoch": 0.14567590260285473,
      "grad_norm": 0.17051447927951813,
      "learning_rate": 9.93691343759727e-05,
      "loss": 0.1664,
      "step": 694
    },
    {
      "epoch": 0.14588581024349287,
      "grad_norm": 0.16692955791950226,
      "learning_rate": 9.936331894864677e-05,
      "loss": 0.1682,
      "step": 695
    },
    {
      "epoch": 0.14609571788413098,
      "grad_norm": 0.20158237218856812,
      "learning_rate": 9.935747701209096e-05,
      "loss": 0.1574,
      "step": 696
    },
    {
      "epoch": 0.1463056255247691,
      "grad_norm": 0.1351911425590515,
      "learning_rate": 9.935160856944257e-05,
      "loss": 0.1618,
      "step": 697
    },
    {
      "epoch": 0.14651553316540722,
      "grad_norm": 0.1611570417881012,
      "learning_rate": 9.934571362385305e-05,
      "loss": 0.164,
      "step": 698
    },
    {
      "epoch": 0.14672544080604533,
      "grad_norm": 0.22555968165397644,
      "learning_rate": 9.933979217848815e-05,
      "loss": 0.2044,
      "step": 699
    },
    {
      "epoch": 0.14693534844668346,
      "grad_norm": 0.17471735179424286,
      "learning_rate": 9.93338442365278e-05,
      "loss": 0.1809,
      "step": 700
    },
    {
      "epoch": 0.14714525608732157,
      "grad_norm": 0.12285126745700836,
      "learning_rate": 9.93278698011662e-05,
      "loss": 0.1644,
      "step": 701
    },
    {
      "epoch": 0.1473551637279597,
      "grad_norm": 0.1364145576953888,
      "learning_rate": 9.932186887561175e-05,
      "loss": 0.1655,
      "step": 702
    },
    {
      "epoch": 0.1475650713685978,
      "grad_norm": 0.14189580082893372,
      "learning_rate": 9.931584146308708e-05,
      "loss": 0.1729,
      "step": 703
    },
    {
      "epoch": 0.14777497900923595,
      "grad_norm": 0.2254076600074768,
      "learning_rate": 9.930978756682905e-05,
      "loss": 0.1731,
      "step": 704
    },
    {
      "epoch": 0.14798488664987405,
      "grad_norm": 0.15398503839969635,
      "learning_rate": 9.930370719008875e-05,
      "loss": 0.1813,
      "step": 705
    },
    {
      "epoch": 0.1481947942905122,
      "grad_norm": 0.15267014503479004,
      "learning_rate": 9.929760033613146e-05,
      "loss": 0.1681,
      "step": 706
    },
    {
      "epoch": 0.1484047019311503,
      "grad_norm": 0.1611442267894745,
      "learning_rate": 9.929146700823671e-05,
      "loss": 0.1925,
      "step": 707
    },
    {
      "epoch": 0.1486146095717884,
      "grad_norm": 0.15641498565673828,
      "learning_rate": 9.928530720969827e-05,
      "loss": 0.1733,
      "step": 708
    },
    {
      "epoch": 0.14882451721242654,
      "grad_norm": 0.12724818289279938,
      "learning_rate": 9.927912094382403e-05,
      "loss": 0.1735,
      "step": 709
    },
    {
      "epoch": 0.14903442485306465,
      "grad_norm": 0.17570020258426666,
      "learning_rate": 9.92729082139362e-05,
      "loss": 0.1732,
      "step": 710
    },
    {
      "epoch": 0.14924433249370278,
      "grad_norm": 0.19477427005767822,
      "learning_rate": 9.926666902337115e-05,
      "loss": 0.1551,
      "step": 711
    },
    {
      "epoch": 0.1494542401343409,
      "grad_norm": 0.16798420250415802,
      "learning_rate": 9.926040337547946e-05,
      "loss": 0.1906,
      "step": 712
    },
    {
      "epoch": 0.14966414777497902,
      "grad_norm": 0.17551501095294952,
      "learning_rate": 9.925411127362594e-05,
      "loss": 0.1747,
      "step": 713
    },
    {
      "epoch": 0.14987405541561713,
      "grad_norm": 0.1842852234840393,
      "learning_rate": 9.924779272118957e-05,
      "loss": 0.1563,
      "step": 714
    },
    {
      "epoch": 0.15008396305625524,
      "grad_norm": 0.1805478185415268,
      "learning_rate": 9.924144772156358e-05,
      "loss": 0.165,
      "step": 715
    },
    {
      "epoch": 0.15029387069689337,
      "grad_norm": 0.21542084217071533,
      "learning_rate": 9.923507627815536e-05,
      "loss": 0.1469,
      "step": 716
    },
    {
      "epoch": 0.15050377833753148,
      "grad_norm": 0.15649262070655823,
      "learning_rate": 9.922867839438654e-05,
      "loss": 0.1706,
      "step": 717
    },
    {
      "epoch": 0.15071368597816961,
      "grad_norm": 0.14313052594661713,
      "learning_rate": 9.92222540736929e-05,
      "loss": 0.1737,
      "step": 718
    },
    {
      "epoch": 0.15092359361880772,
      "grad_norm": 0.13190749287605286,
      "learning_rate": 9.92158033195245e-05,
      "loss": 0.1742,
      "step": 719
    },
    {
      "epoch": 0.15113350125944586,
      "grad_norm": 0.18187177181243896,
      "learning_rate": 9.920932613534549e-05,
      "loss": 0.1748,
      "step": 720
    },
    {
      "epoch": 0.15134340890008396,
      "grad_norm": 0.14302362501621246,
      "learning_rate": 9.920282252463429e-05,
      "loss": 0.1711,
      "step": 721
    },
    {
      "epoch": 0.15155331654072207,
      "grad_norm": 0.16898708045482635,
      "learning_rate": 9.919629249088347e-05,
      "loss": 0.1616,
      "step": 722
    },
    {
      "epoch": 0.1517632241813602,
      "grad_norm": 0.15528154373168945,
      "learning_rate": 9.918973603759984e-05,
      "loss": 0.1652,
      "step": 723
    },
    {
      "epoch": 0.15197313182199831,
      "grad_norm": 0.1718195527791977,
      "learning_rate": 9.918315316830434e-05,
      "loss": 0.1758,
      "step": 724
    },
    {
      "epoch": 0.15218303946263645,
      "grad_norm": 0.15404529869556427,
      "learning_rate": 9.917654388653211e-05,
      "loss": 0.162,
      "step": 725
    },
    {
      "epoch": 0.15239294710327456,
      "grad_norm": 0.19148094952106476,
      "learning_rate": 9.916990819583252e-05,
      "loss": 0.1887,
      "step": 726
    },
    {
      "epoch": 0.1526028547439127,
      "grad_norm": 0.21298371255397797,
      "learning_rate": 9.916324609976906e-05,
      "loss": 0.1712,
      "step": 727
    },
    {
      "epoch": 0.1528127623845508,
      "grad_norm": 0.2041487991809845,
      "learning_rate": 9.915655760191944e-05,
      "loss": 0.167,
      "step": 728
    },
    {
      "epoch": 0.1530226700251889,
      "grad_norm": 0.17242367565631866,
      "learning_rate": 9.914984270587552e-05,
      "loss": 0.163,
      "step": 729
    },
    {
      "epoch": 0.15323257766582704,
      "grad_norm": 0.1646365076303482,
      "learning_rate": 9.914310141524339e-05,
      "loss": 0.1638,
      "step": 730
    },
    {
      "epoch": 0.15344248530646515,
      "grad_norm": 0.1947726458311081,
      "learning_rate": 9.913633373364324e-05,
      "loss": 0.1828,
      "step": 731
    },
    {
      "epoch": 0.15365239294710328,
      "grad_norm": 0.17962804436683655,
      "learning_rate": 9.912953966470948e-05,
      "loss": 0.1699,
      "step": 732
    },
    {
      "epoch": 0.1538623005877414,
      "grad_norm": 0.17121249437332153,
      "learning_rate": 9.912271921209068e-05,
      "loss": 0.1605,
      "step": 733
    },
    {
      "epoch": 0.15407220822837953,
      "grad_norm": 0.17192313075065613,
      "learning_rate": 9.911587237944959e-05,
      "loss": 0.1746,
      "step": 734
    },
    {
      "epoch": 0.15428211586901763,
      "grad_norm": 0.1262722760438919,
      "learning_rate": 9.910899917046311e-05,
      "loss": 0.1677,
      "step": 735
    },
    {
      "epoch": 0.15449202350965574,
      "grad_norm": 0.15364903211593628,
      "learning_rate": 9.910209958882231e-05,
      "loss": 0.1792,
      "step": 736
    },
    {
      "epoch": 0.15470193115029388,
      "grad_norm": 0.17038469016551971,
      "learning_rate": 9.909517363823241e-05,
      "loss": 0.1588,
      "step": 737
    },
    {
      "epoch": 0.15491183879093198,
      "grad_norm": 0.14676110446453094,
      "learning_rate": 9.908822132241281e-05,
      "loss": 0.1639,
      "step": 738
    },
    {
      "epoch": 0.15512174643157012,
      "grad_norm": 0.15300148725509644,
      "learning_rate": 9.908124264509707e-05,
      "loss": 0.1752,
      "step": 739
    },
    {
      "epoch": 0.15533165407220823,
      "grad_norm": 0.1372271180152893,
      "learning_rate": 9.90742376100329e-05,
      "loss": 0.1728,
      "step": 740
    },
    {
      "epoch": 0.15554156171284636,
      "grad_norm": 0.13885709643363953,
      "learning_rate": 9.906720622098215e-05,
      "loss": 0.1625,
      "step": 741
    },
    {
      "epoch": 0.15575146935348447,
      "grad_norm": 0.12582066655158997,
      "learning_rate": 9.906014848172086e-05,
      "loss": 0.17,
      "step": 742
    },
    {
      "epoch": 0.15596137699412257,
      "grad_norm": 0.13245654106140137,
      "learning_rate": 9.905306439603918e-05,
      "loss": 0.1711,
      "step": 743
    },
    {
      "epoch": 0.1561712846347607,
      "grad_norm": 0.12488420307636261,
      "learning_rate": 9.904595396774142e-05,
      "loss": 0.1725,
      "step": 744
    },
    {
      "epoch": 0.15638119227539882,
      "grad_norm": 0.1577821522951126,
      "learning_rate": 9.903881720064606e-05,
      "loss": 0.1707,
      "step": 745
    },
    {
      "epoch": 0.15659109991603695,
      "grad_norm": 0.16098785400390625,
      "learning_rate": 9.903165409858567e-05,
      "loss": 0.1608,
      "step": 746
    },
    {
      "epoch": 0.15680100755667506,
      "grad_norm": 0.14498507976531982,
      "learning_rate": 9.902446466540707e-05,
      "loss": 0.1776,
      "step": 747
    },
    {
      "epoch": 0.1570109151973132,
      "grad_norm": 0.15581083297729492,
      "learning_rate": 9.901724890497109e-05,
      "loss": 0.1746,
      "step": 748
    },
    {
      "epoch": 0.1572208228379513,
      "grad_norm": 0.16081508994102478,
      "learning_rate": 9.901000682115276e-05,
      "loss": 0.1869,
      "step": 749
    },
    {
      "epoch": 0.1574307304785894,
      "grad_norm": 0.16249604523181915,
      "learning_rate": 9.900273841784126e-05,
      "loss": 0.1629,
      "step": 750
    },
    {
      "epoch": 0.15764063811922754,
      "grad_norm": 0.1613830178976059,
      "learning_rate": 9.899544369893992e-05,
      "loss": 0.1588,
      "step": 751
    },
    {
      "epoch": 0.15785054575986565,
      "grad_norm": 0.15741245448589325,
      "learning_rate": 9.898812266836613e-05,
      "loss": 0.1708,
      "step": 752
    },
    {
      "epoch": 0.1580604534005038,
      "grad_norm": 0.12306373566389084,
      "learning_rate": 9.898077533005144e-05,
      "loss": 0.1751,
      "step": 753
    },
    {
      "epoch": 0.1582703610411419,
      "grad_norm": 0.15071968734264374,
      "learning_rate": 9.897340168794155e-05,
      "loss": 0.1686,
      "step": 754
    },
    {
      "epoch": 0.15848026868178003,
      "grad_norm": 0.14102348685264587,
      "learning_rate": 9.896600174599632e-05,
      "loss": 0.1701,
      "step": 755
    },
    {
      "epoch": 0.15869017632241814,
      "grad_norm": 0.16288180649280548,
      "learning_rate": 9.895857550818963e-05,
      "loss": 0.1652,
      "step": 756
    },
    {
      "epoch": 0.15890008396305624,
      "grad_norm": 0.17342409491539001,
      "learning_rate": 9.895112297850956e-05,
      "loss": 0.1702,
      "step": 757
    },
    {
      "epoch": 0.15910999160369438,
      "grad_norm": 0.13485445082187653,
      "learning_rate": 9.894364416095829e-05,
      "loss": 0.1705,
      "step": 758
    },
    {
      "epoch": 0.15931989924433249,
      "grad_norm": 0.17127734422683716,
      "learning_rate": 9.893613905955211e-05,
      "loss": 0.1831,
      "step": 759
    },
    {
      "epoch": 0.15952980688497062,
      "grad_norm": 0.13848379254341125,
      "learning_rate": 9.892860767832144e-05,
      "loss": 0.161,
      "step": 760
    },
    {
      "epoch": 0.15973971452560873,
      "grad_norm": 0.12154096364974976,
      "learning_rate": 9.892105002131081e-05,
      "loss": 0.1637,
      "step": 761
    },
    {
      "epoch": 0.15994962216624686,
      "grad_norm": 0.15433917939662933,
      "learning_rate": 9.891346609257882e-05,
      "loss": 0.1765,
      "step": 762
    },
    {
      "epoch": 0.16015952980688497,
      "grad_norm": 0.16946150362491608,
      "learning_rate": 9.890585589619825e-05,
      "loss": 0.1766,
      "step": 763
    },
    {
      "epoch": 0.16036943744752308,
      "grad_norm": 0.14503143727779388,
      "learning_rate": 9.889821943625594e-05,
      "loss": 0.1458,
      "step": 764
    },
    {
      "epoch": 0.1605793450881612,
      "grad_norm": 0.16491486132144928,
      "learning_rate": 9.889055671685283e-05,
      "loss": 0.1861,
      "step": 765
    },
    {
      "epoch": 0.16078925272879932,
      "grad_norm": 0.15087951719760895,
      "learning_rate": 9.888286774210398e-05,
      "loss": 0.1745,
      "step": 766
    },
    {
      "epoch": 0.16099916036943746,
      "grad_norm": 0.19266510009765625,
      "learning_rate": 9.887515251613857e-05,
      "loss": 0.1711,
      "step": 767
    },
    {
      "epoch": 0.16120906801007556,
      "grad_norm": 0.19419905543327332,
      "learning_rate": 9.886741104309981e-05,
      "loss": 0.1705,
      "step": 768
    },
    {
      "epoch": 0.1614189756507137,
      "grad_norm": 0.13408306241035461,
      "learning_rate": 9.885964332714508e-05,
      "loss": 0.1629,
      "step": 769
    },
    {
      "epoch": 0.1616288832913518,
      "grad_norm": 0.17757900059223175,
      "learning_rate": 9.885184937244581e-05,
      "loss": 0.1643,
      "step": 770
    },
    {
      "epoch": 0.1618387909319899,
      "grad_norm": 0.16376204788684845,
      "learning_rate": 9.884402918318754e-05,
      "loss": 0.1869,
      "step": 771
    },
    {
      "epoch": 0.16204869857262805,
      "grad_norm": 0.1362319141626358,
      "learning_rate": 9.883618276356988e-05,
      "loss": 0.1736,
      "step": 772
    },
    {
      "epoch": 0.16225860621326615,
      "grad_norm": 0.15015824139118195,
      "learning_rate": 9.882831011780653e-05,
      "loss": 0.1598,
      "step": 773
    },
    {
      "epoch": 0.1624685138539043,
      "grad_norm": 0.1815597116947174,
      "learning_rate": 9.882041125012528e-05,
      "loss": 0.1597,
      "step": 774
    },
    {
      "epoch": 0.1626784214945424,
      "grad_norm": 0.16493958234786987,
      "learning_rate": 9.881248616476803e-05,
      "loss": 0.181,
      "step": 775
    },
    {
      "epoch": 0.16288832913518053,
      "grad_norm": 0.1213487982749939,
      "learning_rate": 9.880453486599072e-05,
      "loss": 0.1644,
      "step": 776
    },
    {
      "epoch": 0.16309823677581864,
      "grad_norm": 0.1440252661705017,
      "learning_rate": 9.879655735806337e-05,
      "loss": 0.1657,
      "step": 777
    },
    {
      "epoch": 0.16330814441645675,
      "grad_norm": 0.1406947672367096,
      "learning_rate": 9.878855364527007e-05,
      "loss": 0.172,
      "step": 778
    },
    {
      "epoch": 0.16351805205709488,
      "grad_norm": 0.17073537409305573,
      "learning_rate": 9.878052373190902e-05,
      "loss": 0.1852,
      "step": 779
    },
    {
      "epoch": 0.163727959697733,
      "grad_norm": 0.15924111008644104,
      "learning_rate": 9.877246762229247e-05,
      "loss": 0.18,
      "step": 780
    },
    {
      "epoch": 0.16393786733837112,
      "grad_norm": 0.14216449856758118,
      "learning_rate": 9.876438532074672e-05,
      "loss": 0.1575,
      "step": 781
    },
    {
      "epoch": 0.16414777497900923,
      "grad_norm": 0.12731285393238068,
      "learning_rate": 9.875627683161217e-05,
      "loss": 0.161,
      "step": 782
    },
    {
      "epoch": 0.16435768261964737,
      "grad_norm": 0.16211947798728943,
      "learning_rate": 9.874814215924324e-05,
      "loss": 0.1592,
      "step": 783
    },
    {
      "epoch": 0.16456759026028547,
      "grad_norm": 0.1463523507118225,
      "learning_rate": 9.873998130800844e-05,
      "loss": 0.1773,
      "step": 784
    },
    {
      "epoch": 0.16477749790092358,
      "grad_norm": 0.14601489901542664,
      "learning_rate": 9.873179428229033e-05,
      "loss": 0.1685,
      "step": 785
    },
    {
      "epoch": 0.16498740554156172,
      "grad_norm": 0.1590069830417633,
      "learning_rate": 9.872358108648557e-05,
      "loss": 0.1709,
      "step": 786
    },
    {
      "epoch": 0.16519731318219982,
      "grad_norm": 0.14027854800224304,
      "learning_rate": 9.871534172500479e-05,
      "loss": 0.1842,
      "step": 787
    },
    {
      "epoch": 0.16540722082283796,
      "grad_norm": 0.12902334332466125,
      "learning_rate": 9.870707620227271e-05,
      "loss": 0.1484,
      "step": 788
    },
    {
      "epoch": 0.16561712846347607,
      "grad_norm": 0.13427670300006866,
      "learning_rate": 9.869878452272812e-05,
      "loss": 0.1599,
      "step": 789
    },
    {
      "epoch": 0.1658270361041142,
      "grad_norm": 0.17016825079917908,
      "learning_rate": 9.869046669082386e-05,
      "loss": 0.1506,
      "step": 790
    },
    {
      "epoch": 0.1660369437447523,
      "grad_norm": 0.13757233321666718,
      "learning_rate": 9.868212271102678e-05,
      "loss": 0.1574,
      "step": 791
    },
    {
      "epoch": 0.16624685138539042,
      "grad_norm": 0.19485127925872803,
      "learning_rate": 9.867375258781778e-05,
      "loss": 0.1686,
      "step": 792
    },
    {
      "epoch": 0.16645675902602855,
      "grad_norm": 0.16131381690502167,
      "learning_rate": 9.866535632569182e-05,
      "loss": 0.1693,
      "step": 793
    },
    {
      "epoch": 0.16666666666666666,
      "grad_norm": 0.16035085916519165,
      "learning_rate": 9.865693392915787e-05,
      "loss": 0.1589,
      "step": 794
    },
    {
      "epoch": 0.1668765743073048,
      "grad_norm": 0.16287830471992493,
      "learning_rate": 9.864848540273897e-05,
      "loss": 0.1749,
      "step": 795
    },
    {
      "epoch": 0.1670864819479429,
      "grad_norm": 0.16353754699230194,
      "learning_rate": 9.864001075097214e-05,
      "loss": 0.1846,
      "step": 796
    },
    {
      "epoch": 0.16729638958858103,
      "grad_norm": 0.19826947152614594,
      "learning_rate": 9.863150997840849e-05,
      "loss": 0.1624,
      "step": 797
    },
    {
      "epoch": 0.16750629722921914,
      "grad_norm": 0.14296779036521912,
      "learning_rate": 9.862298308961313e-05,
      "loss": 0.1823,
      "step": 798
    },
    {
      "epoch": 0.16771620486985725,
      "grad_norm": 0.1371513456106186,
      "learning_rate": 9.861443008916517e-05,
      "loss": 0.1742,
      "step": 799
    },
    {
      "epoch": 0.16792611251049538,
      "grad_norm": 0.17043526470661163,
      "learning_rate": 9.86058509816578e-05,
      "loss": 0.1661,
      "step": 800
    },
    {
      "epoch": 0.1681360201511335,
      "grad_norm": 0.17045053839683533,
      "learning_rate": 9.859724577169815e-05,
      "loss": 0.1773,
      "step": 801
    },
    {
      "epoch": 0.16834592779177163,
      "grad_norm": 0.1564718633890152,
      "learning_rate": 9.858861446390748e-05,
      "loss": 0.1636,
      "step": 802
    },
    {
      "epoch": 0.16855583543240973,
      "grad_norm": 0.13335298001766205,
      "learning_rate": 9.857995706292092e-05,
      "loss": 0.18,
      "step": 803
    },
    {
      "epoch": 0.16876574307304787,
      "grad_norm": 0.14552432298660278,
      "learning_rate": 9.857127357338775e-05,
      "loss": 0.1599,
      "step": 804
    },
    {
      "epoch": 0.16897565071368598,
      "grad_norm": 0.1437654346227646,
      "learning_rate": 9.856256399997119e-05,
      "loss": 0.178,
      "step": 805
    },
    {
      "epoch": 0.16918555835432408,
      "grad_norm": 0.143344447016716,
      "learning_rate": 9.855382834734848e-05,
      "loss": 0.159,
      "step": 806
    },
    {
      "epoch": 0.16939546599496222,
      "grad_norm": 0.15504872798919678,
      "learning_rate": 9.854506662021085e-05,
      "loss": 0.1944,
      "step": 807
    },
    {
      "epoch": 0.16960537363560033,
      "grad_norm": 0.12894919514656067,
      "learning_rate": 9.853627882326357e-05,
      "loss": 0.1677,
      "step": 808
    },
    {
      "epoch": 0.16981528127623846,
      "grad_norm": 0.1339186728000641,
      "learning_rate": 9.852746496122587e-05,
      "loss": 0.1666,
      "step": 809
    },
    {
      "epoch": 0.17002518891687657,
      "grad_norm": 0.11661262065172195,
      "learning_rate": 9.8518625038831e-05,
      "loss": 0.1611,
      "step": 810
    },
    {
      "epoch": 0.1702350965575147,
      "grad_norm": 0.14091713726520538,
      "learning_rate": 9.85097590608262e-05,
      "loss": 0.1483,
      "step": 811
    },
    {
      "epoch": 0.1704450041981528,
      "grad_norm": 0.13658110797405243,
      "learning_rate": 9.85008670319727e-05,
      "loss": 0.1667,
      "step": 812
    },
    {
      "epoch": 0.17065491183879095,
      "grad_norm": 0.13288968801498413,
      "learning_rate": 9.849194895704575e-05,
      "loss": 0.1685,
      "step": 813
    },
    {
      "epoch": 0.17086481947942905,
      "grad_norm": 0.1870601624250412,
      "learning_rate": 9.84830048408345e-05,
      "loss": 0.1794,
      "step": 814
    },
    {
      "epoch": 0.17107472712006716,
      "grad_norm": 0.15742088854312897,
      "learning_rate": 9.84740346881422e-05,
      "loss": 0.1546,
      "step": 815
    },
    {
      "epoch": 0.1712846347607053,
      "grad_norm": 0.1625792235136032,
      "learning_rate": 9.846503850378602e-05,
      "loss": 0.1506,
      "step": 816
    },
    {
      "epoch": 0.1714945424013434,
      "grad_norm": 0.1315995752811432,
      "learning_rate": 9.845601629259708e-05,
      "loss": 0.1611,
      "step": 817
    },
    {
      "epoch": 0.17170445004198154,
      "grad_norm": 0.16079425811767578,
      "learning_rate": 9.844696805942055e-05,
      "loss": 0.1588,
      "step": 818
    },
    {
      "epoch": 0.17191435768261965,
      "grad_norm": 0.16755861043930054,
      "learning_rate": 9.843789380911554e-05,
      "loss": 0.1629,
      "step": 819
    },
    {
      "epoch": 0.17212426532325778,
      "grad_norm": 0.12923209369182587,
      "learning_rate": 9.84287935465551e-05,
      "loss": 0.1675,
      "step": 820
    },
    {
      "epoch": 0.1723341729638959,
      "grad_norm": 0.14494512975215912,
      "learning_rate": 9.84196672766263e-05,
      "loss": 0.1674,
      "step": 821
    },
    {
      "epoch": 0.172544080604534,
      "grad_norm": 0.13111597299575806,
      "learning_rate": 9.841051500423014e-05,
      "loss": 0.1835,
      "step": 822
    },
    {
      "epoch": 0.17275398824517213,
      "grad_norm": 0.1468263864517212,
      "learning_rate": 9.840133673428162e-05,
      "loss": 0.1635,
      "step": 823
    },
    {
      "epoch": 0.17296389588581024,
      "grad_norm": 0.14243565499782562,
      "learning_rate": 9.839213247170967e-05,
      "loss": 0.1617,
      "step": 824
    },
    {
      "epoch": 0.17317380352644837,
      "grad_norm": 0.14563241600990295,
      "learning_rate": 9.838290222145718e-05,
      "loss": 0.1614,
      "step": 825
    },
    {
      "epoch": 0.17338371116708648,
      "grad_norm": 0.15182636678218842,
      "learning_rate": 9.837364598848102e-05,
      "loss": 0.1712,
      "step": 826
    },
    {
      "epoch": 0.17359361880772461,
      "grad_norm": 0.15270951390266418,
      "learning_rate": 9.8364363777752e-05,
      "loss": 0.1721,
      "step": 827
    },
    {
      "epoch": 0.17380352644836272,
      "grad_norm": 0.14734511077404022,
      "learning_rate": 9.835505559425487e-05,
      "loss": 0.1751,
      "step": 828
    },
    {
      "epoch": 0.17401343408900083,
      "grad_norm": 0.13699984550476074,
      "learning_rate": 9.834572144298834e-05,
      "loss": 0.1895,
      "step": 829
    },
    {
      "epoch": 0.17422334172963896,
      "grad_norm": 0.1422858089208603,
      "learning_rate": 9.833636132896505e-05,
      "loss": 0.1757,
      "step": 830
    },
    {
      "epoch": 0.17443324937027707,
      "grad_norm": 0.1300913542509079,
      "learning_rate": 9.832697525721161e-05,
      "loss": 0.1774,
      "step": 831
    },
    {
      "epoch": 0.1746431570109152,
      "grad_norm": 0.16247855126857758,
      "learning_rate": 9.831756323276856e-05,
      "loss": 0.1499,
      "step": 832
    },
    {
      "epoch": 0.1748530646515533,
      "grad_norm": 0.13687850534915924,
      "learning_rate": 9.830812526069036e-05,
      "loss": 0.1739,
      "step": 833
    },
    {
      "epoch": 0.17506297229219145,
      "grad_norm": 0.16242316365242004,
      "learning_rate": 9.829866134604543e-05,
      "loss": 0.1703,
      "step": 834
    },
    {
      "epoch": 0.17527287993282956,
      "grad_norm": 0.1983039677143097,
      "learning_rate": 9.82891714939161e-05,
      "loss": 0.1726,
      "step": 835
    },
    {
      "epoch": 0.17548278757346766,
      "grad_norm": 0.17525269091129303,
      "learning_rate": 9.827965570939861e-05,
      "loss": 0.1839,
      "step": 836
    },
    {
      "epoch": 0.1756926952141058,
      "grad_norm": 0.1517772525548935,
      "learning_rate": 9.827011399760319e-05,
      "loss": 0.17,
      "step": 837
    },
    {
      "epoch": 0.1759026028547439,
      "grad_norm": 0.17186671495437622,
      "learning_rate": 9.826054636365396e-05,
      "loss": 0.1526,
      "step": 838
    },
    {
      "epoch": 0.17611251049538204,
      "grad_norm": 0.15066854655742645,
      "learning_rate": 9.825095281268894e-05,
      "loss": 0.1442,
      "step": 839
    },
    {
      "epoch": 0.17632241813602015,
      "grad_norm": 0.14937272667884827,
      "learning_rate": 9.82413333498601e-05,
      "loss": 0.1641,
      "step": 840
    },
    {
      "epoch": 0.17653232577665828,
      "grad_norm": 0.1843566745519638,
      "learning_rate": 9.823168798033328e-05,
      "loss": 0.1674,
      "step": 841
    },
    {
      "epoch": 0.1767422334172964,
      "grad_norm": 0.15555419027805328,
      "learning_rate": 9.82220167092883e-05,
      "loss": 0.1551,
      "step": 842
    },
    {
      "epoch": 0.1769521410579345,
      "grad_norm": 0.1585703343153,
      "learning_rate": 9.821231954191885e-05,
      "loss": 0.1578,
      "step": 843
    },
    {
      "epoch": 0.17716204869857263,
      "grad_norm": 0.1662077009677887,
      "learning_rate": 9.82025964834325e-05,
      "loss": 0.1778,
      "step": 844
    },
    {
      "epoch": 0.17737195633921074,
      "grad_norm": 0.1385970562696457,
      "learning_rate": 9.819284753905078e-05,
      "loss": 0.1653,
      "step": 845
    },
    {
      "epoch": 0.17758186397984888,
      "grad_norm": 0.14429986476898193,
      "learning_rate": 9.81830727140091e-05,
      "loss": 0.1801,
      "step": 846
    },
    {
      "epoch": 0.17779177162048698,
      "grad_norm": 0.16140583157539368,
      "learning_rate": 9.817327201355675e-05,
      "loss": 0.1808,
      "step": 847
    },
    {
      "epoch": 0.17800167926112512,
      "grad_norm": 0.13548845052719116,
      "learning_rate": 9.816344544295692e-05,
      "loss": 0.1567,
      "step": 848
    },
    {
      "epoch": 0.17821158690176322,
      "grad_norm": 0.14740432798862457,
      "learning_rate": 9.815359300748674e-05,
      "loss": 0.1737,
      "step": 849
    },
    {
      "epoch": 0.17842149454240133,
      "grad_norm": 0.15687352418899536,
      "learning_rate": 9.814371471243715e-05,
      "loss": 0.1775,
      "step": 850
    },
    {
      "epoch": 0.17863140218303947,
      "grad_norm": 0.14518170058727264,
      "learning_rate": 9.813381056311307e-05,
      "loss": 0.1536,
      "step": 851
    },
    {
      "epoch": 0.17884130982367757,
      "grad_norm": 0.1569397896528244,
      "learning_rate": 9.812388056483319e-05,
      "loss": 0.1667,
      "step": 852
    },
    {
      "epoch": 0.1790512174643157,
      "grad_norm": 0.13371489942073822,
      "learning_rate": 9.81139247229302e-05,
      "loss": 0.175,
      "step": 853
    },
    {
      "epoch": 0.17926112510495382,
      "grad_norm": 0.15128138661384583,
      "learning_rate": 9.810394304275058e-05,
      "loss": 0.1601,
      "step": 854
    },
    {
      "epoch": 0.17947103274559195,
      "grad_norm": 0.12941020727157593,
      "learning_rate": 9.809393552965476e-05,
      "loss": 0.1579,
      "step": 855
    },
    {
      "epoch": 0.17968094038623006,
      "grad_norm": 0.1644691675901413,
      "learning_rate": 9.808390218901696e-05,
      "loss": 0.1737,
      "step": 856
    },
    {
      "epoch": 0.17989084802686817,
      "grad_norm": 0.16271840035915375,
      "learning_rate": 9.807384302622533e-05,
      "loss": 0.1846,
      "step": 857
    },
    {
      "epoch": 0.1801007556675063,
      "grad_norm": 0.15787339210510254,
      "learning_rate": 9.806375804668189e-05,
      "loss": 0.1733,
      "step": 858
    },
    {
      "epoch": 0.1803106633081444,
      "grad_norm": 0.16740775108337402,
      "learning_rate": 9.805364725580248e-05,
      "loss": 0.1625,
      "step": 859
    },
    {
      "epoch": 0.18052057094878254,
      "grad_norm": 0.11970685422420502,
      "learning_rate": 9.804351065901682e-05,
      "loss": 0.1554,
      "step": 860
    },
    {
      "epoch": 0.18073047858942065,
      "grad_norm": 0.21359467506408691,
      "learning_rate": 9.803334826176852e-05,
      "loss": 0.1658,
      "step": 861
    },
    {
      "epoch": 0.1809403862300588,
      "grad_norm": 0.1699121594429016,
      "learning_rate": 9.8023160069515e-05,
      "loss": 0.1608,
      "step": 862
    },
    {
      "epoch": 0.1811502938706969,
      "grad_norm": 0.15073060989379883,
      "learning_rate": 9.801294608772755e-05,
      "loss": 0.156,
      "step": 863
    },
    {
      "epoch": 0.181360201511335,
      "grad_norm": 0.17342859506607056,
      "learning_rate": 9.800270632189133e-05,
      "loss": 0.1829,
      "step": 864
    },
    {
      "epoch": 0.18157010915197314,
      "grad_norm": 0.20463357865810394,
      "learning_rate": 9.799244077750531e-05,
      "loss": 0.1839,
      "step": 865
    },
    {
      "epoch": 0.18178001679261124,
      "grad_norm": 0.17843082547187805,
      "learning_rate": 9.798214946008234e-05,
      "loss": 0.1707,
      "step": 866
    },
    {
      "epoch": 0.18198992443324938,
      "grad_norm": 0.14360542595386505,
      "learning_rate": 9.797183237514907e-05,
      "loss": 0.1687,
      "step": 867
    },
    {
      "epoch": 0.18219983207388749,
      "grad_norm": 0.15701772272586823,
      "learning_rate": 9.796148952824603e-05,
      "loss": 0.1584,
      "step": 868
    },
    {
      "epoch": 0.18240973971452562,
      "grad_norm": 0.12531954050064087,
      "learning_rate": 9.795112092492755e-05,
      "loss": 0.1467,
      "step": 869
    },
    {
      "epoch": 0.18261964735516373,
      "grad_norm": 0.16143764555454254,
      "learning_rate": 9.794072657076182e-05,
      "loss": 0.1687,
      "step": 870
    },
    {
      "epoch": 0.18282955499580184,
      "grad_norm": 0.13665997982025146,
      "learning_rate": 9.793030647133084e-05,
      "loss": 0.1644,
      "step": 871
    },
    {
      "epoch": 0.18303946263643997,
      "grad_norm": 0.15797339379787445,
      "learning_rate": 9.791986063223045e-05,
      "loss": 0.1737,
      "step": 872
    },
    {
      "epoch": 0.18324937027707808,
      "grad_norm": 0.13726350665092468,
      "learning_rate": 9.790938905907027e-05,
      "loss": 0.1842,
      "step": 873
    },
    {
      "epoch": 0.1834592779177162,
      "grad_norm": 0.13706736266613007,
      "learning_rate": 9.78988917574738e-05,
      "loss": 0.1699,
      "step": 874
    },
    {
      "epoch": 0.18366918555835432,
      "grad_norm": 0.13394543528556824,
      "learning_rate": 9.788836873307835e-05,
      "loss": 0.1757,
      "step": 875
    },
    {
      "epoch": 0.18387909319899245,
      "grad_norm": 0.13909773528575897,
      "learning_rate": 9.7877819991535e-05,
      "loss": 0.1761,
      "step": 876
    },
    {
      "epoch": 0.18408900083963056,
      "grad_norm": 0.1416359543800354,
      "learning_rate": 9.786724553850865e-05,
      "loss": 0.1657,
      "step": 877
    },
    {
      "epoch": 0.18429890848026867,
      "grad_norm": 0.13720481097698212,
      "learning_rate": 9.785664537967806e-05,
      "loss": 0.173,
      "step": 878
    },
    {
      "epoch": 0.1845088161209068,
      "grad_norm": 0.14136558771133423,
      "learning_rate": 9.784601952073573e-05,
      "loss": 0.1625,
      "step": 879
    },
    {
      "epoch": 0.1847187237615449,
      "grad_norm": 0.1424713134765625,
      "learning_rate": 9.783536796738802e-05,
      "loss": 0.1676,
      "step": 880
    },
    {
      "epoch": 0.18492863140218305,
      "grad_norm": 0.163283571600914,
      "learning_rate": 9.782469072535502e-05,
      "loss": 0.1574,
      "step": 881
    },
    {
      "epoch": 0.18513853904282115,
      "grad_norm": 0.16645042598247528,
      "learning_rate": 9.781398780037067e-05,
      "loss": 0.1576,
      "step": 882
    },
    {
      "epoch": 0.1853484466834593,
      "grad_norm": 0.17722882330417633,
      "learning_rate": 9.780325919818268e-05,
      "loss": 0.1815,
      "step": 883
    },
    {
      "epoch": 0.1855583543240974,
      "grad_norm": 0.1435573697090149,
      "learning_rate": 9.779250492455257e-05,
      "loss": 0.168,
      "step": 884
    },
    {
      "epoch": 0.1857682619647355,
      "grad_norm": 0.13388660550117493,
      "learning_rate": 9.778172498525559e-05,
      "loss": 0.1493,
      "step": 885
    },
    {
      "epoch": 0.18597816960537364,
      "grad_norm": 0.15198923647403717,
      "learning_rate": 9.777091938608088e-05,
      "loss": 0.1759,
      "step": 886
    },
    {
      "epoch": 0.18618807724601175,
      "grad_norm": 0.14781691133975983,
      "learning_rate": 9.776008813283125e-05,
      "loss": 0.1521,
      "step": 887
    },
    {
      "epoch": 0.18639798488664988,
      "grad_norm": 0.18348653614521027,
      "learning_rate": 9.774923123132332e-05,
      "loss": 0.1522,
      "step": 888
    },
    {
      "epoch": 0.186607892527288,
      "grad_norm": 0.15346892178058624,
      "learning_rate": 9.773834868738752e-05,
      "loss": 0.1593,
      "step": 889
    },
    {
      "epoch": 0.18681780016792612,
      "grad_norm": 0.1563442200422287,
      "learning_rate": 9.7727440506868e-05,
      "loss": 0.1569,
      "step": 890
    },
    {
      "epoch": 0.18702770780856423,
      "grad_norm": 0.12280000746250153,
      "learning_rate": 9.771650669562274e-05,
      "loss": 0.1593,
      "step": 891
    },
    {
      "epoch": 0.18723761544920234,
      "grad_norm": 0.1622755527496338,
      "learning_rate": 9.770554725952341e-05,
      "loss": 0.1736,
      "step": 892
    },
    {
      "epoch": 0.18744752308984047,
      "grad_norm": 0.19185318052768707,
      "learning_rate": 9.769456220445549e-05,
      "loss": 0.165,
      "step": 893
    },
    {
      "epoch": 0.18765743073047858,
      "grad_norm": 0.16040024161338806,
      "learning_rate": 9.768355153631822e-05,
      "loss": 0.1837,
      "step": 894
    },
    {
      "epoch": 0.18786733837111672,
      "grad_norm": 0.11211330443620682,
      "learning_rate": 9.767251526102456e-05,
      "loss": 0.152,
      "step": 895
    },
    {
      "epoch": 0.18807724601175482,
      "grad_norm": 0.16628898680210114,
      "learning_rate": 9.766145338450125e-05,
      "loss": 0.1729,
      "step": 896
    },
    {
      "epoch": 0.18828715365239296,
      "grad_norm": 0.14260315895080566,
      "learning_rate": 9.765036591268877e-05,
      "loss": 0.165,
      "step": 897
    },
    {
      "epoch": 0.18849706129303107,
      "grad_norm": 0.12734055519104004,
      "learning_rate": 9.763925285154135e-05,
      "loss": 0.1714,
      "step": 898
    },
    {
      "epoch": 0.18870696893366917,
      "grad_norm": 0.1465056985616684,
      "learning_rate": 9.762811420702693e-05,
      "loss": 0.1805,
      "step": 899
    },
    {
      "epoch": 0.1889168765743073,
      "grad_norm": 0.12369433790445328,
      "learning_rate": 9.761694998512727e-05,
      "loss": 0.1737,
      "step": 900
    },
    {
      "epoch": 0.18912678421494541,
      "grad_norm": 0.13893358409404755,
      "learning_rate": 9.760576019183775e-05,
      "loss": 0.1502,
      "step": 901
    },
    {
      "epoch": 0.18933669185558355,
      "grad_norm": 0.13517262041568756,
      "learning_rate": 9.759454483316761e-05,
      "loss": 0.1648,
      "step": 902
    },
    {
      "epoch": 0.18954659949622166,
      "grad_norm": 0.13170479238033295,
      "learning_rate": 9.75833039151397e-05,
      "loss": 0.159,
      "step": 903
    },
    {
      "epoch": 0.1897565071368598,
      "grad_norm": 0.13293495774269104,
      "learning_rate": 9.757203744379067e-05,
      "loss": 0.1667,
      "step": 904
    },
    {
      "epoch": 0.1899664147774979,
      "grad_norm": 0.12381456047296524,
      "learning_rate": 9.756074542517088e-05,
      "loss": 0.1684,
      "step": 905
    },
    {
      "epoch": 0.190176322418136,
      "grad_norm": 0.12754730880260468,
      "learning_rate": 9.75494278653444e-05,
      "loss": 0.1559,
      "step": 906
    },
    {
      "epoch": 0.19038623005877414,
      "grad_norm": 0.17824846506118774,
      "learning_rate": 9.753808477038899e-05,
      "loss": 0.1533,
      "step": 907
    },
    {
      "epoch": 0.19059613769941225,
      "grad_norm": 0.13958828151226044,
      "learning_rate": 9.752671614639619e-05,
      "loss": 0.1494,
      "step": 908
    },
    {
      "epoch": 0.19080604534005038,
      "grad_norm": 0.14593012630939484,
      "learning_rate": 9.75153219994712e-05,
      "loss": 0.1598,
      "step": 909
    },
    {
      "epoch": 0.1910159529806885,
      "grad_norm": 0.19262051582336426,
      "learning_rate": 9.750390233573293e-05,
      "loss": 0.1638,
      "step": 910
    },
    {
      "epoch": 0.19122586062132663,
      "grad_norm": 0.15087257325649261,
      "learning_rate": 9.7492457161314e-05,
      "loss": 0.1644,
      "step": 911
    },
    {
      "epoch": 0.19143576826196473,
      "grad_norm": 0.17646470665931702,
      "learning_rate": 9.748098648236072e-05,
      "loss": 0.1644,
      "step": 912
    },
    {
      "epoch": 0.19164567590260284,
      "grad_norm": 0.15336105227470398,
      "learning_rate": 9.746949030503312e-05,
      "loss": 0.1717,
      "step": 913
    },
    {
      "epoch": 0.19185558354324098,
      "grad_norm": 0.15364870429039001,
      "learning_rate": 9.745796863550492e-05,
      "loss": 0.1721,
      "step": 914
    },
    {
      "epoch": 0.19206549118387908,
      "grad_norm": 0.1769437938928604,
      "learning_rate": 9.74464214799635e-05,
      "loss": 0.1485,
      "step": 915
    },
    {
      "epoch": 0.19227539882451722,
      "grad_norm": 0.18178406357765198,
      "learning_rate": 9.743484884460993e-05,
      "loss": 0.1523,
      "step": 916
    },
    {
      "epoch": 0.19248530646515533,
      "grad_norm": 0.16841888427734375,
      "learning_rate": 9.742325073565905e-05,
      "loss": 0.148,
      "step": 917
    },
    {
      "epoch": 0.19269521410579346,
      "grad_norm": 0.13603579998016357,
      "learning_rate": 9.741162715933924e-05,
      "loss": 0.161,
      "step": 918
    },
    {
      "epoch": 0.19290512174643157,
      "grad_norm": 0.1348285973072052,
      "learning_rate": 9.739997812189265e-05,
      "loss": 0.145,
      "step": 919
    },
    {
      "epoch": 0.19311502938706968,
      "grad_norm": 0.13922441005706787,
      "learning_rate": 9.738830362957508e-05,
      "loss": 0.1607,
      "step": 920
    },
    {
      "epoch": 0.1933249370277078,
      "grad_norm": 0.12017328292131424,
      "learning_rate": 9.7376603688656e-05,
      "loss": 0.152,
      "step": 921
    },
    {
      "epoch": 0.19353484466834592,
      "grad_norm": 0.14594700932502747,
      "learning_rate": 9.736487830541853e-05,
      "loss": 0.1638,
      "step": 922
    },
    {
      "epoch": 0.19374475230898405,
      "grad_norm": 0.14390654861927032,
      "learning_rate": 9.73531274861595e-05,
      "loss": 0.1608,
      "step": 923
    },
    {
      "epoch": 0.19395465994962216,
      "grad_norm": 0.12394702434539795,
      "learning_rate": 9.734135123718933e-05,
      "loss": 0.1612,
      "step": 924
    },
    {
      "epoch": 0.1941645675902603,
      "grad_norm": 0.16754676401615143,
      "learning_rate": 9.732954956483218e-05,
      "loss": 0.1791,
      "step": 925
    },
    {
      "epoch": 0.1943744752308984,
      "grad_norm": 0.2816343605518341,
      "learning_rate": 9.731772247542576e-05,
      "loss": 0.1489,
      "step": 926
    },
    {
      "epoch": 0.19458438287153654,
      "grad_norm": 0.15909412503242493,
      "learning_rate": 9.730586997532155e-05,
      "loss": 0.1531,
      "step": 927
    },
    {
      "epoch": 0.19479429051217464,
      "grad_norm": 0.16626602411270142,
      "learning_rate": 9.729399207088457e-05,
      "loss": 0.1729,
      "step": 928
    },
    {
      "epoch": 0.19500419815281275,
      "grad_norm": 0.14581038057804108,
      "learning_rate": 9.728208876849354e-05,
      "loss": 0.1616,
      "step": 929
    },
    {
      "epoch": 0.1952141057934509,
      "grad_norm": 0.14985312521457672,
      "learning_rate": 9.727016007454079e-05,
      "loss": 0.1583,
      "step": 930
    },
    {
      "epoch": 0.195424013434089,
      "grad_norm": 0.13958559930324554,
      "learning_rate": 9.725820599543234e-05,
      "loss": 0.1646,
      "step": 931
    },
    {
      "epoch": 0.19563392107472713,
      "grad_norm": 0.1607862412929535,
      "learning_rate": 9.724622653758777e-05,
      "loss": 0.1549,
      "step": 932
    },
    {
      "epoch": 0.19584382871536524,
      "grad_norm": 0.17007960379123688,
      "learning_rate": 9.723422170744031e-05,
      "loss": 0.1718,
      "step": 933
    },
    {
      "epoch": 0.19605373635600337,
      "grad_norm": 0.1419927030801773,
      "learning_rate": 9.722219151143688e-05,
      "loss": 0.1689,
      "step": 934
    },
    {
      "epoch": 0.19626364399664148,
      "grad_norm": 0.1631292998790741,
      "learning_rate": 9.721013595603793e-05,
      "loss": 0.1611,
      "step": 935
    },
    {
      "epoch": 0.1964735516372796,
      "grad_norm": 0.19870012998580933,
      "learning_rate": 9.719805504771758e-05,
      "loss": 0.1836,
      "step": 936
    },
    {
      "epoch": 0.19668345927791772,
      "grad_norm": 0.14150285720825195,
      "learning_rate": 9.718594879296355e-05,
      "loss": 0.1718,
      "step": 937
    },
    {
      "epoch": 0.19689336691855583,
      "grad_norm": 0.1416793018579483,
      "learning_rate": 9.717381719827716e-05,
      "loss": 0.1511,
      "step": 938
    },
    {
      "epoch": 0.19710327455919396,
      "grad_norm": 0.14615508913993835,
      "learning_rate": 9.716166027017339e-05,
      "loss": 0.1599,
      "step": 939
    },
    {
      "epoch": 0.19731318219983207,
      "grad_norm": 0.13773533701896667,
      "learning_rate": 9.714947801518076e-05,
      "loss": 0.1765,
      "step": 940
    },
    {
      "epoch": 0.1975230898404702,
      "grad_norm": 0.11306725442409515,
      "learning_rate": 9.713727043984143e-05,
      "loss": 0.1678,
      "step": 941
    },
    {
      "epoch": 0.1977329974811083,
      "grad_norm": 0.14150364696979523,
      "learning_rate": 9.712503755071115e-05,
      "loss": 0.1535,
      "step": 942
    },
    {
      "epoch": 0.19794290512174642,
      "grad_norm": 0.15026667714118958,
      "learning_rate": 9.711277935435925e-05,
      "loss": 0.155,
      "step": 943
    },
    {
      "epoch": 0.19815281276238456,
      "grad_norm": 0.1540324091911316,
      "learning_rate": 9.710049585736866e-05,
      "loss": 0.1866,
      "step": 944
    },
    {
      "epoch": 0.19836272040302266,
      "grad_norm": 0.12859384715557098,
      "learning_rate": 9.708818706633591e-05,
      "loss": 0.1512,
      "step": 945
    },
    {
      "epoch": 0.1985726280436608,
      "grad_norm": 0.14280495047569275,
      "learning_rate": 9.707585298787109e-05,
      "loss": 0.1558,
      "step": 946
    },
    {
      "epoch": 0.1987825356842989,
      "grad_norm": 0.14329122006893158,
      "learning_rate": 9.70634936285979e-05,
      "loss": 0.153,
      "step": 947
    },
    {
      "epoch": 0.19899244332493704,
      "grad_norm": 0.18129687011241913,
      "learning_rate": 9.705110899515359e-05,
      "loss": 0.1592,
      "step": 948
    },
    {
      "epoch": 0.19920235096557515,
      "grad_norm": 0.13191339373588562,
      "learning_rate": 9.7038699094189e-05,
      "loss": 0.1565,
      "step": 949
    },
    {
      "epoch": 0.19941225860621326,
      "grad_norm": 0.1381015032529831,
      "learning_rate": 9.702626393236849e-05,
      "loss": 0.1704,
      "step": 950
    },
    {
      "epoch": 0.1996221662468514,
      "grad_norm": 0.12249460071325302,
      "learning_rate": 9.701380351637007e-05,
      "loss": 0.1534,
      "step": 951
    },
    {
      "epoch": 0.1998320738874895,
      "grad_norm": 0.1820680946111679,
      "learning_rate": 9.700131785288525e-05,
      "loss": 0.1805,
      "step": 952
    },
    {
      "epoch": 0.20004198152812763,
      "grad_norm": 0.13920508325099945,
      "learning_rate": 9.698880694861913e-05,
      "loss": 0.1652,
      "step": 953
    },
    {
      "epoch": 0.20025188916876574,
      "grad_norm": 0.14306975901126862,
      "learning_rate": 9.697627081029033e-05,
      "loss": 0.1397,
      "step": 954
    },
    {
      "epoch": 0.20046179680940387,
      "grad_norm": 0.13642224669456482,
      "learning_rate": 9.696370944463104e-05,
      "loss": 0.174,
      "step": 955
    },
    {
      "epoch": 0.20067170445004198,
      "grad_norm": 0.12083397060632706,
      "learning_rate": 9.695112285838704e-05,
      "loss": 0.1807,
      "step": 956
    },
    {
      "epoch": 0.2008816120906801,
      "grad_norm": 0.17191384732723236,
      "learning_rate": 9.693851105831757e-05,
      "loss": 0.1656,
      "step": 957
    },
    {
      "epoch": 0.20109151973131822,
      "grad_norm": 0.1415136307477951,
      "learning_rate": 9.692587405119549e-05,
      "loss": 0.1726,
      "step": 958
    },
    {
      "epoch": 0.20130142737195633,
      "grad_norm": 0.15210871398448944,
      "learning_rate": 9.691321184380713e-05,
      "loss": 0.1614,
      "step": 959
    },
    {
      "epoch": 0.20151133501259447,
      "grad_norm": 0.14295780658721924,
      "learning_rate": 9.690052444295239e-05,
      "loss": 0.1884,
      "step": 960
    },
    {
      "epoch": 0.20172124265323257,
      "grad_norm": 0.13754574954509735,
      "learning_rate": 9.688781185544471e-05,
      "loss": 0.1719,
      "step": 961
    },
    {
      "epoch": 0.2019311502938707,
      "grad_norm": 0.1425933688879013,
      "learning_rate": 9.687507408811104e-05,
      "loss": 0.1564,
      "step": 962
    },
    {
      "epoch": 0.20214105793450882,
      "grad_norm": 0.1202697604894638,
      "learning_rate": 9.686231114779184e-05,
      "loss": 0.1584,
      "step": 963
    },
    {
      "epoch": 0.20235096557514692,
      "grad_norm": 0.12710309028625488,
      "learning_rate": 9.684952304134111e-05,
      "loss": 0.168,
      "step": 964
    },
    {
      "epoch": 0.20256087321578506,
      "grad_norm": 0.14089150726795197,
      "learning_rate": 9.683670977562633e-05,
      "loss": 0.1852,
      "step": 965
    },
    {
      "epoch": 0.20277078085642317,
      "grad_norm": 0.13602103292942047,
      "learning_rate": 9.682387135752856e-05,
      "loss": 0.1673,
      "step": 966
    },
    {
      "epoch": 0.2029806884970613,
      "grad_norm": 0.13352209329605103,
      "learning_rate": 9.68110077939423e-05,
      "loss": 0.1571,
      "step": 967
    },
    {
      "epoch": 0.2031905961376994,
      "grad_norm": 0.13534000515937805,
      "learning_rate": 9.679811909177556e-05,
      "loss": 0.1706,
      "step": 968
    },
    {
      "epoch": 0.20340050377833754,
      "grad_norm": 0.12475798279047012,
      "learning_rate": 9.67852052579499e-05,
      "loss": 0.1532,
      "step": 969
    },
    {
      "epoch": 0.20361041141897565,
      "grad_norm": 0.1589164137840271,
      "learning_rate": 9.677226629940032e-05,
      "loss": 0.1597,
      "step": 970
    },
    {
      "epoch": 0.20382031905961376,
      "grad_norm": 0.14976951479911804,
      "learning_rate": 9.675930222307537e-05,
      "loss": 0.1731,
      "step": 971
    },
    {
      "epoch": 0.2040302267002519,
      "grad_norm": 0.12664476037025452,
      "learning_rate": 9.6746313035937e-05,
      "loss": 0.1363,
      "step": 972
    },
    {
      "epoch": 0.20424013434089,
      "grad_norm": 0.12275015562772751,
      "learning_rate": 9.673329874496075e-05,
      "loss": 0.1574,
      "step": 973
    },
    {
      "epoch": 0.20445004198152814,
      "grad_norm": 0.13222196698188782,
      "learning_rate": 9.672025935713556e-05,
      "loss": 0.1535,
      "step": 974
    },
    {
      "epoch": 0.20465994962216624,
      "grad_norm": 0.15173639357089996,
      "learning_rate": 9.670719487946389e-05,
      "loss": 0.1792,
      "step": 975
    },
    {
      "epoch": 0.20486985726280438,
      "grad_norm": 0.17748403549194336,
      "learning_rate": 9.669410531896167e-05,
      "loss": 0.1891,
      "step": 976
    },
    {
      "epoch": 0.20507976490344249,
      "grad_norm": 0.14893555641174316,
      "learning_rate": 9.66809906826583e-05,
      "loss": 0.1438,
      "step": 977
    },
    {
      "epoch": 0.2052896725440806,
      "grad_norm": 0.16388140618801117,
      "learning_rate": 9.66678509775966e-05,
      "loss": 0.1641,
      "step": 978
    },
    {
      "epoch": 0.20549958018471873,
      "grad_norm": 0.17024089395999908,
      "learning_rate": 9.665468621083293e-05,
      "loss": 0.1839,
      "step": 979
    },
    {
      "epoch": 0.20570948782535683,
      "grad_norm": 0.14790277183055878,
      "learning_rate": 9.664149638943707e-05,
      "loss": 0.1701,
      "step": 980
    },
    {
      "epoch": 0.20591939546599497,
      "grad_norm": 0.14114373922348022,
      "learning_rate": 9.662828152049223e-05,
      "loss": 0.1699,
      "step": 981
    },
    {
      "epoch": 0.20612930310663308,
      "grad_norm": 0.12556609511375427,
      "learning_rate": 9.661504161109513e-05,
      "loss": 0.1607,
      "step": 982
    },
    {
      "epoch": 0.2063392107472712,
      "grad_norm": 0.11868204176425934,
      "learning_rate": 9.660177666835585e-05,
      "loss": 0.1487,
      "step": 983
    },
    {
      "epoch": 0.20654911838790932,
      "grad_norm": 0.12117776274681091,
      "learning_rate": 9.658848669939805e-05,
      "loss": 0.1668,
      "step": 984
    },
    {
      "epoch": 0.20675902602854743,
      "grad_norm": 0.126139298081398,
      "learning_rate": 9.657517171135866e-05,
      "loss": 0.1535,
      "step": 985
    },
    {
      "epoch": 0.20696893366918556,
      "grad_norm": 0.15042434632778168,
      "learning_rate": 9.656183171138818e-05,
      "loss": 0.168,
      "step": 986
    },
    {
      "epoch": 0.20717884130982367,
      "grad_norm": 0.1438342183828354,
      "learning_rate": 9.65484667066505e-05,
      "loss": 0.1628,
      "step": 987
    },
    {
      "epoch": 0.2073887489504618,
      "grad_norm": 0.1280035823583603,
      "learning_rate": 9.65350767043229e-05,
      "loss": 0.1655,
      "step": 988
    },
    {
      "epoch": 0.2075986565910999,
      "grad_norm": 0.13997893035411835,
      "learning_rate": 9.652166171159614e-05,
      "loss": 0.1766,
      "step": 989
    },
    {
      "epoch": 0.20780856423173805,
      "grad_norm": 0.1653011292219162,
      "learning_rate": 9.650822173567438e-05,
      "loss": 0.1744,
      "step": 990
    },
    {
      "epoch": 0.20801847187237615,
      "grad_norm": 0.1532358080148697,
      "learning_rate": 9.64947567837752e-05,
      "loss": 0.1529,
      "step": 991
    },
    {
      "epoch": 0.20822837951301426,
      "grad_norm": 0.13623711466789246,
      "learning_rate": 9.648126686312955e-05,
      "loss": 0.1632,
      "step": 992
    },
    {
      "epoch": 0.2084382871536524,
      "grad_norm": 0.13220278918743134,
      "learning_rate": 9.646775198098186e-05,
      "loss": 0.1462,
      "step": 993
    },
    {
      "epoch": 0.2086481947942905,
      "grad_norm": 0.16712597012519836,
      "learning_rate": 9.645421214458992e-05,
      "loss": 0.1658,
      "step": 994
    },
    {
      "epoch": 0.20885810243492864,
      "grad_norm": 0.16033975780010223,
      "learning_rate": 9.644064736122493e-05,
      "loss": 0.1523,
      "step": 995
    },
    {
      "epoch": 0.20906801007556675,
      "grad_norm": 0.13283276557922363,
      "learning_rate": 9.64270576381715e-05,
      "loss": 0.1738,
      "step": 996
    },
    {
      "epoch": 0.20927791771620488,
      "grad_norm": 0.2072938233613968,
      "learning_rate": 9.64134429827276e-05,
      "loss": 0.1589,
      "step": 997
    },
    {
      "epoch": 0.209487825356843,
      "grad_norm": 0.15322737395763397,
      "learning_rate": 9.639980340220462e-05,
      "loss": 0.1772,
      "step": 998
    },
    {
      "epoch": 0.2096977329974811,
      "grad_norm": 0.21009685099124908,
      "learning_rate": 9.638613890392734e-05,
      "loss": 0.1619,
      "step": 999
    },
    {
      "epoch": 0.20990764063811923,
      "grad_norm": 0.1521667093038559,
      "learning_rate": 9.63724494952339e-05,
      "loss": 0.1878,
      "step": 1000
    },
    {
      "epoch": 0.21011754827875734,
      "grad_norm": 0.11604970693588257,
      "learning_rate": 9.635873518347581e-05,
      "loss": 0.1621,
      "step": 1001
    },
    {
      "epoch": 0.21032745591939547,
      "grad_norm": 0.16362364590168,
      "learning_rate": 9.634499597601798e-05,
      "loss": 0.1779,
      "step": 1002
    },
    {
      "epoch": 0.21053736356003358,
      "grad_norm": 0.15217158198356628,
      "learning_rate": 9.63312318802387e-05,
      "loss": 0.1657,
      "step": 1003
    },
    {
      "epoch": 0.21074727120067172,
      "grad_norm": 0.15649117529392242,
      "learning_rate": 9.631744290352956e-05,
      "loss": 0.1817,
      "step": 1004
    },
    {
      "epoch": 0.21095717884130982,
      "grad_norm": 0.15730923414230347,
      "learning_rate": 9.63036290532956e-05,
      "loss": 0.1646,
      "step": 1005
    },
    {
      "epoch": 0.21116708648194793,
      "grad_norm": 0.1645728051662445,
      "learning_rate": 9.628979033695513e-05,
      "loss": 0.1546,
      "step": 1006
    },
    {
      "epoch": 0.21137699412258606,
      "grad_norm": 0.13902553915977478,
      "learning_rate": 9.627592676193988e-05,
      "loss": 0.157,
      "step": 1007
    },
    {
      "epoch": 0.21158690176322417,
      "grad_norm": 0.11890498548746109,
      "learning_rate": 9.626203833569491e-05,
      "loss": 0.1668,
      "step": 1008
    },
    {
      "epoch": 0.2117968094038623,
      "grad_norm": 0.14101600646972656,
      "learning_rate": 9.624812506567863e-05,
      "loss": 0.1477,
      "step": 1009
    },
    {
      "epoch": 0.21200671704450041,
      "grad_norm": 0.1529974490404129,
      "learning_rate": 9.623418695936275e-05,
      "loss": 0.1675,
      "step": 1010
    },
    {
      "epoch": 0.21221662468513855,
      "grad_norm": 0.15540415048599243,
      "learning_rate": 9.622022402423239e-05,
      "loss": 0.1797,
      "step": 1011
    },
    {
      "epoch": 0.21242653232577666,
      "grad_norm": 0.16966544091701508,
      "learning_rate": 9.620623626778593e-05,
      "loss": 0.1642,
      "step": 1012
    },
    {
      "epoch": 0.21263643996641476,
      "grad_norm": 0.1261100172996521,
      "learning_rate": 9.619222369753515e-05,
      "loss": 0.1633,
      "step": 1013
    },
    {
      "epoch": 0.2128463476070529,
      "grad_norm": 0.15353427827358246,
      "learning_rate": 9.617818632100508e-05,
      "loss": 0.1638,
      "step": 1014
    },
    {
      "epoch": 0.213056255247691,
      "grad_norm": 0.17301282286643982,
      "learning_rate": 9.616412414573416e-05,
      "loss": 0.1637,
      "step": 1015
    },
    {
      "epoch": 0.21326616288832914,
      "grad_norm": 0.14541825652122498,
      "learning_rate": 9.615003717927406e-05,
      "loss": 0.1677,
      "step": 1016
    },
    {
      "epoch": 0.21347607052896725,
      "grad_norm": 0.13302357494831085,
      "learning_rate": 9.613592542918981e-05,
      "loss": 0.1595,
      "step": 1017
    },
    {
      "epoch": 0.21368597816960538,
      "grad_norm": 0.17005443572998047,
      "learning_rate": 9.612178890305976e-05,
      "loss": 0.1799,
      "step": 1018
    },
    {
      "epoch": 0.2138958858102435,
      "grad_norm": 0.20354509353637695,
      "learning_rate": 9.610762760847553e-05,
      "loss": 0.1549,
      "step": 1019
    },
    {
      "epoch": 0.2141057934508816,
      "grad_norm": 0.16336257755756378,
      "learning_rate": 9.609344155304206e-05,
      "loss": 0.1404,
      "step": 1020
    },
    {
      "epoch": 0.21431570109151973,
      "grad_norm": 0.14682699739933014,
      "learning_rate": 9.607923074437758e-05,
      "loss": 0.1666,
      "step": 1021
    },
    {
      "epoch": 0.21452560873215784,
      "grad_norm": 0.15628460049629211,
      "learning_rate": 9.606499519011367e-05,
      "loss": 0.1723,
      "step": 1022
    },
    {
      "epoch": 0.21473551637279598,
      "grad_norm": 0.18369454145431519,
      "learning_rate": 9.605073489789506e-05,
      "loss": 0.1622,
      "step": 1023
    },
    {
      "epoch": 0.21494542401343408,
      "grad_norm": 0.14383664727210999,
      "learning_rate": 9.603644987537994e-05,
      "loss": 0.1776,
      "step": 1024
    },
    {
      "epoch": 0.21515533165407222,
      "grad_norm": 0.12850511074066162,
      "learning_rate": 9.602214013023963e-05,
      "loss": 0.1552,
      "step": 1025
    },
    {
      "epoch": 0.21536523929471033,
      "grad_norm": 0.12851618230342865,
      "learning_rate": 9.60078056701588e-05,
      "loss": 0.1634,
      "step": 1026
    },
    {
      "epoch": 0.21557514693534843,
      "grad_norm": 0.1252540647983551,
      "learning_rate": 9.599344650283544e-05,
      "loss": 0.1553,
      "step": 1027
    },
    {
      "epoch": 0.21578505457598657,
      "grad_norm": 0.14331619441509247,
      "learning_rate": 9.597906263598067e-05,
      "loss": 0.1681,
      "step": 1028
    },
    {
      "epoch": 0.21599496221662468,
      "grad_norm": 0.13295383751392365,
      "learning_rate": 9.5964654077319e-05,
      "loss": 0.1748,
      "step": 1029
    },
    {
      "epoch": 0.2162048698572628,
      "grad_norm": 0.14106275141239166,
      "learning_rate": 9.595022083458815e-05,
      "loss": 0.1642,
      "step": 1030
    },
    {
      "epoch": 0.21641477749790092,
      "grad_norm": 0.21775981783866882,
      "learning_rate": 9.59357629155391e-05,
      "loss": 0.1465,
      "step": 1031
    },
    {
      "epoch": 0.21662468513853905,
      "grad_norm": 0.19797000288963318,
      "learning_rate": 9.592128032793607e-05,
      "loss": 0.1559,
      "step": 1032
    },
    {
      "epoch": 0.21683459277917716,
      "grad_norm": 0.14408251643180847,
      "learning_rate": 9.590677307955656e-05,
      "loss": 0.1696,
      "step": 1033
    },
    {
      "epoch": 0.2170445004198153,
      "grad_norm": 0.14695291221141815,
      "learning_rate": 9.589224117819128e-05,
      "loss": 0.1415,
      "step": 1034
    },
    {
      "epoch": 0.2172544080604534,
      "grad_norm": 0.1755986511707306,
      "learning_rate": 9.58776846316442e-05,
      "loss": 0.1565,
      "step": 1035
    },
    {
      "epoch": 0.2174643157010915,
      "grad_norm": 0.2155688852071762,
      "learning_rate": 9.58631034477325e-05,
      "loss": 0.1631,
      "step": 1036
    },
    {
      "epoch": 0.21767422334172964,
      "grad_norm": 0.16168251633644104,
      "learning_rate": 9.584849763428665e-05,
      "loss": 0.174,
      "step": 1037
    },
    {
      "epoch": 0.21788413098236775,
      "grad_norm": 0.14332053065299988,
      "learning_rate": 9.583386719915026e-05,
      "loss": 0.1733,
      "step": 1038
    },
    {
      "epoch": 0.2180940386230059,
      "grad_norm": 0.1143033355474472,
      "learning_rate": 9.581921215018023e-05,
      "loss": 0.1431,
      "step": 1039
    },
    {
      "epoch": 0.218303946263644,
      "grad_norm": 0.16438762843608856,
      "learning_rate": 9.580453249524667e-05,
      "loss": 0.1731,
      "step": 1040
    },
    {
      "epoch": 0.21851385390428213,
      "grad_norm": 0.16054236888885498,
      "learning_rate": 9.578982824223285e-05,
      "loss": 0.1706,
      "step": 1041
    },
    {
      "epoch": 0.21872376154492024,
      "grad_norm": 0.14396212995052338,
      "learning_rate": 9.577509939903535e-05,
      "loss": 0.1694,
      "step": 1042
    },
    {
      "epoch": 0.21893366918555834,
      "grad_norm": 0.14557866752147675,
      "learning_rate": 9.576034597356382e-05,
      "loss": 0.1692,
      "step": 1043
    },
    {
      "epoch": 0.21914357682619648,
      "grad_norm": 0.1378258913755417,
      "learning_rate": 9.574556797374124e-05,
      "loss": 0.1616,
      "step": 1044
    },
    {
      "epoch": 0.2193534844668346,
      "grad_norm": 0.11917531490325928,
      "learning_rate": 9.573076540750371e-05,
      "loss": 0.1523,
      "step": 1045
    },
    {
      "epoch": 0.21956339210747272,
      "grad_norm": 0.12740926444530487,
      "learning_rate": 9.571593828280058e-05,
      "loss": 0.1581,
      "step": 1046
    },
    {
      "epoch": 0.21977329974811083,
      "grad_norm": 0.12773928046226501,
      "learning_rate": 9.570108660759432e-05,
      "loss": 0.1657,
      "step": 1047
    },
    {
      "epoch": 0.21998320738874896,
      "grad_norm": 0.15594597160816193,
      "learning_rate": 9.568621038986061e-05,
      "loss": 0.175,
      "step": 1048
    },
    {
      "epoch": 0.22019311502938707,
      "grad_norm": 0.1410975605249405,
      "learning_rate": 9.567130963758834e-05,
      "loss": 0.1527,
      "step": 1049
    },
    {
      "epoch": 0.22040302267002518,
      "grad_norm": 0.1446082442998886,
      "learning_rate": 9.565638435877957e-05,
      "loss": 0.1553,
      "step": 1050
    },
    {
      "epoch": 0.2206129303106633,
      "grad_norm": 0.14107242226600647,
      "learning_rate": 9.564143456144949e-05,
      "loss": 0.1628,
      "step": 1051
    },
    {
      "epoch": 0.22082283795130142,
      "grad_norm": 0.15437398850917816,
      "learning_rate": 9.562646025362649e-05,
      "loss": 0.1704,
      "step": 1052
    },
    {
      "epoch": 0.22103274559193956,
      "grad_norm": 0.13233573734760284,
      "learning_rate": 9.561146144335212e-05,
      "loss": 0.1588,
      "step": 1053
    },
    {
      "epoch": 0.22124265323257766,
      "grad_norm": 0.13578198850154877,
      "learning_rate": 9.559643813868107e-05,
      "loss": 0.1649,
      "step": 1054
    },
    {
      "epoch": 0.2214525608732158,
      "grad_norm": 0.17958106100559235,
      "learning_rate": 9.55813903476812e-05,
      "loss": 0.1707,
      "step": 1055
    },
    {
      "epoch": 0.2216624685138539,
      "grad_norm": 0.14567284286022186,
      "learning_rate": 9.556631807843352e-05,
      "loss": 0.1622,
      "step": 1056
    },
    {
      "epoch": 0.221872376154492,
      "grad_norm": 0.12384296208620071,
      "learning_rate": 9.55512213390322e-05,
      "loss": 0.1763,
      "step": 1057
    },
    {
      "epoch": 0.22208228379513015,
      "grad_norm": 0.14242222905158997,
      "learning_rate": 9.553610013758453e-05,
      "loss": 0.1695,
      "step": 1058
    },
    {
      "epoch": 0.22229219143576825,
      "grad_norm": 0.13062936067581177,
      "learning_rate": 9.552095448221093e-05,
      "loss": 0.1631,
      "step": 1059
    },
    {
      "epoch": 0.2225020990764064,
      "grad_norm": 0.158433735370636,
      "learning_rate": 9.550578438104494e-05,
      "loss": 0.1679,
      "step": 1060
    },
    {
      "epoch": 0.2227120067170445,
      "grad_norm": 0.14844875037670135,
      "learning_rate": 9.549058984223329e-05,
      "loss": 0.1606,
      "step": 1061
    },
    {
      "epoch": 0.22292191435768263,
      "grad_norm": 0.1360979825258255,
      "learning_rate": 9.547537087393579e-05,
      "loss": 0.1528,
      "step": 1062
    },
    {
      "epoch": 0.22313182199832074,
      "grad_norm": 0.1500774621963501,
      "learning_rate": 9.546012748432535e-05,
      "loss": 0.1639,
      "step": 1063
    },
    {
      "epoch": 0.22334172963895885,
      "grad_norm": 0.1147661879658699,
      "learning_rate": 9.544485968158803e-05,
      "loss": 0.1729,
      "step": 1064
    },
    {
      "epoch": 0.22355163727959698,
      "grad_norm": 0.13184726238250732,
      "learning_rate": 9.542956747392299e-05,
      "loss": 0.1733,
      "step": 1065
    },
    {
      "epoch": 0.2237615449202351,
      "grad_norm": 0.14958997070789337,
      "learning_rate": 9.541425086954251e-05,
      "loss": 0.1621,
      "step": 1066
    },
    {
      "epoch": 0.22397145256087322,
      "grad_norm": 0.13268455862998962,
      "learning_rate": 9.539890987667193e-05,
      "loss": 0.1542,
      "step": 1067
    },
    {
      "epoch": 0.22418136020151133,
      "grad_norm": 0.12983419001102448,
      "learning_rate": 9.538354450354974e-05,
      "loss": 0.1608,
      "step": 1068
    },
    {
      "epoch": 0.22439126784214947,
      "grad_norm": 0.11506141722202301,
      "learning_rate": 9.536815475842747e-05,
      "loss": 0.1521,
      "step": 1069
    },
    {
      "epoch": 0.22460117548278757,
      "grad_norm": 0.16174156963825226,
      "learning_rate": 9.535274064956976e-05,
      "loss": 0.176,
      "step": 1070
    },
    {
      "epoch": 0.22481108312342568,
      "grad_norm": 0.14904946088790894,
      "learning_rate": 9.533730218525438e-05,
      "loss": 0.1743,
      "step": 1071
    },
    {
      "epoch": 0.22502099076406382,
      "grad_norm": 0.1392981857061386,
      "learning_rate": 9.53218393737721e-05,
      "loss": 0.1619,
      "step": 1072
    },
    {
      "epoch": 0.22523089840470192,
      "grad_norm": 0.1343872994184494,
      "learning_rate": 9.530635222342682e-05,
      "loss": 0.1436,
      "step": 1073
    },
    {
      "epoch": 0.22544080604534006,
      "grad_norm": 0.13077078759670258,
      "learning_rate": 9.529084074253549e-05,
      "loss": 0.1501,
      "step": 1074
    },
    {
      "epoch": 0.22565071368597817,
      "grad_norm": 0.13800117373466492,
      "learning_rate": 9.527530493942812e-05,
      "loss": 0.1634,
      "step": 1075
    },
    {
      "epoch": 0.2258606213266163,
      "grad_norm": 0.15503515303134918,
      "learning_rate": 9.525974482244782e-05,
      "loss": 0.1682,
      "step": 1076
    },
    {
      "epoch": 0.2260705289672544,
      "grad_norm": 0.13676904141902924,
      "learning_rate": 9.524416039995068e-05,
      "loss": 0.1648,
      "step": 1077
    },
    {
      "epoch": 0.22628043660789252,
      "grad_norm": 0.1415819525718689,
      "learning_rate": 9.522855168030595e-05,
      "loss": 0.1732,
      "step": 1078
    },
    {
      "epoch": 0.22649034424853065,
      "grad_norm": 0.14641550183296204,
      "learning_rate": 9.521291867189581e-05,
      "loss": 0.1535,
      "step": 1079
    },
    {
      "epoch": 0.22670025188916876,
      "grad_norm": 0.13009630143642426,
      "learning_rate": 9.519726138311558e-05,
      "loss": 0.1624,
      "step": 1080
    },
    {
      "epoch": 0.2269101595298069,
      "grad_norm": 0.12057724595069885,
      "learning_rate": 9.51815798223736e-05,
      "loss": 0.1626,
      "step": 1081
    },
    {
      "epoch": 0.227120067170445,
      "grad_norm": 0.13820677995681763,
      "learning_rate": 9.516587399809116e-05,
      "loss": 0.1566,
      "step": 1082
    },
    {
      "epoch": 0.22732997481108314,
      "grad_norm": 0.19474509358406067,
      "learning_rate": 9.515014391870269e-05,
      "loss": 0.1591,
      "step": 1083
    },
    {
      "epoch": 0.22753988245172124,
      "grad_norm": 0.13335439562797546,
      "learning_rate": 9.513438959265561e-05,
      "loss": 0.1559,
      "step": 1084
    },
    {
      "epoch": 0.22774979009235935,
      "grad_norm": 0.16238383948802948,
      "learning_rate": 9.511861102841036e-05,
      "loss": 0.1701,
      "step": 1085
    },
    {
      "epoch": 0.22795969773299748,
      "grad_norm": 0.129498690366745,
      "learning_rate": 9.510280823444034e-05,
      "loss": 0.1542,
      "step": 1086
    },
    {
      "epoch": 0.2281696053736356,
      "grad_norm": 0.13683976233005524,
      "learning_rate": 9.508698121923204e-05,
      "loss": 0.1647,
      "step": 1087
    },
    {
      "epoch": 0.22837951301427373,
      "grad_norm": 0.15396280586719513,
      "learning_rate": 9.507112999128493e-05,
      "loss": 0.1519,
      "step": 1088
    },
    {
      "epoch": 0.22858942065491183,
      "grad_norm": 0.12210803478956223,
      "learning_rate": 9.50552545591115e-05,
      "loss": 0.1579,
      "step": 1089
    },
    {
      "epoch": 0.22879932829554997,
      "grad_norm": 0.1511714905500412,
      "learning_rate": 9.503935493123719e-05,
      "loss": 0.155,
      "step": 1090
    },
    {
      "epoch": 0.22900923593618808,
      "grad_norm": 0.14200669527053833,
      "learning_rate": 9.502343111620047e-05,
      "loss": 0.174,
      "step": 1091
    },
    {
      "epoch": 0.22921914357682618,
      "grad_norm": 0.12680545449256897,
      "learning_rate": 9.500748312255281e-05,
      "loss": 0.1844,
      "step": 1092
    },
    {
      "epoch": 0.22942905121746432,
      "grad_norm": 0.14180205762386322,
      "learning_rate": 9.499151095885861e-05,
      "loss": 0.1619,
      "step": 1093
    },
    {
      "epoch": 0.22963895885810243,
      "grad_norm": 0.12461036443710327,
      "learning_rate": 9.497551463369532e-05,
      "loss": 0.1558,
      "step": 1094
    },
    {
      "epoch": 0.22984886649874056,
      "grad_norm": 0.17455211281776428,
      "learning_rate": 9.495949415565332e-05,
      "loss": 0.172,
      "step": 1095
    },
    {
      "epoch": 0.23005877413937867,
      "grad_norm": 0.15062375366687775,
      "learning_rate": 9.494344953333599e-05,
      "loss": 0.1656,
      "step": 1096
    },
    {
      "epoch": 0.2302686817800168,
      "grad_norm": 0.14742077887058258,
      "learning_rate": 9.492738077535963e-05,
      "loss": 0.1751,
      "step": 1097
    },
    {
      "epoch": 0.2304785894206549,
      "grad_norm": 0.1343587338924408,
      "learning_rate": 9.491128789035355e-05,
      "loss": 0.16,
      "step": 1098
    },
    {
      "epoch": 0.23068849706129302,
      "grad_norm": 0.13116249442100525,
      "learning_rate": 9.489517088696e-05,
      "loss": 0.1495,
      "step": 1099
    },
    {
      "epoch": 0.23089840470193115,
      "grad_norm": 0.17240563035011292,
      "learning_rate": 9.487902977383415e-05,
      "loss": 0.171,
      "step": 1100
    },
    {
      "epoch": 0.23110831234256926,
      "grad_norm": 0.1725616306066513,
      "learning_rate": 9.486286455964417e-05,
      "loss": 0.1672,
      "step": 1101
    },
    {
      "epoch": 0.2313182199832074,
      "grad_norm": 0.16195368766784668,
      "learning_rate": 9.484667525307116e-05,
      "loss": 0.1697,
      "step": 1102
    },
    {
      "epoch": 0.2315281276238455,
      "grad_norm": 0.15474824607372284,
      "learning_rate": 9.483046186280912e-05,
      "loss": 0.1757,
      "step": 1103
    },
    {
      "epoch": 0.23173803526448364,
      "grad_norm": 0.13763071596622467,
      "learning_rate": 9.481422439756503e-05,
      "loss": 0.1545,
      "step": 1104
    },
    {
      "epoch": 0.23194794290512175,
      "grad_norm": 0.15999558568000793,
      "learning_rate": 9.479796286605876e-05,
      "loss": 0.168,
      "step": 1105
    },
    {
      "epoch": 0.23215785054575985,
      "grad_norm": 0.13550569117069244,
      "learning_rate": 9.478167727702315e-05,
      "loss": 0.1642,
      "step": 1106
    },
    {
      "epoch": 0.232367758186398,
      "grad_norm": 0.1440218985080719,
      "learning_rate": 9.47653676392039e-05,
      "loss": 0.1763,
      "step": 1107
    },
    {
      "epoch": 0.2325776658270361,
      "grad_norm": 0.13457611203193665,
      "learning_rate": 9.47490339613597e-05,
      "loss": 0.1604,
      "step": 1108
    },
    {
      "epoch": 0.23278757346767423,
      "grad_norm": 0.13920889794826508,
      "learning_rate": 9.473267625226207e-05,
      "loss": 0.158,
      "step": 1109
    },
    {
      "epoch": 0.23299748110831234,
      "grad_norm": 0.1562851071357727,
      "learning_rate": 9.47162945206955e-05,
      "loss": 0.1647,
      "step": 1110
    },
    {
      "epoch": 0.23320738874895047,
      "grad_norm": 0.13289715349674225,
      "learning_rate": 9.469988877545735e-05,
      "loss": 0.1535,
      "step": 1111
    },
    {
      "epoch": 0.23341729638958858,
      "grad_norm": 0.13001225888729095,
      "learning_rate": 9.468345902535787e-05,
      "loss": 0.1513,
      "step": 1112
    },
    {
      "epoch": 0.2336272040302267,
      "grad_norm": 0.1498699188232422,
      "learning_rate": 9.466700527922021e-05,
      "loss": 0.1711,
      "step": 1113
    },
    {
      "epoch": 0.23383711167086482,
      "grad_norm": 0.14534981548786163,
      "learning_rate": 9.465052754588042e-05,
      "loss": 0.1663,
      "step": 1114
    },
    {
      "epoch": 0.23404701931150293,
      "grad_norm": 0.18299245834350586,
      "learning_rate": 9.463402583418741e-05,
      "loss": 0.1478,
      "step": 1115
    },
    {
      "epoch": 0.23425692695214106,
      "grad_norm": 0.12957628071308136,
      "learning_rate": 9.461750015300296e-05,
      "loss": 0.1528,
      "step": 1116
    },
    {
      "epoch": 0.23446683459277917,
      "grad_norm": 0.13237446546554565,
      "learning_rate": 9.460095051120179e-05,
      "loss": 0.1749,
      "step": 1117
    },
    {
      "epoch": 0.2346767422334173,
      "grad_norm": 0.12967370450496674,
      "learning_rate": 9.458437691767138e-05,
      "loss": 0.1663,
      "step": 1118
    },
    {
      "epoch": 0.23488664987405541,
      "grad_norm": 0.14041563868522644,
      "learning_rate": 9.456777938131216e-05,
      "loss": 0.1581,
      "step": 1119
    },
    {
      "epoch": 0.23509655751469352,
      "grad_norm": 0.1613098829984665,
      "learning_rate": 9.455115791103736e-05,
      "loss": 0.1581,
      "step": 1120
    },
    {
      "epoch": 0.23530646515533166,
      "grad_norm": 0.12044283002614975,
      "learning_rate": 9.453451251577312e-05,
      "loss": 0.1667,
      "step": 1121
    },
    {
      "epoch": 0.23551637279596976,
      "grad_norm": 0.11583617329597473,
      "learning_rate": 9.451784320445838e-05,
      "loss": 0.1703,
      "step": 1122
    },
    {
      "epoch": 0.2357262804366079,
      "grad_norm": 0.13826625049114227,
      "learning_rate": 9.450114998604493e-05,
      "loss": 0.1634,
      "step": 1123
    },
    {
      "epoch": 0.235936188077246,
      "grad_norm": 0.1239803358912468,
      "learning_rate": 9.448443286949745e-05,
      "loss": 0.1593,
      "step": 1124
    },
    {
      "epoch": 0.23614609571788414,
      "grad_norm": 0.11571444571018219,
      "learning_rate": 9.446769186379338e-05,
      "loss": 0.1718,
      "step": 1125
    },
    {
      "epoch": 0.23635600335852225,
      "grad_norm": 0.13045620918273926,
      "learning_rate": 9.445092697792302e-05,
      "loss": 0.1767,
      "step": 1126
    },
    {
      "epoch": 0.23656591099916036,
      "grad_norm": 0.14445173740386963,
      "learning_rate": 9.44341382208895e-05,
      "loss": 0.1518,
      "step": 1127
    },
    {
      "epoch": 0.2367758186397985,
      "grad_norm": 0.13226744532585144,
      "learning_rate": 9.441732560170879e-05,
      "loss": 0.1575,
      "step": 1128
    },
    {
      "epoch": 0.2369857262804366,
      "grad_norm": 0.12810592353343964,
      "learning_rate": 9.440048912940961e-05,
      "loss": 0.1772,
      "step": 1129
    },
    {
      "epoch": 0.23719563392107473,
      "grad_norm": 0.12598370015621185,
      "learning_rate": 9.438362881303358e-05,
      "loss": 0.1663,
      "step": 1130
    },
    {
      "epoch": 0.23740554156171284,
      "grad_norm": 0.11832118779420853,
      "learning_rate": 9.436674466163503e-05,
      "loss": 0.1495,
      "step": 1131
    },
    {
      "epoch": 0.23761544920235098,
      "grad_norm": 0.13778379559516907,
      "learning_rate": 9.434983668428116e-05,
      "loss": 0.1637,
      "step": 1132
    },
    {
      "epoch": 0.23782535684298908,
      "grad_norm": 0.12428466975688934,
      "learning_rate": 9.433290489005193e-05,
      "loss": 0.1552,
      "step": 1133
    },
    {
      "epoch": 0.2380352644836272,
      "grad_norm": 0.11398124694824219,
      "learning_rate": 9.431594928804011e-05,
      "loss": 0.1814,
      "step": 1134
    },
    {
      "epoch": 0.23824517212426533,
      "grad_norm": 0.12184252589941025,
      "learning_rate": 9.429896988735123e-05,
      "loss": 0.1618,
      "step": 1135
    },
    {
      "epoch": 0.23845507976490343,
      "grad_norm": 0.12840035557746887,
      "learning_rate": 9.428196669710363e-05,
      "loss": 0.1858,
      "step": 1136
    },
    {
      "epoch": 0.23866498740554157,
      "grad_norm": 0.12335941940546036,
      "learning_rate": 9.42649397264284e-05,
      "loss": 0.1605,
      "step": 1137
    },
    {
      "epoch": 0.23887489504617967,
      "grad_norm": 0.13650208711624146,
      "learning_rate": 9.42478889844694e-05,
      "loss": 0.1811,
      "step": 1138
    },
    {
      "epoch": 0.2390848026868178,
      "grad_norm": 0.14578799903392792,
      "learning_rate": 9.423081448038334e-05,
      "loss": 0.1781,
      "step": 1139
    },
    {
      "epoch": 0.23929471032745592,
      "grad_norm": 0.13492166996002197,
      "learning_rate": 9.421371622333952e-05,
      "loss": 0.1644,
      "step": 1140
    },
    {
      "epoch": 0.23950461796809405,
      "grad_norm": 0.12785859405994415,
      "learning_rate": 9.419659422252017e-05,
      "loss": 0.1614,
      "step": 1141
    },
    {
      "epoch": 0.23971452560873216,
      "grad_norm": 0.1437363475561142,
      "learning_rate": 9.417944848712014e-05,
      "loss": 0.1737,
      "step": 1142
    },
    {
      "epoch": 0.23992443324937027,
      "grad_norm": 0.12084683775901794,
      "learning_rate": 9.416227902634711e-05,
      "loss": 0.16,
      "step": 1143
    },
    {
      "epoch": 0.2401343408900084,
      "grad_norm": 0.13587456941604614,
      "learning_rate": 9.414508584942147e-05,
      "loss": 0.1666,
      "step": 1144
    },
    {
      "epoch": 0.2403442485306465,
      "grad_norm": 0.1418214589357376,
      "learning_rate": 9.412786896557638e-05,
      "loss": 0.1526,
      "step": 1145
    },
    {
      "epoch": 0.24055415617128464,
      "grad_norm": 0.13639256358146667,
      "learning_rate": 9.411062838405762e-05,
      "loss": 0.1739,
      "step": 1146
    },
    {
      "epoch": 0.24076406381192275,
      "grad_norm": 0.1474551260471344,
      "learning_rate": 9.409336411412386e-05,
      "loss": 0.1626,
      "step": 1147
    },
    {
      "epoch": 0.2409739714525609,
      "grad_norm": 0.14461228251457214,
      "learning_rate": 9.407607616504635e-05,
      "loss": 0.1693,
      "step": 1148
    },
    {
      "epoch": 0.241183879093199,
      "grad_norm": 0.15202075242996216,
      "learning_rate": 9.405876454610915e-05,
      "loss": 0.1607,
      "step": 1149
    },
    {
      "epoch": 0.2413937867338371,
      "grad_norm": 0.14336565136909485,
      "learning_rate": 9.404142926660897e-05,
      "loss": 0.1363,
      "step": 1150
    },
    {
      "epoch": 0.24160369437447524,
      "grad_norm": 0.15390141308307648,
      "learning_rate": 9.402407033585527e-05,
      "loss": 0.1666,
      "step": 1151
    },
    {
      "epoch": 0.24181360201511334,
      "grad_norm": 0.16369935870170593,
      "learning_rate": 9.400668776317017e-05,
      "loss": 0.1584,
      "step": 1152
    },
    {
      "epoch": 0.24202350965575148,
      "grad_norm": 0.13057595491409302,
      "learning_rate": 9.398928155788853e-05,
      "loss": 0.1638,
      "step": 1153
    },
    {
      "epoch": 0.24223341729638959,
      "grad_norm": 0.13949787616729736,
      "learning_rate": 9.397185172935787e-05,
      "loss": 0.1558,
      "step": 1154
    },
    {
      "epoch": 0.24244332493702772,
      "grad_norm": 0.1358940303325653,
      "learning_rate": 9.395439828693841e-05,
      "loss": 0.1482,
      "step": 1155
    },
    {
      "epoch": 0.24265323257766583,
      "grad_norm": 0.1706434041261673,
      "learning_rate": 9.393692124000305e-05,
      "loss": 0.1663,
      "step": 1156
    },
    {
      "epoch": 0.24286314021830394,
      "grad_norm": 0.16532990336418152,
      "learning_rate": 9.391942059793734e-05,
      "loss": 0.169,
      "step": 1157
    },
    {
      "epoch": 0.24307304785894207,
      "grad_norm": 0.13570407032966614,
      "learning_rate": 9.390189637013958e-05,
      "loss": 0.1546,
      "step": 1158
    },
    {
      "epoch": 0.24328295549958018,
      "grad_norm": 0.12366579473018646,
      "learning_rate": 9.388434856602061e-05,
      "loss": 0.1517,
      "step": 1159
    },
    {
      "epoch": 0.2434928631402183,
      "grad_norm": 0.13216842710971832,
      "learning_rate": 9.386677719500406e-05,
      "loss": 0.1765,
      "step": 1160
    },
    {
      "epoch": 0.24370277078085642,
      "grad_norm": 0.15686704218387604,
      "learning_rate": 9.38491822665261e-05,
      "loss": 0.1783,
      "step": 1161
    },
    {
      "epoch": 0.24391267842149456,
      "grad_norm": 0.12082766741514206,
      "learning_rate": 9.383156379003566e-05,
      "loss": 0.1667,
      "step": 1162
    },
    {
      "epoch": 0.24412258606213266,
      "grad_norm": 0.12626822292804718,
      "learning_rate": 9.381392177499426e-05,
      "loss": 0.1589,
      "step": 1163
    },
    {
      "epoch": 0.24433249370277077,
      "grad_norm": 0.13412562012672424,
      "learning_rate": 9.379625623087603e-05,
      "loss": 0.1548,
      "step": 1164
    },
    {
      "epoch": 0.2445424013434089,
      "grad_norm": 0.14654812216758728,
      "learning_rate": 9.37785671671678e-05,
      "loss": 0.1521,
      "step": 1165
    },
    {
      "epoch": 0.244752308984047,
      "grad_norm": 0.14660955965518951,
      "learning_rate": 9.376085459336898e-05,
      "loss": 0.173,
      "step": 1166
    },
    {
      "epoch": 0.24496221662468515,
      "grad_norm": 0.1366211324930191,
      "learning_rate": 9.374311851899166e-05,
      "loss": 0.1795,
      "step": 1167
    },
    {
      "epoch": 0.24517212426532325,
      "grad_norm": 0.1468217670917511,
      "learning_rate": 9.37253589535605e-05,
      "loss": 0.1606,
      "step": 1168
    },
    {
      "epoch": 0.2453820319059614,
      "grad_norm": 0.13741639256477356,
      "learning_rate": 9.370757590661277e-05,
      "loss": 0.1653,
      "step": 1169
    },
    {
      "epoch": 0.2455919395465995,
      "grad_norm": 0.14487077295780182,
      "learning_rate": 9.368976938769839e-05,
      "loss": 0.1638,
      "step": 1170
    },
    {
      "epoch": 0.2458018471872376,
      "grad_norm": 0.14959833025932312,
      "learning_rate": 9.367193940637988e-05,
      "loss": 0.1839,
      "step": 1171
    },
    {
      "epoch": 0.24601175482787574,
      "grad_norm": 0.14737145602703094,
      "learning_rate": 9.365408597223233e-05,
      "loss": 0.1737,
      "step": 1172
    },
    {
      "epoch": 0.24622166246851385,
      "grad_norm": 0.1258353590965271,
      "learning_rate": 9.363620909484345e-05,
      "loss": 0.1574,
      "step": 1173
    },
    {
      "epoch": 0.24643157010915198,
      "grad_norm": 0.12692254781723022,
      "learning_rate": 9.361830878381352e-05,
      "loss": 0.1676,
      "step": 1174
    },
    {
      "epoch": 0.2466414777497901,
      "grad_norm": 0.13535474240779877,
      "learning_rate": 9.360038504875545e-05,
      "loss": 0.162,
      "step": 1175
    },
    {
      "epoch": 0.24685138539042822,
      "grad_norm": 0.1385834813117981,
      "learning_rate": 9.358243789929465e-05,
      "loss": 0.1611,
      "step": 1176
    },
    {
      "epoch": 0.24706129303106633,
      "grad_norm": 0.15537822246551514,
      "learning_rate": 9.356446734506918e-05,
      "loss": 0.1561,
      "step": 1177
    },
    {
      "epoch": 0.24727120067170444,
      "grad_norm": 0.13240060210227966,
      "learning_rate": 9.354647339572961e-05,
      "loss": 0.1566,
      "step": 1178
    },
    {
      "epoch": 0.24748110831234257,
      "grad_norm": 0.12547266483306885,
      "learning_rate": 9.352845606093914e-05,
      "loss": 0.1743,
      "step": 1179
    },
    {
      "epoch": 0.24769101595298068,
      "grad_norm": 0.132894828915596,
      "learning_rate": 9.351041535037347e-05,
      "loss": 0.1733,
      "step": 1180
    },
    {
      "epoch": 0.24790092359361882,
      "grad_norm": 0.1465143859386444,
      "learning_rate": 9.349235127372086e-05,
      "loss": 0.1776,
      "step": 1181
    },
    {
      "epoch": 0.24811083123425692,
      "grad_norm": 0.15798325836658478,
      "learning_rate": 9.347426384068216e-05,
      "loss": 0.1811,
      "step": 1182
    },
    {
      "epoch": 0.24832073887489506,
      "grad_norm": 0.11733090877532959,
      "learning_rate": 9.345615306097071e-05,
      "loss": 0.158,
      "step": 1183
    },
    {
      "epoch": 0.24853064651553317,
      "grad_norm": 0.11072125285863876,
      "learning_rate": 9.343801894431244e-05,
      "loss": 0.1688,
      "step": 1184
    },
    {
      "epoch": 0.24874055415617127,
      "grad_norm": 0.11508002877235413,
      "learning_rate": 9.341986150044574e-05,
      "loss": 0.1538,
      "step": 1185
    },
    {
      "epoch": 0.2489504617968094,
      "grad_norm": 0.13166356086730957,
      "learning_rate": 9.340168073912162e-05,
      "loss": 0.1489,
      "step": 1186
    },
    {
      "epoch": 0.24916036943744752,
      "grad_norm": 0.14776529371738434,
      "learning_rate": 9.338347667010352e-05,
      "loss": 0.1459,
      "step": 1187
    },
    {
      "epoch": 0.24937027707808565,
      "grad_norm": 0.13042989373207092,
      "learning_rate": 9.336524930316749e-05,
      "loss": 0.1595,
      "step": 1188
    },
    {
      "epoch": 0.24958018471872376,
      "grad_norm": 0.11727176606655121,
      "learning_rate": 9.334699864810201e-05,
      "loss": 0.1525,
      "step": 1189
    },
    {
      "epoch": 0.2497900923593619,
      "grad_norm": 0.14481998980045319,
      "learning_rate": 9.332872471470811e-05,
      "loss": 0.1627,
      "step": 1190
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.13064126670360565,
      "learning_rate": 9.331042751279929e-05,
      "loss": 0.1711,
      "step": 1191
    },
    {
      "epoch": 0.2502099076406381,
      "grad_norm": 0.1243947371840477,
      "learning_rate": 9.329210705220159e-05,
      "loss": 0.1592,
      "step": 1192
    },
    {
      "epoch": 0.2504198152812762,
      "grad_norm": 0.10476188361644745,
      "learning_rate": 9.327376334275351e-05,
      "loss": 0.1548,
      "step": 1193
    },
    {
      "epoch": 0.2506297229219144,
      "grad_norm": 0.1116849035024643,
      "learning_rate": 9.325539639430602e-05,
      "loss": 0.1621,
      "step": 1194
    },
    {
      "epoch": 0.2508396305625525,
      "grad_norm": 0.134584441781044,
      "learning_rate": 9.323700621672263e-05,
      "loss": 0.1657,
      "step": 1195
    },
    {
      "epoch": 0.2510495382031906,
      "grad_norm": 0.14008194208145142,
      "learning_rate": 9.321859281987926e-05,
      "loss": 0.1569,
      "step": 1196
    },
    {
      "epoch": 0.2512594458438287,
      "grad_norm": 0.136927530169487,
      "learning_rate": 9.320015621366432e-05,
      "loss": 0.1612,
      "step": 1197
    },
    {
      "epoch": 0.25146935348446686,
      "grad_norm": 0.12949542701244354,
      "learning_rate": 9.31816964079787e-05,
      "loss": 0.1612,
      "step": 1198
    },
    {
      "epoch": 0.25167926112510497,
      "grad_norm": 0.131092369556427,
      "learning_rate": 9.316321341273574e-05,
      "loss": 0.1663,
      "step": 1199
    },
    {
      "epoch": 0.2518891687657431,
      "grad_norm": 0.15064486861228943,
      "learning_rate": 9.314470723786121e-05,
      "loss": 0.1626,
      "step": 1200
    },
    {
      "epoch": 0.2520990764063812,
      "grad_norm": 0.14182250201702118,
      "learning_rate": 9.312617789329339e-05,
      "loss": 0.1775,
      "step": 1201
    },
    {
      "epoch": 0.2523089840470193,
      "grad_norm": 0.14778292179107666,
      "learning_rate": 9.310762538898292e-05,
      "loss": 0.1699,
      "step": 1202
    },
    {
      "epoch": 0.25251889168765745,
      "grad_norm": 0.13292577862739563,
      "learning_rate": 9.308904973489292e-05,
      "loss": 0.1598,
      "step": 1203
    },
    {
      "epoch": 0.25272879932829556,
      "grad_norm": 0.14361746609210968,
      "learning_rate": 9.307045094099898e-05,
      "loss": 0.1701,
      "step": 1204
    },
    {
      "epoch": 0.25293870696893367,
      "grad_norm": 0.18234290182590485,
      "learning_rate": 9.305182901728904e-05,
      "loss": 0.1454,
      "step": 1205
    },
    {
      "epoch": 0.2531486146095718,
      "grad_norm": 0.13576486706733704,
      "learning_rate": 9.30331839737635e-05,
      "loss": 0.1751,
      "step": 1206
    },
    {
      "epoch": 0.2533585222502099,
      "grad_norm": 0.14371106028556824,
      "learning_rate": 9.30145158204352e-05,
      "loss": 0.1625,
      "step": 1207
    },
    {
      "epoch": 0.25356842989084805,
      "grad_norm": 0.14439460635185242,
      "learning_rate": 9.299582456732933e-05,
      "loss": 0.1687,
      "step": 1208
    },
    {
      "epoch": 0.25377833753148615,
      "grad_norm": 0.12489113956689835,
      "learning_rate": 9.297711022448357e-05,
      "loss": 0.1578,
      "step": 1209
    },
    {
      "epoch": 0.25398824517212426,
      "grad_norm": 0.11665599048137665,
      "learning_rate": 9.29583728019479e-05,
      "loss": 0.1585,
      "step": 1210
    },
    {
      "epoch": 0.25419815281276237,
      "grad_norm": 0.1369842141866684,
      "learning_rate": 9.293961230978476e-05,
      "loss": 0.1685,
      "step": 1211
    },
    {
      "epoch": 0.25440806045340053,
      "grad_norm": 0.1267922818660736,
      "learning_rate": 9.292082875806895e-05,
      "loss": 0.1512,
      "step": 1212
    },
    {
      "epoch": 0.25461796809403864,
      "grad_norm": 0.13632832467556,
      "learning_rate": 9.29020221568877e-05,
      "loss": 0.1588,
      "step": 1213
    },
    {
      "epoch": 0.25482787573467675,
      "grad_norm": 0.13703875243663788,
      "learning_rate": 9.288319251634057e-05,
      "loss": 0.1653,
      "step": 1214
    },
    {
      "epoch": 0.25503778337531485,
      "grad_norm": 0.1690177023410797,
      "learning_rate": 9.28643398465395e-05,
      "loss": 0.1452,
      "step": 1215
    },
    {
      "epoch": 0.25524769101595296,
      "grad_norm": 0.13149164617061615,
      "learning_rate": 9.28454641576088e-05,
      "loss": 0.1585,
      "step": 1216
    },
    {
      "epoch": 0.2554575986565911,
      "grad_norm": 0.11899828165769577,
      "learning_rate": 9.282656545968516e-05,
      "loss": 0.1666,
      "step": 1217
    },
    {
      "epoch": 0.25566750629722923,
      "grad_norm": 0.1439366489648819,
      "learning_rate": 9.28076437629176e-05,
      "loss": 0.17,
      "step": 1218
    },
    {
      "epoch": 0.25587741393786734,
      "grad_norm": 0.1383795440196991,
      "learning_rate": 9.278869907746751e-05,
      "loss": 0.1609,
      "step": 1219
    },
    {
      "epoch": 0.25608732157850544,
      "grad_norm": 0.13621094822883606,
      "learning_rate": 9.276973141350862e-05,
      "loss": 0.1657,
      "step": 1220
    },
    {
      "epoch": 0.25629722921914355,
      "grad_norm": 0.10558905452489853,
      "learning_rate": 9.2750740781227e-05,
      "loss": 0.1548,
      "step": 1221
    },
    {
      "epoch": 0.2565071368597817,
      "grad_norm": 0.1380019336938858,
      "learning_rate": 9.273172719082105e-05,
      "loss": 0.1647,
      "step": 1222
    },
    {
      "epoch": 0.2567170445004198,
      "grad_norm": 0.1091807410120964,
      "learning_rate": 9.27126906525015e-05,
      "loss": 0.1598,
      "step": 1223
    },
    {
      "epoch": 0.25692695214105793,
      "grad_norm": 0.13246595859527588,
      "learning_rate": 9.269363117649139e-05,
      "loss": 0.1684,
      "step": 1224
    },
    {
      "epoch": 0.25713685978169604,
      "grad_norm": 0.1521276980638504,
      "learning_rate": 9.267454877302614e-05,
      "loss": 0.1745,
      "step": 1225
    },
    {
      "epoch": 0.2573467674223342,
      "grad_norm": 0.13144977390766144,
      "learning_rate": 9.26554434523534e-05,
      "loss": 0.1652,
      "step": 1226
    },
    {
      "epoch": 0.2575566750629723,
      "grad_norm": 0.14672890305519104,
      "learning_rate": 9.263631522473317e-05,
      "loss": 0.1563,
      "step": 1227
    },
    {
      "epoch": 0.2577665827036104,
      "grad_norm": 0.1289316564798355,
      "learning_rate": 9.261716410043774e-05,
      "loss": 0.1594,
      "step": 1228
    },
    {
      "epoch": 0.2579764903442485,
      "grad_norm": 0.13553071022033691,
      "learning_rate": 9.259799008975173e-05,
      "loss": 0.1457,
      "step": 1229
    },
    {
      "epoch": 0.25818639798488663,
      "grad_norm": 0.11398561298847198,
      "learning_rate": 9.257879320297198e-05,
      "loss": 0.1583,
      "step": 1230
    },
    {
      "epoch": 0.2583963056255248,
      "grad_norm": 0.13140664994716644,
      "learning_rate": 9.255957345040769e-05,
      "loss": 0.1667,
      "step": 1231
    },
    {
      "epoch": 0.2586062132661629,
      "grad_norm": 0.11999331414699554,
      "learning_rate": 9.25403308423803e-05,
      "loss": 0.1656,
      "step": 1232
    },
    {
      "epoch": 0.258816120906801,
      "grad_norm": 0.14149609208106995,
      "learning_rate": 9.25210653892235e-05,
      "loss": 0.1605,
      "step": 1233
    },
    {
      "epoch": 0.2590260285474391,
      "grad_norm": 0.1261209398508072,
      "learning_rate": 9.250177710128332e-05,
      "loss": 0.1594,
      "step": 1234
    },
    {
      "epoch": 0.2592359361880772,
      "grad_norm": 0.15169532597064972,
      "learning_rate": 9.248246598891798e-05,
      "loss": 0.1744,
      "step": 1235
    },
    {
      "epoch": 0.2594458438287154,
      "grad_norm": 0.12228883057832718,
      "learning_rate": 9.246313206249802e-05,
      "loss": 0.1753,
      "step": 1236
    },
    {
      "epoch": 0.2596557514693535,
      "grad_norm": 0.14006085693836212,
      "learning_rate": 9.244377533240618e-05,
      "loss": 0.1614,
      "step": 1237
    },
    {
      "epoch": 0.2598656591099916,
      "grad_norm": 0.11327286064624786,
      "learning_rate": 9.242439580903745e-05,
      "loss": 0.1447,
      "step": 1238
    },
    {
      "epoch": 0.2600755667506297,
      "grad_norm": 0.11881330609321594,
      "learning_rate": 9.24049935027991e-05,
      "loss": 0.153,
      "step": 1239
    },
    {
      "epoch": 0.26028547439126787,
      "grad_norm": 0.11056238412857056,
      "learning_rate": 9.238556842411061e-05,
      "loss": 0.1607,
      "step": 1240
    },
    {
      "epoch": 0.260495382031906,
      "grad_norm": 0.1527404934167862,
      "learning_rate": 9.23661205834037e-05,
      "loss": 0.182,
      "step": 1241
    },
    {
      "epoch": 0.2607052896725441,
      "grad_norm": 0.11118711531162262,
      "learning_rate": 9.234664999112229e-05,
      "loss": 0.1756,
      "step": 1242
    },
    {
      "epoch": 0.2609151973131822,
      "grad_norm": 0.1669243574142456,
      "learning_rate": 9.232715665772252e-05,
      "loss": 0.1475,
      "step": 1243
    },
    {
      "epoch": 0.2611251049538203,
      "grad_norm": 0.14391769468784332,
      "learning_rate": 9.230764059367276e-05,
      "loss": 0.1807,
      "step": 1244
    },
    {
      "epoch": 0.26133501259445846,
      "grad_norm": 0.11989136785268784,
      "learning_rate": 9.228810180945362e-05,
      "loss": 0.1689,
      "step": 1245
    },
    {
      "epoch": 0.26154492023509657,
      "grad_norm": 0.1314937323331833,
      "learning_rate": 9.226854031555783e-05,
      "loss": 0.1672,
      "step": 1246
    },
    {
      "epoch": 0.2617548278757347,
      "grad_norm": 0.13055700063705444,
      "learning_rate": 9.224895612249039e-05,
      "loss": 0.1807,
      "step": 1247
    },
    {
      "epoch": 0.2619647355163728,
      "grad_norm": 0.1459030658006668,
      "learning_rate": 9.222934924076843e-05,
      "loss": 0.1669,
      "step": 1248
    },
    {
      "epoch": 0.2621746431570109,
      "grad_norm": 0.14526896178722382,
      "learning_rate": 9.22097196809213e-05,
      "loss": 0.1546,
      "step": 1249
    },
    {
      "epoch": 0.26238455079764905,
      "grad_norm": 0.12496902793645859,
      "learning_rate": 9.219006745349053e-05,
      "loss": 0.1733,
      "step": 1250
    },
    {
      "epoch": 0.26259445843828716,
      "grad_norm": 0.12503020465373993,
      "learning_rate": 9.217039256902983e-05,
      "loss": 0.1697,
      "step": 1251
    },
    {
      "epoch": 0.26280436607892527,
      "grad_norm": 0.1265639215707779,
      "learning_rate": 9.215069503810505e-05,
      "loss": 0.1688,
      "step": 1252
    },
    {
      "epoch": 0.2630142737195634,
      "grad_norm": 0.1370651125907898,
      "learning_rate": 9.213097487129418e-05,
      "loss": 0.138,
      "step": 1253
    },
    {
      "epoch": 0.26322418136020154,
      "grad_norm": 0.1354241967201233,
      "learning_rate": 9.211123207918746e-05,
      "loss": 0.1563,
      "step": 1254
    },
    {
      "epoch": 0.26343408900083964,
      "grad_norm": 0.13176575303077698,
      "learning_rate": 9.209146667238719e-05,
      "loss": 0.1683,
      "step": 1255
    },
    {
      "epoch": 0.26364399664147775,
      "grad_norm": 0.1484147608280182,
      "learning_rate": 9.207167866150783e-05,
      "loss": 0.1617,
      "step": 1256
    },
    {
      "epoch": 0.26385390428211586,
      "grad_norm": 0.1321289986371994,
      "learning_rate": 9.205186805717603e-05,
      "loss": 0.1592,
      "step": 1257
    },
    {
      "epoch": 0.26406381192275397,
      "grad_norm": 0.13387440145015717,
      "learning_rate": 9.203203487003049e-05,
      "loss": 0.1758,
      "step": 1258
    },
    {
      "epoch": 0.26427371956339213,
      "grad_norm": 0.15501335263252258,
      "learning_rate": 9.201217911072213e-05,
      "loss": 0.1748,
      "step": 1259
    },
    {
      "epoch": 0.26448362720403024,
      "grad_norm": 0.13924849033355713,
      "learning_rate": 9.199230078991393e-05,
      "loss": 0.1526,
      "step": 1260
    },
    {
      "epoch": 0.26469353484466834,
      "grad_norm": 0.155021071434021,
      "learning_rate": 9.197239991828099e-05,
      "loss": 0.1566,
      "step": 1261
    },
    {
      "epoch": 0.26490344248530645,
      "grad_norm": 0.10661531239748001,
      "learning_rate": 9.195247650651055e-05,
      "loss": 0.159,
      "step": 1262
    },
    {
      "epoch": 0.26511335012594456,
      "grad_norm": 0.15488192439079285,
      "learning_rate": 9.193253056530194e-05,
      "loss": 0.1772,
      "step": 1263
    },
    {
      "epoch": 0.2653232577665827,
      "grad_norm": 0.11265954375267029,
      "learning_rate": 9.191256210536655e-05,
      "loss": 0.1549,
      "step": 1264
    },
    {
      "epoch": 0.26553316540722083,
      "grad_norm": 0.1617230325937271,
      "learning_rate": 9.189257113742793e-05,
      "loss": 0.1737,
      "step": 1265
    },
    {
      "epoch": 0.26574307304785894,
      "grad_norm": 0.11783073842525482,
      "learning_rate": 9.187255767222169e-05,
      "loss": 0.1615,
      "step": 1266
    },
    {
      "epoch": 0.26595298068849704,
      "grad_norm": 0.14353324472904205,
      "learning_rate": 9.185252172049553e-05,
      "loss": 0.1578,
      "step": 1267
    },
    {
      "epoch": 0.2661628883291352,
      "grad_norm": 0.1482020765542984,
      "learning_rate": 9.183246329300917e-05,
      "loss": 0.1483,
      "step": 1268
    },
    {
      "epoch": 0.2663727959697733,
      "grad_norm": 0.1308319866657257,
      "learning_rate": 9.181238240053446e-05,
      "loss": 0.1742,
      "step": 1269
    },
    {
      "epoch": 0.2665827036104114,
      "grad_norm": 0.13996942341327667,
      "learning_rate": 9.179227905385532e-05,
      "loss": 0.1649,
      "step": 1270
    },
    {
      "epoch": 0.2667926112510495,
      "grad_norm": 0.13277165591716766,
      "learning_rate": 9.177215326376769e-05,
      "loss": 0.1732,
      "step": 1271
    },
    {
      "epoch": 0.26700251889168763,
      "grad_norm": 0.12486343085765839,
      "learning_rate": 9.175200504107957e-05,
      "loss": 0.1493,
      "step": 1272
    },
    {
      "epoch": 0.2672124265323258,
      "grad_norm": 0.12765567004680634,
      "learning_rate": 9.173183439661103e-05,
      "loss": 0.1533,
      "step": 1273
    },
    {
      "epoch": 0.2674223341729639,
      "grad_norm": 0.11738407611846924,
      "learning_rate": 9.171164134119417e-05,
      "loss": 0.1551,
      "step": 1274
    },
    {
      "epoch": 0.267632241813602,
      "grad_norm": 0.12661415338516235,
      "learning_rate": 9.16914258856731e-05,
      "loss": 0.1655,
      "step": 1275
    },
    {
      "epoch": 0.2678421494542401,
      "grad_norm": 0.10266046971082687,
      "learning_rate": 9.167118804090401e-05,
      "loss": 0.1506,
      "step": 1276
    },
    {
      "epoch": 0.2680520570948782,
      "grad_norm": 0.13607458770275116,
      "learning_rate": 9.165092781775506e-05,
      "loss": 0.1621,
      "step": 1277
    },
    {
      "epoch": 0.2682619647355164,
      "grad_norm": 0.14382155239582062,
      "learning_rate": 9.163064522710649e-05,
      "loss": 0.1756,
      "step": 1278
    },
    {
      "epoch": 0.2684718723761545,
      "grad_norm": 0.14696674048900604,
      "learning_rate": 9.161034027985047e-05,
      "loss": 0.1506,
      "step": 1279
    },
    {
      "epoch": 0.2686817800167926,
      "grad_norm": 0.12927404046058655,
      "learning_rate": 9.159001298689122e-05,
      "loss": 0.1693,
      "step": 1280
    },
    {
      "epoch": 0.2688916876574307,
      "grad_norm": 0.1403432935476303,
      "learning_rate": 9.1569663359145e-05,
      "loss": 0.178,
      "step": 1281
    },
    {
      "epoch": 0.2691015952980689,
      "grad_norm": 0.12026988714933395,
      "learning_rate": 9.154929140754e-05,
      "loss": 0.1611,
      "step": 1282
    },
    {
      "epoch": 0.269311502938707,
      "grad_norm": 0.1373865157365799,
      "learning_rate": 9.152889714301645e-05,
      "loss": 0.1696,
      "step": 1283
    },
    {
      "epoch": 0.2695214105793451,
      "grad_norm": 0.13119877874851227,
      "learning_rate": 9.150848057652648e-05,
      "loss": 0.178,
      "step": 1284
    },
    {
      "epoch": 0.2697313182199832,
      "grad_norm": 0.12198714911937714,
      "learning_rate": 9.148804171903432e-05,
      "loss": 0.1579,
      "step": 1285
    },
    {
      "epoch": 0.2699412258606213,
      "grad_norm": 0.12499682605266571,
      "learning_rate": 9.146758058151603e-05,
      "loss": 0.1669,
      "step": 1286
    },
    {
      "epoch": 0.27015113350125947,
      "grad_norm": 0.12025036662817001,
      "learning_rate": 9.144709717495978e-05,
      "loss": 0.1602,
      "step": 1287
    },
    {
      "epoch": 0.2703610411418976,
      "grad_norm": 0.1325342059135437,
      "learning_rate": 9.142659151036558e-05,
      "loss": 0.1559,
      "step": 1288
    },
    {
      "epoch": 0.2705709487825357,
      "grad_norm": 0.14255130290985107,
      "learning_rate": 9.140606359874546e-05,
      "loss": 0.1441,
      "step": 1289
    },
    {
      "epoch": 0.2707808564231738,
      "grad_norm": 0.13800451159477234,
      "learning_rate": 9.138551345112337e-05,
      "loss": 0.1543,
      "step": 1290
    },
    {
      "epoch": 0.2709907640638119,
      "grad_norm": 0.10976230353116989,
      "learning_rate": 9.136494107853521e-05,
      "loss": 0.1513,
      "step": 1291
    },
    {
      "epoch": 0.27120067170445006,
      "grad_norm": 0.1330142766237259,
      "learning_rate": 9.134434649202882e-05,
      "loss": 0.1527,
      "step": 1292
    },
    {
      "epoch": 0.27141057934508817,
      "grad_norm": 0.13712665438652039,
      "learning_rate": 9.132372970266397e-05,
      "loss": 0.1621,
      "step": 1293
    },
    {
      "epoch": 0.2716204869857263,
      "grad_norm": 0.15420737862586975,
      "learning_rate": 9.130309072151231e-05,
      "loss": 0.1773,
      "step": 1294
    },
    {
      "epoch": 0.2718303946263644,
      "grad_norm": 0.14155341684818268,
      "learning_rate": 9.12824295596575e-05,
      "loss": 0.1615,
      "step": 1295
    },
    {
      "epoch": 0.27204030226700254,
      "grad_norm": 0.14036308228969574,
      "learning_rate": 9.126174622819499e-05,
      "loss": 0.1496,
      "step": 1296
    },
    {
      "epoch": 0.27225020990764065,
      "grad_norm": 0.13568180799484253,
      "learning_rate": 9.124104073823225e-05,
      "loss": 0.1562,
      "step": 1297
    },
    {
      "epoch": 0.27246011754827876,
      "grad_norm": 0.12056776136159897,
      "learning_rate": 9.122031310088861e-05,
      "loss": 0.1722,
      "step": 1298
    },
    {
      "epoch": 0.27267002518891686,
      "grad_norm": 0.139260396361351,
      "learning_rate": 9.119956332729524e-05,
      "loss": 0.1797,
      "step": 1299
    },
    {
      "epoch": 0.27287993282955497,
      "grad_norm": 0.16409984230995178,
      "learning_rate": 9.117879142859527e-05,
      "loss": 0.1661,
      "step": 1300
    },
    {
      "epoch": 0.27308984047019313,
      "grad_norm": 0.1330333650112152,
      "learning_rate": 9.115799741594368e-05,
      "loss": 0.1794,
      "step": 1301
    },
    {
      "epoch": 0.27329974811083124,
      "grad_norm": 0.12173225730657578,
      "learning_rate": 9.113718130050734e-05,
      "loss": 0.1449,
      "step": 1302
    },
    {
      "epoch": 0.27350965575146935,
      "grad_norm": 0.10996419191360474,
      "learning_rate": 9.111634309346496e-05,
      "loss": 0.1626,
      "step": 1303
    },
    {
      "epoch": 0.27371956339210746,
      "grad_norm": 0.13602428138256073,
      "learning_rate": 9.109548280600715e-05,
      "loss": 0.1769,
      "step": 1304
    },
    {
      "epoch": 0.2739294710327456,
      "grad_norm": 0.11914105713367462,
      "learning_rate": 9.107460044933635e-05,
      "loss": 0.1664,
      "step": 1305
    },
    {
      "epoch": 0.2741393786733837,
      "grad_norm": 0.12417447566986084,
      "learning_rate": 9.105369603466686e-05,
      "loss": 0.1455,
      "step": 1306
    },
    {
      "epoch": 0.27434928631402183,
      "grad_norm": 0.13093994557857513,
      "learning_rate": 9.103276957322484e-05,
      "loss": 0.1721,
      "step": 1307
    },
    {
      "epoch": 0.27455919395465994,
      "grad_norm": 0.13235576450824738,
      "learning_rate": 9.101182107624827e-05,
      "loss": 0.1614,
      "step": 1308
    },
    {
      "epoch": 0.27476910159529805,
      "grad_norm": 0.11165154725313187,
      "learning_rate": 9.099085055498696e-05,
      "loss": 0.1656,
      "step": 1309
    },
    {
      "epoch": 0.2749790092359362,
      "grad_norm": 0.10962437093257904,
      "learning_rate": 9.096985802070259e-05,
      "loss": 0.1633,
      "step": 1310
    },
    {
      "epoch": 0.2751889168765743,
      "grad_norm": 0.12218458205461502,
      "learning_rate": 9.09488434846686e-05,
      "loss": 0.1542,
      "step": 1311
    },
    {
      "epoch": 0.2753988245172124,
      "grad_norm": 0.11284459382295609,
      "learning_rate": 9.09278069581703e-05,
      "loss": 0.1591,
      "step": 1312
    },
    {
      "epoch": 0.27560873215785053,
      "grad_norm": 0.10450571775436401,
      "learning_rate": 9.090674845250475e-05,
      "loss": 0.1571,
      "step": 1313
    },
    {
      "epoch": 0.27581863979848864,
      "grad_norm": 0.12627463042736053,
      "learning_rate": 9.088566797898089e-05,
      "loss": 0.1613,
      "step": 1314
    },
    {
      "epoch": 0.2760285474391268,
      "grad_norm": 0.1578662246465683,
      "learning_rate": 9.08645655489194e-05,
      "loss": 0.1506,
      "step": 1315
    },
    {
      "epoch": 0.2762384550797649,
      "grad_norm": 0.11859230697154999,
      "learning_rate": 9.084344117365276e-05,
      "loss": 0.1502,
      "step": 1316
    },
    {
      "epoch": 0.276448362720403,
      "grad_norm": 0.11622706055641174,
      "learning_rate": 9.082229486452524e-05,
      "loss": 0.1674,
      "step": 1317
    },
    {
      "epoch": 0.2766582703610411,
      "grad_norm": 0.12626247107982635,
      "learning_rate": 9.08011266328929e-05,
      "loss": 0.1522,
      "step": 1318
    },
    {
      "epoch": 0.2768681780016793,
      "grad_norm": 0.13337282836437225,
      "learning_rate": 9.077993649012357e-05,
      "loss": 0.1647,
      "step": 1319
    },
    {
      "epoch": 0.2770780856423174,
      "grad_norm": 0.1255621761083603,
      "learning_rate": 9.075872444759683e-05,
      "loss": 0.1564,
      "step": 1320
    },
    {
      "epoch": 0.2772879932829555,
      "grad_norm": 0.12364380061626434,
      "learning_rate": 9.073749051670403e-05,
      "loss": 0.1555,
      "step": 1321
    },
    {
      "epoch": 0.2774979009235936,
      "grad_norm": 0.11966480314731598,
      "learning_rate": 9.07162347088483e-05,
      "loss": 0.163,
      "step": 1322
    },
    {
      "epoch": 0.2777078085642317,
      "grad_norm": 0.10513235628604889,
      "learning_rate": 9.069495703544443e-05,
      "loss": 0.1578,
      "step": 1323
    },
    {
      "epoch": 0.2779177162048699,
      "grad_norm": 0.10422056168317795,
      "learning_rate": 9.06736575079191e-05,
      "loss": 0.1435,
      "step": 1324
    },
    {
      "epoch": 0.278127623845508,
      "grad_norm": 0.12099135667085648,
      "learning_rate": 9.065233613771059e-05,
      "loss": 0.1606,
      "step": 1325
    },
    {
      "epoch": 0.2783375314861461,
      "grad_norm": 0.10527081042528152,
      "learning_rate": 9.063099293626898e-05,
      "loss": 0.1606,
      "step": 1326
    },
    {
      "epoch": 0.2785474391267842,
      "grad_norm": 0.15932826697826385,
      "learning_rate": 9.060962791505605e-05,
      "loss": 0.1626,
      "step": 1327
    },
    {
      "epoch": 0.2787573467674223,
      "grad_norm": 0.14808166027069092,
      "learning_rate": 9.05882410855453e-05,
      "loss": 0.1648,
      "step": 1328
    },
    {
      "epoch": 0.27896725440806047,
      "grad_norm": 0.1467159539461136,
      "learning_rate": 9.056683245922196e-05,
      "loss": 0.1533,
      "step": 1329
    },
    {
      "epoch": 0.2791771620486986,
      "grad_norm": 0.13326837122440338,
      "learning_rate": 9.054540204758295e-05,
      "loss": 0.1638,
      "step": 1330
    },
    {
      "epoch": 0.2793870696893367,
      "grad_norm": 0.14456981420516968,
      "learning_rate": 9.052394986213688e-05,
      "loss": 0.1639,
      "step": 1331
    },
    {
      "epoch": 0.2795969773299748,
      "grad_norm": 0.1200864389538765,
      "learning_rate": 9.050247591440407e-05,
      "loss": 0.1556,
      "step": 1332
    },
    {
      "epoch": 0.27980688497061296,
      "grad_norm": 0.1340358853340149,
      "learning_rate": 9.04809802159165e-05,
      "loss": 0.1627,
      "step": 1333
    },
    {
      "epoch": 0.28001679261125106,
      "grad_norm": 0.14233215153217316,
      "learning_rate": 9.045946277821791e-05,
      "loss": 0.1604,
      "step": 1334
    },
    {
      "epoch": 0.28022670025188917,
      "grad_norm": 0.15358757972717285,
      "learning_rate": 9.04379236128636e-05,
      "loss": 0.1709,
      "step": 1335
    },
    {
      "epoch": 0.2804366078925273,
      "grad_norm": 0.14173360168933868,
      "learning_rate": 9.041636273142061e-05,
      "loss": 0.1659,
      "step": 1336
    },
    {
      "epoch": 0.2806465155331654,
      "grad_norm": 0.13441811501979828,
      "learning_rate": 9.039478014546761e-05,
      "loss": 0.171,
      "step": 1337
    },
    {
      "epoch": 0.28085642317380355,
      "grad_norm": 0.12063397467136383,
      "learning_rate": 9.037317586659498e-05,
      "loss": 0.1514,
      "step": 1338
    },
    {
      "epoch": 0.28106633081444166,
      "grad_norm": 0.15047864615917206,
      "learning_rate": 9.035154990640466e-05,
      "loss": 0.179,
      "step": 1339
    },
    {
      "epoch": 0.28127623845507976,
      "grad_norm": 0.1325361728668213,
      "learning_rate": 9.032990227651034e-05,
      "loss": 0.1659,
      "step": 1340
    },
    {
      "epoch": 0.28148614609571787,
      "grad_norm": 0.14775165915489197,
      "learning_rate": 9.030823298853725e-05,
      "loss": 0.1732,
      "step": 1341
    },
    {
      "epoch": 0.281696053736356,
      "grad_norm": 0.1677199900150299,
      "learning_rate": 9.028654205412233e-05,
      "loss": 0.1696,
      "step": 1342
    },
    {
      "epoch": 0.28190596137699414,
      "grad_norm": 0.1279178410768509,
      "learning_rate": 9.026482948491407e-05,
      "loss": 0.1529,
      "step": 1343
    },
    {
      "epoch": 0.28211586901763225,
      "grad_norm": 0.12682883441448212,
      "learning_rate": 9.024309529257262e-05,
      "loss": 0.1513,
      "step": 1344
    },
    {
      "epoch": 0.28232577665827036,
      "grad_norm": 0.10392288863658905,
      "learning_rate": 9.022133948876975e-05,
      "loss": 0.1611,
      "step": 1345
    },
    {
      "epoch": 0.28253568429890846,
      "grad_norm": 0.11686202883720398,
      "learning_rate": 9.019956208518883e-05,
      "loss": 0.1672,
      "step": 1346
    },
    {
      "epoch": 0.2827455919395466,
      "grad_norm": 0.12934426963329315,
      "learning_rate": 9.017776309352481e-05,
      "loss": 0.1433,
      "step": 1347
    },
    {
      "epoch": 0.28295549958018473,
      "grad_norm": 0.11639254540205002,
      "learning_rate": 9.015594252548426e-05,
      "loss": 0.1707,
      "step": 1348
    },
    {
      "epoch": 0.28316540722082284,
      "grad_norm": 0.12937045097351074,
      "learning_rate": 9.013410039278531e-05,
      "loss": 0.1553,
      "step": 1349
    },
    {
      "epoch": 0.28337531486146095,
      "grad_norm": 0.1247095987200737,
      "learning_rate": 9.01122367071577e-05,
      "loss": 0.1679,
      "step": 1350
    },
    {
      "epoch": 0.28358522250209905,
      "grad_norm": 0.12007705122232437,
      "learning_rate": 9.00903514803427e-05,
      "loss": 0.1554,
      "step": 1351
    },
    {
      "epoch": 0.2837951301427372,
      "grad_norm": 0.17036741971969604,
      "learning_rate": 9.00684447240932e-05,
      "loss": 0.1698,
      "step": 1352
    },
    {
      "epoch": 0.2840050377833753,
      "grad_norm": 0.11295659095048904,
      "learning_rate": 9.004651645017363e-05,
      "loss": 0.1635,
      "step": 1353
    },
    {
      "epoch": 0.28421494542401343,
      "grad_norm": 0.1332911103963852,
      "learning_rate": 9.002456667035997e-05,
      "loss": 0.169,
      "step": 1354
    },
    {
      "epoch": 0.28442485306465154,
      "grad_norm": 0.1429576724767685,
      "learning_rate": 9.000259539643972e-05,
      "loss": 0.1641,
      "step": 1355
    },
    {
      "epoch": 0.28463476070528965,
      "grad_norm": 0.14206334948539734,
      "learning_rate": 8.998060264021201e-05,
      "loss": 0.1692,
      "step": 1356
    },
    {
      "epoch": 0.2848446683459278,
      "grad_norm": 0.11948953568935394,
      "learning_rate": 8.995858841348743e-05,
      "loss": 0.1594,
      "step": 1357
    },
    {
      "epoch": 0.2850545759865659,
      "grad_norm": 0.12463600933551788,
      "learning_rate": 8.993655272808811e-05,
      "loss": 0.1657,
      "step": 1358
    },
    {
      "epoch": 0.285264483627204,
      "grad_norm": 0.15918885171413422,
      "learning_rate": 8.991449559584775e-05,
      "loss": 0.1532,
      "step": 1359
    },
    {
      "epoch": 0.28547439126784213,
      "grad_norm": 0.12887489795684814,
      "learning_rate": 8.989241702861149e-05,
      "loss": 0.1724,
      "step": 1360
    },
    {
      "epoch": 0.2856842989084803,
      "grad_norm": 0.1456436812877655,
      "learning_rate": 8.987031703823606e-05,
      "loss": 0.1529,
      "step": 1361
    },
    {
      "epoch": 0.2858942065491184,
      "grad_norm": 0.16147381067276,
      "learning_rate": 8.984819563658964e-05,
      "loss": 0.1655,
      "step": 1362
    },
    {
      "epoch": 0.2861041141897565,
      "grad_norm": 0.18096451461315155,
      "learning_rate": 8.982605283555192e-05,
      "loss": 0.152,
      "step": 1363
    },
    {
      "epoch": 0.2863140218303946,
      "grad_norm": 0.12937211990356445,
      "learning_rate": 8.980388864701412e-05,
      "loss": 0.1627,
      "step": 1364
    },
    {
      "epoch": 0.2865239294710327,
      "grad_norm": 0.16723109781742096,
      "learning_rate": 8.97817030828789e-05,
      "loss": 0.1666,
      "step": 1365
    },
    {
      "epoch": 0.2867338371116709,
      "grad_norm": 0.1125638410449028,
      "learning_rate": 8.97594961550604e-05,
      "loss": 0.1514,
      "step": 1366
    },
    {
      "epoch": 0.286943744752309,
      "grad_norm": 0.12200607359409332,
      "learning_rate": 8.973726787548427e-05,
      "loss": 0.1593,
      "step": 1367
    },
    {
      "epoch": 0.2871536523929471,
      "grad_norm": 0.12197429686784744,
      "learning_rate": 8.971501825608762e-05,
      "loss": 0.1662,
      "step": 1368
    },
    {
      "epoch": 0.2873635600335852,
      "grad_norm": 0.11950329691171646,
      "learning_rate": 8.969274730881893e-05,
      "loss": 0.1619,
      "step": 1369
    },
    {
      "epoch": 0.2875734676742233,
      "grad_norm": 0.13871550559997559,
      "learning_rate": 8.96704550456383e-05,
      "loss": 0.1467,
      "step": 1370
    },
    {
      "epoch": 0.2877833753148615,
      "grad_norm": 0.11093387007713318,
      "learning_rate": 8.964814147851713e-05,
      "loss": 0.1703,
      "step": 1371
    },
    {
      "epoch": 0.2879932829554996,
      "grad_norm": 0.13531048595905304,
      "learning_rate": 8.962580661943831e-05,
      "loss": 0.1587,
      "step": 1372
    },
    {
      "epoch": 0.2882031905961377,
      "grad_norm": 0.11192765831947327,
      "learning_rate": 8.96034504803962e-05,
      "loss": 0.1537,
      "step": 1373
    },
    {
      "epoch": 0.2884130982367758,
      "grad_norm": 0.12293984740972519,
      "learning_rate": 8.958107307339653e-05,
      "loss": 0.1644,
      "step": 1374
    },
    {
      "epoch": 0.28862300587741396,
      "grad_norm": 0.12963594496250153,
      "learning_rate": 8.955867441045652e-05,
      "loss": 0.16,
      "step": 1375
    },
    {
      "epoch": 0.28883291351805207,
      "grad_norm": 0.11687605082988739,
      "learning_rate": 8.95362545036047e-05,
      "loss": 0.157,
      "step": 1376
    },
    {
      "epoch": 0.2890428211586902,
      "grad_norm": 0.11696937680244446,
      "learning_rate": 8.951381336488114e-05,
      "loss": 0.1572,
      "step": 1377
    },
    {
      "epoch": 0.2892527287993283,
      "grad_norm": 0.12374909222126007,
      "learning_rate": 8.949135100633718e-05,
      "loss": 0.1584,
      "step": 1378
    },
    {
      "epoch": 0.2894626364399664,
      "grad_norm": 0.133868008852005,
      "learning_rate": 8.946886744003569e-05,
      "loss": 0.1415,
      "step": 1379
    },
    {
      "epoch": 0.28967254408060455,
      "grad_norm": 0.11039866507053375,
      "learning_rate": 8.94463626780508e-05,
      "loss": 0.1783,
      "step": 1380
    },
    {
      "epoch": 0.28988245172124266,
      "grad_norm": 0.1570889800786972,
      "learning_rate": 8.94238367324681e-05,
      "loss": 0.1707,
      "step": 1381
    },
    {
      "epoch": 0.29009235936188077,
      "grad_norm": 0.1414021998643875,
      "learning_rate": 8.940128961538454e-05,
      "loss": 0.1339,
      "step": 1382
    },
    {
      "epoch": 0.2903022670025189,
      "grad_norm": 0.12162122875452042,
      "learning_rate": 8.937872133890843e-05,
      "loss": 0.1595,
      "step": 1383
    },
    {
      "epoch": 0.290512174643157,
      "grad_norm": 0.1278192549943924,
      "learning_rate": 8.935613191515946e-05,
      "loss": 0.1386,
      "step": 1384
    },
    {
      "epoch": 0.29072208228379515,
      "grad_norm": 0.12220022082328796,
      "learning_rate": 8.933352135626867e-05,
      "loss": 0.1663,
      "step": 1385
    },
    {
      "epoch": 0.29093198992443325,
      "grad_norm": 0.15517336130142212,
      "learning_rate": 8.931088967437844e-05,
      "loss": 0.1596,
      "step": 1386
    },
    {
      "epoch": 0.29114189756507136,
      "grad_norm": 0.13512621819972992,
      "learning_rate": 8.928823688164249e-05,
      "loss": 0.1412,
      "step": 1387
    },
    {
      "epoch": 0.29135180520570947,
      "grad_norm": 0.10588439553976059,
      "learning_rate": 8.92655629902259e-05,
      "loss": 0.1502,
      "step": 1388
    },
    {
      "epoch": 0.29156171284634763,
      "grad_norm": 0.16245464980602264,
      "learning_rate": 8.924286801230506e-05,
      "loss": 0.1669,
      "step": 1389
    },
    {
      "epoch": 0.29177162048698574,
      "grad_norm": 0.12498240917921066,
      "learning_rate": 8.92201519600677e-05,
      "loss": 0.1518,
      "step": 1390
    },
    {
      "epoch": 0.29198152812762385,
      "grad_norm": 0.13893572986125946,
      "learning_rate": 8.919741484571286e-05,
      "loss": 0.18,
      "step": 1391
    },
    {
      "epoch": 0.29219143576826195,
      "grad_norm": 0.1361815184354782,
      "learning_rate": 8.917465668145086e-05,
      "loss": 0.166,
      "step": 1392
    },
    {
      "epoch": 0.29240134340890006,
      "grad_norm": 0.12212132662534714,
      "learning_rate": 8.915187747950339e-05,
      "loss": 0.1606,
      "step": 1393
    },
    {
      "epoch": 0.2926112510495382,
      "grad_norm": 0.12465030699968338,
      "learning_rate": 8.912907725210342e-05,
      "loss": 0.1454,
      "step": 1394
    },
    {
      "epoch": 0.29282115869017633,
      "grad_norm": 0.15186457335948944,
      "learning_rate": 8.910625601149512e-05,
      "loss": 0.1581,
      "step": 1395
    },
    {
      "epoch": 0.29303106633081444,
      "grad_norm": 0.1369972676038742,
      "learning_rate": 8.908341376993409e-05,
      "loss": 0.1636,
      "step": 1396
    },
    {
      "epoch": 0.29324097397145255,
      "grad_norm": 0.12631604075431824,
      "learning_rate": 8.90605505396871e-05,
      "loss": 0.1536,
      "step": 1397
    },
    {
      "epoch": 0.29345088161209065,
      "grad_norm": 0.13023649156093597,
      "learning_rate": 8.903766633303222e-05,
      "loss": 0.1615,
      "step": 1398
    },
    {
      "epoch": 0.2936607892527288,
      "grad_norm": 0.13841018080711365,
      "learning_rate": 8.90147611622588e-05,
      "loss": 0.1756,
      "step": 1399
    },
    {
      "epoch": 0.2938706968933669,
      "grad_norm": 0.14210903644561768,
      "learning_rate": 8.899183503966746e-05,
      "loss": 0.1661,
      "step": 1400
    },
    {
      "epoch": 0.29408060453400503,
      "grad_norm": 0.13981324434280396,
      "learning_rate": 8.896888797757003e-05,
      "loss": 0.1547,
      "step": 1401
    },
    {
      "epoch": 0.29429051217464314,
      "grad_norm": 0.11732388287782669,
      "learning_rate": 8.89459199882896e-05,
      "loss": 0.1648,
      "step": 1402
    },
    {
      "epoch": 0.2945004198152813,
      "grad_norm": 0.13775314390659332,
      "learning_rate": 8.892293108416055e-05,
      "loss": 0.171,
      "step": 1403
    },
    {
      "epoch": 0.2947103274559194,
      "grad_norm": 0.11232218891382217,
      "learning_rate": 8.889992127752839e-05,
      "loss": 0.1813,
      "step": 1404
    },
    {
      "epoch": 0.2949202350965575,
      "grad_norm": 0.12403729557991028,
      "learning_rate": 8.887689058074994e-05,
      "loss": 0.1694,
      "step": 1405
    },
    {
      "epoch": 0.2951301427371956,
      "grad_norm": 0.12742987275123596,
      "learning_rate": 8.885383900619321e-05,
      "loss": 0.1584,
      "step": 1406
    },
    {
      "epoch": 0.29534005037783373,
      "grad_norm": 0.12318763881921768,
      "learning_rate": 8.883076656623741e-05,
      "loss": 0.1484,
      "step": 1407
    },
    {
      "epoch": 0.2955499580184719,
      "grad_norm": 0.1265798807144165,
      "learning_rate": 8.880767327327297e-05,
      "loss": 0.1708,
      "step": 1408
    },
    {
      "epoch": 0.29575986565911,
      "grad_norm": 0.12394752353429794,
      "learning_rate": 8.878455913970154e-05,
      "loss": 0.1599,
      "step": 1409
    },
    {
      "epoch": 0.2959697732997481,
      "grad_norm": 0.1458372324705124,
      "learning_rate": 8.876142417793591e-05,
      "loss": 0.1751,
      "step": 1410
    },
    {
      "epoch": 0.2961796809403862,
      "grad_norm": 0.14673961699008942,
      "learning_rate": 8.87382684004001e-05,
      "loss": 0.1651,
      "step": 1411
    },
    {
      "epoch": 0.2963895885810244,
      "grad_norm": 0.15783408284187317,
      "learning_rate": 8.871509181952925e-05,
      "loss": 0.1788,
      "step": 1412
    },
    {
      "epoch": 0.2965994962216625,
      "grad_norm": 0.1385459154844284,
      "learning_rate": 8.869189444776979e-05,
      "loss": 0.158,
      "step": 1413
    },
    {
      "epoch": 0.2968094038623006,
      "grad_norm": 0.12669824063777924,
      "learning_rate": 8.866867629757916e-05,
      "loss": 0.1556,
      "step": 1414
    },
    {
      "epoch": 0.2970193115029387,
      "grad_norm": 0.12607593834400177,
      "learning_rate": 8.86454373814261e-05,
      "loss": 0.145,
      "step": 1415
    },
    {
      "epoch": 0.2972292191435768,
      "grad_norm": 0.15276843309402466,
      "learning_rate": 8.86221777117904e-05,
      "loss": 0.1644,
      "step": 1416
    },
    {
      "epoch": 0.29743912678421497,
      "grad_norm": 0.154046893119812,
      "learning_rate": 8.859889730116304e-05,
      "loss": 0.162,
      "step": 1417
    },
    {
      "epoch": 0.2976490344248531,
      "grad_norm": 0.13313519954681396,
      "learning_rate": 8.857559616204613e-05,
      "loss": 0.1659,
      "step": 1418
    },
    {
      "epoch": 0.2978589420654912,
      "grad_norm": 0.11869316548109055,
      "learning_rate": 8.855227430695294e-05,
      "loss": 0.149,
      "step": 1419
    },
    {
      "epoch": 0.2980688497061293,
      "grad_norm": 0.13355213403701782,
      "learning_rate": 8.852893174840782e-05,
      "loss": 0.1674,
      "step": 1420
    },
    {
      "epoch": 0.2982787573467674,
      "grad_norm": 0.11961048096418381,
      "learning_rate": 8.850556849894625e-05,
      "loss": 0.1596,
      "step": 1421
    },
    {
      "epoch": 0.29848866498740556,
      "grad_norm": 0.15859396755695343,
      "learning_rate": 8.848218457111483e-05,
      "loss": 0.1688,
      "step": 1422
    },
    {
      "epoch": 0.29869857262804367,
      "grad_norm": 0.12413342297077179,
      "learning_rate": 8.845877997747127e-05,
      "loss": 0.1764,
      "step": 1423
    },
    {
      "epoch": 0.2989084802686818,
      "grad_norm": 0.1361013799905777,
      "learning_rate": 8.843535473058437e-05,
      "loss": 0.166,
      "step": 1424
    },
    {
      "epoch": 0.2991183879093199,
      "grad_norm": 0.17420732975006104,
      "learning_rate": 8.841190884303402e-05,
      "loss": 0.1699,
      "step": 1425
    },
    {
      "epoch": 0.29932829554995805,
      "grad_norm": 0.13347145915031433,
      "learning_rate": 8.838844232741122e-05,
      "loss": 0.1642,
      "step": 1426
    },
    {
      "epoch": 0.29953820319059615,
      "grad_norm": 0.1271234154701233,
      "learning_rate": 8.8364955196318e-05,
      "loss": 0.1425,
      "step": 1427
    },
    {
      "epoch": 0.29974811083123426,
      "grad_norm": 0.17551806569099426,
      "learning_rate": 8.834144746236748e-05,
      "loss": 0.1771,
      "step": 1428
    },
    {
      "epoch": 0.29995801847187237,
      "grad_norm": 0.16244301199913025,
      "learning_rate": 8.831791913818387e-05,
      "loss": 0.1598,
      "step": 1429
    },
    {
      "epoch": 0.3001679261125105,
      "grad_norm": 0.1273384541273117,
      "learning_rate": 8.829437023640239e-05,
      "loss": 0.1448,
      "step": 1430
    },
    {
      "epoch": 0.30037783375314864,
      "grad_norm": 0.10530916601419449,
      "learning_rate": 8.827080076966938e-05,
      "loss": 0.1502,
      "step": 1431
    },
    {
      "epoch": 0.30058774139378674,
      "grad_norm": 0.10858159512281418,
      "learning_rate": 8.824721075064215e-05,
      "loss": 0.1674,
      "step": 1432
    },
    {
      "epoch": 0.30079764903442485,
      "grad_norm": 0.12510398030281067,
      "learning_rate": 8.82236001919891e-05,
      "loss": 0.1739,
      "step": 1433
    },
    {
      "epoch": 0.30100755667506296,
      "grad_norm": 0.12490396946668625,
      "learning_rate": 8.819996910638962e-05,
      "loss": 0.1441,
      "step": 1434
    },
    {
      "epoch": 0.30121746431570107,
      "grad_norm": 0.13503123819828033,
      "learning_rate": 8.817631750653412e-05,
      "loss": 0.1545,
      "step": 1435
    },
    {
      "epoch": 0.30142737195633923,
      "grad_norm": 0.12778516113758087,
      "learning_rate": 8.815264540512411e-05,
      "loss": 0.1618,
      "step": 1436
    },
    {
      "epoch": 0.30163727959697734,
      "grad_norm": 0.12481486797332764,
      "learning_rate": 8.812895281487201e-05,
      "loss": 0.1541,
      "step": 1437
    },
    {
      "epoch": 0.30184718723761544,
      "grad_norm": 0.15225140750408173,
      "learning_rate": 8.810523974850131e-05,
      "loss": 0.1737,
      "step": 1438
    },
    {
      "epoch": 0.30205709487825355,
      "grad_norm": 0.12728790938854218,
      "learning_rate": 8.808150621874643e-05,
      "loss": 0.1627,
      "step": 1439
    },
    {
      "epoch": 0.3022670025188917,
      "grad_norm": 0.1369866281747818,
      "learning_rate": 8.805775223835284e-05,
      "loss": 0.1685,
      "step": 1440
    },
    {
      "epoch": 0.3024769101595298,
      "grad_norm": 0.13604697585105896,
      "learning_rate": 8.803397782007697e-05,
      "loss": 0.1623,
      "step": 1441
    },
    {
      "epoch": 0.30268681780016793,
      "grad_norm": 0.12286148965358734,
      "learning_rate": 8.801018297668622e-05,
      "loss": 0.1544,
      "step": 1442
    },
    {
      "epoch": 0.30289672544080604,
      "grad_norm": 0.12319783121347427,
      "learning_rate": 8.798636772095896e-05,
      "loss": 0.1559,
      "step": 1443
    },
    {
      "epoch": 0.30310663308144414,
      "grad_norm": 0.13532286882400513,
      "learning_rate": 8.796253206568454e-05,
      "loss": 0.1462,
      "step": 1444
    },
    {
      "epoch": 0.3033165407220823,
      "grad_norm": 0.14293217658996582,
      "learning_rate": 8.793867602366326e-05,
      "loss": 0.167,
      "step": 1445
    },
    {
      "epoch": 0.3035264483627204,
      "grad_norm": 0.1477852612733841,
      "learning_rate": 8.791479960770633e-05,
      "loss": 0.166,
      "step": 1446
    },
    {
      "epoch": 0.3037363560033585,
      "grad_norm": 0.1326388120651245,
      "learning_rate": 8.789090283063595e-05,
      "loss": 0.1737,
      "step": 1447
    },
    {
      "epoch": 0.30394626364399663,
      "grad_norm": 0.14401552081108093,
      "learning_rate": 8.786698570528522e-05,
      "loss": 0.1645,
      "step": 1448
    },
    {
      "epoch": 0.30415617128463474,
      "grad_norm": 0.13078097999095917,
      "learning_rate": 8.784304824449819e-05,
      "loss": 0.1712,
      "step": 1449
    },
    {
      "epoch": 0.3043660789252729,
      "grad_norm": 0.12756723165512085,
      "learning_rate": 8.781909046112983e-05,
      "loss": 0.1606,
      "step": 1450
    },
    {
      "epoch": 0.304575986565911,
      "grad_norm": 0.13873012363910675,
      "learning_rate": 8.779511236804603e-05,
      "loss": 0.1682,
      "step": 1451
    },
    {
      "epoch": 0.3047858942065491,
      "grad_norm": 0.12623098492622375,
      "learning_rate": 8.777111397812354e-05,
      "loss": 0.1617,
      "step": 1452
    },
    {
      "epoch": 0.3049958018471872,
      "grad_norm": 0.1419798731803894,
      "learning_rate": 8.774709530425006e-05,
      "loss": 0.1446,
      "step": 1453
    },
    {
      "epoch": 0.3052057094878254,
      "grad_norm": 0.1524171084165573,
      "learning_rate": 8.772305635932416e-05,
      "loss": 0.1606,
      "step": 1454
    },
    {
      "epoch": 0.3054156171284635,
      "grad_norm": 0.1456252783536911,
      "learning_rate": 8.769899715625533e-05,
      "loss": 0.1491,
      "step": 1455
    },
    {
      "epoch": 0.3056255247691016,
      "grad_norm": 0.16533705592155457,
      "learning_rate": 8.767491770796388e-05,
      "loss": 0.1455,
      "step": 1456
    },
    {
      "epoch": 0.3058354324097397,
      "grad_norm": 0.1010533794760704,
      "learning_rate": 8.765081802738104e-05,
      "loss": 0.1599,
      "step": 1457
    },
    {
      "epoch": 0.3060453400503778,
      "grad_norm": 0.1445004791021347,
      "learning_rate": 8.762669812744887e-05,
      "loss": 0.1788,
      "step": 1458
    },
    {
      "epoch": 0.306255247691016,
      "grad_norm": 0.1463691145181656,
      "learning_rate": 8.760255802112032e-05,
      "loss": 0.1554,
      "step": 1459
    },
    {
      "epoch": 0.3064651553316541,
      "grad_norm": 0.14523173868656158,
      "learning_rate": 8.757839772135919e-05,
      "loss": 0.1768,
      "step": 1460
    },
    {
      "epoch": 0.3066750629722922,
      "grad_norm": 0.12887312471866608,
      "learning_rate": 8.75542172411401e-05,
      "loss": 0.1385,
      "step": 1461
    },
    {
      "epoch": 0.3068849706129303,
      "grad_norm": 0.12244197726249695,
      "learning_rate": 8.753001659344852e-05,
      "loss": 0.177,
      "step": 1462
    },
    {
      "epoch": 0.3070948782535684,
      "grad_norm": 0.1390761286020279,
      "learning_rate": 8.750579579128077e-05,
      "loss": 0.152,
      "step": 1463
    },
    {
      "epoch": 0.30730478589420657,
      "grad_norm": 0.1180199533700943,
      "learning_rate": 8.748155484764393e-05,
      "loss": 0.1822,
      "step": 1464
    },
    {
      "epoch": 0.3075146935348447,
      "grad_norm": 0.12501199543476105,
      "learning_rate": 8.745729377555598e-05,
      "loss": 0.1585,
      "step": 1465
    },
    {
      "epoch": 0.3077246011754828,
      "grad_norm": 0.1500568687915802,
      "learning_rate": 8.743301258804567e-05,
      "loss": 0.1571,
      "step": 1466
    },
    {
      "epoch": 0.3079345088161209,
      "grad_norm": 0.10966819524765015,
      "learning_rate": 8.740871129815253e-05,
      "loss": 0.1654,
      "step": 1467
    },
    {
      "epoch": 0.30814441645675905,
      "grad_norm": 0.14994241297245026,
      "learning_rate": 8.73843899189269e-05,
      "loss": 0.156,
      "step": 1468
    },
    {
      "epoch": 0.30835432409739716,
      "grad_norm": 0.13417619466781616,
      "learning_rate": 8.736004846342996e-05,
      "loss": 0.16,
      "step": 1469
    },
    {
      "epoch": 0.30856423173803527,
      "grad_norm": 0.12869034707546234,
      "learning_rate": 8.733568694473359e-05,
      "loss": 0.1526,
      "step": 1470
    },
    {
      "epoch": 0.3087741393786734,
      "grad_norm": 0.14110779762268066,
      "learning_rate": 8.731130537592048e-05,
      "loss": 0.1677,
      "step": 1471
    },
    {
      "epoch": 0.3089840470193115,
      "grad_norm": 0.11759759485721588,
      "learning_rate": 8.728690377008412e-05,
      "loss": 0.1583,
      "step": 1472
    },
    {
      "epoch": 0.30919395465994964,
      "grad_norm": 0.13631226122379303,
      "learning_rate": 8.726248214032871e-05,
      "loss": 0.151,
      "step": 1473
    },
    {
      "epoch": 0.30940386230058775,
      "grad_norm": 0.13227972388267517,
      "learning_rate": 8.723804049976922e-05,
      "loss": 0.1616,
      "step": 1474
    },
    {
      "epoch": 0.30961376994122586,
      "grad_norm": 0.15408119559288025,
      "learning_rate": 8.721357886153137e-05,
      "loss": 0.1731,
      "step": 1475
    },
    {
      "epoch": 0.30982367758186397,
      "grad_norm": 0.14567722380161285,
      "learning_rate": 8.718909723875163e-05,
      "loss": 0.1555,
      "step": 1476
    },
    {
      "epoch": 0.3100335852225021,
      "grad_norm": 0.1292102187871933,
      "learning_rate": 8.71645956445772e-05,
      "loss": 0.1566,
      "step": 1477
    },
    {
      "epoch": 0.31024349286314024,
      "grad_norm": 0.12155589461326599,
      "learning_rate": 8.714007409216595e-05,
      "loss": 0.168,
      "step": 1478
    },
    {
      "epoch": 0.31045340050377834,
      "grad_norm": 0.1343718022108078,
      "learning_rate": 8.711553259468657e-05,
      "loss": 0.1673,
      "step": 1479
    },
    {
      "epoch": 0.31066330814441645,
      "grad_norm": 0.11374703049659729,
      "learning_rate": 8.709097116531835e-05,
      "loss": 0.1655,
      "step": 1480
    },
    {
      "epoch": 0.31087321578505456,
      "grad_norm": 0.13738465309143066,
      "learning_rate": 8.706638981725139e-05,
      "loss": 0.1573,
      "step": 1481
    },
    {
      "epoch": 0.3110831234256927,
      "grad_norm": 0.13506565988063812,
      "learning_rate": 8.70417885636864e-05,
      "loss": 0.1649,
      "step": 1482
    },
    {
      "epoch": 0.3112930310663308,
      "grad_norm": 0.1244024857878685,
      "learning_rate": 8.701716741783483e-05,
      "loss": 0.1605,
      "step": 1483
    },
    {
      "epoch": 0.31150293870696893,
      "grad_norm": 0.15039315819740295,
      "learning_rate": 8.69925263929188e-05,
      "loss": 0.1634,
      "step": 1484
    },
    {
      "epoch": 0.31171284634760704,
      "grad_norm": 0.12742501497268677,
      "learning_rate": 8.69678655021711e-05,
      "loss": 0.1649,
      "step": 1485
    },
    {
      "epoch": 0.31192275398824515,
      "grad_norm": 0.1437070369720459,
      "learning_rate": 8.694318475883518e-05,
      "loss": 0.1565,
      "step": 1486
    },
    {
      "epoch": 0.3121326616288833,
      "grad_norm": 0.13695649802684784,
      "learning_rate": 8.691848417616517e-05,
      "loss": 0.1628,
      "step": 1487
    },
    {
      "epoch": 0.3123425692695214,
      "grad_norm": 0.12154117226600647,
      "learning_rate": 8.689376376742586e-05,
      "loss": 0.159,
      "step": 1488
    },
    {
      "epoch": 0.3125524769101595,
      "grad_norm": 0.12522564828395844,
      "learning_rate": 8.686902354589266e-05,
      "loss": 0.1567,
      "step": 1489
    },
    {
      "epoch": 0.31276238455079763,
      "grad_norm": 0.14949142932891846,
      "learning_rate": 8.684426352485165e-05,
      "loss": 0.1602,
      "step": 1490
    },
    {
      "epoch": 0.31297229219143574,
      "grad_norm": 0.13096019625663757,
      "learning_rate": 8.68194837175995e-05,
      "loss": 0.157,
      "step": 1491
    },
    {
      "epoch": 0.3131821998320739,
      "grad_norm": 0.12589357793331146,
      "learning_rate": 8.679468413744356e-05,
      "loss": 0.1711,
      "step": 1492
    },
    {
      "epoch": 0.313392107472712,
      "grad_norm": 0.1338101178407669,
      "learning_rate": 8.676986479770175e-05,
      "loss": 0.1628,
      "step": 1493
    },
    {
      "epoch": 0.3136020151133501,
      "grad_norm": 0.13691024482250214,
      "learning_rate": 8.674502571170262e-05,
      "loss": 0.1604,
      "step": 1494
    },
    {
      "epoch": 0.3138119227539882,
      "grad_norm": 0.14150124788284302,
      "learning_rate": 8.672016689278535e-05,
      "loss": 0.1707,
      "step": 1495
    },
    {
      "epoch": 0.3140218303946264,
      "grad_norm": 0.13233508169651031,
      "learning_rate": 8.669528835429969e-05,
      "loss": 0.1421,
      "step": 1496
    },
    {
      "epoch": 0.3142317380352645,
      "grad_norm": 0.11874301731586456,
      "learning_rate": 8.667039010960596e-05,
      "loss": 0.1576,
      "step": 1497
    },
    {
      "epoch": 0.3144416456759026,
      "grad_norm": 0.10090667009353638,
      "learning_rate": 8.66454721720751e-05,
      "loss": 0.158,
      "step": 1498
    },
    {
      "epoch": 0.3146515533165407,
      "grad_norm": 0.126051664352417,
      "learning_rate": 8.662053455508862e-05,
      "loss": 0.1725,
      "step": 1499
    },
    {
      "epoch": 0.3148614609571788,
      "grad_norm": 0.11627226322889328,
      "learning_rate": 8.659557727203858e-05,
      "loss": 0.1572,
      "step": 1500
    },
    {
      "epoch": 0.315071368597817,
      "grad_norm": 0.12283121049404144,
      "learning_rate": 8.657060033632763e-05,
      "loss": 0.1633,
      "step": 1501
    },
    {
      "epoch": 0.3152812762384551,
      "grad_norm": 0.11624766141176224,
      "learning_rate": 8.654560376136891e-05,
      "loss": 0.1762,
      "step": 1502
    },
    {
      "epoch": 0.3154911838790932,
      "grad_norm": 0.13653838634490967,
      "learning_rate": 8.65205875605862e-05,
      "loss": 0.1784,
      "step": 1503
    },
    {
      "epoch": 0.3157010915197313,
      "grad_norm": 0.12123393267393112,
      "learning_rate": 8.649555174741374e-05,
      "loss": 0.1551,
      "step": 1504
    },
    {
      "epoch": 0.3159109991603694,
      "grad_norm": 0.12023040652275085,
      "learning_rate": 8.647049633529635e-05,
      "loss": 0.174,
      "step": 1505
    },
    {
      "epoch": 0.3161209068010076,
      "grad_norm": 0.12046807259321213,
      "learning_rate": 8.644542133768936e-05,
      "loss": 0.1703,
      "step": 1506
    },
    {
      "epoch": 0.3163308144416457,
      "grad_norm": 0.11224795132875443,
      "learning_rate": 8.64203267680586e-05,
      "loss": 0.1435,
      "step": 1507
    },
    {
      "epoch": 0.3165407220822838,
      "grad_norm": 0.1572083830833435,
      "learning_rate": 8.639521263988044e-05,
      "loss": 0.153,
      "step": 1508
    },
    {
      "epoch": 0.3167506297229219,
      "grad_norm": 0.1247357502579689,
      "learning_rate": 8.637007896664173e-05,
      "loss": 0.1566,
      "step": 1509
    },
    {
      "epoch": 0.31696053736356006,
      "grad_norm": 0.12798605859279633,
      "learning_rate": 8.634492576183984e-05,
      "loss": 0.1526,
      "step": 1510
    },
    {
      "epoch": 0.31717044500419816,
      "grad_norm": 0.1474895477294922,
      "learning_rate": 8.631975303898261e-05,
      "loss": 0.169,
      "step": 1511
    },
    {
      "epoch": 0.31738035264483627,
      "grad_norm": 0.11636695265769958,
      "learning_rate": 8.629456081158838e-05,
      "loss": 0.1613,
      "step": 1512
    },
    {
      "epoch": 0.3175902602854744,
      "grad_norm": 0.11031098663806915,
      "learning_rate": 8.626934909318591e-05,
      "loss": 0.1576,
      "step": 1513
    },
    {
      "epoch": 0.3178001679261125,
      "grad_norm": 0.12879802286624908,
      "learning_rate": 8.624411789731452e-05,
      "loss": 0.1556,
      "step": 1514
    },
    {
      "epoch": 0.31801007556675065,
      "grad_norm": 0.1240730881690979,
      "learning_rate": 8.621886723752392e-05,
      "loss": 0.1633,
      "step": 1515
    },
    {
      "epoch": 0.31821998320738876,
      "grad_norm": 0.10870420932769775,
      "learning_rate": 8.619359712737427e-05,
      "loss": 0.1385,
      "step": 1516
    },
    {
      "epoch": 0.31842989084802686,
      "grad_norm": 0.11011097580194473,
      "learning_rate": 8.616830758043622e-05,
      "loss": 0.1655,
      "step": 1517
    },
    {
      "epoch": 0.31863979848866497,
      "grad_norm": 0.17963182926177979,
      "learning_rate": 8.614299861029083e-05,
      "loss": 0.1651,
      "step": 1518
    },
    {
      "epoch": 0.31884970612930313,
      "grad_norm": 0.12011148035526276,
      "learning_rate": 8.61176702305296e-05,
      "loss": 0.1657,
      "step": 1519
    },
    {
      "epoch": 0.31905961376994124,
      "grad_norm": 0.1365176886320114,
      "learning_rate": 8.609232245475443e-05,
      "loss": 0.1839,
      "step": 1520
    },
    {
      "epoch": 0.31926952141057935,
      "grad_norm": 0.1179775819182396,
      "learning_rate": 8.606695529657767e-05,
      "loss": 0.1581,
      "step": 1521
    },
    {
      "epoch": 0.31947942905121746,
      "grad_norm": 0.15118494629859924,
      "learning_rate": 8.604156876962206e-05,
      "loss": 0.1831,
      "step": 1522
    },
    {
      "epoch": 0.31968933669185556,
      "grad_norm": 0.14029067754745483,
      "learning_rate": 8.601616288752073e-05,
      "loss": 0.1708,
      "step": 1523
    },
    {
      "epoch": 0.3198992443324937,
      "grad_norm": 0.14659684896469116,
      "learning_rate": 8.599073766391725e-05,
      "loss": 0.1599,
      "step": 1524
    },
    {
      "epoch": 0.32010915197313183,
      "grad_norm": 0.1260727345943451,
      "learning_rate": 8.596529311246551e-05,
      "loss": 0.1623,
      "step": 1525
    },
    {
      "epoch": 0.32031905961376994,
      "grad_norm": 0.12665653228759766,
      "learning_rate": 8.593982924682986e-05,
      "loss": 0.1672,
      "step": 1526
    },
    {
      "epoch": 0.32052896725440805,
      "grad_norm": 0.12388867139816284,
      "learning_rate": 8.591434608068493e-05,
      "loss": 0.1738,
      "step": 1527
    },
    {
      "epoch": 0.32073887489504616,
      "grad_norm": 0.12167280912399292,
      "learning_rate": 8.58888436277158e-05,
      "loss": 0.1627,
      "step": 1528
    },
    {
      "epoch": 0.3209487825356843,
      "grad_norm": 0.12876589596271515,
      "learning_rate": 8.586332190161786e-05,
      "loss": 0.1646,
      "step": 1529
    },
    {
      "epoch": 0.3211586901763224,
      "grad_norm": 0.12370327860116959,
      "learning_rate": 8.583778091609683e-05,
      "loss": 0.156,
      "step": 1530
    },
    {
      "epoch": 0.32136859781696053,
      "grad_norm": 0.11784835159778595,
      "learning_rate": 8.581222068486884e-05,
      "loss": 0.1664,
      "step": 1531
    },
    {
      "epoch": 0.32157850545759864,
      "grad_norm": 0.14363597333431244,
      "learning_rate": 8.578664122166029e-05,
      "loss": 0.1724,
      "step": 1532
    },
    {
      "epoch": 0.3217884130982368,
      "grad_norm": 0.11469676345586777,
      "learning_rate": 8.576104254020796e-05,
      "loss": 0.1469,
      "step": 1533
    },
    {
      "epoch": 0.3219983207388749,
      "grad_norm": 0.12534739077091217,
      "learning_rate": 8.573542465425892e-05,
      "loss": 0.1687,
      "step": 1534
    },
    {
      "epoch": 0.322208228379513,
      "grad_norm": 0.12772725522518158,
      "learning_rate": 8.570978757757057e-05,
      "loss": 0.159,
      "step": 1535
    },
    {
      "epoch": 0.3224181360201511,
      "grad_norm": 0.14892329275608063,
      "learning_rate": 8.568413132391058e-05,
      "loss": 0.1586,
      "step": 1536
    },
    {
      "epoch": 0.32262804366078923,
      "grad_norm": 0.1334220916032791,
      "learning_rate": 8.565845590705695e-05,
      "loss": 0.1692,
      "step": 1537
    },
    {
      "epoch": 0.3228379513014274,
      "grad_norm": 0.1165885403752327,
      "learning_rate": 8.563276134079798e-05,
      "loss": 0.1673,
      "step": 1538
    },
    {
      "epoch": 0.3230478589420655,
      "grad_norm": 0.11613230407238007,
      "learning_rate": 8.560704763893224e-05,
      "loss": 0.1512,
      "step": 1539
    },
    {
      "epoch": 0.3232577665827036,
      "grad_norm": 0.11873317509889603,
      "learning_rate": 8.558131481526856e-05,
      "loss": 0.1556,
      "step": 1540
    },
    {
      "epoch": 0.3234676742233417,
      "grad_norm": 0.12491460889577866,
      "learning_rate": 8.555556288362605e-05,
      "loss": 0.1728,
      "step": 1541
    },
    {
      "epoch": 0.3236775818639798,
      "grad_norm": 0.1342668980360031,
      "learning_rate": 8.552979185783412e-05,
      "loss": 0.1746,
      "step": 1542
    },
    {
      "epoch": 0.323887489504618,
      "grad_norm": 0.1082087978720665,
      "learning_rate": 8.550400175173236e-05,
      "loss": 0.1668,
      "step": 1543
    },
    {
      "epoch": 0.3240973971452561,
      "grad_norm": 0.12464690953493118,
      "learning_rate": 8.547819257917065e-05,
      "loss": 0.1642,
      "step": 1544
    },
    {
      "epoch": 0.3243073047858942,
      "grad_norm": 0.11872904002666473,
      "learning_rate": 8.545236435400914e-05,
      "loss": 0.1586,
      "step": 1545
    },
    {
      "epoch": 0.3245172124265323,
      "grad_norm": 0.14073589444160461,
      "learning_rate": 8.542651709011814e-05,
      "loss": 0.1733,
      "step": 1546
    },
    {
      "epoch": 0.32472712006717047,
      "grad_norm": 0.11652261763811111,
      "learning_rate": 8.540065080137824e-05,
      "loss": 0.1689,
      "step": 1547
    },
    {
      "epoch": 0.3249370277078086,
      "grad_norm": 0.14177857339382172,
      "learning_rate": 8.537476550168022e-05,
      "loss": 0.1504,
      "step": 1548
    },
    {
      "epoch": 0.3251469353484467,
      "grad_norm": 0.11179566383361816,
      "learning_rate": 8.534886120492509e-05,
      "loss": 0.1504,
      "step": 1549
    },
    {
      "epoch": 0.3253568429890848,
      "grad_norm": 0.141026109457016,
      "learning_rate": 8.532293792502403e-05,
      "loss": 0.1642,
      "step": 1550
    },
    {
      "epoch": 0.3255667506297229,
      "grad_norm": 0.11241170763969421,
      "learning_rate": 8.529699567589844e-05,
      "loss": 0.1662,
      "step": 1551
    },
    {
      "epoch": 0.32577665827036106,
      "grad_norm": 0.11532200872898102,
      "learning_rate": 8.52710344714799e-05,
      "loss": 0.1588,
      "step": 1552
    },
    {
      "epoch": 0.32598656591099917,
      "grad_norm": 0.13776087760925293,
      "learning_rate": 8.524505432571019e-05,
      "loss": 0.1654,
      "step": 1553
    },
    {
      "epoch": 0.3261964735516373,
      "grad_norm": 0.11376765370368958,
      "learning_rate": 8.521905525254123e-05,
      "loss": 0.1536,
      "step": 1554
    },
    {
      "epoch": 0.3264063811922754,
      "grad_norm": 0.11074686050415039,
      "learning_rate": 8.519303726593508e-05,
      "loss": 0.1589,
      "step": 1555
    },
    {
      "epoch": 0.3266162888329135,
      "grad_norm": 0.11213414371013641,
      "learning_rate": 8.516700037986406e-05,
      "loss": 0.1548,
      "step": 1556
    },
    {
      "epoch": 0.32682619647355166,
      "grad_norm": 0.14128370583057404,
      "learning_rate": 8.514094460831052e-05,
      "loss": 0.1556,
      "step": 1557
    },
    {
      "epoch": 0.32703610411418976,
      "grad_norm": 0.1197216585278511,
      "learning_rate": 8.5114869965267e-05,
      "loss": 0.146,
      "step": 1558
    },
    {
      "epoch": 0.32724601175482787,
      "grad_norm": 0.12456157803535461,
      "learning_rate": 8.508877646473623e-05,
      "loss": 0.1586,
      "step": 1559
    },
    {
      "epoch": 0.327455919395466,
      "grad_norm": 0.11759025603532791,
      "learning_rate": 8.506266412073099e-05,
      "loss": 0.1564,
      "step": 1560
    },
    {
      "epoch": 0.32766582703610414,
      "grad_norm": 0.1229950562119484,
      "learning_rate": 8.503653294727418e-05,
      "loss": 0.159,
      "step": 1561
    },
    {
      "epoch": 0.32787573467674225,
      "grad_norm": 0.11062034219503403,
      "learning_rate": 8.501038295839887e-05,
      "loss": 0.1536,
      "step": 1562
    },
    {
      "epoch": 0.32808564231738035,
      "grad_norm": 0.1223401203751564,
      "learning_rate": 8.498421416814818e-05,
      "loss": 0.1473,
      "step": 1563
    },
    {
      "epoch": 0.32829554995801846,
      "grad_norm": 0.13364095985889435,
      "learning_rate": 8.495802659057536e-05,
      "loss": 0.169,
      "step": 1564
    },
    {
      "epoch": 0.32850545759865657,
      "grad_norm": 0.1344279795885086,
      "learning_rate": 8.493182023974372e-05,
      "loss": 0.1537,
      "step": 1565
    },
    {
      "epoch": 0.32871536523929473,
      "grad_norm": 0.17328642308712006,
      "learning_rate": 8.490559512972671e-05,
      "loss": 0.1476,
      "step": 1566
    },
    {
      "epoch": 0.32892527287993284,
      "grad_norm": 0.1484242081642151,
      "learning_rate": 8.487935127460779e-05,
      "loss": 0.1666,
      "step": 1567
    },
    {
      "epoch": 0.32913518052057095,
      "grad_norm": 0.17023669183254242,
      "learning_rate": 8.485308868848049e-05,
      "loss": 0.1653,
      "step": 1568
    },
    {
      "epoch": 0.32934508816120905,
      "grad_norm": 0.14054429531097412,
      "learning_rate": 8.482680738544843e-05,
      "loss": 0.1675,
      "step": 1569
    },
    {
      "epoch": 0.32955499580184716,
      "grad_norm": 0.1140383705496788,
      "learning_rate": 8.480050737962531e-05,
      "loss": 0.1612,
      "step": 1570
    },
    {
      "epoch": 0.3297649034424853,
      "grad_norm": 0.13256023824214935,
      "learning_rate": 8.477418868513476e-05,
      "loss": 0.1595,
      "step": 1571
    },
    {
      "epoch": 0.32997481108312343,
      "grad_norm": 0.13051024079322815,
      "learning_rate": 8.474785131611058e-05,
      "loss": 0.1621,
      "step": 1572
    },
    {
      "epoch": 0.33018471872376154,
      "grad_norm": 0.11545020341873169,
      "learning_rate": 8.472149528669651e-05,
      "loss": 0.1683,
      "step": 1573
    },
    {
      "epoch": 0.33039462636439965,
      "grad_norm": 0.13742703199386597,
      "learning_rate": 8.469512061104635e-05,
      "loss": 0.1611,
      "step": 1574
    },
    {
      "epoch": 0.3306045340050378,
      "grad_norm": 0.13308970630168915,
      "learning_rate": 8.46687273033239e-05,
      "loss": 0.148,
      "step": 1575
    },
    {
      "epoch": 0.3308144416456759,
      "grad_norm": 0.13026942312717438,
      "learning_rate": 8.464231537770298e-05,
      "loss": 0.1772,
      "step": 1576
    },
    {
      "epoch": 0.331024349286314,
      "grad_norm": 0.11242230236530304,
      "learning_rate": 8.461588484836738e-05,
      "loss": 0.1632,
      "step": 1577
    },
    {
      "epoch": 0.33123425692695213,
      "grad_norm": 0.13627904653549194,
      "learning_rate": 8.45894357295109e-05,
      "loss": 0.1644,
      "step": 1578
    },
    {
      "epoch": 0.33144416456759024,
      "grad_norm": 0.11995008587837219,
      "learning_rate": 8.456296803533734e-05,
      "loss": 0.1532,
      "step": 1579
    },
    {
      "epoch": 0.3316540722082284,
      "grad_norm": 0.12165706604719162,
      "learning_rate": 8.453648178006044e-05,
      "loss": 0.1596,
      "step": 1580
    },
    {
      "epoch": 0.3318639798488665,
      "grad_norm": 0.1462729126214981,
      "learning_rate": 8.450997697790393e-05,
      "loss": 0.1572,
      "step": 1581
    },
    {
      "epoch": 0.3320738874895046,
      "grad_norm": 0.11989284306764603,
      "learning_rate": 8.448345364310149e-05,
      "loss": 0.1631,
      "step": 1582
    },
    {
      "epoch": 0.3322837951301427,
      "grad_norm": 0.1307172328233719,
      "learning_rate": 8.445691178989674e-05,
      "loss": 0.1639,
      "step": 1583
    },
    {
      "epoch": 0.33249370277078083,
      "grad_norm": 0.13907490670681,
      "learning_rate": 8.44303514325433e-05,
      "loss": 0.1466,
      "step": 1584
    },
    {
      "epoch": 0.332703610411419,
      "grad_norm": 0.21999351680278778,
      "learning_rate": 8.440377258530465e-05,
      "loss": 0.1596,
      "step": 1585
    },
    {
      "epoch": 0.3329135180520571,
      "grad_norm": 0.11224652826786041,
      "learning_rate": 8.437717526245428e-05,
      "loss": 0.1543,
      "step": 1586
    },
    {
      "epoch": 0.3331234256926952,
      "grad_norm": 0.12237223982810974,
      "learning_rate": 8.435055947827552e-05,
      "loss": 0.149,
      "step": 1587
    },
    {
      "epoch": 0.3333333333333333,
      "grad_norm": 0.1236414983868599,
      "learning_rate": 8.432392524706168e-05,
      "loss": 0.1421,
      "step": 1588
    },
    {
      "epoch": 0.3335432409739715,
      "grad_norm": 0.12933699786663055,
      "learning_rate": 8.429727258311593e-05,
      "loss": 0.152,
      "step": 1589
    },
    {
      "epoch": 0.3337531486146096,
      "grad_norm": 0.14159703254699707,
      "learning_rate": 8.427060150075137e-05,
      "loss": 0.1705,
      "step": 1590
    },
    {
      "epoch": 0.3339630562552477,
      "grad_norm": 0.1670205146074295,
      "learning_rate": 8.424391201429099e-05,
      "loss": 0.1544,
      "step": 1591
    },
    {
      "epoch": 0.3341729638958858,
      "grad_norm": 0.11887665838003159,
      "learning_rate": 8.421720413806764e-05,
      "loss": 0.147,
      "step": 1592
    },
    {
      "epoch": 0.3343828715365239,
      "grad_norm": 0.11611668765544891,
      "learning_rate": 8.419047788642407e-05,
      "loss": 0.1758,
      "step": 1593
    },
    {
      "epoch": 0.33459277917716207,
      "grad_norm": 0.15789294242858887,
      "learning_rate": 8.416373327371287e-05,
      "loss": 0.1533,
      "step": 1594
    },
    {
      "epoch": 0.3348026868178002,
      "grad_norm": 0.13557223975658417,
      "learning_rate": 8.413697031429653e-05,
      "loss": 0.1588,
      "step": 1595
    },
    {
      "epoch": 0.3350125944584383,
      "grad_norm": 0.29409486055374146,
      "learning_rate": 8.411018902254736e-05,
      "loss": 0.1686,
      "step": 1596
    },
    {
      "epoch": 0.3352225020990764,
      "grad_norm": 0.13911570608615875,
      "learning_rate": 8.408338941284752e-05,
      "loss": 0.168,
      "step": 1597
    },
    {
      "epoch": 0.3354324097397145,
      "grad_norm": 0.1477447748184204,
      "learning_rate": 8.405657149958902e-05,
      "loss": 0.1611,
      "step": 1598
    },
    {
      "epoch": 0.33564231738035266,
      "grad_norm": 0.13031038641929626,
      "learning_rate": 8.402973529717368e-05,
      "loss": 0.1676,
      "step": 1599
    },
    {
      "epoch": 0.33585222502099077,
      "grad_norm": 0.12024478614330292,
      "learning_rate": 8.400288082001319e-05,
      "loss": 0.1567,
      "step": 1600
    },
    {
      "epoch": 0.3360621326616289,
      "grad_norm": 0.12338093668222427,
      "learning_rate": 8.397600808252897e-05,
      "loss": 0.1585,
      "step": 1601
    },
    {
      "epoch": 0.336272040302267,
      "grad_norm": 0.12878558039665222,
      "learning_rate": 8.394911709915232e-05,
      "loss": 0.1624,
      "step": 1602
    },
    {
      "epoch": 0.33648194794290515,
      "grad_norm": 0.12042135745286942,
      "learning_rate": 8.392220788432431e-05,
      "loss": 0.1719,
      "step": 1603
    },
    {
      "epoch": 0.33669185558354325,
      "grad_norm": 0.14576420187950134,
      "learning_rate": 8.389528045249579e-05,
      "loss": 0.1669,
      "step": 1604
    },
    {
      "epoch": 0.33690176322418136,
      "grad_norm": 0.13084185123443604,
      "learning_rate": 8.386833481812744e-05,
      "loss": 0.1556,
      "step": 1605
    },
    {
      "epoch": 0.33711167086481947,
      "grad_norm": 0.15233568847179413,
      "learning_rate": 8.384137099568965e-05,
      "loss": 0.1674,
      "step": 1606
    },
    {
      "epoch": 0.3373215785054576,
      "grad_norm": 0.15160681307315826,
      "learning_rate": 8.381438899966261e-05,
      "loss": 0.1727,
      "step": 1607
    },
    {
      "epoch": 0.33753148614609574,
      "grad_norm": 0.13782384991645813,
      "learning_rate": 8.378738884453627e-05,
      "loss": 0.1754,
      "step": 1608
    },
    {
      "epoch": 0.33774139378673385,
      "grad_norm": 0.12946204841136932,
      "learning_rate": 8.376037054481034e-05,
      "loss": 0.176,
      "step": 1609
    },
    {
      "epoch": 0.33795130142737195,
      "grad_norm": 0.11807133257389069,
      "learning_rate": 8.373333411499426e-05,
      "loss": 0.1465,
      "step": 1610
    },
    {
      "epoch": 0.33816120906801006,
      "grad_norm": 0.1325267106294632,
      "learning_rate": 8.370627956960721e-05,
      "loss": 0.1475,
      "step": 1611
    },
    {
      "epoch": 0.33837111670864817,
      "grad_norm": 0.12463680654764175,
      "learning_rate": 8.36792069231781e-05,
      "loss": 0.1691,
      "step": 1612
    },
    {
      "epoch": 0.33858102434928633,
      "grad_norm": 0.1311996728181839,
      "learning_rate": 8.365211619024555e-05,
      "loss": 0.1562,
      "step": 1613
    },
    {
      "epoch": 0.33879093198992444,
      "grad_norm": 0.13213030993938446,
      "learning_rate": 8.362500738535792e-05,
      "loss": 0.1611,
      "step": 1614
    },
    {
      "epoch": 0.33900083963056254,
      "grad_norm": 0.1333753913640976,
      "learning_rate": 8.359788052307324e-05,
      "loss": 0.1612,
      "step": 1615
    },
    {
      "epoch": 0.33921074727120065,
      "grad_norm": 0.11926085501909256,
      "learning_rate": 8.357073561795928e-05,
      "loss": 0.1621,
      "step": 1616
    },
    {
      "epoch": 0.3394206549118388,
      "grad_norm": 0.14635923504829407,
      "learning_rate": 8.354357268459343e-05,
      "loss": 0.1716,
      "step": 1617
    },
    {
      "epoch": 0.3396305625524769,
      "grad_norm": 0.1318664848804474,
      "learning_rate": 8.351639173756284e-05,
      "loss": 0.171,
      "step": 1618
    },
    {
      "epoch": 0.33984047019311503,
      "grad_norm": 0.13774079084396362,
      "learning_rate": 8.34891927914643e-05,
      "loss": 0.1527,
      "step": 1619
    },
    {
      "epoch": 0.34005037783375314,
      "grad_norm": 0.1164809837937355,
      "learning_rate": 8.346197586090426e-05,
      "loss": 0.161,
      "step": 1620
    },
    {
      "epoch": 0.34026028547439124,
      "grad_norm": 0.13626568019390106,
      "learning_rate": 8.343474096049881e-05,
      "loss": 0.1573,
      "step": 1621
    },
    {
      "epoch": 0.3404701931150294,
      "grad_norm": 0.10750927031040192,
      "learning_rate": 8.340748810487375e-05,
      "loss": 0.1516,
      "step": 1622
    },
    {
      "epoch": 0.3406801007556675,
      "grad_norm": 0.1209016814827919,
      "learning_rate": 8.338021730866448e-05,
      "loss": 0.1604,
      "step": 1623
    },
    {
      "epoch": 0.3408900083963056,
      "grad_norm": 0.12268110364675522,
      "learning_rate": 8.335292858651601e-05,
      "loss": 0.1723,
      "step": 1624
    },
    {
      "epoch": 0.34109991603694373,
      "grad_norm": 0.1716935932636261,
      "learning_rate": 8.332562195308305e-05,
      "loss": 0.1617,
      "step": 1625
    },
    {
      "epoch": 0.3413098236775819,
      "grad_norm": 0.12799715995788574,
      "learning_rate": 8.329829742302984e-05,
      "loss": 0.1677,
      "step": 1626
    },
    {
      "epoch": 0.34151973131822,
      "grad_norm": 0.12748028337955475,
      "learning_rate": 8.327095501103029e-05,
      "loss": 0.1587,
      "step": 1627
    },
    {
      "epoch": 0.3417296389588581,
      "grad_norm": 0.15869416296482086,
      "learning_rate": 8.324359473176793e-05,
      "loss": 0.1644,
      "step": 1628
    },
    {
      "epoch": 0.3419395465994962,
      "grad_norm": 0.14217017590999603,
      "learning_rate": 8.321621659993583e-05,
      "loss": 0.1688,
      "step": 1629
    },
    {
      "epoch": 0.3421494542401343,
      "grad_norm": 0.11858684569597244,
      "learning_rate": 8.318882063023669e-05,
      "loss": 0.1513,
      "step": 1630
    },
    {
      "epoch": 0.3423593618807725,
      "grad_norm": 0.11514747142791748,
      "learning_rate": 8.316140683738274e-05,
      "loss": 0.146,
      "step": 1631
    },
    {
      "epoch": 0.3425692695214106,
      "grad_norm": 0.13995614647865295,
      "learning_rate": 8.313397523609585e-05,
      "loss": 0.1567,
      "step": 1632
    },
    {
      "epoch": 0.3427791771620487,
      "grad_norm": 0.10559704899787903,
      "learning_rate": 8.310652584110738e-05,
      "loss": 0.1577,
      "step": 1633
    },
    {
      "epoch": 0.3429890848026868,
      "grad_norm": 0.11493990570306778,
      "learning_rate": 8.30790586671583e-05,
      "loss": 0.1431,
      "step": 1634
    },
    {
      "epoch": 0.3431989924433249,
      "grad_norm": 0.14548490941524506,
      "learning_rate": 8.305157372899913e-05,
      "loss": 0.1791,
      "step": 1635
    },
    {
      "epoch": 0.3434089000839631,
      "grad_norm": 0.13234320282936096,
      "learning_rate": 8.302407104138988e-05,
      "loss": 0.1568,
      "step": 1636
    },
    {
      "epoch": 0.3436188077246012,
      "grad_norm": 0.1269916445016861,
      "learning_rate": 8.299655061910012e-05,
      "loss": 0.1494,
      "step": 1637
    },
    {
      "epoch": 0.3438287153652393,
      "grad_norm": 0.12883053719997406,
      "learning_rate": 8.296901247690896e-05,
      "loss": 0.1633,
      "step": 1638
    },
    {
      "epoch": 0.3440386230058774,
      "grad_norm": 0.12479805201292038,
      "learning_rate": 8.294145662960502e-05,
      "loss": 0.1744,
      "step": 1639
    },
    {
      "epoch": 0.34424853064651556,
      "grad_norm": 0.13455580174922943,
      "learning_rate": 8.29138830919864e-05,
      "loss": 0.1787,
      "step": 1640
    },
    {
      "epoch": 0.34445843828715367,
      "grad_norm": 0.11923965066671371,
      "learning_rate": 8.288629187886073e-05,
      "loss": 0.1703,
      "step": 1641
    },
    {
      "epoch": 0.3446683459277918,
      "grad_norm": 0.1303246170282364,
      "learning_rate": 8.285868300504511e-05,
      "loss": 0.1669,
      "step": 1642
    },
    {
      "epoch": 0.3448782535684299,
      "grad_norm": 0.09619605541229248,
      "learning_rate": 8.283105648536616e-05,
      "loss": 0.163,
      "step": 1643
    },
    {
      "epoch": 0.345088161209068,
      "grad_norm": 0.11518269032239914,
      "learning_rate": 8.280341233465992e-05,
      "loss": 0.1511,
      "step": 1644
    },
    {
      "epoch": 0.34529806884970615,
      "grad_norm": 0.10729756206274033,
      "learning_rate": 8.277575056777195e-05,
      "loss": 0.1671,
      "step": 1645
    },
    {
      "epoch": 0.34550797649034426,
      "grad_norm": 0.13708291947841644,
      "learning_rate": 8.274807119955725e-05,
      "loss": 0.1719,
      "step": 1646
    },
    {
      "epoch": 0.34571788413098237,
      "grad_norm": 0.11630459129810333,
      "learning_rate": 8.272037424488028e-05,
      "loss": 0.1505,
      "step": 1647
    },
    {
      "epoch": 0.3459277917716205,
      "grad_norm": 0.15268699824810028,
      "learning_rate": 8.26926597186149e-05,
      "loss": 0.1714,
      "step": 1648
    },
    {
      "epoch": 0.3461376994122586,
      "grad_norm": 0.1177772805094719,
      "learning_rate": 8.266492763564451e-05,
      "loss": 0.1651,
      "step": 1649
    },
    {
      "epoch": 0.34634760705289674,
      "grad_norm": 0.12576864659786224,
      "learning_rate": 8.263717801086182e-05,
      "loss": 0.1542,
      "step": 1650
    },
    {
      "epoch": 0.34655751469353485,
      "grad_norm": 0.10988842695951462,
      "learning_rate": 8.260941085916901e-05,
      "loss": 0.1436,
      "step": 1651
    },
    {
      "epoch": 0.34676742233417296,
      "grad_norm": 0.12295406311750412,
      "learning_rate": 8.258162619547771e-05,
      "loss": 0.1611,
      "step": 1652
    },
    {
      "epoch": 0.34697732997481107,
      "grad_norm": 0.12765416502952576,
      "learning_rate": 8.25538240347089e-05,
      "loss": 0.1707,
      "step": 1653
    },
    {
      "epoch": 0.34718723761544923,
      "grad_norm": 0.128097802400589,
      "learning_rate": 8.252600439179295e-05,
      "loss": 0.1777,
      "step": 1654
    },
    {
      "epoch": 0.34739714525608734,
      "grad_norm": 0.12253068387508392,
      "learning_rate": 8.24981672816697e-05,
      "loss": 0.1625,
      "step": 1655
    },
    {
      "epoch": 0.34760705289672544,
      "grad_norm": 0.12777206301689148,
      "learning_rate": 8.247031271928826e-05,
      "loss": 0.1822,
      "step": 1656
    },
    {
      "epoch": 0.34781696053736355,
      "grad_norm": 0.10205968469381332,
      "learning_rate": 8.24424407196072e-05,
      "loss": 0.1459,
      "step": 1657
    },
    {
      "epoch": 0.34802686817800166,
      "grad_norm": 0.11570538580417633,
      "learning_rate": 8.241455129759441e-05,
      "loss": 0.1719,
      "step": 1658
    },
    {
      "epoch": 0.3482367758186398,
      "grad_norm": 0.12801894545555115,
      "learning_rate": 8.238664446822715e-05,
      "loss": 0.1628,
      "step": 1659
    },
    {
      "epoch": 0.34844668345927793,
      "grad_norm": 0.13114169239997864,
      "learning_rate": 8.2358720246492e-05,
      "loss": 0.1768,
      "step": 1660
    },
    {
      "epoch": 0.34865659109991604,
      "grad_norm": 0.14204199612140656,
      "learning_rate": 8.233077864738495e-05,
      "loss": 0.1359,
      "step": 1661
    },
    {
      "epoch": 0.34886649874055414,
      "grad_norm": 0.13539434969425201,
      "learning_rate": 8.230281968591125e-05,
      "loss": 0.1696,
      "step": 1662
    },
    {
      "epoch": 0.34907640638119225,
      "grad_norm": 0.10501176118850708,
      "learning_rate": 8.22748433770855e-05,
      "loss": 0.1593,
      "step": 1663
    },
    {
      "epoch": 0.3492863140218304,
      "grad_norm": 0.10992568731307983,
      "learning_rate": 8.224684973593161e-05,
      "loss": 0.167,
      "step": 1664
    },
    {
      "epoch": 0.3494962216624685,
      "grad_norm": 0.10650011897087097,
      "learning_rate": 8.221883877748285e-05,
      "loss": 0.1533,
      "step": 1665
    },
    {
      "epoch": 0.3497061293031066,
      "grad_norm": 0.10955175757408142,
      "learning_rate": 8.219081051678167e-05,
      "loss": 0.1684,
      "step": 1666
    },
    {
      "epoch": 0.34991603694374473,
      "grad_norm": 0.11864706128835678,
      "learning_rate": 8.216276496887995e-05,
      "loss": 0.1662,
      "step": 1667
    },
    {
      "epoch": 0.3501259445843829,
      "grad_norm": 0.11310485005378723,
      "learning_rate": 8.213470214883876e-05,
      "loss": 0.1551,
      "step": 1668
    },
    {
      "epoch": 0.350335852225021,
      "grad_norm": 0.11496865004301071,
      "learning_rate": 8.210662207172848e-05,
      "loss": 0.1599,
      "step": 1669
    },
    {
      "epoch": 0.3505457598656591,
      "grad_norm": 0.16101498901844025,
      "learning_rate": 8.207852475262876e-05,
      "loss": 0.1599,
      "step": 1670
    },
    {
      "epoch": 0.3507556675062972,
      "grad_norm": 0.12098553776741028,
      "learning_rate": 8.205041020662849e-05,
      "loss": 0.1541,
      "step": 1671
    },
    {
      "epoch": 0.3509655751469353,
      "grad_norm": 0.14668092131614685,
      "learning_rate": 8.202227844882583e-05,
      "loss": 0.1625,
      "step": 1672
    },
    {
      "epoch": 0.3511754827875735,
      "grad_norm": 0.11160324513912201,
      "learning_rate": 8.199412949432817e-05,
      "loss": 0.1534,
      "step": 1673
    },
    {
      "epoch": 0.3513853904282116,
      "grad_norm": 0.1357201784849167,
      "learning_rate": 8.196596335825217e-05,
      "loss": 0.1702,
      "step": 1674
    },
    {
      "epoch": 0.3515952980688497,
      "grad_norm": 0.11901015788316727,
      "learning_rate": 8.193778005572365e-05,
      "loss": 0.1581,
      "step": 1675
    },
    {
      "epoch": 0.3518052057094878,
      "grad_norm": 0.12563678622245789,
      "learning_rate": 8.19095796018777e-05,
      "loss": 0.1681,
      "step": 1676
    },
    {
      "epoch": 0.3520151133501259,
      "grad_norm": 0.12322632968425751,
      "learning_rate": 8.188136201185863e-05,
      "loss": 0.1503,
      "step": 1677
    },
    {
      "epoch": 0.3522250209907641,
      "grad_norm": 0.13357025384902954,
      "learning_rate": 8.185312730081991e-05,
      "loss": 0.1704,
      "step": 1678
    },
    {
      "epoch": 0.3524349286314022,
      "grad_norm": 0.12442915886640549,
      "learning_rate": 8.182487548392422e-05,
      "loss": 0.163,
      "step": 1679
    },
    {
      "epoch": 0.3526448362720403,
      "grad_norm": 0.12422217428684235,
      "learning_rate": 8.179660657634342e-05,
      "loss": 0.1492,
      "step": 1680
    },
    {
      "epoch": 0.3528547439126784,
      "grad_norm": 0.12983065843582153,
      "learning_rate": 8.176832059325861e-05,
      "loss": 0.1509,
      "step": 1681
    },
    {
      "epoch": 0.35306465155331657,
      "grad_norm": 0.10647305846214294,
      "learning_rate": 8.174001754985996e-05,
      "loss": 0.1775,
      "step": 1682
    },
    {
      "epoch": 0.3532745591939547,
      "grad_norm": 0.1141040176153183,
      "learning_rate": 8.171169746134686e-05,
      "loss": 0.1617,
      "step": 1683
    },
    {
      "epoch": 0.3534844668345928,
      "grad_norm": 0.13015101850032806,
      "learning_rate": 8.168336034292786e-05,
      "loss": 0.16,
      "step": 1684
    },
    {
      "epoch": 0.3536943744752309,
      "grad_norm": 0.1377038061618805,
      "learning_rate": 8.165500620982062e-05,
      "loss": 0.1524,
      "step": 1685
    },
    {
      "epoch": 0.353904282115869,
      "grad_norm": 0.1097029447555542,
      "learning_rate": 8.162663507725198e-05,
      "loss": 0.1697,
      "step": 1686
    },
    {
      "epoch": 0.35411418975650716,
      "grad_norm": 0.1175478920340538,
      "learning_rate": 8.159824696045788e-05,
      "loss": 0.1826,
      "step": 1687
    },
    {
      "epoch": 0.35432409739714527,
      "grad_norm": 0.10951685905456543,
      "learning_rate": 8.156984187468337e-05,
      "loss": 0.156,
      "step": 1688
    },
    {
      "epoch": 0.3545340050377834,
      "grad_norm": 0.11676156520843506,
      "learning_rate": 8.154141983518265e-05,
      "loss": 0.1443,
      "step": 1689
    },
    {
      "epoch": 0.3547439126784215,
      "grad_norm": 0.10480359196662903,
      "learning_rate": 8.151298085721899e-05,
      "loss": 0.1675,
      "step": 1690
    },
    {
      "epoch": 0.3549538203190596,
      "grad_norm": 0.11646043509244919,
      "learning_rate": 8.148452495606478e-05,
      "loss": 0.1576,
      "step": 1691
    },
    {
      "epoch": 0.35516372795969775,
      "grad_norm": 0.1242692843079567,
      "learning_rate": 8.14560521470015e-05,
      "loss": 0.1486,
      "step": 1692
    },
    {
      "epoch": 0.35537363560033586,
      "grad_norm": 0.09616512805223465,
      "learning_rate": 8.142756244531967e-05,
      "loss": 0.1522,
      "step": 1693
    },
    {
      "epoch": 0.35558354324097396,
      "grad_norm": 0.139311745762825,
      "learning_rate": 8.13990558663189e-05,
      "loss": 0.1613,
      "step": 1694
    },
    {
      "epoch": 0.35579345088161207,
      "grad_norm": 0.11522821336984634,
      "learning_rate": 8.137053242530795e-05,
      "loss": 0.1603,
      "step": 1695
    },
    {
      "epoch": 0.35600335852225024,
      "grad_norm": 0.11472437530755997,
      "learning_rate": 8.134199213760447e-05,
      "loss": 0.1537,
      "step": 1696
    },
    {
      "epoch": 0.35621326616288834,
      "grad_norm": 0.14459721744060516,
      "learning_rate": 8.13134350185353e-05,
      "loss": 0.151,
      "step": 1697
    },
    {
      "epoch": 0.35642317380352645,
      "grad_norm": 0.12816378474235535,
      "learning_rate": 8.128486108343625e-05,
      "loss": 0.1677,
      "step": 1698
    },
    {
      "epoch": 0.35663308144416456,
      "grad_norm": 0.11897154897451401,
      "learning_rate": 8.125627034765218e-05,
      "loss": 0.1802,
      "step": 1699
    },
    {
      "epoch": 0.35684298908480266,
      "grad_norm": 0.12595590949058533,
      "learning_rate": 8.122766282653696e-05,
      "loss": 0.1668,
      "step": 1700
    },
    {
      "epoch": 0.3570528967254408,
      "grad_norm": 0.1388867050409317,
      "learning_rate": 8.119903853545348e-05,
      "loss": 0.1826,
      "step": 1701
    },
    {
      "epoch": 0.35726280436607893,
      "grad_norm": 0.1385848969221115,
      "learning_rate": 8.117039748977366e-05,
      "loss": 0.1581,
      "step": 1702
    },
    {
      "epoch": 0.35747271200671704,
      "grad_norm": 0.14706510305404663,
      "learning_rate": 8.114173970487838e-05,
      "loss": 0.1618,
      "step": 1703
    },
    {
      "epoch": 0.35768261964735515,
      "grad_norm": 0.15850181877613068,
      "learning_rate": 8.111306519615751e-05,
      "loss": 0.1527,
      "step": 1704
    },
    {
      "epoch": 0.35789252728799326,
      "grad_norm": 0.11479350924491882,
      "learning_rate": 8.108437397900995e-05,
      "loss": 0.166,
      "step": 1705
    },
    {
      "epoch": 0.3581024349286314,
      "grad_norm": 0.14089131355285645,
      "learning_rate": 8.10556660688435e-05,
      "loss": 0.1599,
      "step": 1706
    },
    {
      "epoch": 0.3583123425692695,
      "grad_norm": 0.11980936676263809,
      "learning_rate": 8.102694148107499e-05,
      "loss": 0.1685,
      "step": 1707
    },
    {
      "epoch": 0.35852225020990763,
      "grad_norm": 0.12391865998506546,
      "learning_rate": 8.099820023113018e-05,
      "loss": 0.164,
      "step": 1708
    },
    {
      "epoch": 0.35873215785054574,
      "grad_norm": 0.11180952191352844,
      "learning_rate": 8.096944233444379e-05,
      "loss": 0.1498,
      "step": 1709
    },
    {
      "epoch": 0.3589420654911839,
      "grad_norm": 0.12841565907001495,
      "learning_rate": 8.094066780645943e-05,
      "loss": 0.1606,
      "step": 1710
    },
    {
      "epoch": 0.359151973131822,
      "grad_norm": 0.11051752418279648,
      "learning_rate": 8.09118766626297e-05,
      "loss": 0.1519,
      "step": 1711
    },
    {
      "epoch": 0.3593618807724601,
      "grad_norm": 0.12816421687602997,
      "learning_rate": 8.088306891841609e-05,
      "loss": 0.1657,
      "step": 1712
    },
    {
      "epoch": 0.3595717884130982,
      "grad_norm": 0.10819518566131592,
      "learning_rate": 8.085424458928904e-05,
      "loss": 0.1805,
      "step": 1713
    },
    {
      "epoch": 0.35978169605373633,
      "grad_norm": 0.10757758468389511,
      "learning_rate": 8.082540369072786e-05,
      "loss": 0.1524,
      "step": 1714
    },
    {
      "epoch": 0.3599916036943745,
      "grad_norm": 0.13418933749198914,
      "learning_rate": 8.079654623822077e-05,
      "loss": 0.1623,
      "step": 1715
    },
    {
      "epoch": 0.3602015113350126,
      "grad_norm": 0.11752188950777054,
      "learning_rate": 8.076767224726487e-05,
      "loss": 0.1736,
      "step": 1716
    },
    {
      "epoch": 0.3604114189756507,
      "grad_norm": 0.1459878832101822,
      "learning_rate": 8.073878173336618e-05,
      "loss": 0.1502,
      "step": 1717
    },
    {
      "epoch": 0.3606213266162888,
      "grad_norm": 0.10812754929065704,
      "learning_rate": 8.070987471203955e-05,
      "loss": 0.1484,
      "step": 1718
    },
    {
      "epoch": 0.3608312342569269,
      "grad_norm": 0.1168069913983345,
      "learning_rate": 8.06809511988087e-05,
      "loss": 0.1355,
      "step": 1719
    },
    {
      "epoch": 0.3610411418975651,
      "grad_norm": 0.133521169424057,
      "learning_rate": 8.065201120920625e-05,
      "loss": 0.1707,
      "step": 1720
    },
    {
      "epoch": 0.3612510495382032,
      "grad_norm": 0.12936338782310486,
      "learning_rate": 8.062305475877358e-05,
      "loss": 0.1518,
      "step": 1721
    },
    {
      "epoch": 0.3614609571788413,
      "grad_norm": 0.11167915165424347,
      "learning_rate": 8.059408186306104e-05,
      "loss": 0.1365,
      "step": 1722
    },
    {
      "epoch": 0.3616708648194794,
      "grad_norm": 0.1492997705936432,
      "learning_rate": 8.056509253762768e-05,
      "loss": 0.1719,
      "step": 1723
    },
    {
      "epoch": 0.3618807724601176,
      "grad_norm": 0.1078883707523346,
      "learning_rate": 8.053608679804144e-05,
      "loss": 0.1466,
      "step": 1724
    },
    {
      "epoch": 0.3620906801007557,
      "grad_norm": 0.11772193014621735,
      "learning_rate": 8.050706465987907e-05,
      "loss": 0.1607,
      "step": 1725
    },
    {
      "epoch": 0.3623005877413938,
      "grad_norm": 0.11475538462400436,
      "learning_rate": 8.047802613872615e-05,
      "loss": 0.1727,
      "step": 1726
    },
    {
      "epoch": 0.3625104953820319,
      "grad_norm": 0.10466550290584564,
      "learning_rate": 8.044897125017696e-05,
      "loss": 0.1586,
      "step": 1727
    },
    {
      "epoch": 0.36272040302267,
      "grad_norm": 0.13003644347190857,
      "learning_rate": 8.041990000983472e-05,
      "loss": 0.168,
      "step": 1728
    },
    {
      "epoch": 0.36293031066330816,
      "grad_norm": 0.11580996215343475,
      "learning_rate": 8.039081243331128e-05,
      "loss": 0.1532,
      "step": 1729
    },
    {
      "epoch": 0.36314021830394627,
      "grad_norm": 0.11512108892202377,
      "learning_rate": 8.036170853622738e-05,
      "loss": 0.153,
      "step": 1730
    },
    {
      "epoch": 0.3633501259445844,
      "grad_norm": 0.14258354902267456,
      "learning_rate": 8.033258833421246e-05,
      "loss": 0.1542,
      "step": 1731
    },
    {
      "epoch": 0.3635600335852225,
      "grad_norm": 0.11443816125392914,
      "learning_rate": 8.030345184290473e-05,
      "loss": 0.1523,
      "step": 1732
    },
    {
      "epoch": 0.36376994122586065,
      "grad_norm": 0.098850317299366,
      "learning_rate": 8.027429907795116e-05,
      "loss": 0.1413,
      "step": 1733
    },
    {
      "epoch": 0.36397984886649876,
      "grad_norm": 0.09741919487714767,
      "learning_rate": 8.024513005500744e-05,
      "loss": 0.1482,
      "step": 1734
    },
    {
      "epoch": 0.36418975650713686,
      "grad_norm": 0.12056808918714523,
      "learning_rate": 8.021594478973801e-05,
      "loss": 0.1621,
      "step": 1735
    },
    {
      "epoch": 0.36439966414777497,
      "grad_norm": 0.10978755354881287,
      "learning_rate": 8.018674329781603e-05,
      "loss": 0.1578,
      "step": 1736
    },
    {
      "epoch": 0.3646095717884131,
      "grad_norm": 0.12888531386852264,
      "learning_rate": 8.015752559492335e-05,
      "loss": 0.1552,
      "step": 1737
    },
    {
      "epoch": 0.36481947942905124,
      "grad_norm": 0.12127751111984253,
      "learning_rate": 8.012829169675056e-05,
      "loss": 0.165,
      "step": 1738
    },
    {
      "epoch": 0.36502938706968935,
      "grad_norm": 0.10316291451454163,
      "learning_rate": 8.009904161899692e-05,
      "loss": 0.1306,
      "step": 1739
    },
    {
      "epoch": 0.36523929471032746,
      "grad_norm": 0.12621663510799408,
      "learning_rate": 8.006977537737043e-05,
      "loss": 0.1586,
      "step": 1740
    },
    {
      "epoch": 0.36544920235096556,
      "grad_norm": 0.13226893544197083,
      "learning_rate": 8.004049298758767e-05,
      "loss": 0.1432,
      "step": 1741
    },
    {
      "epoch": 0.36565910999160367,
      "grad_norm": 0.11602537333965302,
      "learning_rate": 8.001119446537397e-05,
      "loss": 0.152,
      "step": 1742
    },
    {
      "epoch": 0.36586901763224183,
      "grad_norm": 0.10899031907320023,
      "learning_rate": 7.998187982646332e-05,
      "loss": 0.1489,
      "step": 1743
    },
    {
      "epoch": 0.36607892527287994,
      "grad_norm": 0.11780837923288345,
      "learning_rate": 7.995254908659836e-05,
      "loss": 0.166,
      "step": 1744
    },
    {
      "epoch": 0.36628883291351805,
      "grad_norm": 0.13604290783405304,
      "learning_rate": 7.992320226153032e-05,
      "loss": 0.1731,
      "step": 1745
    },
    {
      "epoch": 0.36649874055415615,
      "grad_norm": 0.1298649162054062,
      "learning_rate": 7.989383936701916e-05,
      "loss": 0.151,
      "step": 1746
    },
    {
      "epoch": 0.3667086481947943,
      "grad_norm": 0.12146075814962387,
      "learning_rate": 7.986446041883342e-05,
      "loss": 0.1674,
      "step": 1747
    },
    {
      "epoch": 0.3669185558354324,
      "grad_norm": 0.13171765208244324,
      "learning_rate": 7.983506543275025e-05,
      "loss": 0.147,
      "step": 1748
    },
    {
      "epoch": 0.36712846347607053,
      "grad_norm": 0.13356606662273407,
      "learning_rate": 7.980565442455545e-05,
      "loss": 0.1724,
      "step": 1749
    },
    {
      "epoch": 0.36733837111670864,
      "grad_norm": 0.13275739550590515,
      "learning_rate": 7.977622741004338e-05,
      "loss": 0.1586,
      "step": 1750
    },
    {
      "epoch": 0.36754827875734675,
      "grad_norm": 0.1208769828081131,
      "learning_rate": 7.974678440501703e-05,
      "loss": 0.1558,
      "step": 1751
    },
    {
      "epoch": 0.3677581863979849,
      "grad_norm": 0.15097500383853912,
      "learning_rate": 7.971732542528798e-05,
      "loss": 0.157,
      "step": 1752
    },
    {
      "epoch": 0.367968094038623,
      "grad_norm": 0.13939009606838226,
      "learning_rate": 7.968785048667634e-05,
      "loss": 0.1483,
      "step": 1753
    },
    {
      "epoch": 0.3681780016792611,
      "grad_norm": 0.1147853285074234,
      "learning_rate": 7.965835960501086e-05,
      "loss": 0.1638,
      "step": 1754
    },
    {
      "epoch": 0.36838790931989923,
      "grad_norm": 0.10615186393260956,
      "learning_rate": 7.96288527961288e-05,
      "loss": 0.1674,
      "step": 1755
    },
    {
      "epoch": 0.36859781696053734,
      "grad_norm": 0.10610269010066986,
      "learning_rate": 7.959933007587598e-05,
      "loss": 0.1598,
      "step": 1756
    },
    {
      "epoch": 0.3688077246011755,
      "grad_norm": 0.11840075999498367,
      "learning_rate": 7.956979146010683e-05,
      "loss": 0.1673,
      "step": 1757
    },
    {
      "epoch": 0.3690176322418136,
      "grad_norm": 0.1263597160577774,
      "learning_rate": 7.954023696468418e-05,
      "loss": 0.1332,
      "step": 1758
    },
    {
      "epoch": 0.3692275398824517,
      "grad_norm": 0.13218241930007935,
      "learning_rate": 7.951066660547952e-05,
      "loss": 0.165,
      "step": 1759
    },
    {
      "epoch": 0.3694374475230898,
      "grad_norm": 0.11182461678981781,
      "learning_rate": 7.948108039837281e-05,
      "loss": 0.175,
      "step": 1760
    },
    {
      "epoch": 0.369647355163728,
      "grad_norm": 0.12093862891197205,
      "learning_rate": 7.94514783592525e-05,
      "loss": 0.17,
      "step": 1761
    },
    {
      "epoch": 0.3698572628043661,
      "grad_norm": 0.1129969134926796,
      "learning_rate": 7.942186050401556e-05,
      "loss": 0.1605,
      "step": 1762
    },
    {
      "epoch": 0.3700671704450042,
      "grad_norm": 0.11315370351076126,
      "learning_rate": 7.939222684856747e-05,
      "loss": 0.1724,
      "step": 1763
    },
    {
      "epoch": 0.3702770780856423,
      "grad_norm": 0.12104691565036774,
      "learning_rate": 7.936257740882218e-05,
      "loss": 0.1693,
      "step": 1764
    },
    {
      "epoch": 0.3704869857262804,
      "grad_norm": 0.12000355124473572,
      "learning_rate": 7.933291220070209e-05,
      "loss": 0.1855,
      "step": 1765
    },
    {
      "epoch": 0.3706968933669186,
      "grad_norm": 0.11460446566343307,
      "learning_rate": 7.930323124013812e-05,
      "loss": 0.1531,
      "step": 1766
    },
    {
      "epoch": 0.3709068010075567,
      "grad_norm": 0.12112744152545929,
      "learning_rate": 7.927353454306962e-05,
      "loss": 0.166,
      "step": 1767
    },
    {
      "epoch": 0.3711167086481948,
      "grad_norm": 0.11343761533498764,
      "learning_rate": 7.92438221254444e-05,
      "loss": 0.1718,
      "step": 1768
    },
    {
      "epoch": 0.3713266162888329,
      "grad_norm": 0.09574848413467407,
      "learning_rate": 7.921409400321869e-05,
      "loss": 0.1552,
      "step": 1769
    },
    {
      "epoch": 0.371536523929471,
      "grad_norm": 0.12521985173225403,
      "learning_rate": 7.918435019235719e-05,
      "loss": 0.173,
      "step": 1770
    },
    {
      "epoch": 0.37174643157010917,
      "grad_norm": 0.13208284974098206,
      "learning_rate": 7.915459070883297e-05,
      "loss": 0.1461,
      "step": 1771
    },
    {
      "epoch": 0.3719563392107473,
      "grad_norm": 0.13827428221702576,
      "learning_rate": 7.91248155686276e-05,
      "loss": 0.1763,
      "step": 1772
    },
    {
      "epoch": 0.3721662468513854,
      "grad_norm": 0.11231647431850433,
      "learning_rate": 7.909502478773099e-05,
      "loss": 0.1507,
      "step": 1773
    },
    {
      "epoch": 0.3723761544920235,
      "grad_norm": 0.13902410864830017,
      "learning_rate": 7.906521838214147e-05,
      "loss": 0.1658,
      "step": 1774
    },
    {
      "epoch": 0.37258606213266166,
      "grad_norm": 0.1284346878528595,
      "learning_rate": 7.903539636786574e-05,
      "loss": 0.1473,
      "step": 1775
    },
    {
      "epoch": 0.37279596977329976,
      "grad_norm": 0.09582658857107162,
      "learning_rate": 7.900555876091894e-05,
      "loss": 0.1578,
      "step": 1776
    },
    {
      "epoch": 0.37300587741393787,
      "grad_norm": 0.11832595616579056,
      "learning_rate": 7.897570557732452e-05,
      "loss": 0.1528,
      "step": 1777
    },
    {
      "epoch": 0.373215785054576,
      "grad_norm": 0.12966689467430115,
      "learning_rate": 7.894583683311434e-05,
      "loss": 0.1861,
      "step": 1778
    },
    {
      "epoch": 0.3734256926952141,
      "grad_norm": 0.11145227402448654,
      "learning_rate": 7.891595254432857e-05,
      "loss": 0.1555,
      "step": 1779
    },
    {
      "epoch": 0.37363560033585225,
      "grad_norm": 0.13669519126415253,
      "learning_rate": 7.888605272701578e-05,
      "loss": 0.1618,
      "step": 1780
    },
    {
      "epoch": 0.37384550797649035,
      "grad_norm": 0.12473908811807632,
      "learning_rate": 7.885613739723285e-05,
      "loss": 0.1604,
      "step": 1781
    },
    {
      "epoch": 0.37405541561712846,
      "grad_norm": 0.10881557315587997,
      "learning_rate": 7.8826206571045e-05,
      "loss": 0.1701,
      "step": 1782
    },
    {
      "epoch": 0.37426532325776657,
      "grad_norm": 0.11604516953229904,
      "learning_rate": 7.879626026452576e-05,
      "loss": 0.1467,
      "step": 1783
    },
    {
      "epoch": 0.3744752308984047,
      "grad_norm": 0.11594154685735703,
      "learning_rate": 7.876629849375696e-05,
      "loss": 0.181,
      "step": 1784
    },
    {
      "epoch": 0.37468513853904284,
      "grad_norm": 0.11912267655134201,
      "learning_rate": 7.87363212748288e-05,
      "loss": 0.1451,
      "step": 1785
    },
    {
      "epoch": 0.37489504617968095,
      "grad_norm": 0.11206775903701782,
      "learning_rate": 7.87063286238397e-05,
      "loss": 0.1706,
      "step": 1786
    },
    {
      "epoch": 0.37510495382031905,
      "grad_norm": 0.12397405505180359,
      "learning_rate": 7.86763205568964e-05,
      "loss": 0.1599,
      "step": 1787
    },
    {
      "epoch": 0.37531486146095716,
      "grad_norm": 0.1209942027926445,
      "learning_rate": 7.864629709011395e-05,
      "loss": 0.139,
      "step": 1788
    },
    {
      "epoch": 0.3755247691015953,
      "grad_norm": 0.10950622707605362,
      "learning_rate": 7.86162582396156e-05,
      "loss": 0.1625,
      "step": 1789
    },
    {
      "epoch": 0.37573467674223343,
      "grad_norm": 0.12611068785190582,
      "learning_rate": 7.85862040215329e-05,
      "loss": 0.1719,
      "step": 1790
    },
    {
      "epoch": 0.37594458438287154,
      "grad_norm": 0.1125999391078949,
      "learning_rate": 7.855613445200567e-05,
      "loss": 0.1732,
      "step": 1791
    },
    {
      "epoch": 0.37615449202350965,
      "grad_norm": 0.12845373153686523,
      "learning_rate": 7.852604954718195e-05,
      "loss": 0.1477,
      "step": 1792
    },
    {
      "epoch": 0.37636439966414775,
      "grad_norm": 0.17373661696910858,
      "learning_rate": 7.8495949323218e-05,
      "loss": 0.158,
      "step": 1793
    },
    {
      "epoch": 0.3765743073047859,
      "grad_norm": 0.10421296954154968,
      "learning_rate": 7.846583379627836e-05,
      "loss": 0.1683,
      "step": 1794
    },
    {
      "epoch": 0.376784214945424,
      "grad_norm": 0.1311960518360138,
      "learning_rate": 7.843570298253577e-05,
      "loss": 0.1361,
      "step": 1795
    },
    {
      "epoch": 0.37699412258606213,
      "grad_norm": 0.17410288751125336,
      "learning_rate": 7.840555689817111e-05,
      "loss": 0.1678,
      "step": 1796
    },
    {
      "epoch": 0.37720403022670024,
      "grad_norm": 0.11635412275791168,
      "learning_rate": 7.837539555937354e-05,
      "loss": 0.1582,
      "step": 1797
    },
    {
      "epoch": 0.37741393786733834,
      "grad_norm": 0.1161051094532013,
      "learning_rate": 7.83452189823404e-05,
      "loss": 0.1686,
      "step": 1798
    },
    {
      "epoch": 0.3776238455079765,
      "grad_norm": 0.11930254846811295,
      "learning_rate": 7.831502718327719e-05,
      "loss": 0.1717,
      "step": 1799
    },
    {
      "epoch": 0.3778337531486146,
      "grad_norm": 0.12649065256118774,
      "learning_rate": 7.82848201783976e-05,
      "loss": 0.1418,
      "step": 1800
    },
    {
      "epoch": 0.3780436607892527,
      "grad_norm": 0.1233733594417572,
      "learning_rate": 7.825459798392349e-05,
      "loss": 0.1412,
      "step": 1801
    },
    {
      "epoch": 0.37825356842989083,
      "grad_norm": 0.12751632928848267,
      "learning_rate": 7.822436061608485e-05,
      "loss": 0.1591,
      "step": 1802
    },
    {
      "epoch": 0.378463476070529,
      "grad_norm": 0.11433545500040054,
      "learning_rate": 7.819410809111984e-05,
      "loss": 0.1608,
      "step": 1803
    },
    {
      "epoch": 0.3786733837111671,
      "grad_norm": 0.12332860380411148,
      "learning_rate": 7.816384042527479e-05,
      "loss": 0.1544,
      "step": 1804
    },
    {
      "epoch": 0.3788832913518052,
      "grad_norm": 0.14029990136623383,
      "learning_rate": 7.813355763480407e-05,
      "loss": 0.1652,
      "step": 1805
    },
    {
      "epoch": 0.3790931989924433,
      "grad_norm": 0.12804871797561646,
      "learning_rate": 7.810325973597027e-05,
      "loss": 0.1464,
      "step": 1806
    },
    {
      "epoch": 0.3793031066330814,
      "grad_norm": 0.10844779014587402,
      "learning_rate": 7.807294674504405e-05,
      "loss": 0.1517,
      "step": 1807
    },
    {
      "epoch": 0.3795130142737196,
      "grad_norm": 0.1057022362947464,
      "learning_rate": 7.804261867830418e-05,
      "loss": 0.1558,
      "step": 1808
    },
    {
      "epoch": 0.3797229219143577,
      "grad_norm": 0.1034378632903099,
      "learning_rate": 7.801227555203751e-05,
      "loss": 0.1519,
      "step": 1809
    },
    {
      "epoch": 0.3799328295549958,
      "grad_norm": 0.13120600581169128,
      "learning_rate": 7.7981917382539e-05,
      "loss": 0.1689,
      "step": 1810
    },
    {
      "epoch": 0.3801427371956339,
      "grad_norm": 0.14298629760742188,
      "learning_rate": 7.795154418611169e-05,
      "loss": 0.1609,
      "step": 1811
    },
    {
      "epoch": 0.380352644836272,
      "grad_norm": 0.1206570416688919,
      "learning_rate": 7.792115597906666e-05,
      "loss": 0.1481,
      "step": 1812
    },
    {
      "epoch": 0.3805625524769102,
      "grad_norm": 0.12700314819812775,
      "learning_rate": 7.789075277772309e-05,
      "loss": 0.1526,
      "step": 1813
    },
    {
      "epoch": 0.3807724601175483,
      "grad_norm": 0.13419455289840698,
      "learning_rate": 7.786033459840817e-05,
      "loss": 0.1665,
      "step": 1814
    },
    {
      "epoch": 0.3809823677581864,
      "grad_norm": 0.11350502818822861,
      "learning_rate": 7.78299014574572e-05,
      "loss": 0.1293,
      "step": 1815
    },
    {
      "epoch": 0.3811922753988245,
      "grad_norm": 0.12667128443717957,
      "learning_rate": 7.779945337121343e-05,
      "loss": 0.1474,
      "step": 1816
    },
    {
      "epoch": 0.38140218303946266,
      "grad_norm": 0.14447607100009918,
      "learning_rate": 7.77689903560282e-05,
      "loss": 0.1642,
      "step": 1817
    },
    {
      "epoch": 0.38161209068010077,
      "grad_norm": 0.1254795789718628,
      "learning_rate": 7.773851242826083e-05,
      "loss": 0.1565,
      "step": 1818
    },
    {
      "epoch": 0.3818219983207389,
      "grad_norm": 0.1166321262717247,
      "learning_rate": 7.770801960427869e-05,
      "loss": 0.1583,
      "step": 1819
    },
    {
      "epoch": 0.382031905961377,
      "grad_norm": 0.11959258466959,
      "learning_rate": 7.767751190045706e-05,
      "loss": 0.1623,
      "step": 1820
    },
    {
      "epoch": 0.3822418136020151,
      "grad_norm": 0.09385194629430771,
      "learning_rate": 7.764698933317935e-05,
      "loss": 0.1525,
      "step": 1821
    },
    {
      "epoch": 0.38245172124265325,
      "grad_norm": 0.1065850630402565,
      "learning_rate": 7.761645191883684e-05,
      "loss": 0.1538,
      "step": 1822
    },
    {
      "epoch": 0.38266162888329136,
      "grad_norm": 0.13342434167861938,
      "learning_rate": 7.758589967382883e-05,
      "loss": 0.1488,
      "step": 1823
    },
    {
      "epoch": 0.38287153652392947,
      "grad_norm": 0.13576741516590118,
      "learning_rate": 7.755533261456256e-05,
      "loss": 0.1693,
      "step": 1824
    },
    {
      "epoch": 0.3830814441645676,
      "grad_norm": 0.12756936252117157,
      "learning_rate": 7.752475075745325e-05,
      "loss": 0.1573,
      "step": 1825
    },
    {
      "epoch": 0.3832913518052057,
      "grad_norm": 0.1328425407409668,
      "learning_rate": 7.749415411892404e-05,
      "loss": 0.142,
      "step": 1826
    },
    {
      "epoch": 0.38350125944584385,
      "grad_norm": 0.1451667994260788,
      "learning_rate": 7.746354271540605e-05,
      "loss": 0.1544,
      "step": 1827
    },
    {
      "epoch": 0.38371116708648195,
      "grad_norm": 0.13675998151302338,
      "learning_rate": 7.743291656333826e-05,
      "loss": 0.1576,
      "step": 1828
    },
    {
      "epoch": 0.38392107472712006,
      "grad_norm": 0.11071248352527618,
      "learning_rate": 7.740227567916767e-05,
      "loss": 0.1636,
      "step": 1829
    },
    {
      "epoch": 0.38413098236775817,
      "grad_norm": 0.13009920716285706,
      "learning_rate": 7.73716200793491e-05,
      "loss": 0.1705,
      "step": 1830
    },
    {
      "epoch": 0.38434089000839633,
      "grad_norm": 0.13077101111412048,
      "learning_rate": 7.734094978034532e-05,
      "loss": 0.1715,
      "step": 1831
    },
    {
      "epoch": 0.38455079764903444,
      "grad_norm": 0.11513860523700714,
      "learning_rate": 7.731026479862696e-05,
      "loss": 0.1584,
      "step": 1832
    },
    {
      "epoch": 0.38476070528967254,
      "grad_norm": 0.10579150915145874,
      "learning_rate": 7.727956515067259e-05,
      "loss": 0.1618,
      "step": 1833
    },
    {
      "epoch": 0.38497061293031065,
      "grad_norm": 0.1078706756234169,
      "learning_rate": 7.724885085296858e-05,
      "loss": 0.167,
      "step": 1834
    },
    {
      "epoch": 0.38518052057094876,
      "grad_norm": 0.131106436252594,
      "learning_rate": 7.721812192200924e-05,
      "loss": 0.1525,
      "step": 1835
    },
    {
      "epoch": 0.3853904282115869,
      "grad_norm": 0.15161815285682678,
      "learning_rate": 7.718737837429668e-05,
      "loss": 0.1523,
      "step": 1836
    },
    {
      "epoch": 0.38560033585222503,
      "grad_norm": 0.11949379742145538,
      "learning_rate": 7.715662022634094e-05,
      "loss": 0.1543,
      "step": 1837
    },
    {
      "epoch": 0.38581024349286314,
      "grad_norm": 0.16841568052768707,
      "learning_rate": 7.712584749465981e-05,
      "loss": 0.1754,
      "step": 1838
    },
    {
      "epoch": 0.38602015113350124,
      "grad_norm": 0.1119508296251297,
      "learning_rate": 7.709506019577895e-05,
      "loss": 0.16,
      "step": 1839
    },
    {
      "epoch": 0.38623005877413935,
      "grad_norm": 0.12148314714431763,
      "learning_rate": 7.706425834623184e-05,
      "loss": 0.1545,
      "step": 1840
    },
    {
      "epoch": 0.3864399664147775,
      "grad_norm": 0.13948237895965576,
      "learning_rate": 7.703344196255979e-05,
      "loss": 0.156,
      "step": 1841
    },
    {
      "epoch": 0.3866498740554156,
      "grad_norm": 0.11464816331863403,
      "learning_rate": 7.700261106131191e-05,
      "loss": 0.172,
      "step": 1842
    },
    {
      "epoch": 0.38685978169605373,
      "grad_norm": 0.11518428474664688,
      "learning_rate": 7.697176565904507e-05,
      "loss": 0.171,
      "step": 1843
    },
    {
      "epoch": 0.38706968933669184,
      "grad_norm": 0.11466710269451141,
      "learning_rate": 7.694090577232398e-05,
      "loss": 0.1545,
      "step": 1844
    },
    {
      "epoch": 0.38727959697733,
      "grad_norm": 0.13466320931911469,
      "learning_rate": 7.691003141772108e-05,
      "loss": 0.1487,
      "step": 1845
    },
    {
      "epoch": 0.3874895046179681,
      "grad_norm": 0.1101074293255806,
      "learning_rate": 7.687914261181664e-05,
      "loss": 0.1659,
      "step": 1846
    },
    {
      "epoch": 0.3876994122586062,
      "grad_norm": 0.12081106752157211,
      "learning_rate": 7.684823937119862e-05,
      "loss": 0.1558,
      "step": 1847
    },
    {
      "epoch": 0.3879093198992443,
      "grad_norm": 0.12375544011592865,
      "learning_rate": 7.681732171246276e-05,
      "loss": 0.1626,
      "step": 1848
    },
    {
      "epoch": 0.3881192275398824,
      "grad_norm": 0.13245245814323425,
      "learning_rate": 7.678638965221262e-05,
      "loss": 0.1445,
      "step": 1849
    },
    {
      "epoch": 0.3883291351805206,
      "grad_norm": 0.12885937094688416,
      "learning_rate": 7.675544320705932e-05,
      "loss": 0.151,
      "step": 1850
    },
    {
      "epoch": 0.3885390428211587,
      "grad_norm": 0.15650451183319092,
      "learning_rate": 7.67244823936219e-05,
      "loss": 0.1456,
      "step": 1851
    },
    {
      "epoch": 0.3887489504617968,
      "grad_norm": 0.13623982667922974,
      "learning_rate": 7.669350722852693e-05,
      "loss": 0.1726,
      "step": 1852
    },
    {
      "epoch": 0.3889588581024349,
      "grad_norm": 0.1482025533914566,
      "learning_rate": 7.666251772840887e-05,
      "loss": 0.1594,
      "step": 1853
    },
    {
      "epoch": 0.3891687657430731,
      "grad_norm": 0.12205847352743149,
      "learning_rate": 7.663151390990975e-05,
      "loss": 0.1656,
      "step": 1854
    },
    {
      "epoch": 0.3893786733837112,
      "grad_norm": 0.11122436821460724,
      "learning_rate": 7.660049578967934e-05,
      "loss": 0.1674,
      "step": 1855
    },
    {
      "epoch": 0.3895885810243493,
      "grad_norm": 0.13141264021396637,
      "learning_rate": 7.656946338437505e-05,
      "loss": 0.1543,
      "step": 1856
    },
    {
      "epoch": 0.3897984886649874,
      "grad_norm": 0.12650364637374878,
      "learning_rate": 7.653841671066203e-05,
      "loss": 0.1469,
      "step": 1857
    },
    {
      "epoch": 0.3900083963056255,
      "grad_norm": 0.12014992535114288,
      "learning_rate": 7.650735578521302e-05,
      "loss": 0.1541,
      "step": 1858
    },
    {
      "epoch": 0.39021830394626367,
      "grad_norm": 0.1951693892478943,
      "learning_rate": 7.647628062470848e-05,
      "loss": 0.1653,
      "step": 1859
    },
    {
      "epoch": 0.3904282115869018,
      "grad_norm": 0.1394760012626648,
      "learning_rate": 7.644519124583643e-05,
      "loss": 0.1573,
      "step": 1860
    },
    {
      "epoch": 0.3906381192275399,
      "grad_norm": 0.14827001094818115,
      "learning_rate": 7.641408766529266e-05,
      "loss": 0.1585,
      "step": 1861
    },
    {
      "epoch": 0.390848026868178,
      "grad_norm": 0.1227707639336586,
      "learning_rate": 7.63829698997804e-05,
      "loss": 0.1548,
      "step": 1862
    },
    {
      "epoch": 0.3910579345088161,
      "grad_norm": 0.13828247785568237,
      "learning_rate": 7.635183796601068e-05,
      "loss": 0.1789,
      "step": 1863
    },
    {
      "epoch": 0.39126784214945426,
      "grad_norm": 0.13925831019878387,
      "learning_rate": 7.632069188070204e-05,
      "loss": 0.1699,
      "step": 1864
    },
    {
      "epoch": 0.39147774979009237,
      "grad_norm": 0.13822948932647705,
      "learning_rate": 7.628953166058062e-05,
      "loss": 0.1674,
      "step": 1865
    },
    {
      "epoch": 0.3916876574307305,
      "grad_norm": 0.14011485874652863,
      "learning_rate": 7.625835732238018e-05,
      "loss": 0.1778,
      "step": 1866
    },
    {
      "epoch": 0.3918975650713686,
      "grad_norm": 0.1188986524939537,
      "learning_rate": 7.622716888284205e-05,
      "loss": 0.1563,
      "step": 1867
    },
    {
      "epoch": 0.39210747271200674,
      "grad_norm": 0.10080984979867935,
      "learning_rate": 7.619596635871513e-05,
      "loss": 0.169,
      "step": 1868
    },
    {
      "epoch": 0.39231738035264485,
      "grad_norm": 0.11640693992376328,
      "learning_rate": 7.616474976675592e-05,
      "loss": 0.1699,
      "step": 1869
    },
    {
      "epoch": 0.39252728799328296,
      "grad_norm": 0.1325574517250061,
      "learning_rate": 7.613351912372842e-05,
      "loss": 0.1727,
      "step": 1870
    },
    {
      "epoch": 0.39273719563392107,
      "grad_norm": 0.12673214077949524,
      "learning_rate": 7.610227444640417e-05,
      "loss": 0.1579,
      "step": 1871
    },
    {
      "epoch": 0.3929471032745592,
      "grad_norm": 0.10220383107662201,
      "learning_rate": 7.607101575156232e-05,
      "loss": 0.1559,
      "step": 1872
    },
    {
      "epoch": 0.39315701091519734,
      "grad_norm": 0.11445717513561249,
      "learning_rate": 7.603974305598948e-05,
      "loss": 0.1609,
      "step": 1873
    },
    {
      "epoch": 0.39336691855583544,
      "grad_norm": 0.10287608951330185,
      "learning_rate": 7.600845637647981e-05,
      "loss": 0.1442,
      "step": 1874
    },
    {
      "epoch": 0.39357682619647355,
      "grad_norm": 0.11501127481460571,
      "learning_rate": 7.597715572983499e-05,
      "loss": 0.162,
      "step": 1875
    },
    {
      "epoch": 0.39378673383711166,
      "grad_norm": 0.18507914245128632,
      "learning_rate": 7.594584113286413e-05,
      "loss": 0.1611,
      "step": 1876
    },
    {
      "epoch": 0.39399664147774976,
      "grad_norm": 0.11232730746269226,
      "learning_rate": 7.591451260238394e-05,
      "loss": 0.1555,
      "step": 1877
    },
    {
      "epoch": 0.39420654911838793,
      "grad_norm": 0.0994970053434372,
      "learning_rate": 7.588317015521853e-05,
      "loss": 0.1331,
      "step": 1878
    },
    {
      "epoch": 0.39441645675902604,
      "grad_norm": 0.11033397912979126,
      "learning_rate": 7.58518138081995e-05,
      "loss": 0.1611,
      "step": 1879
    },
    {
      "epoch": 0.39462636439966414,
      "grad_norm": 0.12892796099185944,
      "learning_rate": 7.582044357816593e-05,
      "loss": 0.1597,
      "step": 1880
    },
    {
      "epoch": 0.39483627204030225,
      "grad_norm": 0.13372941315174103,
      "learning_rate": 7.578905948196437e-05,
      "loss": 0.1605,
      "step": 1881
    },
    {
      "epoch": 0.3950461796809404,
      "grad_norm": 0.11890514194965363,
      "learning_rate": 7.57576615364488e-05,
      "loss": 0.1598,
      "step": 1882
    },
    {
      "epoch": 0.3952560873215785,
      "grad_norm": 0.14822334051132202,
      "learning_rate": 7.57262497584806e-05,
      "loss": 0.1801,
      "step": 1883
    },
    {
      "epoch": 0.3954659949622166,
      "grad_norm": 0.1431664526462555,
      "learning_rate": 7.569482416492863e-05,
      "loss": 0.1632,
      "step": 1884
    },
    {
      "epoch": 0.39567590260285473,
      "grad_norm": 0.11748656630516052,
      "learning_rate": 7.566338477266914e-05,
      "loss": 0.1479,
      "step": 1885
    },
    {
      "epoch": 0.39588581024349284,
      "grad_norm": 0.11131531000137329,
      "learning_rate": 7.563193159858584e-05,
      "loss": 0.1545,
      "step": 1886
    },
    {
      "epoch": 0.396095717884131,
      "grad_norm": 0.11406468600034714,
      "learning_rate": 7.560046465956975e-05,
      "loss": 0.1505,
      "step": 1887
    },
    {
      "epoch": 0.3963056255247691,
      "grad_norm": 0.12816153466701508,
      "learning_rate": 7.556898397251938e-05,
      "loss": 0.1709,
      "step": 1888
    },
    {
      "epoch": 0.3965155331654072,
      "grad_norm": 0.11210453510284424,
      "learning_rate": 7.553748955434054e-05,
      "loss": 0.1575,
      "step": 1889
    },
    {
      "epoch": 0.3967254408060453,
      "grad_norm": 0.11992809921503067,
      "learning_rate": 7.55059814219465e-05,
      "loss": 0.1396,
      "step": 1890
    },
    {
      "epoch": 0.39693534844668343,
      "grad_norm": 0.11076368391513824,
      "learning_rate": 7.54744595922578e-05,
      "loss": 0.1574,
      "step": 1891
    },
    {
      "epoch": 0.3971452560873216,
      "grad_norm": 0.12722079455852509,
      "learning_rate": 7.544292408220241e-05,
      "loss": 0.1595,
      "step": 1892
    },
    {
      "epoch": 0.3973551637279597,
      "grad_norm": 0.12932024896144867,
      "learning_rate": 7.541137490871559e-05,
      "loss": 0.1822,
      "step": 1893
    },
    {
      "epoch": 0.3975650713685978,
      "grad_norm": 0.1031857281923294,
      "learning_rate": 7.537981208874002e-05,
      "loss": 0.1624,
      "step": 1894
    },
    {
      "epoch": 0.3977749790092359,
      "grad_norm": 0.10605569183826447,
      "learning_rate": 7.534823563922558e-05,
      "loss": 0.1553,
      "step": 1895
    },
    {
      "epoch": 0.3979848866498741,
      "grad_norm": 0.11709260195493698,
      "learning_rate": 7.531664557712962e-05,
      "loss": 0.1636,
      "step": 1896
    },
    {
      "epoch": 0.3981947942905122,
      "grad_norm": 0.12204380333423615,
      "learning_rate": 7.528504191941668e-05,
      "loss": 0.1446,
      "step": 1897
    },
    {
      "epoch": 0.3984047019311503,
      "grad_norm": 0.1183619275689125,
      "learning_rate": 7.525342468305865e-05,
      "loss": 0.1715,
      "step": 1898
    },
    {
      "epoch": 0.3986146095717884,
      "grad_norm": 0.15471158921718597,
      "learning_rate": 7.522179388503471e-05,
      "loss": 0.1648,
      "step": 1899
    },
    {
      "epoch": 0.3988245172124265,
      "grad_norm": 0.11495169252157211,
      "learning_rate": 7.519014954233136e-05,
      "loss": 0.1752,
      "step": 1900
    },
    {
      "epoch": 0.3990344248530647,
      "grad_norm": 0.1107286810874939,
      "learning_rate": 7.515849167194227e-05,
      "loss": 0.1535,
      "step": 1901
    },
    {
      "epoch": 0.3992443324937028,
      "grad_norm": 0.13978683948516846,
      "learning_rate": 7.51268202908685e-05,
      "loss": 0.1535,
      "step": 1902
    },
    {
      "epoch": 0.3994542401343409,
      "grad_norm": 0.13453927636146545,
      "learning_rate": 7.509513541611826e-05,
      "loss": 0.154,
      "step": 1903
    },
    {
      "epoch": 0.399664147774979,
      "grad_norm": 0.14580851793289185,
      "learning_rate": 7.506343706470707e-05,
      "loss": 0.1758,
      "step": 1904
    },
    {
      "epoch": 0.3998740554156171,
      "grad_norm": 0.14356370270252228,
      "learning_rate": 7.503172525365766e-05,
      "loss": 0.1781,
      "step": 1905
    },
    {
      "epoch": 0.40008396305625527,
      "grad_norm": 0.1313529908657074,
      "learning_rate": 7.500000000000001e-05,
      "loss": 0.1627,
      "step": 1906
    },
    {
      "epoch": 0.4002938706968934,
      "grad_norm": 0.11975359171628952,
      "learning_rate": 7.49682613207713e-05,
      "loss": 0.1542,
      "step": 1907
    },
    {
      "epoch": 0.4005037783375315,
      "grad_norm": 0.1087488904595375,
      "learning_rate": 7.493650923301592e-05,
      "loss": 0.1641,
      "step": 1908
    },
    {
      "epoch": 0.4007136859781696,
      "grad_norm": 0.1269085705280304,
      "learning_rate": 7.490474375378546e-05,
      "loss": 0.1627,
      "step": 1909
    },
    {
      "epoch": 0.40092359361880775,
      "grad_norm": 0.1457659900188446,
      "learning_rate": 7.487296490013875e-05,
      "loss": 0.1808,
      "step": 1910
    },
    {
      "epoch": 0.40113350125944586,
      "grad_norm": 0.1324755698442459,
      "learning_rate": 7.484117268914171e-05,
      "loss": 0.1625,
      "step": 1911
    },
    {
      "epoch": 0.40134340890008396,
      "grad_norm": 0.14338278770446777,
      "learning_rate": 7.48093671378675e-05,
      "loss": 0.1551,
      "step": 1912
    },
    {
      "epoch": 0.40155331654072207,
      "grad_norm": 0.13183006644248962,
      "learning_rate": 7.477754826339645e-05,
      "loss": 0.1611,
      "step": 1913
    },
    {
      "epoch": 0.4017632241813602,
      "grad_norm": 0.11733553558588028,
      "learning_rate": 7.474571608281599e-05,
      "loss": 0.1514,
      "step": 1914
    },
    {
      "epoch": 0.40197313182199834,
      "grad_norm": 0.12164483219385147,
      "learning_rate": 7.471387061322074e-05,
      "loss": 0.1609,
      "step": 1915
    },
    {
      "epoch": 0.40218303946263645,
      "grad_norm": 0.1412167251110077,
      "learning_rate": 7.468201187171245e-05,
      "loss": 0.1594,
      "step": 1916
    },
    {
      "epoch": 0.40239294710327456,
      "grad_norm": 0.14494703710079193,
      "learning_rate": 7.46501398754e-05,
      "loss": 0.1702,
      "step": 1917
    },
    {
      "epoch": 0.40260285474391266,
      "grad_norm": 0.1307775378227234,
      "learning_rate": 7.461825464139934e-05,
      "loss": 0.1626,
      "step": 1918
    },
    {
      "epoch": 0.40281276238455077,
      "grad_norm": 0.11550642549991608,
      "learning_rate": 7.458635618683362e-05,
      "loss": 0.1708,
      "step": 1919
    },
    {
      "epoch": 0.40302267002518893,
      "grad_norm": 0.11663535982370377,
      "learning_rate": 7.4554444528833e-05,
      "loss": 0.1467,
      "step": 1920
    },
    {
      "epoch": 0.40323257766582704,
      "grad_norm": 0.11326033622026443,
      "learning_rate": 7.452251968453481e-05,
      "loss": 0.1555,
      "step": 1921
    },
    {
      "epoch": 0.40344248530646515,
      "grad_norm": 0.10349076241254807,
      "learning_rate": 7.449058167108337e-05,
      "loss": 0.1725,
      "step": 1922
    },
    {
      "epoch": 0.40365239294710326,
      "grad_norm": 0.12648595869541168,
      "learning_rate": 7.445863050563017e-05,
      "loss": 0.1535,
      "step": 1923
    },
    {
      "epoch": 0.4038623005877414,
      "grad_norm": 0.12781694531440735,
      "learning_rate": 7.442666620533368e-05,
      "loss": 0.1718,
      "step": 1924
    },
    {
      "epoch": 0.4040722082283795,
      "grad_norm": 0.14075273275375366,
      "learning_rate": 7.439468878735947e-05,
      "loss": 0.1496,
      "step": 1925
    },
    {
      "epoch": 0.40428211586901763,
      "grad_norm": 0.10260710120201111,
      "learning_rate": 7.436269826888015e-05,
      "loss": 0.1569,
      "step": 1926
    },
    {
      "epoch": 0.40449202350965574,
      "grad_norm": 0.15809062123298645,
      "learning_rate": 7.433069466707537e-05,
      "loss": 0.1714,
      "step": 1927
    },
    {
      "epoch": 0.40470193115029385,
      "grad_norm": 0.12049717456102371,
      "learning_rate": 7.429867799913177e-05,
      "loss": 0.1624,
      "step": 1928
    },
    {
      "epoch": 0.404911838790932,
      "grad_norm": 0.11036177724599838,
      "learning_rate": 7.426664828224304e-05,
      "loss": 0.1672,
      "step": 1929
    },
    {
      "epoch": 0.4051217464315701,
      "grad_norm": 0.12535598874092102,
      "learning_rate": 7.423460553360988e-05,
      "loss": 0.1677,
      "step": 1930
    },
    {
      "epoch": 0.4053316540722082,
      "grad_norm": 0.11823215335607529,
      "learning_rate": 7.420254977043998e-05,
      "loss": 0.159,
      "step": 1931
    },
    {
      "epoch": 0.40554156171284633,
      "grad_norm": 0.13293218612670898,
      "learning_rate": 7.4170481009948e-05,
      "loss": 0.1688,
      "step": 1932
    },
    {
      "epoch": 0.40575146935348444,
      "grad_norm": 0.11967580020427704,
      "learning_rate": 7.41383992693556e-05,
      "loss": 0.1649,
      "step": 1933
    },
    {
      "epoch": 0.4059613769941226,
      "grad_norm": 0.12903308868408203,
      "learning_rate": 7.410630456589141e-05,
      "loss": 0.1527,
      "step": 1934
    },
    {
      "epoch": 0.4061712846347607,
      "grad_norm": 0.13130050897598267,
      "learning_rate": 7.407419691679104e-05,
      "loss": 0.1607,
      "step": 1935
    },
    {
      "epoch": 0.4063811922753988,
      "grad_norm": 0.12664926052093506,
      "learning_rate": 7.404207633929699e-05,
      "loss": 0.151,
      "step": 1936
    },
    {
      "epoch": 0.4065910999160369,
      "grad_norm": 0.11870288848876953,
      "learning_rate": 7.400994285065878e-05,
      "loss": 0.151,
      "step": 1937
    },
    {
      "epoch": 0.4068010075566751,
      "grad_norm": 0.10910684615373611,
      "learning_rate": 7.397779646813282e-05,
      "loss": 0.1393,
      "step": 1938
    },
    {
      "epoch": 0.4070109151973132,
      "grad_norm": 0.13992692530155182,
      "learning_rate": 7.394563720898245e-05,
      "loss": 0.18,
      "step": 1939
    },
    {
      "epoch": 0.4072208228379513,
      "grad_norm": 0.1333228498697281,
      "learning_rate": 7.39134650904779e-05,
      "loss": 0.1536,
      "step": 1940
    },
    {
      "epoch": 0.4074307304785894,
      "grad_norm": 0.11630193144083023,
      "learning_rate": 7.388128012989636e-05,
      "loss": 0.1603,
      "step": 1941
    },
    {
      "epoch": 0.4076406381192275,
      "grad_norm": 0.12996290624141693,
      "learning_rate": 7.38490823445219e-05,
      "loss": 0.1532,
      "step": 1942
    },
    {
      "epoch": 0.4078505457598657,
      "grad_norm": 0.12526121735572815,
      "learning_rate": 7.381687175164544e-05,
      "loss": 0.1684,
      "step": 1943
    },
    {
      "epoch": 0.4080604534005038,
      "grad_norm": 0.13063152134418488,
      "learning_rate": 7.378464836856481e-05,
      "loss": 0.1637,
      "step": 1944
    },
    {
      "epoch": 0.4082703610411419,
      "grad_norm": 0.11410169303417206,
      "learning_rate": 7.375241221258471e-05,
      "loss": 0.1621,
      "step": 1945
    },
    {
      "epoch": 0.40848026868178,
      "grad_norm": 0.1353740096092224,
      "learning_rate": 7.37201633010167e-05,
      "loss": 0.1274,
      "step": 1946
    },
    {
      "epoch": 0.4086901763224181,
      "grad_norm": 0.1723635345697403,
      "learning_rate": 7.368790165117916e-05,
      "loss": 0.1589,
      "step": 1947
    },
    {
      "epoch": 0.40890008396305627,
      "grad_norm": 0.111173614859581,
      "learning_rate": 7.365562728039734e-05,
      "loss": 0.1626,
      "step": 1948
    },
    {
      "epoch": 0.4091099916036944,
      "grad_norm": 0.10644244402647018,
      "learning_rate": 7.362334020600334e-05,
      "loss": 0.1709,
      "step": 1949
    },
    {
      "epoch": 0.4093198992443325,
      "grad_norm": 0.12528212368488312,
      "learning_rate": 7.359104044533601e-05,
      "loss": 0.1688,
      "step": 1950
    },
    {
      "epoch": 0.4095298068849706,
      "grad_norm": 0.11388932913541794,
      "learning_rate": 7.35587280157411e-05,
      "loss": 0.1624,
      "step": 1951
    },
    {
      "epoch": 0.40973971452560876,
      "grad_norm": 0.11479576677083969,
      "learning_rate": 7.35264029345711e-05,
      "loss": 0.1728,
      "step": 1952
    },
    {
      "epoch": 0.40994962216624686,
      "grad_norm": 0.11558470875024796,
      "learning_rate": 7.349406521918533e-05,
      "loss": 0.1286,
      "step": 1953
    },
    {
      "epoch": 0.41015952980688497,
      "grad_norm": 0.12116646766662598,
      "learning_rate": 7.346171488694988e-05,
      "loss": 0.1704,
      "step": 1954
    },
    {
      "epoch": 0.4103694374475231,
      "grad_norm": 0.12693926692008972,
      "learning_rate": 7.34293519552376e-05,
      "loss": 0.165,
      "step": 1955
    },
    {
      "epoch": 0.4105793450881612,
      "grad_norm": 0.11315865814685822,
      "learning_rate": 7.339697644142815e-05,
      "loss": 0.1562,
      "step": 1956
    },
    {
      "epoch": 0.41078925272879935,
      "grad_norm": 0.13625770807266235,
      "learning_rate": 7.33645883629079e-05,
      "loss": 0.1792,
      "step": 1957
    },
    {
      "epoch": 0.41099916036943746,
      "grad_norm": 0.11685556173324585,
      "learning_rate": 7.333218773707e-05,
      "loss": 0.1626,
      "step": 1958
    },
    {
      "epoch": 0.41120906801007556,
      "grad_norm": 0.13109751045703888,
      "learning_rate": 7.329977458131431e-05,
      "loss": 0.1563,
      "step": 1959
    },
    {
      "epoch": 0.41141897565071367,
      "grad_norm": 0.1340368390083313,
      "learning_rate": 7.326734891304745e-05,
      "loss": 0.1559,
      "step": 1960
    },
    {
      "epoch": 0.41162888329135183,
      "grad_norm": 0.12216036021709442,
      "learning_rate": 7.323491074968275e-05,
      "loss": 0.1604,
      "step": 1961
    },
    {
      "epoch": 0.41183879093198994,
      "grad_norm": 0.11884403228759766,
      "learning_rate": 7.320246010864023e-05,
      "loss": 0.1572,
      "step": 1962
    },
    {
      "epoch": 0.41204869857262805,
      "grad_norm": 0.11110524088144302,
      "learning_rate": 7.316999700734663e-05,
      "loss": 0.1606,
      "step": 1963
    },
    {
      "epoch": 0.41225860621326615,
      "grad_norm": 0.12798446416854858,
      "learning_rate": 7.313752146323539e-05,
      "loss": 0.1625,
      "step": 1964
    },
    {
      "epoch": 0.41246851385390426,
      "grad_norm": 0.11968263983726501,
      "learning_rate": 7.31050334937466e-05,
      "loss": 0.1749,
      "step": 1965
    },
    {
      "epoch": 0.4126784214945424,
      "grad_norm": 0.10930939763784409,
      "learning_rate": 7.307253311632704e-05,
      "loss": 0.1722,
      "step": 1966
    },
    {
      "epoch": 0.41288832913518053,
      "grad_norm": 0.12315979599952698,
      "learning_rate": 7.304002034843015e-05,
      "loss": 0.1776,
      "step": 1967
    },
    {
      "epoch": 0.41309823677581864,
      "grad_norm": 0.1320822834968567,
      "learning_rate": 7.300749520751608e-05,
      "loss": 0.1539,
      "step": 1968
    },
    {
      "epoch": 0.41330814441645675,
      "grad_norm": 0.12075715512037277,
      "learning_rate": 7.297495771105152e-05,
      "loss": 0.1414,
      "step": 1969
    },
    {
      "epoch": 0.41351805205709485,
      "grad_norm": 0.1056932806968689,
      "learning_rate": 7.294240787650986e-05,
      "loss": 0.1485,
      "step": 1970
    },
    {
      "epoch": 0.413727959697733,
      "grad_norm": 0.10487929731607437,
      "learning_rate": 7.290984572137111e-05,
      "loss": 0.1728,
      "step": 1971
    },
    {
      "epoch": 0.4139378673383711,
      "grad_norm": 0.10982023179531097,
      "learning_rate": 7.287727126312191e-05,
      "loss": 0.1608,
      "step": 1972
    },
    {
      "epoch": 0.41414777497900923,
      "grad_norm": 0.09598106890916824,
      "learning_rate": 7.284468451925546e-05,
      "loss": 0.1503,
      "step": 1973
    },
    {
      "epoch": 0.41435768261964734,
      "grad_norm": 0.14356835186481476,
      "learning_rate": 7.281208550727159e-05,
      "loss": 0.1495,
      "step": 1974
    },
    {
      "epoch": 0.4145675902602855,
      "grad_norm": 0.10420846194028854,
      "learning_rate": 7.277947424467672e-05,
      "loss": 0.1586,
      "step": 1975
    },
    {
      "epoch": 0.4147774979009236,
      "grad_norm": 0.12487095594406128,
      "learning_rate": 7.274685074898386e-05,
      "loss": 0.175,
      "step": 1976
    },
    {
      "epoch": 0.4149874055415617,
      "grad_norm": 0.11194780468940735,
      "learning_rate": 7.271421503771251e-05,
      "loss": 0.1574,
      "step": 1977
    },
    {
      "epoch": 0.4151973131821998,
      "grad_norm": 0.12557700276374817,
      "learning_rate": 7.268156712838888e-05,
      "loss": 0.1529,
      "step": 1978
    },
    {
      "epoch": 0.41540722082283793,
      "grad_norm": 0.10330946743488312,
      "learning_rate": 7.264890703854559e-05,
      "loss": 0.1311,
      "step": 1979
    },
    {
      "epoch": 0.4156171284634761,
      "grad_norm": 0.11576497554779053,
      "learning_rate": 7.261623478572185e-05,
      "loss": 0.1518,
      "step": 1980
    },
    {
      "epoch": 0.4158270361041142,
      "grad_norm": 0.1512029618024826,
      "learning_rate": 7.258355038746345e-05,
      "loss": 0.1629,
      "step": 1981
    },
    {
      "epoch": 0.4160369437447523,
      "grad_norm": 0.142082080245018,
      "learning_rate": 7.25508538613226e-05,
      "loss": 0.1607,
      "step": 1982
    },
    {
      "epoch": 0.4162468513853904,
      "grad_norm": 0.13623762130737305,
      "learning_rate": 7.251814522485813e-05,
      "loss": 0.1695,
      "step": 1983
    },
    {
      "epoch": 0.4164567590260285,
      "grad_norm": 0.1192348450422287,
      "learning_rate": 7.248542449563529e-05,
      "loss": 0.1668,
      "step": 1984
    },
    {
      "epoch": 0.4166666666666667,
      "grad_norm": 0.12478368729352951,
      "learning_rate": 7.245269169122588e-05,
      "loss": 0.1427,
      "step": 1985
    },
    {
      "epoch": 0.4168765743073048,
      "grad_norm": 0.1585802137851715,
      "learning_rate": 7.241994682920816e-05,
      "loss": 0.1778,
      "step": 1986
    },
    {
      "epoch": 0.4170864819479429,
      "grad_norm": 0.10179055482149124,
      "learning_rate": 7.238718992716687e-05,
      "loss": 0.1358,
      "step": 1987
    },
    {
      "epoch": 0.417296389588581,
      "grad_norm": 0.11688593029975891,
      "learning_rate": 7.23544210026932e-05,
      "loss": 0.1603,
      "step": 1988
    },
    {
      "epoch": 0.41750629722921917,
      "grad_norm": 0.1158967837691307,
      "learning_rate": 7.232164007338484e-05,
      "loss": 0.1408,
      "step": 1989
    },
    {
      "epoch": 0.4177162048698573,
      "grad_norm": 0.1265357881784439,
      "learning_rate": 7.228884715684588e-05,
      "loss": 0.1402,
      "step": 1990
    },
    {
      "epoch": 0.4179261125104954,
      "grad_norm": 0.10598758608102798,
      "learning_rate": 7.225604227068684e-05,
      "loss": 0.1431,
      "step": 1991
    },
    {
      "epoch": 0.4181360201511335,
      "grad_norm": 0.12342094630002975,
      "learning_rate": 7.222322543252474e-05,
      "loss": 0.1539,
      "step": 1992
    },
    {
      "epoch": 0.4183459277917716,
      "grad_norm": 0.12338914722204208,
      "learning_rate": 7.219039665998295e-05,
      "loss": 0.1719,
      "step": 1993
    },
    {
      "epoch": 0.41855583543240976,
      "grad_norm": 0.11439438164234161,
      "learning_rate": 7.215755597069126e-05,
      "loss": 0.1639,
      "step": 1994
    },
    {
      "epoch": 0.41876574307304787,
      "grad_norm": 0.12324292212724686,
      "learning_rate": 7.212470338228589e-05,
      "loss": 0.1759,
      "step": 1995
    },
    {
      "epoch": 0.418975650713686,
      "grad_norm": 0.10522071272134781,
      "learning_rate": 7.209183891240941e-05,
      "loss": 0.1475,
      "step": 1996
    },
    {
      "epoch": 0.4191855583543241,
      "grad_norm": 0.09488867968320847,
      "learning_rate": 7.205896257871082e-05,
      "loss": 0.1564,
      "step": 1997
    },
    {
      "epoch": 0.4193954659949622,
      "grad_norm": 0.12088136374950409,
      "learning_rate": 7.202607439884543e-05,
      "loss": 0.1671,
      "step": 1998
    },
    {
      "epoch": 0.41960537363560035,
      "grad_norm": 0.11763715744018555,
      "learning_rate": 7.199317439047499e-05,
      "loss": 0.1463,
      "step": 1999
    },
    {
      "epoch": 0.41981528127623846,
      "grad_norm": 0.10898829251527786,
      "learning_rate": 7.196026257126749e-05,
      "loss": 0.1697,
      "step": 2000
    }
  ],
  "logging_steps": 1.0,
  "max_steps": 4764,
  "num_input_tokens_seen": 0,
  "num_train_epochs": 1,
  "save_steps": 1000,
  "stateful_callbacks": {
    "TrainerControl": {
      "args": {
        "should_epoch_stop": false,
        "should_evaluate": false,
        "should_log": false,
        "should_save": true,
        "should_training_stop": false
      },
      "attributes": {}
    }
  },
  "total_flos": 1.429436679625638e+18,
  "train_batch_size": 1,
  "trial_name": null,
  "trial_params": null
}