{
  "best_metric": null,
  "best_model_checkpoint": null,
  "epoch": 2.6755852842809364,
  "eval_steps": 500,
  "global_step": 2000,
  "is_hyper_param_search": false,
  "is_local_process_zero": true,
  "is_world_process_zero": true,
  "log_history": [
    {
      "epoch": 0.0013377926421404682,
      "grad_norm": 8.909825325012207,
      "learning_rate": 5e-06,
      "loss": 2.9362,
      "step": 1
    },
    {
      "epoch": 0.0026755852842809363,
      "grad_norm": 11.420860290527344,
      "learning_rate": 1e-05,
      "loss": 3.1217,
      "step": 2
    },
    {
      "epoch": 0.004013377926421404,
      "grad_norm": 9.437948226928711,
      "learning_rate": 1.5e-05,
      "loss": 2.9222,
      "step": 3
    },
    {
      "epoch": 0.005351170568561873,
      "grad_norm": 6.445765972137451,
      "learning_rate": 2e-05,
      "loss": 3.0696,
      "step": 4
    },
    {
      "epoch": 0.006688963210702341,
      "grad_norm": 4.019280433654785,
      "learning_rate": 2.5e-05,
      "loss": 2.8349,
      "step": 5
    },
    {
      "epoch": 0.008026755852842809,
      "grad_norm": 3.1477231979370117,
      "learning_rate": 3e-05,
      "loss": 2.5674,
      "step": 6
    },
    {
      "epoch": 0.009364548494983277,
      "grad_norm": 6.760244846343994,
      "learning_rate": 3.5e-05,
      "loss": 2.4331,
      "step": 7
    },
    {
      "epoch": 0.010702341137123745,
      "grad_norm": 2.688317060470581,
      "learning_rate": 4e-05,
      "loss": 2.5435,
      "step": 8
    },
    {
      "epoch": 0.012040133779264214,
      "grad_norm": 2.7984418869018555,
      "learning_rate": 4.5e-05,
      "loss": 2.2427,
      "step": 9
    },
    {
      "epoch": 0.013377926421404682,
      "grad_norm": 8.174599647521973,
      "learning_rate": 5e-05,
      "loss": 2.7228,
      "step": 10
    },
    {
      "epoch": 0.01471571906354515,
      "grad_norm": 1.9308923482894897,
      "learning_rate": 5.500000000000001e-05,
      "loss": 2.6318,
      "step": 11
    },
    {
      "epoch": 0.016053511705685617,
      "grad_norm": 2.391382932662964,
      "learning_rate": 6e-05,
      "loss": 2.1344,
      "step": 12
    },
    {
      "epoch": 0.017391304347826087,
      "grad_norm": 3.2888343334198,
      "learning_rate": 6.500000000000001e-05,
      "loss": 2.3624,
      "step": 13
    },
    {
      "epoch": 0.018729096989966554,
      "grad_norm": 2.822356939315796,
      "learning_rate": 7e-05,
      "loss": 2.1097,
      "step": 14
    },
    {
      "epoch": 0.020066889632107024,
      "grad_norm": 2.7134530544281006,
      "learning_rate": 7.500000000000001e-05,
      "loss": 2.2967,
      "step": 15
    },
    {
      "epoch": 0.02140468227424749,
      "grad_norm": 1.22122323513031,
      "learning_rate": 8e-05,
      "loss": 2.2179,
      "step": 16
    },
    {
      "epoch": 0.02274247491638796,
      "grad_norm": 1.7385879755020142,
      "learning_rate": 8.5e-05,
      "loss": 2.1567,
      "step": 17
    },
    {
      "epoch": 0.024080267558528427,
      "grad_norm": 1.5407123565673828,
      "learning_rate": 9e-05,
      "loss": 2.1197,
      "step": 18
    },
    {
      "epoch": 0.025418060200668897,
      "grad_norm": 1.191502571105957,
      "learning_rate": 9.5e-05,
      "loss": 2.2177,
      "step": 19
    },
    {
      "epoch": 0.026755852842809364,
      "grad_norm": 1.3412432670593262,
      "learning_rate": 0.0001,
      "loss": 2.0945,
      "step": 20
    },
    {
      "epoch": 0.028093645484949834,
      "grad_norm": 1.3845086097717285,
      "learning_rate": 9.995497523638001e-05,
      "loss": 2.03,
      "step": 21
    },
    {
      "epoch": 0.0294314381270903,
      "grad_norm": 1.0299506187438965,
      "learning_rate": 9.990995047276002e-05,
      "loss": 2.0461,
      "step": 22
    },
    {
      "epoch": 0.03076923076923077,
      "grad_norm": 1.3790980577468872,
      "learning_rate": 9.986492570914003e-05,
      "loss": 2.4299,
      "step": 23
    },
    {
      "epoch": 0.032107023411371234,
      "grad_norm": 1.0298329591751099,
      "learning_rate": 9.981990094552004e-05,
      "loss": 2.3317,
      "step": 24
    },
    {
      "epoch": 0.033444816053511704,
      "grad_norm": 1.129258155822754,
      "learning_rate": 9.977487618190005e-05,
      "loss": 2.0976,
      "step": 25
    },
    {
      "epoch": 0.034782608695652174,
      "grad_norm": 2.688994884490967,
      "learning_rate": 9.972985141828006e-05,
      "loss": 1.8448,
      "step": 26
    },
    {
      "epoch": 0.036120401337792644,
      "grad_norm": 0.8690654039382935,
      "learning_rate": 9.968482665466006e-05,
      "loss": 1.9989,
      "step": 27
    },
    {
      "epoch": 0.03745819397993311,
      "grad_norm": 1.0267047882080078,
      "learning_rate": 9.963980189104007e-05,
      "loss": 2.2423,
      "step": 28
    },
    {
      "epoch": 0.03879598662207358,
      "grad_norm": 0.9660020470619202,
      "learning_rate": 9.95947771274201e-05,
      "loss": 2.3757,
      "step": 29
    },
    {
      "epoch": 0.04013377926421405,
      "grad_norm": 0.8873792886734009,
      "learning_rate": 9.954975236380009e-05,
      "loss": 2.1465,
      "step": 30
    },
    {
      "epoch": 0.04147157190635452,
      "grad_norm": 4.266314506530762,
      "learning_rate": 9.95047276001801e-05,
      "loss": 2.1787,
      "step": 31
    },
    {
      "epoch": 0.04280936454849498,
      "grad_norm": 0.9623901844024658,
      "learning_rate": 9.945970283656011e-05,
      "loss": 2.1484,
      "step": 32
    },
    {
      "epoch": 0.04414715719063545,
      "grad_norm": 1.0520215034484863,
      "learning_rate": 9.941467807294013e-05,
      "loss": 2.2169,
      "step": 33
    },
    {
      "epoch": 0.04548494983277592,
      "grad_norm": 1.2446262836456299,
      "learning_rate": 9.936965330932014e-05,
      "loss": 2.4512,
      "step": 34
    },
    {
      "epoch": 0.046822742474916385,
      "grad_norm": 8.27096176147461,
      "learning_rate": 9.932462854570013e-05,
      "loss": 2.3712,
      "step": 35
    },
    {
      "epoch": 0.048160535117056855,
      "grad_norm": 1.682218074798584,
      "learning_rate": 9.927960378208014e-05,
      "loss": 2.1458,
      "step": 36
    },
    {
      "epoch": 0.049498327759197325,
      "grad_norm": 2.1129493713378906,
      "learning_rate": 9.923457901846016e-05,
      "loss": 2.0079,
      "step": 37
    },
    {
      "epoch": 0.050836120401337795,
      "grad_norm": 2.0923821926116943,
      "learning_rate": 9.918955425484017e-05,
      "loss": 1.9663,
      "step": 38
    },
    {
      "epoch": 0.05217391304347826,
      "grad_norm": 1.2081389427185059,
      "learning_rate": 9.914452949122017e-05,
      "loss": 1.9386,
      "step": 39
    },
    {
      "epoch": 0.05351170568561873,
      "grad_norm": 1.0379135608673096,
      "learning_rate": 9.909950472760019e-05,
      "loss": 2.0566,
      "step": 40
    },
    {
      "epoch": 0.0548494983277592,
      "grad_norm": 0.940050482749939,
      "learning_rate": 9.90544799639802e-05,
      "loss": 2.1417,
      "step": 41
    },
    {
      "epoch": 0.05618729096989967,
      "grad_norm": 0.8876029849052429,
      "learning_rate": 9.900945520036021e-05,
      "loss": 1.8754,
      "step": 42
    },
    {
      "epoch": 0.05752508361204013,
      "grad_norm": 0.8591219186782837,
      "learning_rate": 9.89644304367402e-05,
      "loss": 2.4408,
      "step": 43
    },
    {
      "epoch": 0.0588628762541806,
      "grad_norm": 1.1327685117721558,
      "learning_rate": 9.891940567312022e-05,
      "loss": 1.9898,
      "step": 44
    },
    {
      "epoch": 0.06020066889632107,
      "grad_norm": 0.9538917541503906,
      "learning_rate": 9.887438090950023e-05,
      "loss": 2.2119,
      "step": 45
    },
    {
      "epoch": 0.06153846153846154,
      "grad_norm": 1.2696901559829712,
      "learning_rate": 9.882935614588024e-05,
      "loss": 2.3051,
      "step": 46
    },
    {
      "epoch": 0.06287625418060201,
      "grad_norm": 0.92132169008255,
      "learning_rate": 9.878433138226025e-05,
      "loss": 2.4412,
      "step": 47
    },
    {
      "epoch": 0.06421404682274247,
      "grad_norm": 0.8168394565582275,
      "learning_rate": 9.873930661864026e-05,
      "loss": 2.0003,
      "step": 48
    },
    {
      "epoch": 0.06555183946488294,
      "grad_norm": 0.9975659251213074,
      "learning_rate": 9.869428185502027e-05,
      "loss": 2.3379,
      "step": 49
    },
    {
      "epoch": 0.06688963210702341,
      "grad_norm": 0.8086885213851929,
      "learning_rate": 9.864925709140028e-05,
      "loss": 2.0088,
      "step": 50
    },
    {
      "epoch": 0.06822742474916388,
      "grad_norm": 3.218966007232666,
      "learning_rate": 9.860423232778028e-05,
      "loss": 2.2297,
      "step": 51
    },
    {
      "epoch": 0.06956521739130435,
      "grad_norm": 0.8951109647750854,
      "learning_rate": 9.855920756416029e-05,
      "loss": 2.2432,
      "step": 52
    },
    {
      "epoch": 0.07090301003344482,
      "grad_norm": 0.8055227994918823,
      "learning_rate": 9.85141828005403e-05,
      "loss": 2.2097,
      "step": 53
    },
    {
      "epoch": 0.07224080267558529,
      "grad_norm": 0.7354897856712341,
      "learning_rate": 9.846915803692031e-05,
      "loss": 2.1813,
      "step": 54
    },
    {
      "epoch": 0.07357859531772576,
      "grad_norm": 1.6674742698669434,
      "learning_rate": 9.842413327330032e-05,
      "loss": 2.1639,
      "step": 55
    },
    {
      "epoch": 0.07491638795986622,
      "grad_norm": 30.052200317382812,
      "learning_rate": 9.837910850968033e-05,
      "loss": 2.9772,
      "step": 56
    },
    {
      "epoch": 0.07625418060200669,
      "grad_norm": 1.2769989967346191,
      "learning_rate": 9.833408374606034e-05,
      "loss": 2.1162,
      "step": 57
    },
    {
      "epoch": 0.07759197324414716,
      "grad_norm": 0.7706617116928101,
      "learning_rate": 9.828905898244036e-05,
      "loss": 2.2085,
      "step": 58
    },
    {
      "epoch": 0.07892976588628763,
      "grad_norm": 1.0603933334350586,
      "learning_rate": 9.824403421882035e-05,
      "loss": 1.5831,
      "step": 59
    },
    {
      "epoch": 0.0802675585284281,
      "grad_norm": 3.992206573486328,
      "learning_rate": 9.819900945520036e-05,
      "loss": 2.5495,
      "step": 60
    },
    {
      "epoch": 0.08160535117056857,
      "grad_norm": 0.6373528242111206,
      "learning_rate": 9.815398469158037e-05,
      "loss": 1.9227,
      "step": 61
    },
    {
      "epoch": 0.08294314381270904,
      "grad_norm": 0.8140257000923157,
      "learning_rate": 9.810895992796039e-05,
      "loss": 2.1962,
      "step": 62
    },
    {
      "epoch": 0.08428093645484949,
      "grad_norm": 1.3295791149139404,
      "learning_rate": 9.806393516434039e-05,
      "loss": 2.1865,
      "step": 63
    },
    {
      "epoch": 0.08561872909698996,
      "grad_norm": 1.176526427268982,
      "learning_rate": 9.80189104007204e-05,
      "loss": 2.0073,
      "step": 64
    },
    {
      "epoch": 0.08695652173913043,
      "grad_norm": 0.8747398257255554,
      "learning_rate": 9.797388563710042e-05,
      "loss": 2.332,
      "step": 65
    },
    {
      "epoch": 0.0882943143812709,
      "grad_norm": 0.7917225956916809,
      "learning_rate": 9.792886087348043e-05,
      "loss": 2.1264,
      "step": 66
    },
    {
      "epoch": 0.08963210702341137,
      "grad_norm": 0.6571652889251709,
      "learning_rate": 9.788383610986042e-05,
      "loss": 2.1755,
      "step": 67
    },
    {
      "epoch": 0.09096989966555184,
      "grad_norm": 0.7766551971435547,
      "learning_rate": 9.783881134624043e-05,
      "loss": 2.0205,
      "step": 68
    },
    {
      "epoch": 0.09230769230769231,
      "grad_norm": 0.8635064363479614,
      "learning_rate": 9.779378658262045e-05,
      "loss": 1.6129,
      "step": 69
    },
    {
      "epoch": 0.09364548494983277,
      "grad_norm": 1.7755194902420044,
      "learning_rate": 9.774876181900046e-05,
      "loss": 2.2523,
      "step": 70
    },
    {
      "epoch": 0.09498327759197324,
      "grad_norm": 0.8652092814445496,
      "learning_rate": 9.770373705538046e-05,
      "loss": 2.1025,
      "step": 71
    },
    {
      "epoch": 0.09632107023411371,
      "grad_norm": 0.7990190982818604,
      "learning_rate": 9.765871229176046e-05,
      "loss": 2.1477,
      "step": 72
    },
    {
      "epoch": 0.09765886287625418,
      "grad_norm": 3.000373125076294,
      "learning_rate": 9.761368752814049e-05,
      "loss": 1.8605,
      "step": 73
    },
    {
      "epoch": 0.09899665551839465,
      "grad_norm": 0.7712721228599548,
      "learning_rate": 9.75686627645205e-05,
      "loss": 2.2457,
      "step": 74
    },
    {
      "epoch": 0.10033444816053512,
      "grad_norm": 0.6429623365402222,
      "learning_rate": 9.752363800090049e-05,
      "loss": 2.1461,
      "step": 75
    },
    {
      "epoch": 0.10167224080267559,
      "grad_norm": 0.8219184875488281,
      "learning_rate": 9.747861323728051e-05,
      "loss": 2.1473,
      "step": 76
    },
    {
      "epoch": 0.10301003344481606,
      "grad_norm": 0.8963313102722168,
      "learning_rate": 9.743358847366052e-05,
      "loss": 1.9817,
      "step": 77
    },
    {
      "epoch": 0.10434782608695652,
      "grad_norm": 1.6411892175674438,
      "learning_rate": 9.738856371004053e-05,
      "loss": 1.8849,
      "step": 78
    },
    {
      "epoch": 0.10568561872909699,
      "grad_norm": 0.8379344344139099,
      "learning_rate": 9.734353894642053e-05,
      "loss": 2.226,
      "step": 79
    },
    {
      "epoch": 0.10702341137123746,
      "grad_norm": 1.2867939472198486,
      "learning_rate": 9.729851418280055e-05,
      "loss": 2.3098,
      "step": 80
    },
    {
      "epoch": 0.10836120401337793,
      "grad_norm": 0.8235902190208435,
      "learning_rate": 9.725348941918056e-05,
      "loss": 2.0342,
      "step": 81
    },
    {
      "epoch": 0.1096989966555184,
      "grad_norm": 2.49385142326355,
      "learning_rate": 9.720846465556056e-05,
      "loss": 1.7611,
      "step": 82
    },
    {
      "epoch": 0.11103678929765887,
      "grad_norm": 0.8160635232925415,
      "learning_rate": 9.716343989194057e-05,
      "loss": 2.1493,
      "step": 83
    },
    {
      "epoch": 0.11237458193979934,
      "grad_norm": 0.739782452583313,
      "learning_rate": 9.711841512832058e-05,
      "loss": 2.1616,
      "step": 84
    },
    {
      "epoch": 0.11371237458193979,
      "grad_norm": 2.4206647872924805,
      "learning_rate": 9.707339036470059e-05,
      "loss": 2.0272,
      "step": 85
    },
    {
      "epoch": 0.11505016722408026,
      "grad_norm": 1.1969271898269653,
      "learning_rate": 9.70283656010806e-05,
      "loss": 2.2563,
      "step": 86
    },
    {
      "epoch": 0.11638795986622073,
      "grad_norm": 1.1001033782958984,
      "learning_rate": 9.698334083746061e-05,
      "loss": 2.2338,
      "step": 87
    },
    {
      "epoch": 0.1177257525083612,
      "grad_norm": 0.660322368144989,
      "learning_rate": 9.693831607384062e-05,
      "loss": 2.1892,
      "step": 88
    },
    {
      "epoch": 0.11906354515050167,
      "grad_norm": 0.7543382048606873,
      "learning_rate": 9.689329131022062e-05,
      "loss": 2.1808,
      "step": 89
    },
    {
      "epoch": 0.12040133779264214,
      "grad_norm": 0.6813179850578308,
      "learning_rate": 9.684826654660063e-05,
      "loss": 2.4539,
      "step": 90
    },
    {
      "epoch": 0.12173913043478261,
      "grad_norm": 0.7714611887931824,
      "learning_rate": 9.680324178298064e-05,
      "loss": 2.1391,
      "step": 91
    },
    {
      "epoch": 0.12307692307692308,
      "grad_norm": 0.8586478233337402,
      "learning_rate": 9.675821701936065e-05,
      "loss": 2.2578,
      "step": 92
    },
    {
      "epoch": 0.12441471571906354,
      "grad_norm": 0.6438114643096924,
      "learning_rate": 9.671319225574066e-05,
      "loss": 1.9993,
      "step": 93
    },
    {
      "epoch": 0.12575250836120402,
      "grad_norm": 0.7894206643104553,
      "learning_rate": 9.666816749212068e-05,
      "loss": 1.7321,
      "step": 94
    },
    {
      "epoch": 0.12709030100334448,
      "grad_norm": 1.143427848815918,
      "learning_rate": 9.662314272850068e-05,
      "loss": 1.93,
      "step": 95
    },
    {
      "epoch": 0.12842809364548494,
      "grad_norm": 0.6587496399879456,
      "learning_rate": 9.657811796488068e-05,
      "loss": 1.6182,
      "step": 96
    },
    {
      "epoch": 0.12976588628762542,
      "grad_norm": 0.8149839639663696,
      "learning_rate": 9.653309320126069e-05,
      "loss": 1.957,
      "step": 97
    },
    {
      "epoch": 0.13110367892976588,
      "grad_norm": 0.6830907464027405,
      "learning_rate": 9.648806843764072e-05,
      "loss": 2.6174,
      "step": 98
    },
    {
      "epoch": 0.13244147157190636,
      "grad_norm": 0.8395177721977234,
      "learning_rate": 9.644304367402071e-05,
      "loss": 2.2876,
      "step": 99
    },
    {
      "epoch": 0.13377926421404682,
      "grad_norm": 0.723379373550415,
      "learning_rate": 9.639801891040072e-05,
      "loss": 1.9577,
      "step": 100
    },
    {
      "epoch": 0.1351170568561873,
      "grad_norm": 1.305256962776184,
      "learning_rate": 9.635299414678074e-05,
      "loss": 1.787,
      "step": 101
    },
    {
      "epoch": 0.13645484949832776,
      "grad_norm": 1.9760712385177612,
      "learning_rate": 9.630796938316075e-05,
      "loss": 1.877,
      "step": 102
    },
    {
      "epoch": 0.13779264214046824,
      "grad_norm": 1.0222994089126587,
      "learning_rate": 9.626294461954074e-05,
      "loss": 2.1143,
      "step": 103
    },
    {
      "epoch": 0.1391304347826087,
      "grad_norm": 1.173228144645691,
      "learning_rate": 9.621791985592075e-05,
      "loss": 2.4299,
      "step": 104
    },
    {
      "epoch": 0.14046822742474915,
      "grad_norm": 0.8381727337837219,
      "learning_rate": 9.617289509230078e-05,
      "loss": 2.0116,
      "step": 105
    },
    {
      "epoch": 0.14180602006688964,
      "grad_norm": 0.8412324786186218,
      "learning_rate": 9.612787032868078e-05,
      "loss": 2.1727,
      "step": 106
    },
    {
      "epoch": 0.1431438127090301,
      "grad_norm": 1.3966891765594482,
      "learning_rate": 9.608284556506079e-05,
      "loss": 1.7429,
      "step": 107
    },
    {
      "epoch": 0.14448160535117058,
      "grad_norm": 0.860106885433197,
      "learning_rate": 9.603782080144079e-05,
      "loss": 2.2786,
      "step": 108
    },
    {
      "epoch": 0.14581939799331103,
      "grad_norm": 2.258957862854004,
      "learning_rate": 9.599279603782081e-05,
      "loss": 1.8781,
      "step": 109
    },
    {
      "epoch": 0.14715719063545152,
      "grad_norm": 0.8375409245491028,
      "learning_rate": 9.594777127420082e-05,
      "loss": 2.0776,
      "step": 110
    },
    {
      "epoch": 0.14849498327759197,
      "grad_norm": 0.6946620345115662,
      "learning_rate": 9.590274651058083e-05,
      "loss": 1.9919,
      "step": 111
    },
    {
      "epoch": 0.14983277591973243,
      "grad_norm": 0.7830926775932312,
      "learning_rate": 9.585772174696084e-05,
      "loss": 2.1972,
      "step": 112
    },
    {
      "epoch": 0.15117056856187291,
      "grad_norm": 0.7598808407783508,
      "learning_rate": 9.581269698334084e-05,
      "loss": 2.2484,
      "step": 113
    },
    {
      "epoch": 0.15250836120401337,
      "grad_norm": 1.3570846319198608,
      "learning_rate": 9.576767221972085e-05,
      "loss": 2.4063,
      "step": 114
    },
    {
      "epoch": 0.15384615384615385,
      "grad_norm": 0.9569585919380188,
      "learning_rate": 9.572264745610086e-05,
      "loss": 2.0047,
      "step": 115
    },
    {
      "epoch": 0.1551839464882943,
      "grad_norm": 0.6475653648376465,
      "learning_rate": 9.567762269248087e-05,
      "loss": 2.1048,
      "step": 116
    },
    {
      "epoch": 0.1565217391304348,
      "grad_norm": 1.0348844528198242,
      "learning_rate": 9.563259792886088e-05,
      "loss": 2.1492,
      "step": 117
    },
    {
      "epoch": 0.15785953177257525,
      "grad_norm": 1.5975306034088135,
      "learning_rate": 9.558757316524089e-05,
      "loss": 2.2867,
      "step": 118
    },
    {
      "epoch": 0.1591973244147157,
      "grad_norm": 0.7629224061965942,
      "learning_rate": 9.55425484016209e-05,
      "loss": 2.5327,
      "step": 119
    },
    {
      "epoch": 0.1605351170568562,
      "grad_norm": 7.820376396179199,
      "learning_rate": 9.54975236380009e-05,
      "loss": 1.9083,
      "step": 120
    },
    {
      "epoch": 0.16187290969899665,
      "grad_norm": 0.7340837121009827,
      "learning_rate": 9.545249887438091e-05,
      "loss": 2.1577,
      "step": 121
    },
    {
      "epoch": 0.16321070234113713,
      "grad_norm": 0.772908627986908,
      "learning_rate": 9.540747411076092e-05,
      "loss": 2.0281,
      "step": 122
    },
    {
      "epoch": 0.1645484949832776,
      "grad_norm": 0.6994909048080444,
      "learning_rate": 9.536244934714093e-05,
      "loss": 1.7673,
      "step": 123
    },
    {
      "epoch": 0.16588628762541807,
      "grad_norm": 4.215782642364502,
      "learning_rate": 9.531742458352094e-05,
      "loss": 2.0693,
      "step": 124
    },
    {
      "epoch": 0.16722408026755853,
      "grad_norm": 0.8319055438041687,
      "learning_rate": 9.527239981990095e-05,
      "loss": 2.1013,
      "step": 125
    },
    {
      "epoch": 0.16856187290969898,
      "grad_norm": 0.720665454864502,
      "learning_rate": 9.522737505628096e-05,
      "loss": 2.4987,
      "step": 126
    },
    {
      "epoch": 0.16989966555183947,
      "grad_norm": 0.7007707953453064,
      "learning_rate": 9.518235029266098e-05,
      "loss": 1.9992,
      "step": 127
    },
    {
      "epoch": 0.17123745819397992,
      "grad_norm": 1.1250567436218262,
      "learning_rate": 9.513732552904097e-05,
      "loss": 2.3214,
      "step": 128
    },
    {
      "epoch": 0.1725752508361204,
      "grad_norm": 1.054378628730774,
      "learning_rate": 9.509230076542098e-05,
      "loss": 2.3111,
      "step": 129
    },
    {
      "epoch": 0.17391304347826086,
      "grad_norm": 0.8713865876197815,
      "learning_rate": 9.5047276001801e-05,
      "loss": 1.4661,
      "step": 130
    },
    {
      "epoch": 0.17525083612040135,
      "grad_norm": 1.2222834825515747,
      "learning_rate": 9.500225123818101e-05,
      "loss": 2.0313,
      "step": 131
    },
    {
      "epoch": 0.1765886287625418,
      "grad_norm": 0.7791270017623901,
      "learning_rate": 9.495722647456101e-05,
      "loss": 1.8948,
      "step": 132
    },
    {
      "epoch": 0.17792642140468226,
      "grad_norm": 0.7660050392150879,
      "learning_rate": 9.491220171094102e-05,
      "loss": 1.5712,
      "step": 133
    },
    {
      "epoch": 0.17926421404682275,
      "grad_norm": 2.549402952194214,
      "learning_rate": 9.486717694732104e-05,
      "loss": 2.3461,
      "step": 134
    },
    {
      "epoch": 0.1806020066889632,
      "grad_norm": 0.8216731548309326,
      "learning_rate": 9.482215218370105e-05,
      "loss": 2.1653,
      "step": 135
    },
    {
      "epoch": 0.18193979933110369,
      "grad_norm": 1.1273239850997925,
      "learning_rate": 9.477712742008104e-05,
      "loss": 2.0003,
      "step": 136
    },
    {
      "epoch": 0.18327759197324414,
      "grad_norm": 1.4746992588043213,
      "learning_rate": 9.473210265646106e-05,
      "loss": 1.8176,
      "step": 137
    },
    {
      "epoch": 0.18461538461538463,
      "grad_norm": 0.6299614906311035,
      "learning_rate": 9.468707789284107e-05,
      "loss": 1.9021,
      "step": 138
    },
    {
      "epoch": 0.18595317725752508,
      "grad_norm": 0.6752821207046509,
      "learning_rate": 9.464205312922108e-05,
      "loss": 1.5829,
      "step": 139
    },
    {
      "epoch": 0.18729096989966554,
      "grad_norm": 0.8192165493965149,
      "learning_rate": 9.459702836560108e-05,
      "loss": 2.093,
      "step": 140
    },
    {
      "epoch": 0.18862876254180602,
      "grad_norm": 0.7584500312805176,
      "learning_rate": 9.45520036019811e-05,
      "loss": 2.2507,
      "step": 141
    },
    {
      "epoch": 0.18996655518394648,
      "grad_norm": 0.6749038100242615,
      "learning_rate": 9.45069788383611e-05,
      "loss": 2.1704,
      "step": 142
    },
    {
      "epoch": 0.19130434782608696,
      "grad_norm": 8.133386611938477,
      "learning_rate": 9.446195407474112e-05,
      "loss": 2.121,
      "step": 143
    },
    {
      "epoch": 0.19264214046822742,
      "grad_norm": 0.7231702208518982,
      "learning_rate": 9.441692931112111e-05,
      "loss": 2.1596,
      "step": 144
    },
    {
      "epoch": 0.1939799331103679,
      "grad_norm": 1.4290648698806763,
      "learning_rate": 9.437190454750113e-05,
      "loss": 2.0232,
      "step": 145
    },
    {
      "epoch": 0.19531772575250836,
      "grad_norm": 0.7707876563072205,
      "learning_rate": 9.432687978388114e-05,
      "loss": 2.408,
      "step": 146
    },
    {
      "epoch": 0.19665551839464884,
      "grad_norm": 0.6868302822113037,
      "learning_rate": 9.428185502026115e-05,
      "loss": 2.1884,
      "step": 147
    },
    {
      "epoch": 0.1979933110367893,
      "grad_norm": 1.0088449716567993,
      "learning_rate": 9.423683025664116e-05,
      "loss": 2.3152,
      "step": 148
    },
    {
      "epoch": 0.19933110367892976,
      "grad_norm": 0.7136611938476562,
      "learning_rate": 9.419180549302117e-05,
      "loss": 2.0639,
      "step": 149
    },
    {
      "epoch": 0.20066889632107024,
      "grad_norm": 0.7562902569770813,
      "learning_rate": 9.414678072940118e-05,
      "loss": 2.1516,
      "step": 150
    },
    {
      "epoch": 0.2020066889632107,
      "grad_norm": 0.7779235243797302,
      "learning_rate": 9.410175596578118e-05,
      "loss": 2.0172,
      "step": 151
    },
    {
      "epoch": 0.20334448160535118,
      "grad_norm": 0.7270732522010803,
      "learning_rate": 9.405673120216119e-05,
      "loss": 2.5632,
      "step": 152
    },
    {
      "epoch": 0.20468227424749164,
      "grad_norm": 6.782646179199219,
      "learning_rate": 9.40117064385412e-05,
      "loss": 1.9557,
      "step": 153
    },
    {
      "epoch": 0.20602006688963212,
      "grad_norm": 0.8244476318359375,
      "learning_rate": 9.396668167492121e-05,
      "loss": 2.2257,
      "step": 154
    },
    {
      "epoch": 0.20735785953177258,
      "grad_norm": 0.7395358681678772,
      "learning_rate": 9.392165691130123e-05,
      "loss": 1.8494,
      "step": 155
    },
    {
      "epoch": 0.20869565217391303,
      "grad_norm": 1.2709623575210571,
      "learning_rate": 9.387663214768123e-05,
      "loss": 2.579,
      "step": 156
    },
    {
      "epoch": 0.21003344481605352,
      "grad_norm": 0.8247620463371277,
      "learning_rate": 9.383160738406124e-05,
      "loss": 2.2475,
      "step": 157
    },
    {
      "epoch": 0.21137123745819397,
      "grad_norm": 0.7007076144218445,
      "learning_rate": 9.378658262044124e-05,
      "loss": 2.1498,
      "step": 158
    },
    {
      "epoch": 0.21270903010033446,
      "grad_norm": 0.6729087829589844,
      "learning_rate": 9.374155785682127e-05,
      "loss": 1.8659,
      "step": 159
    },
    {
      "epoch": 0.2140468227424749,
      "grad_norm": 0.8817248344421387,
      "learning_rate": 9.369653309320126e-05,
      "loss": 2.2277,
      "step": 160
    },
    {
      "epoch": 0.2153846153846154,
      "grad_norm": 0.660122275352478,
      "learning_rate": 9.365150832958127e-05,
      "loss": 2.3238,
      "step": 161
    },
    {
      "epoch": 0.21672240802675585,
      "grad_norm": 0.7546793818473816,
      "learning_rate": 9.360648356596128e-05,
      "loss": 2.3036,
      "step": 162
    },
    {
      "epoch": 0.2180602006688963,
      "grad_norm": 1.4873573780059814,
      "learning_rate": 9.35614588023413e-05,
      "loss": 2.0092,
      "step": 163
    },
    {
      "epoch": 0.2193979933110368,
      "grad_norm": 0.8722738027572632,
      "learning_rate": 9.35164340387213e-05,
      "loss": 1.6106,
      "step": 164
    },
    {
      "epoch": 0.22073578595317725,
      "grad_norm": 0.7043030858039856,
      "learning_rate": 9.34714092751013e-05,
      "loss": 1.7849,
      "step": 165
    },
    {
      "epoch": 0.22207357859531773,
      "grad_norm": 0.8679320216178894,
      "learning_rate": 9.342638451148133e-05,
      "loss": 1.816,
      "step": 166
    },
    {
      "epoch": 0.2234113712374582,
      "grad_norm": 0.6752539873123169,
      "learning_rate": 9.338135974786133e-05,
      "loss": 1.9456,
      "step": 167
    },
    {
      "epoch": 0.22474916387959867,
      "grad_norm": 0.8796198964118958,
      "learning_rate": 9.333633498424133e-05,
      "loss": 2.0501,
      "step": 168
    },
    {
      "epoch": 0.22608695652173913,
      "grad_norm": 1.6531866788864136,
      "learning_rate": 9.329131022062134e-05,
      "loss": 1.9587,
      "step": 169
    },
    {
      "epoch": 0.22742474916387959,
      "grad_norm": 0.6475427746772766,
      "learning_rate": 9.324628545700136e-05,
      "loss": 2.0387,
      "step": 170
    },
    {
      "epoch": 0.22876254180602007,
      "grad_norm": 0.79718416929245,
      "learning_rate": 9.320126069338137e-05,
      "loss": 1.9722,
      "step": 171
    },
    {
      "epoch": 0.23010033444816053,
      "grad_norm": 0.6928245425224304,
      "learning_rate": 9.315623592976136e-05,
      "loss": 2.2319,
      "step": 172
    },
    {
      "epoch": 0.231438127090301,
      "grad_norm": 0.8756166100502014,
      "learning_rate": 9.311121116614139e-05,
      "loss": 2.2756,
      "step": 173
    },
    {
      "epoch": 0.23277591973244147,
      "grad_norm": 0.7787219285964966,
      "learning_rate": 9.30661864025214e-05,
      "loss": 1.9229,
      "step": 174
    },
    {
      "epoch": 0.23411371237458195,
      "grad_norm": 1.0068119764328003,
      "learning_rate": 9.30211616389014e-05,
      "loss": 1.5486,
      "step": 175
    },
    {
      "epoch": 0.2354515050167224,
      "grad_norm": 0.6787288188934326,
      "learning_rate": 9.29761368752814e-05,
      "loss": 2.2945,
      "step": 176
    },
    {
      "epoch": 0.23678929765886286,
      "grad_norm": 0.6843809485435486,
      "learning_rate": 9.293111211166142e-05,
      "loss": 2.148,
      "step": 177
    },
    {
      "epoch": 0.23812709030100335,
      "grad_norm": 0.6215639114379883,
      "learning_rate": 9.288608734804143e-05,
      "loss": 2.2645,
      "step": 178
    },
    {
      "epoch": 0.2394648829431438,
      "grad_norm": 0.6082992553710938,
      "learning_rate": 9.284106258442144e-05,
      "loss": 2.0593,
      "step": 179
    },
    {
      "epoch": 0.2408026755852843,
      "grad_norm": 1.1877750158309937,
      "learning_rate": 9.279603782080145e-05,
      "loss": 2.0483,
      "step": 180
    },
    {
      "epoch": 0.24214046822742474,
      "grad_norm": 1.0917643308639526,
      "learning_rate": 9.275101305718146e-05,
      "loss": 2.2408,
      "step": 181
    },
    {
      "epoch": 0.24347826086956523,
      "grad_norm": 1.3530385494232178,
      "learning_rate": 9.270598829356146e-05,
      "loss": 2.0472,
      "step": 182
    },
    {
      "epoch": 0.24481605351170568,
      "grad_norm": 0.7971617579460144,
      "learning_rate": 9.266096352994147e-05,
      "loss": 2.2679,
      "step": 183
    },
    {
      "epoch": 0.24615384615384617,
      "grad_norm": 0.9468250870704651,
      "learning_rate": 9.261593876632148e-05,
      "loss": 2.2528,
      "step": 184
    },
    {
      "epoch": 0.24749163879598662,
      "grad_norm": 0.9091652035713196,
      "learning_rate": 9.257091400270149e-05,
      "loss": 2.2659,
      "step": 185
    },
    {
      "epoch": 0.24882943143812708,
      "grad_norm": 0.708520770072937,
      "learning_rate": 9.25258892390815e-05,
      "loss": 1.8455,
      "step": 186
    },
    {
      "epoch": 0.25016722408026754,
      "grad_norm": 1.3594329357147217,
      "learning_rate": 9.24808644754615e-05,
      "loss": 1.8394,
      "step": 187
    },
    {
      "epoch": 0.25150501672240805,
      "grad_norm": 0.7608739733695984,
      "learning_rate": 9.243583971184152e-05,
      "loss": 2.3127,
      "step": 188
    },
    {
      "epoch": 0.2528428093645485,
      "grad_norm": 0.8208618760108948,
      "learning_rate": 9.239081494822152e-05,
      "loss": 2.0739,
      "step": 189
    },
    {
      "epoch": 0.25418060200668896,
      "grad_norm": 4.823574542999268,
      "learning_rate": 9.234579018460153e-05,
      "loss": 2.3519,
      "step": 190
    },
    {
      "epoch": 0.2555183946488294,
      "grad_norm": 1.364988923072815,
      "learning_rate": 9.230076542098155e-05,
      "loss": 2.1007,
      "step": 191
    },
    {
      "epoch": 0.2568561872909699,
      "grad_norm": 0.6934626698493958,
      "learning_rate": 9.225574065736155e-05,
      "loss": 2.1424,
      "step": 192
    },
    {
      "epoch": 0.2581939799331104,
      "grad_norm": 0.6626648306846619,
      "learning_rate": 9.221071589374156e-05,
      "loss": 1.6056,
      "step": 193
    },
    {
      "epoch": 0.25953177257525084,
      "grad_norm": 0.6442323923110962,
      "learning_rate": 9.216569113012157e-05,
      "loss": 2.1698,
      "step": 194
    },
    {
      "epoch": 0.2608695652173913,
      "grad_norm": 1.5775187015533447,
      "learning_rate": 9.212066636650159e-05,
      "loss": 2.2164,
      "step": 195
    },
    {
      "epoch": 0.26220735785953175,
      "grad_norm": 0.5490379929542542,
      "learning_rate": 9.207564160288158e-05,
      "loss": 2.5528,
      "step": 196
    },
    {
      "epoch": 0.26354515050167227,
      "grad_norm": 0.6842154860496521,
      "learning_rate": 9.203061683926159e-05,
      "loss": 2.1731,
      "step": 197
    },
    {
      "epoch": 0.2648829431438127,
      "grad_norm": 0.5994325280189514,
      "learning_rate": 9.19855920756416e-05,
      "loss": 2.2696,
      "step": 198
    },
    {
      "epoch": 0.2662207357859532,
      "grad_norm": 0.7480664253234863,
      "learning_rate": 9.194056731202162e-05,
      "loss": 2.1391,
      "step": 199
    },
    {
      "epoch": 0.26755852842809363,
      "grad_norm": 0.7724633812904358,
      "learning_rate": 9.189554254840163e-05,
      "loss": 2.4159,
      "step": 200
    },
    {
      "epoch": 0.2688963210702341,
      "grad_norm": 0.7571853995323181,
      "learning_rate": 9.185051778478163e-05,
      "loss": 1.6681,
      "step": 201
    },
    {
      "epoch": 0.2702341137123746,
      "grad_norm": 0.7273426651954651,
      "learning_rate": 9.180549302116165e-05,
      "loss": 2.529,
      "step": 202
    },
    {
      "epoch": 0.27157190635451506,
      "grad_norm": 0.8330695629119873,
      "learning_rate": 9.176046825754166e-05,
      "loss": 2.2551,
      "step": 203
    },
    {
      "epoch": 0.2729096989966555,
      "grad_norm": 0.6210842132568359,
      "learning_rate": 9.171544349392167e-05,
      "loss": 2.0951,
      "step": 204
    },
    {
      "epoch": 0.27424749163879597,
      "grad_norm": 0.6045663356781006,
      "learning_rate": 9.167041873030166e-05,
      "loss": 1.6797,
      "step": 205
    },
    {
      "epoch": 0.2755852842809365,
      "grad_norm": 0.7505138516426086,
      "learning_rate": 9.162539396668168e-05,
      "loss": 1.862,
      "step": 206
    },
    {
      "epoch": 0.27692307692307694,
      "grad_norm": 0.77651047706604,
      "learning_rate": 9.158036920306169e-05,
      "loss": 2.2327,
      "step": 207
    },
    {
      "epoch": 0.2782608695652174,
      "grad_norm": 0.7325896620750427,
      "learning_rate": 9.15353444394417e-05,
      "loss": 1.8543,
      "step": 208
    },
    {
      "epoch": 0.27959866220735785,
      "grad_norm": 0.7572547793388367,
      "learning_rate": 9.149031967582171e-05,
      "loss": 2.2101,
      "step": 209
    },
    {
      "epoch": 0.2809364548494983,
      "grad_norm": 0.6754357814788818,
      "learning_rate": 9.144529491220172e-05,
      "loss": 2.2774,
      "step": 210
    },
    {
      "epoch": 0.2822742474916388,
      "grad_norm": 0.7105769515037537,
      "learning_rate": 9.140027014858173e-05,
      "loss": 2.0553,
      "step": 211
    },
    {
      "epoch": 0.2836120401337793,
      "grad_norm": 0.5948776602745056,
      "learning_rate": 9.135524538496173e-05,
      "loss": 2.1455,
      "step": 212
    },
    {
      "epoch": 0.28494983277591973,
      "grad_norm": 0.6252536177635193,
      "learning_rate": 9.131022062134174e-05,
      "loss": 1.7525,
      "step": 213
    },
    {
      "epoch": 0.2862876254180602,
      "grad_norm": 0.5089226961135864,
      "learning_rate": 9.126519585772175e-05,
      "loss": 1.8441,
      "step": 214
    },
    {
      "epoch": 0.28762541806020064,
      "grad_norm": 0.7862734198570251,
      "learning_rate": 9.122017109410176e-05,
      "loss": 2.1025,
      "step": 215
    },
    {
      "epoch": 0.28896321070234116,
      "grad_norm": 1.1257396936416626,
      "learning_rate": 9.117514633048177e-05,
      "loss": 1.8203,
      "step": 216
    },
    {
      "epoch": 0.2903010033444816,
      "grad_norm": 0.7163311243057251,
      "learning_rate": 9.113012156686178e-05,
      "loss": 2.513,
      "step": 217
    },
    {
      "epoch": 0.29163879598662207,
      "grad_norm": 0.5419270396232605,
      "learning_rate": 9.108509680324179e-05,
      "loss": 2.233,
      "step": 218
    },
    {
      "epoch": 0.2929765886287625,
      "grad_norm": 0.6821555495262146,
      "learning_rate": 9.10400720396218e-05,
      "loss": 2.0814,
      "step": 219
    },
    {
      "epoch": 0.29431438127090304,
      "grad_norm": 0.7533456087112427,
      "learning_rate": 9.09950472760018e-05,
      "loss": 1.9754,
      "step": 220
    },
    {
      "epoch": 0.2956521739130435,
      "grad_norm": 1.0426667928695679,
      "learning_rate": 9.095002251238181e-05,
      "loss": 1.8963,
      "step": 221
    },
    {
      "epoch": 0.29698996655518395,
      "grad_norm": 0.725765585899353,
      "learning_rate": 9.090499774876182e-05,
      "loss": 2.0239,
      "step": 222
    },
    {
      "epoch": 0.2983277591973244,
      "grad_norm": 0.5403796434402466,
      "learning_rate": 9.085997298514183e-05,
      "loss": 1.8004,
      "step": 223
    },
    {
      "epoch": 0.29966555183946486,
      "grad_norm": 0.7222073078155518,
      "learning_rate": 9.081494822152185e-05,
      "loss": 1.9841,
      "step": 224
    },
    {
      "epoch": 0.3010033444816054,
      "grad_norm": 1.1086597442626953,
      "learning_rate": 9.076992345790185e-05,
      "loss": 1.9085,
      "step": 225
    },
    {
      "epoch": 0.30234113712374583,
      "grad_norm": 1.6131938695907593,
      "learning_rate": 9.072489869428186e-05,
      "loss": 2.2787,
      "step": 226
    },
    {
      "epoch": 0.3036789297658863,
      "grad_norm": 0.686077356338501,
      "learning_rate": 9.067987393066188e-05,
      "loss": 1.8902,
      "step": 227
    },
    {
      "epoch": 0.30501672240802674,
      "grad_norm": 0.6241892576217651,
      "learning_rate": 9.063484916704189e-05,
      "loss": 2.2297,
      "step": 228
    },
    {
      "epoch": 0.3063545150501672,
      "grad_norm": 0.522357165813446,
      "learning_rate": 9.058982440342188e-05,
      "loss": 2.291,
      "step": 229
    },
    {
      "epoch": 0.3076923076923077,
      "grad_norm": 0.5207152962684631,
      "learning_rate": 9.054479963980189e-05,
      "loss": 2.0928,
      "step": 230
    },
    {
      "epoch": 0.30903010033444817,
      "grad_norm": 0.6844892501831055,
      "learning_rate": 9.049977487618191e-05,
      "loss": 2.187,
      "step": 231
    },
    {
      "epoch": 0.3103678929765886,
      "grad_norm": 0.7186736464500427,
      "learning_rate": 9.045475011256192e-05,
      "loss": 2.0874,
      "step": 232
    },
    {
      "epoch": 0.3117056856187291,
      "grad_norm": 0.7187355160713196,
      "learning_rate": 9.040972534894192e-05,
      "loss": 1.9839,
      "step": 233
    },
    {
      "epoch": 0.3130434782608696,
      "grad_norm": 0.5916269421577454,
      "learning_rate": 9.036470058532192e-05,
      "loss": 2.1944,
      "step": 234
    },
    {
      "epoch": 0.31438127090301005,
      "grad_norm": 0.6687160134315491,
      "learning_rate": 9.031967582170195e-05,
      "loss": 1.9731,
      "step": 235
    },
    {
      "epoch": 0.3157190635451505,
      "grad_norm": 0.7072650790214539,
      "learning_rate": 9.027465105808195e-05,
      "loss": 1.8072,
      "step": 236
    },
    {
      "epoch": 0.31705685618729096,
      "grad_norm": 1.014660120010376,
      "learning_rate": 9.022962629446195e-05,
      "loss": 1.8828,
      "step": 237
    },
    {
      "epoch": 0.3183946488294314,
      "grad_norm": 1.4133497476577759,
      "learning_rate": 9.018460153084197e-05,
      "loss": 2.312,
      "step": 238
    },
    {
      "epoch": 0.3197324414715719,
      "grad_norm": 0.8173983693122864,
      "learning_rate": 9.013957676722198e-05,
      "loss": 2.0622,
      "step": 239
    },
    {
      "epoch": 0.3210702341137124,
      "grad_norm": 0.5996513962745667,
      "learning_rate": 9.009455200360199e-05,
      "loss": 2.2583,
      "step": 240
    },
    {
      "epoch": 0.32240802675585284,
      "grad_norm": 0.9969286918640137,
      "learning_rate": 9.004952723998198e-05,
      "loss": 2.1541,
      "step": 241
    },
    {
      "epoch": 0.3237458193979933,
      "grad_norm": 1.5697412490844727,
      "learning_rate": 9.0004502476362e-05,
      "loss": 1.8842,
      "step": 242
    },
    {
      "epoch": 0.3250836120401338,
      "grad_norm": 0.844559907913208,
      "learning_rate": 8.995947771274201e-05,
      "loss": 2.2494,
      "step": 243
    },
    {
      "epoch": 0.32642140468227426,
      "grad_norm": 0.8279430866241455,
      "learning_rate": 8.991445294912202e-05,
      "loss": 2.0218,
      "step": 244
    },
    {
      "epoch": 0.3277591973244147,
      "grad_norm": 0.5577782392501831,
      "learning_rate": 8.986942818550203e-05,
      "loss": 2.1826,
      "step": 245
    },
    {
      "epoch": 0.3290969899665552,
      "grad_norm": 0.6767871975898743,
      "learning_rate": 8.982440342188204e-05,
      "loss": 1.6723,
      "step": 246
    },
    {
      "epoch": 0.33043478260869563,
      "grad_norm": 0.7155217528343201,
      "learning_rate": 8.977937865826205e-05,
      "loss": 1.9627,
      "step": 247
    },
    {
      "epoch": 0.33177257525083614,
      "grad_norm": 0.7340490818023682,
      "learning_rate": 8.973435389464206e-05,
      "loss": 2.1307,
      "step": 248
    },
    {
      "epoch": 0.3331103678929766,
      "grad_norm": 1.0434596538543701,
      "learning_rate": 8.968932913102207e-05,
      "loss": 1.7668,
      "step": 249
    },
    {
      "epoch": 0.33444816053511706,
      "grad_norm": 0.7316417694091797,
      "learning_rate": 8.964430436740207e-05,
      "loss": 2.0947,
      "step": 250
    },
    {
      "epoch": 0.3357859531772575,
      "grad_norm": 0.6342126131057739,
      "learning_rate": 8.959927960378208e-05,
      "loss": 2.1559,
      "step": 251
    },
    {
      "epoch": 0.33712374581939797,
      "grad_norm": 0.6641095280647278,
      "learning_rate": 8.955425484016209e-05,
      "loss": 2.1736,
      "step": 252
    },
    {
      "epoch": 0.3384615384615385,
      "grad_norm": 0.6803715229034424,
      "learning_rate": 8.95092300765421e-05,
      "loss": 2.3727,
      "step": 253
    },
    {
      "epoch": 0.33979933110367894,
      "grad_norm": 0.6433061957359314,
      "learning_rate": 8.946420531292211e-05,
      "loss": 2.0082,
      "step": 254
    },
    {
      "epoch": 0.3411371237458194,
      "grad_norm": 0.7468423247337341,
      "learning_rate": 8.941918054930212e-05,
      "loss": 2.2016,
      "step": 255
    },
    {
      "epoch": 0.34247491638795985,
      "grad_norm": 1.1716619729995728,
      "learning_rate": 8.937415578568214e-05,
      "loss": 2.1824,
      "step": 256
    },
    {
      "epoch": 0.34381270903010036,
      "grad_norm": 3.4333786964416504,
      "learning_rate": 8.932913102206213e-05,
      "loss": 1.9569,
      "step": 257
    },
    {
      "epoch": 0.3451505016722408,
      "grad_norm": 0.7647579312324524,
      "learning_rate": 8.928410625844214e-05,
      "loss": 2.1416,
      "step": 258
    },
    {
      "epoch": 0.3464882943143813,
      "grad_norm": 0.7500723600387573,
      "learning_rate": 8.923908149482215e-05,
      "loss": 2.1495,
      "step": 259
    },
    {
      "epoch": 0.34782608695652173,
      "grad_norm": 0.7312958240509033,
      "learning_rate": 8.919405673120217e-05,
      "loss": 2.136,
      "step": 260
    },
    {
      "epoch": 0.3491638795986622,
      "grad_norm": 0.7841158509254456,
      "learning_rate": 8.914903196758217e-05,
      "loss": 1.837,
      "step": 261
    },
    {
      "epoch": 0.3505016722408027,
      "grad_norm": 1.176447868347168,
      "learning_rate": 8.910400720396218e-05,
      "loss": 2.1528,
      "step": 262
    },
    {
      "epoch": 0.35183946488294315,
      "grad_norm": 0.694179117679596,
      "learning_rate": 8.90589824403422e-05,
      "loss": 2.1408,
      "step": 263
    },
    {
      "epoch": 0.3531772575250836,
      "grad_norm": 0.6742585301399231,
      "learning_rate": 8.901395767672221e-05,
      "loss": 2.3309,
      "step": 264
    },
    {
      "epoch": 0.35451505016722407,
      "grad_norm": 10.892035484313965,
      "learning_rate": 8.89689329131022e-05,
      "loss": 2.6335,
      "step": 265
    },
    {
      "epoch": 0.3558528428093645,
      "grad_norm": 0.718368411064148,
      "learning_rate": 8.892390814948221e-05,
      "loss": 2.2006,
      "step": 266
    },
    {
      "epoch": 0.35719063545150503,
      "grad_norm": 0.9842844605445862,
      "learning_rate": 8.887888338586223e-05,
      "loss": 1.8219,
      "step": 267
    },
    {
      "epoch": 0.3585284280936455,
      "grad_norm": 0.6319013237953186,
      "learning_rate": 8.883385862224224e-05,
      "loss": 2.0513,
      "step": 268
    },
    {
      "epoch": 0.35986622073578595,
      "grad_norm": 0.7445176243782043,
      "learning_rate": 8.878883385862224e-05,
      "loss": 1.9484,
      "step": 269
    },
    {
      "epoch": 0.3612040133779264,
      "grad_norm": 0.7307112216949463,
      "learning_rate": 8.874380909500225e-05,
      "loss": 2.2374,
      "step": 270
    },
    {
      "epoch": 0.3625418060200669,
      "grad_norm": 0.6374862790107727,
      "learning_rate": 8.869878433138227e-05,
      "loss": 2.0784,
      "step": 271
    },
    {
      "epoch": 0.36387959866220737,
      "grad_norm": 0.6017893552780151,
      "learning_rate": 8.865375956776228e-05,
      "loss": 2.0,
      "step": 272
    },
    {
      "epoch": 0.3652173913043478,
      "grad_norm": 1.2405426502227783,
      "learning_rate": 8.860873480414229e-05,
      "loss": 2.3654,
      "step": 273
    },
    {
      "epoch": 0.3665551839464883,
      "grad_norm": 0.8560160398483276,
      "learning_rate": 8.85637100405223e-05,
      "loss": 2.181,
      "step": 274
    },
    {
      "epoch": 0.36789297658862874,
      "grad_norm": 1.1745796203613281,
      "learning_rate": 8.85186852769023e-05,
      "loss": 2.2952,
      "step": 275
    },
    {
      "epoch": 0.36923076923076925,
      "grad_norm": 0.6238117814064026,
      "learning_rate": 8.847366051328231e-05,
      "loss": 2.2203,
      "step": 276
    },
    {
      "epoch": 0.3705685618729097,
      "grad_norm": 1.2902181148529053,
      "learning_rate": 8.842863574966232e-05,
      "loss": 2.07,
      "step": 277
    },
    {
      "epoch": 0.37190635451505016,
      "grad_norm": 2.4692065715789795,
      "learning_rate": 8.838361098604233e-05,
      "loss": 1.4593,
      "step": 278
    },
    {
      "epoch": 0.3732441471571906,
      "grad_norm": 0.7107476592063904,
      "learning_rate": 8.833858622242234e-05,
      "loss": 2.1197,
      "step": 279
    },
    {
      "epoch": 0.3745819397993311,
      "grad_norm": 0.6064779162406921,
      "learning_rate": 8.829356145880235e-05,
      "loss": 2.1686,
      "step": 280
    },
    {
      "epoch": 0.3759197324414716,
      "grad_norm": 0.9871097207069397,
      "learning_rate": 8.824853669518235e-05,
      "loss": 2.1,
      "step": 281
    },
    {
      "epoch": 0.37725752508361204,
      "grad_norm": 0.7217766642570496,
      "learning_rate": 8.820351193156236e-05,
      "loss": 1.8677,
      "step": 282
    },
    {
      "epoch": 0.3785953177257525,
      "grad_norm": 0.6632222533226013,
      "learning_rate": 8.815848716794237e-05,
      "loss": 1.9146,
      "step": 283
    },
    {
      "epoch": 0.37993311036789296,
      "grad_norm": 1.2834292650222778,
      "learning_rate": 8.811346240432238e-05,
      "loss": 2.2682,
      "step": 284
    },
    {
      "epoch": 0.38127090301003347,
      "grad_norm": 0.6314417719841003,
      "learning_rate": 8.806843764070239e-05,
      "loss": 2.0938,
      "step": 285
    },
    {
      "epoch": 0.3826086956521739,
      "grad_norm": 0.8179631233215332,
      "learning_rate": 8.80234128770824e-05,
      "loss": 2.308,
      "step": 286
    },
    {
      "epoch": 0.3839464882943144,
      "grad_norm": 0.6403224468231201,
      "learning_rate": 8.79783881134624e-05,
      "loss": 2.186,
      "step": 287
    },
    {
      "epoch": 0.38528428093645484,
      "grad_norm": 0.6534366011619568,
      "learning_rate": 8.793336334984241e-05,
      "loss": 2.0692,
      "step": 288
    },
    {
      "epoch": 0.3866220735785953,
      "grad_norm": 0.7949046492576599,
      "learning_rate": 8.788833858622242e-05,
      "loss": 2.1176,
      "step": 289
    },
    {
      "epoch": 0.3879598662207358,
      "grad_norm": 0.5939118266105652,
      "learning_rate": 8.784331382260243e-05,
      "loss": 2.3326,
      "step": 290
    },
    {
      "epoch": 0.38929765886287626,
      "grad_norm": 0.6752482652664185,
      "learning_rate": 8.779828905898244e-05,
      "loss": 2.1355,
      "step": 291
    },
    {
      "epoch": 0.3906354515050167,
      "grad_norm": 1.1655668020248413,
      "learning_rate": 8.775326429536246e-05,
      "loss": 1.7848,
      "step": 292
    },
    {
      "epoch": 0.3919732441471572,
      "grad_norm": 1.0938507318496704,
      "learning_rate": 8.770823953174246e-05,
      "loss": 2.0154,
      "step": 293
    },
    {
      "epoch": 0.3933110367892977,
      "grad_norm": 1.0485789775848389,
      "learning_rate": 8.766321476812247e-05,
      "loss": 2.2559,
      "step": 294
    },
    {
      "epoch": 0.39464882943143814,
      "grad_norm": 1.0306947231292725,
      "learning_rate": 8.761819000450247e-05,
      "loss": 2.403,
      "step": 295
    },
    {
      "epoch": 0.3959866220735786,
      "grad_norm": 2.1360738277435303,
      "learning_rate": 8.75731652408825e-05,
      "loss": 2.0527,
      "step": 296
    },
    {
      "epoch": 0.39732441471571905,
      "grad_norm": 0.8952328562736511,
      "learning_rate": 8.75281404772625e-05,
      "loss": 2.2782,
      "step": 297
    },
    {
      "epoch": 0.3986622073578595,
      "grad_norm": 0.5768120288848877,
      "learning_rate": 8.74831157136425e-05,
      "loss": 2.527,
      "step": 298
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.754932165145874,
      "learning_rate": 8.743809095002252e-05,
      "loss": 2.3223,
      "step": 299
    },
    {
      "epoch": 0.4013377926421405,
      "grad_norm": 1.3988687992095947,
      "learning_rate": 8.739306618640253e-05,
      "loss": 2.2288,
      "step": 300
    },
    {
      "epoch": 0.40267558528428093,
      "grad_norm": 0.6435593366622925,
      "learning_rate": 8.734804142278254e-05,
      "loss": 1.9748,
      "step": 301
    },
    {
      "epoch": 0.4040133779264214,
      "grad_norm": 2.387169361114502,
      "learning_rate": 8.730301665916253e-05,
      "loss": 2.2313,
      "step": 302
    },
    {
      "epoch": 0.40535117056856185,
      "grad_norm": 0.6832499504089355,
      "learning_rate": 8.725799189554256e-05,
      "loss": 2.2805,
      "step": 303
    },
    {
      "epoch": 0.40668896321070236,
      "grad_norm": 0.7087513208389282,
      "learning_rate": 8.721296713192257e-05,
      "loss": 2.0363,
      "step": 304
    },
    {
      "epoch": 0.4080267558528428,
      "grad_norm": 0.9904339909553528,
      "learning_rate": 8.716794236830257e-05,
      "loss": 2.4282,
      "step": 305
    },
    {
      "epoch": 0.40936454849498327,
      "grad_norm": 3.171907901763916,
      "learning_rate": 8.712291760468257e-05,
      "loss": 2.412,
      "step": 306
    },
    {
      "epoch": 0.4107023411371237,
      "grad_norm": 0.6599994897842407,
      "learning_rate": 8.707789284106259e-05,
      "loss": 2.0818,
      "step": 307
    },
    {
      "epoch": 0.41204013377926424,
      "grad_norm": 0.7640592455863953,
      "learning_rate": 8.70328680774426e-05,
      "loss": 2.0231,
      "step": 308
    },
    {
      "epoch": 0.4133779264214047,
      "grad_norm": 0.7022128701210022,
      "learning_rate": 8.698784331382261e-05,
      "loss": 1.9285,
      "step": 309
    },
    {
      "epoch": 0.41471571906354515,
      "grad_norm": 8.844080924987793,
      "learning_rate": 8.694281855020262e-05,
      "loss": 2.3078,
      "step": 310
    },
    {
      "epoch": 0.4160535117056856,
      "grad_norm": 0.6928530335426331,
      "learning_rate": 8.689779378658263e-05,
      "loss": 2.276,
      "step": 311
    },
    {
      "epoch": 0.41739130434782606,
      "grad_norm": 0.808559238910675,
      "learning_rate": 8.685276902296263e-05,
      "loss": 2.2516,
      "step": 312
    },
    {
      "epoch": 0.4187290969899666,
      "grad_norm": 0.5992230176925659,
      "learning_rate": 8.680774425934264e-05,
      "loss": 2.3784,
      "step": 313
    },
    {
      "epoch": 0.42006688963210703,
      "grad_norm": 0.618697464466095,
      "learning_rate": 8.676271949572265e-05,
      "loss": 2.0398,
      "step": 314
    },
    {
      "epoch": 0.4214046822742475,
      "grad_norm": 0.5807810425758362,
      "learning_rate": 8.671769473210266e-05,
      "loss": 2.2591,
      "step": 315
    },
    {
      "epoch": 0.42274247491638794,
      "grad_norm": 1.4132471084594727,
      "learning_rate": 8.667266996848267e-05,
      "loss": 2.3054,
      "step": 316
    },
    {
      "epoch": 0.4240802675585284,
      "grad_norm": 2.2352471351623535,
      "learning_rate": 8.662764520486268e-05,
      "loss": 1.8411,
      "step": 317
    },
    {
      "epoch": 0.4254180602006689,
      "grad_norm": 1.1899638175964355,
      "learning_rate": 8.658262044124269e-05,
      "loss": 2.1932,
      "step": 318
    },
    {
      "epoch": 0.42675585284280937,
      "grad_norm": 0.7901405096054077,
      "learning_rate": 8.65375956776227e-05,
      "loss": 2.2161,
      "step": 319
    },
    {
      "epoch": 0.4280936454849498,
      "grad_norm": 0.5505680441856384,
      "learning_rate": 8.64925709140027e-05,
      "loss": 2.1754,
      "step": 320
    },
    {
      "epoch": 0.4294314381270903,
      "grad_norm": 0.8369104266166687,
      "learning_rate": 8.644754615038273e-05,
      "loss": 1.6397,
      "step": 321
    },
    {
      "epoch": 0.4307692307692308,
      "grad_norm": 0.6455018520355225,
      "learning_rate": 8.640252138676272e-05,
      "loss": 2.1317,
      "step": 322
    },
    {
      "epoch": 0.43210702341137125,
      "grad_norm": 15.443414688110352,
      "learning_rate": 8.635749662314273e-05,
      "loss": 2.2802,
      "step": 323
    },
    {
      "epoch": 0.4334448160535117,
      "grad_norm": 0.7857627272605896,
      "learning_rate": 8.631247185952274e-05,
      "loss": 2.2325,
      "step": 324
    },
    {
      "epoch": 0.43478260869565216,
      "grad_norm": 0.9558283090591431,
      "learning_rate": 8.626744709590276e-05,
      "loss": 1.9514,
      "step": 325
    },
    {
      "epoch": 0.4361204013377926,
      "grad_norm": 2.7177233695983887,
      "learning_rate": 8.622242233228275e-05,
      "loss": 2.1328,
      "step": 326
    },
    {
      "epoch": 0.43745819397993313,
      "grad_norm": 2.361937999725342,
      "learning_rate": 8.617739756866276e-05,
      "loss": 2.4997,
      "step": 327
    },
    {
      "epoch": 0.4387959866220736,
      "grad_norm": 0.877702534198761,
      "learning_rate": 8.613237280504279e-05,
      "loss": 2.3157,
      "step": 328
    },
    {
      "epoch": 0.44013377926421404,
      "grad_norm": 1.4664796590805054,
      "learning_rate": 8.60873480414228e-05,
      "loss": 2.024,
      "step": 329
    },
    {
      "epoch": 0.4414715719063545,
      "grad_norm": 0.627921462059021,
      "learning_rate": 8.604232327780279e-05,
      "loss": 2.0251,
      "step": 330
    },
    {
      "epoch": 0.442809364548495,
      "grad_norm": 0.7147241830825806,
      "learning_rate": 8.59972985141828e-05,
      "loss": 1.9449,
      "step": 331
    },
    {
      "epoch": 0.44414715719063547,
      "grad_norm": 0.7197164297103882,
      "learning_rate": 8.595227375056282e-05,
      "loss": 2.1838,
      "step": 332
    },
    {
      "epoch": 0.4454849498327759,
      "grad_norm": 0.6572954654693604,
      "learning_rate": 8.590724898694283e-05,
      "loss": 2.0866,
      "step": 333
    },
    {
      "epoch": 0.4468227424749164,
      "grad_norm": 1.1570994853973389,
      "learning_rate": 8.586222422332282e-05,
      "loss": 2.3039,
      "step": 334
    },
    {
      "epoch": 0.44816053511705684,
      "grad_norm": 0.6246607899665833,
      "learning_rate": 8.581719945970285e-05,
      "loss": 2.1241,
      "step": 335
    },
    {
      "epoch": 0.44949832775919735,
      "grad_norm": 0.6910645365715027,
      "learning_rate": 8.577217469608285e-05,
      "loss": 2.3911,
      "step": 336
    },
    {
      "epoch": 0.4508361204013378,
      "grad_norm": 0.978664219379425,
      "learning_rate": 8.572714993246286e-05,
      "loss": 1.7931,
      "step": 337
    },
    {
      "epoch": 0.45217391304347826,
      "grad_norm": 0.7342723608016968,
      "learning_rate": 8.568212516884286e-05,
      "loss": 1.6711,
      "step": 338
    },
    {
      "epoch": 0.4535117056856187,
      "grad_norm": 0.8861130475997925,
      "learning_rate": 8.563710040522288e-05,
      "loss": 2.1125,
      "step": 339
    },
    {
      "epoch": 0.45484949832775917,
      "grad_norm": 0.6609489917755127,
      "learning_rate": 8.559207564160289e-05,
      "loss": 1.9707,
      "step": 340
    },
    {
      "epoch": 0.4561872909698997,
      "grad_norm": 0.6270145177841187,
      "learning_rate": 8.55470508779829e-05,
      "loss": 2.0042,
      "step": 341
    },
    {
      "epoch": 0.45752508361204014,
      "grad_norm": 0.9418214559555054,
      "learning_rate": 8.550202611436289e-05,
      "loss": 2.2311,
      "step": 342
    },
    {
      "epoch": 0.4588628762541806,
      "grad_norm": 0.9658025503158569,
      "learning_rate": 8.545700135074291e-05,
      "loss": 2.3304,
      "step": 343
    },
    {
      "epoch": 0.46020066889632105,
      "grad_norm": 0.769410252571106,
      "learning_rate": 8.541197658712292e-05,
      "loss": 2.3133,
      "step": 344
    },
    {
      "epoch": 0.46153846153846156,
      "grad_norm": 0.5739223957061768,
      "learning_rate": 8.536695182350293e-05,
      "loss": 2.2093,
      "step": 345
    },
    {
      "epoch": 0.462876254180602,
      "grad_norm": 0.6792606711387634,
      "learning_rate": 8.532192705988294e-05,
      "loss": 2.2756,
      "step": 346
    },
    {
      "epoch": 0.4642140468227425,
      "grad_norm": 1.3889212608337402,
      "learning_rate": 8.527690229626295e-05,
      "loss": 1.7942,
      "step": 347
    },
    {
      "epoch": 0.46555183946488293,
      "grad_norm": 0.7244587540626526,
      "learning_rate": 8.523187753264296e-05,
      "loss": 1.8507,
      "step": 348
    },
    {
      "epoch": 0.4668896321070234,
      "grad_norm": 1.222921371459961,
      "learning_rate": 8.518685276902297e-05,
      "loss": 1.653,
      "step": 349
    },
    {
      "epoch": 0.4682274247491639,
      "grad_norm": 0.6867721080780029,
      "learning_rate": 8.514182800540297e-05,
      "loss": 2.1023,
      "step": 350
    },
    {
      "epoch": 0.46956521739130436,
      "grad_norm": 0.9452881813049316,
      "learning_rate": 8.509680324178298e-05,
      "loss": 2.4444,
      "step": 351
    },
    {
      "epoch": 0.4709030100334448,
      "grad_norm": 0.6819020509719849,
      "learning_rate": 8.505177847816299e-05,
      "loss": 1.7956,
      "step": 352
    },
    {
      "epoch": 0.47224080267558527,
      "grad_norm": 0.6449626088142395,
      "learning_rate": 8.500675371454301e-05,
      "loss": 2.1945,
      "step": 353
    },
    {
      "epoch": 0.4735785953177257,
      "grad_norm": 0.9270897507667542,
      "learning_rate": 8.496172895092301e-05,
      "loss": 2.0861,
      "step": 354
    },
    {
      "epoch": 0.47491638795986624,
      "grad_norm": 0.8355360627174377,
      "learning_rate": 8.491670418730302e-05,
      "loss": 2.0304,
      "step": 355
    },
    {
      "epoch": 0.4762541806020067,
      "grad_norm": 0.7179239988327026,
      "learning_rate": 8.487167942368303e-05,
      "loss": 2.038,
      "step": 356
    },
    {
      "epoch": 0.47759197324414715,
      "grad_norm": 0.6889925003051758,
      "learning_rate": 8.482665466006305e-05,
      "loss": 2.0335,
      "step": 357
    },
    {
      "epoch": 0.4789297658862876,
      "grad_norm": 0.6966556906700134,
      "learning_rate": 8.478162989644304e-05,
      "loss": 2.1169,
      "step": 358
    },
    {
      "epoch": 0.4802675585284281,
      "grad_norm": 0.6920545697212219,
      "learning_rate": 8.473660513282305e-05,
      "loss": 1.5785,
      "step": 359
    },
    {
      "epoch": 0.4816053511705686,
      "grad_norm": 2.2333033084869385,
      "learning_rate": 8.469158036920306e-05,
      "loss": 1.952,
      "step": 360
    },
    {
      "epoch": 0.48294314381270903,
      "grad_norm": 0.7082642912864685,
      "learning_rate": 8.464655560558308e-05,
      "loss": 1.8666,
      "step": 361
    },
    {
      "epoch": 0.4842809364548495,
      "grad_norm": 0.6602755784988403,
      "learning_rate": 8.460153084196308e-05,
      "loss": 2.1019,
      "step": 362
    },
    {
      "epoch": 0.48561872909698994,
      "grad_norm": 0.5931785106658936,
      "learning_rate": 8.455650607834309e-05,
      "loss": 2.1189,
      "step": 363
    },
    {
      "epoch": 0.48695652173913045,
      "grad_norm": 0.9318033456802368,
      "learning_rate": 8.451148131472311e-05,
      "loss": 1.6414,
      "step": 364
    },
    {
      "epoch": 0.4882943143812709,
      "grad_norm": 0.7986612915992737,
      "learning_rate": 8.446645655110312e-05,
      "loss": 2.1038,
      "step": 365
    },
    {
      "epoch": 0.48963210702341137,
      "grad_norm": 0.7577380537986755,
      "learning_rate": 8.442143178748311e-05,
      "loss": 2.2667,
      "step": 366
    },
    {
      "epoch": 0.4909698996655518,
      "grad_norm": 0.7104119658470154,
      "learning_rate": 8.437640702386312e-05,
      "loss": 1.9144,
      "step": 367
    },
    {
      "epoch": 0.49230769230769234,
      "grad_norm": 0.6269575953483582,
      "learning_rate": 8.433138226024314e-05,
      "loss": 2.2801,
      "step": 368
    },
    {
      "epoch": 0.4936454849498328,
      "grad_norm": 0.6898888349533081,
      "learning_rate": 8.428635749662315e-05,
      "loss": 2.0484,
      "step": 369
    },
    {
      "epoch": 0.49498327759197325,
      "grad_norm": 0.797180712223053,
      "learning_rate": 8.424133273300316e-05,
      "loss": 2.0078,
      "step": 370
    },
    {
      "epoch": 0.4963210702341137,
      "grad_norm": 1.7789745330810547,
      "learning_rate": 8.419630796938317e-05,
      "loss": 2.3178,
      "step": 371
    },
    {
      "epoch": 0.49765886287625416,
      "grad_norm": 0.699863851070404,
      "learning_rate": 8.415128320576318e-05,
      "loss": 1.7836,
      "step": 372
    },
    {
      "epoch": 0.49899665551839467,
      "grad_norm": 0.720728874206543,
      "learning_rate": 8.410625844214319e-05,
      "loss": 1.9491,
      "step": 373
    },
    {
      "epoch": 0.5003344481605351,
      "grad_norm": 0.5678146481513977,
      "learning_rate": 8.40612336785232e-05,
      "loss": 2.2176,
      "step": 374
    },
    {
      "epoch": 0.5016722408026756,
      "grad_norm": 0.5399537682533264,
      "learning_rate": 8.40162089149032e-05,
      "loss": 1.6731,
      "step": 375
    },
    {
      "epoch": 0.5030100334448161,
      "grad_norm": 0.5497046709060669,
      "learning_rate": 8.397118415128321e-05,
      "loss": 2.2015,
      "step": 376
    },
    {
      "epoch": 0.5043478260869565,
      "grad_norm": 0.6270225644111633,
      "learning_rate": 8.392615938766322e-05,
      "loss": 2.183,
      "step": 377
    },
    {
      "epoch": 0.505685618729097,
      "grad_norm": 1.1238292455673218,
      "learning_rate": 8.388113462404323e-05,
      "loss": 1.8471,
      "step": 378
    },
    {
      "epoch": 0.5070234113712374,
      "grad_norm": 1.303147315979004,
      "learning_rate": 8.383610986042324e-05,
      "loss": 1.9869,
      "step": 379
    },
    {
      "epoch": 0.5083612040133779,
      "grad_norm": 1.5513795614242554,
      "learning_rate": 8.379108509680325e-05,
      "loss": 2.4587,
      "step": 380
    },
    {
      "epoch": 0.5096989966555184,
      "grad_norm": 0.7095016241073608,
      "learning_rate": 8.374606033318325e-05,
      "loss": 2.2674,
      "step": 381
    },
    {
      "epoch": 0.5110367892976588,
      "grad_norm": 0.7620111703872681,
      "learning_rate": 8.370103556956326e-05,
      "loss": 1.7031,
      "step": 382
    },
    {
      "epoch": 0.5123745819397993,
      "grad_norm": 0.7550622820854187,
      "learning_rate": 8.365601080594327e-05,
      "loss": 1.3312,
      "step": 383
    },
    {
      "epoch": 0.5137123745819397,
      "grad_norm": 0.5809555053710938,
      "learning_rate": 8.361098604232328e-05,
      "loss": 1.9467,
      "step": 384
    },
    {
      "epoch": 0.5150501672240803,
      "grad_norm": 0.7727431654930115,
      "learning_rate": 8.356596127870329e-05,
      "loss": 2.2737,
      "step": 385
    },
    {
      "epoch": 0.5163879598662208,
      "grad_norm": 0.6559184789657593,
      "learning_rate": 8.35209365150833e-05,
      "loss": 1.7666,
      "step": 386
    },
    {
      "epoch": 0.5177257525083612,
      "grad_norm": 0.5735934376716614,
      "learning_rate": 8.34759117514633e-05,
      "loss": 1.666,
      "step": 387
    },
    {
      "epoch": 0.5190635451505017,
      "grad_norm": 0.7658690810203552,
      "learning_rate": 8.343088698784331e-05,
      "loss": 2.1627,
      "step": 388
    },
    {
      "epoch": 0.5204013377926422,
      "grad_norm": 2.0972108840942383,
      "learning_rate": 8.338586222422334e-05,
      "loss": 1.8791,
      "step": 389
    },
    {
      "epoch": 0.5217391304347826,
      "grad_norm": 1.761249303817749,
      "learning_rate": 8.334083746060334e-05,
      "loss": 2.3226,
      "step": 390
    },
    {
      "epoch": 0.5230769230769231,
      "grad_norm": 0.7268783450126648,
      "learning_rate": 8.329581269698334e-05,
      "loss": 1.5883,
      "step": 391
    },
    {
      "epoch": 0.5244147157190635,
      "grad_norm": 0.6746870279312134,
      "learning_rate": 8.325078793336335e-05,
      "loss": 2.0599,
      "step": 392
    },
    {
      "epoch": 0.525752508361204,
      "grad_norm": 1.0724132061004639,
      "learning_rate": 8.320576316974337e-05,
      "loss": 2.1421,
      "step": 393
    },
    {
      "epoch": 0.5270903010033445,
      "grad_norm": 1.250876545906067,
      "learning_rate": 8.316073840612338e-05,
      "loss": 1.8395,
      "step": 394
    },
    {
      "epoch": 0.5284280936454849,
      "grad_norm": 0.6708439588546753,
      "learning_rate": 8.311571364250337e-05,
      "loss": 2.0683,
      "step": 395
    },
    {
      "epoch": 0.5297658862876254,
      "grad_norm": 0.5887036919593811,
      "learning_rate": 8.307068887888338e-05,
      "loss": 2.127,
      "step": 396
    },
    {
      "epoch": 0.5311036789297658,
      "grad_norm": 0.5943152904510498,
      "learning_rate": 8.30256641152634e-05,
      "loss": 1.7854,
      "step": 397
    },
    {
      "epoch": 0.5324414715719064,
      "grad_norm": 1.3241448402404785,
      "learning_rate": 8.298063935164341e-05,
      "loss": 2.1065,
      "step": 398
    },
    {
      "epoch": 0.5337792642140469,
      "grad_norm": 0.6816858053207397,
      "learning_rate": 8.293561458802341e-05,
      "loss": 2.102,
      "step": 399
    },
    {
      "epoch": 0.5351170568561873,
      "grad_norm": 0.6760022640228271,
      "learning_rate": 8.289058982440343e-05,
      "loss": 2.212,
      "step": 400
    },
    {
      "epoch": 0.5364548494983278,
      "grad_norm": 0.6194477081298828,
      "learning_rate": 8.284556506078344e-05,
      "loss": 2.1954,
      "step": 401
    },
    {
      "epoch": 0.5377926421404682,
      "grad_norm": 0.7603004574775696,
      "learning_rate": 8.280054029716345e-05,
      "loss": 1.7049,
      "step": 402
    },
    {
      "epoch": 0.5391304347826087,
      "grad_norm": 0.7374423146247864,
      "learning_rate": 8.275551553354344e-05,
      "loss": 2.2612,
      "step": 403
    },
    {
      "epoch": 0.5404682274247492,
      "grad_norm": 1.0149405002593994,
      "learning_rate": 8.271049076992347e-05,
      "loss": 2.253,
      "step": 404
    },
    {
      "epoch": 0.5418060200668896,
      "grad_norm": 3.6246485710144043,
      "learning_rate": 8.266546600630347e-05,
      "loss": 2.2174,
      "step": 405
    },
    {
      "epoch": 0.5431438127090301,
      "grad_norm": 0.8437290787696838,
      "learning_rate": 8.262044124268348e-05,
      "loss": 2.3438,
      "step": 406
    },
    {
      "epoch": 0.5444816053511705,
      "grad_norm": 1.9442620277404785,
      "learning_rate": 8.257541647906349e-05,
      "loss": 2.1367,
      "step": 407
    },
    {
      "epoch": 0.545819397993311,
      "grad_norm": 0.688398540019989,
      "learning_rate": 8.25303917154435e-05,
      "loss": 2.0249,
      "step": 408
    },
    {
      "epoch": 0.5471571906354515,
      "grad_norm": 1.0315051078796387,
      "learning_rate": 8.248536695182351e-05,
      "loss": 2.2462,
      "step": 409
    },
    {
      "epoch": 0.5484949832775919,
      "grad_norm": 0.5843492746353149,
      "learning_rate": 8.244034218820352e-05,
      "loss": 2.283,
      "step": 410
    },
    {
      "epoch": 0.5498327759197325,
      "grad_norm": 0.6606890559196472,
      "learning_rate": 8.239531742458353e-05,
      "loss": 2.1271,
      "step": 411
    },
    {
      "epoch": 0.551170568561873,
      "grad_norm": 1.0793790817260742,
      "learning_rate": 8.235029266096353e-05,
      "loss": 1.8262,
      "step": 412
    },
    {
      "epoch": 0.5525083612040134,
      "grad_norm": 0.7052550315856934,
      "learning_rate": 8.230526789734354e-05,
      "loss": 2.0957,
      "step": 413
    },
    {
      "epoch": 0.5538461538461539,
      "grad_norm": 0.5878149271011353,
      "learning_rate": 8.226024313372355e-05,
      "loss": 2.0454,
      "step": 414
    },
    {
      "epoch": 0.5551839464882943,
      "grad_norm": 0.7138032913208008,
      "learning_rate": 8.221521837010356e-05,
      "loss": 1.3474,
      "step": 415
    },
    {
      "epoch": 0.5565217391304348,
      "grad_norm": 0.7557210922241211,
      "learning_rate": 8.217019360648357e-05,
      "loss": 1.8938,
      "step": 416
    },
    {
      "epoch": 0.5578595317725753,
      "grad_norm": 2.0228683948516846,
      "learning_rate": 8.212516884286358e-05,
      "loss": 1.7706,
      "step": 417
    },
    {
      "epoch": 0.5591973244147157,
      "grad_norm": 1.115006685256958,
      "learning_rate": 8.20801440792436e-05,
      "loss": 1.9386,
      "step": 418
    },
    {
      "epoch": 0.5605351170568562,
      "grad_norm": 0.6833808422088623,
      "learning_rate": 8.20351193156236e-05,
      "loss": 2.2975,
      "step": 419
    },
    {
      "epoch": 0.5618729096989966,
      "grad_norm": 0.7261201739311218,
      "learning_rate": 8.19900945520036e-05,
      "loss": 1.9896,
      "step": 420
    },
    {
      "epoch": 0.5632107023411371,
      "grad_norm": 0.5823450088500977,
      "learning_rate": 8.194506978838361e-05,
      "loss": 2.0871,
      "step": 421
    },
    {
      "epoch": 0.5645484949832776,
      "grad_norm": 1.1205475330352783,
      "learning_rate": 8.190004502476363e-05,
      "loss": 1.6322,
      "step": 422
    },
    {
      "epoch": 0.565886287625418,
      "grad_norm": 1.0447696447372437,
      "learning_rate": 8.185502026114363e-05,
      "loss": 1.9021,
      "step": 423
    },
    {
      "epoch": 0.5672240802675586,
      "grad_norm": 0.7044191956520081,
      "learning_rate": 8.180999549752364e-05,
      "loss": 2.1777,
      "step": 424
    },
    {
      "epoch": 0.568561872909699,
      "grad_norm": 0.729425311088562,
      "learning_rate": 8.176497073390366e-05,
      "loss": 2.1134,
      "step": 425
    },
    {
      "epoch": 0.5698996655518395,
      "grad_norm": 0.5844915509223938,
      "learning_rate": 8.171994597028367e-05,
      "loss": 2.1628,
      "step": 426
    },
    {
      "epoch": 0.57123745819398,
      "grad_norm": 0.7304236888885498,
      "learning_rate": 8.167492120666366e-05,
      "loss": 2.0648,
      "step": 427
    },
    {
      "epoch": 0.5725752508361204,
      "grad_norm": 0.6491372585296631,
      "learning_rate": 8.162989644304367e-05,
      "loss": 2.2494,
      "step": 428
    },
    {
      "epoch": 0.5739130434782609,
      "grad_norm": 1.2159571647644043,
      "learning_rate": 8.15848716794237e-05,
      "loss": 1.9377,
      "step": 429
    },
    {
      "epoch": 0.5752508361204013,
      "grad_norm": 0.6987440586090088,
      "learning_rate": 8.15398469158037e-05,
      "loss": 2.0905,
      "step": 430
    },
    {
      "epoch": 0.5765886287625418,
      "grad_norm": 0.7087085843086243,
      "learning_rate": 8.14948221521837e-05,
      "loss": 2.0827,
      "step": 431
    },
    {
      "epoch": 0.5779264214046823,
      "grad_norm": 0.6850048303604126,
      "learning_rate": 8.14497973885637e-05,
      "loss": 2.0597,
      "step": 432
    },
    {
      "epoch": 0.5792642140468227,
      "grad_norm": 0.6765483617782593,
      "learning_rate": 8.140477262494373e-05,
      "loss": 2.1179,
      "step": 433
    },
    {
      "epoch": 0.5806020066889632,
      "grad_norm": 0.7441734075546265,
      "learning_rate": 8.135974786132374e-05,
      "loss": 1.92,
      "step": 434
    },
    {
      "epoch": 0.5819397993311036,
      "grad_norm": 0.5449488759040833,
      "learning_rate": 8.131472309770373e-05,
      "loss": 1.6044,
      "step": 435
    },
    {
      "epoch": 0.5832775919732441,
      "grad_norm": 1.5431065559387207,
      "learning_rate": 8.126969833408375e-05,
      "loss": 2.3955,
      "step": 436
    },
    {
      "epoch": 0.5846153846153846,
      "grad_norm": 0.9129299521446228,
      "learning_rate": 8.122467357046376e-05,
      "loss": 2.1519,
      "step": 437
    },
    {
      "epoch": 0.585953177257525,
      "grad_norm": 0.811627209186554,
      "learning_rate": 8.117964880684377e-05,
      "loss": 1.9321,
      "step": 438
    },
    {
      "epoch": 0.5872909698996656,
      "grad_norm": 0.7428467273712158,
      "learning_rate": 8.113462404322378e-05,
      "loss": 1.9116,
      "step": 439
    },
    {
      "epoch": 0.5886287625418061,
      "grad_norm": 0.639267086982727,
      "learning_rate": 8.108959927960379e-05,
      "loss": 2.1715,
      "step": 440
    },
    {
      "epoch": 0.5899665551839465,
      "grad_norm": 0.6075000762939453,
      "learning_rate": 8.10445745159838e-05,
      "loss": 2.1022,
      "step": 441
    },
    {
      "epoch": 0.591304347826087,
      "grad_norm": 0.9672107696533203,
      "learning_rate": 8.09995497523638e-05,
      "loss": 2.2003,
      "step": 442
    },
    {
      "epoch": 0.5926421404682274,
      "grad_norm": 0.6427491307258606,
      "learning_rate": 8.095452498874381e-05,
      "loss": 2.1691,
      "step": 443
    },
    {
      "epoch": 0.5939799331103679,
      "grad_norm": 0.6234250664710999,
      "learning_rate": 8.090950022512382e-05,
      "loss": 2.0638,
      "step": 444
    },
    {
      "epoch": 0.5953177257525084,
      "grad_norm": 0.7238093018531799,
      "learning_rate": 8.086447546150383e-05,
      "loss": 1.9306,
      "step": 445
    },
    {
      "epoch": 0.5966555183946488,
      "grad_norm": 0.8159562349319458,
      "learning_rate": 8.081945069788384e-05,
      "loss": 2.1521,
      "step": 446
    },
    {
      "epoch": 0.5979933110367893,
      "grad_norm": 0.7235854268074036,
      "learning_rate": 8.077442593426385e-05,
      "loss": 2.0037,
      "step": 447
    },
    {
      "epoch": 0.5993311036789297,
      "grad_norm": 0.6752901673316956,
      "learning_rate": 8.072940117064386e-05,
      "loss": 1.8655,
      "step": 448
    },
    {
      "epoch": 0.6006688963210702,
      "grad_norm": 0.6684075593948364,
      "learning_rate": 8.068437640702387e-05,
      "loss": 1.8252,
      "step": 449
    },
    {
      "epoch": 0.6020066889632107,
      "grad_norm": 1.1071723699569702,
      "learning_rate": 8.063935164340387e-05,
      "loss": 1.9672,
      "step": 450
    },
    {
      "epoch": 0.6033444816053511,
      "grad_norm": 1.1972233057022095,
      "learning_rate": 8.059432687978388e-05,
      "loss": 1.9875,
      "step": 451
    },
    {
      "epoch": 0.6046822742474917,
      "grad_norm": 0.9624909162521362,
      "learning_rate": 8.054930211616389e-05,
      "loss": 2.1065,
      "step": 452
    },
    {
      "epoch": 0.6060200668896321,
      "grad_norm": 1.122995138168335,
      "learning_rate": 8.05042773525439e-05,
      "loss": 1.8011,
      "step": 453
    },
    {
      "epoch": 0.6073578595317726,
      "grad_norm": 0.7040053009986877,
      "learning_rate": 8.045925258892392e-05,
      "loss": 2.256,
      "step": 454
    },
    {
      "epoch": 0.6086956521739131,
      "grad_norm": 0.6881429553031921,
      "learning_rate": 8.041422782530392e-05,
      "loss": 2.0402,
      "step": 455
    },
    {
      "epoch": 0.6100334448160535,
      "grad_norm": 0.7463430166244507,
      "learning_rate": 8.036920306168393e-05,
      "loss": 2.0015,
      "step": 456
    },
    {
      "epoch": 0.611371237458194,
      "grad_norm": 1.0377933979034424,
      "learning_rate": 8.032417829806393e-05,
      "loss": 2.1018,
      "step": 457
    },
    {
      "epoch": 0.6127090301003344,
      "grad_norm": 0.9001895785331726,
      "learning_rate": 8.027915353444396e-05,
      "loss": 2.3027,
      "step": 458
    },
    {
      "epoch": 0.6140468227424749,
      "grad_norm": 5.800236225128174,
      "learning_rate": 8.023412877082395e-05,
      "loss": 1.8693,
      "step": 459
    },
    {
      "epoch": 0.6153846153846154,
      "grad_norm": 0.8842706680297852,
      "learning_rate": 8.018910400720396e-05,
      "loss": 2.0381,
      "step": 460
    },
    {
      "epoch": 0.6167224080267558,
      "grad_norm": 0.8452533483505249,
      "learning_rate": 8.014407924358398e-05,
      "loss": 2.3479,
      "step": 461
    },
    {
      "epoch": 0.6180602006688963,
      "grad_norm": 0.6322413682937622,
      "learning_rate": 8.009905447996399e-05,
      "loss": 2.3138,
      "step": 462
    },
    {
      "epoch": 0.6193979933110368,
      "grad_norm": 0.6229380369186401,
      "learning_rate": 8.0054029716344e-05,
      "loss": 2.2976,
      "step": 463
    },
    {
      "epoch": 0.6207357859531772,
      "grad_norm": 0.6563906073570251,
      "learning_rate": 8.0009004952724e-05,
      "loss": 1.8494,
      "step": 464
    },
    {
      "epoch": 0.6220735785953178,
      "grad_norm": 0.7984792590141296,
      "learning_rate": 7.996398018910402e-05,
      "loss": 2.2518,
      "step": 465
    },
    {
      "epoch": 0.6234113712374582,
      "grad_norm": 0.7305912375450134,
      "learning_rate": 7.991895542548402e-05,
      "loss": 2.4169,
      "step": 466
    },
    {
      "epoch": 0.6247491638795987,
      "grad_norm": 0.6074129343032837,
      "learning_rate": 7.987393066186403e-05,
      "loss": 1.9538,
      "step": 467
    },
    {
      "epoch": 0.6260869565217392,
      "grad_norm": 0.9291728138923645,
      "learning_rate": 7.982890589824403e-05,
      "loss": 2.1267,
      "step": 468
    },
    {
      "epoch": 0.6274247491638796,
      "grad_norm": 0.754958987236023,
      "learning_rate": 7.978388113462405e-05,
      "loss": 1.5873,
      "step": 469
    },
    {
      "epoch": 0.6287625418060201,
      "grad_norm": 0.6374825835227966,
      "learning_rate": 7.973885637100406e-05,
      "loss": 2.3097,
      "step": 470
    },
    {
      "epoch": 0.6301003344481605,
      "grad_norm": 0.6864813566207886,
      "learning_rate": 7.969383160738407e-05,
      "loss": 2.144,
      "step": 471
    },
    {
      "epoch": 0.631438127090301,
      "grad_norm": 0.7135668992996216,
      "learning_rate": 7.964880684376408e-05,
      "loss": 2.1547,
      "step": 472
    },
    {
      "epoch": 0.6327759197324415,
      "grad_norm": 0.7295858263969421,
      "learning_rate": 7.960378208014408e-05,
      "loss": 2.1507,
      "step": 473
    },
    {
      "epoch": 0.6341137123745819,
      "grad_norm": 0.6061326861381531,
      "learning_rate": 7.95587573165241e-05,
      "loss": 2.1332,
      "step": 474
    },
    {
      "epoch": 0.6354515050167224,
      "grad_norm": 0.5801835656166077,
      "learning_rate": 7.95137325529041e-05,
      "loss": 2.3511,
      "step": 475
    },
    {
      "epoch": 0.6367892976588628,
      "grad_norm": 0.880810022354126,
      "learning_rate": 7.946870778928411e-05,
      "loss": 1.8928,
      "step": 476
    },
    {
      "epoch": 0.6381270903010033,
      "grad_norm": 0.6856786012649536,
      "learning_rate": 7.942368302566412e-05,
      "loss": 2.2853,
      "step": 477
    },
    {
      "epoch": 0.6394648829431439,
      "grad_norm": 0.8725216388702393,
      "learning_rate": 7.937865826204413e-05,
      "loss": 2.0342,
      "step": 478
    },
    {
      "epoch": 0.6408026755852843,
      "grad_norm": 0.6242175102233887,
      "learning_rate": 7.933363349842414e-05,
      "loss": 2.2387,
      "step": 479
    },
    {
      "epoch": 0.6421404682274248,
      "grad_norm": 0.6608566045761108,
      "learning_rate": 7.928860873480414e-05,
      "loss": 1.9173,
      "step": 480
    },
    {
      "epoch": 0.6434782608695652,
      "grad_norm": 0.7391323447227478,
      "learning_rate": 7.924358397118415e-05,
      "loss": 1.9619,
      "step": 481
    },
    {
      "epoch": 0.6448160535117057,
      "grad_norm": 3.0505142211914062,
      "learning_rate": 7.919855920756416e-05,
      "loss": 1.6748,
      "step": 482
    },
    {
      "epoch": 0.6461538461538462,
      "grad_norm": 0.6999809145927429,
      "learning_rate": 7.915353444394417e-05,
      "loss": 2.0176,
      "step": 483
    },
    {
      "epoch": 0.6474916387959866,
      "grad_norm": 0.5944308638572693,
      "learning_rate": 7.910850968032418e-05,
      "loss": 2.2038,
      "step": 484
    },
    {
      "epoch": 0.6488294314381271,
      "grad_norm": 0.5410728454589844,
      "learning_rate": 7.906348491670419e-05,
      "loss": 2.2845,
      "step": 485
    },
    {
      "epoch": 0.6501672240802676,
      "grad_norm": 1.1120905876159668,
      "learning_rate": 7.90184601530842e-05,
      "loss": 2.2103,
      "step": 486
    },
    {
      "epoch": 0.651505016722408,
      "grad_norm": 0.7039376497268677,
      "learning_rate": 7.897343538946422e-05,
      "loss": 2.0109,
      "step": 487
    },
    {
      "epoch": 0.6528428093645485,
      "grad_norm": 0.5916752815246582,
      "learning_rate": 7.892841062584421e-05,
      "loss": 1.8574,
      "step": 488
    },
    {
      "epoch": 0.6541806020066889,
      "grad_norm": 0.6692883968353271,
      "learning_rate": 7.888338586222422e-05,
      "loss": 2.0059,
      "step": 489
    },
    {
      "epoch": 0.6555183946488294,
      "grad_norm": 0.7776148319244385,
      "learning_rate": 7.883836109860424e-05,
      "loss": 1.5996,
      "step": 490
    },
    {
      "epoch": 0.65685618729097,
      "grad_norm": 0.6935766935348511,
      "learning_rate": 7.879333633498425e-05,
      "loss": 2.0559,
      "step": 491
    },
    {
      "epoch": 0.6581939799331104,
      "grad_norm": 0.6067877411842346,
      "learning_rate": 7.874831157136425e-05,
      "loss": 2.1315,
      "step": 492
    },
    {
      "epoch": 0.6595317725752509,
      "grad_norm": 0.8041082620620728,
      "learning_rate": 7.870328680774426e-05,
      "loss": 2.221,
      "step": 493
    },
    {
      "epoch": 0.6608695652173913,
      "grad_norm": 0.8216106295585632,
      "learning_rate": 7.865826204412428e-05,
      "loss": 1.5927,
      "step": 494
    },
    {
      "epoch": 0.6622073578595318,
      "grad_norm": 0.6823213696479797,
      "learning_rate": 7.861323728050429e-05,
      "loss": 2.0342,
      "step": 495
    },
    {
      "epoch": 0.6635451505016723,
      "grad_norm": 0.8577303290367126,
      "learning_rate": 7.856821251688428e-05,
      "loss": 2.1152,
      "step": 496
    },
    {
      "epoch": 0.6648829431438127,
      "grad_norm": 0.6728313565254211,
      "learning_rate": 7.85231877532643e-05,
      "loss": 2.0798,
      "step": 497
    },
    {
      "epoch": 0.6662207357859532,
      "grad_norm": 0.7217726707458496,
      "learning_rate": 7.847816298964431e-05,
      "loss": 2.0957,
      "step": 498
    },
    {
      "epoch": 0.6675585284280936,
      "grad_norm": 0.6146303415298462,
      "learning_rate": 7.843313822602432e-05,
      "loss": 2.0668,
      "step": 499
    },
    {
      "epoch": 0.6688963210702341,
      "grad_norm": 0.9603506922721863,
      "learning_rate": 7.838811346240432e-05,
      "loss": 2.2326,
      "step": 500
    },
    {
      "epoch": 0.6702341137123746,
      "grad_norm": 0.5818818807601929,
      "learning_rate": 7.834308869878434e-05,
      "loss": 2.182,
      "step": 501
    },
    {
      "epoch": 0.671571906354515,
      "grad_norm": 0.6627907752990723,
      "learning_rate": 7.829806393516435e-05,
      "loss": 2.3262,
      "step": 502
    },
    {
      "epoch": 0.6729096989966555,
      "grad_norm": 0.7816573977470398,
      "learning_rate": 7.825303917154436e-05,
      "loss": 1.7924,
      "step": 503
    },
    {
      "epoch": 0.6742474916387959,
      "grad_norm": 0.7198929786682129,
      "learning_rate": 7.820801440792435e-05,
      "loss": 2.2945,
      "step": 504
    },
    {
      "epoch": 0.6755852842809364,
      "grad_norm": 0.6396793723106384,
      "learning_rate": 7.816298964430437e-05,
      "loss": 2.0723,
      "step": 505
    },
    {
      "epoch": 0.676923076923077,
      "grad_norm": 1.1741186380386353,
      "learning_rate": 7.811796488068438e-05,
      "loss": 1.5519,
      "step": 506
    },
    {
      "epoch": 0.6782608695652174,
      "grad_norm": 1.1423894166946411,
      "learning_rate": 7.807294011706439e-05,
      "loss": 1.9561,
      "step": 507
    },
    {
      "epoch": 0.6795986622073579,
      "grad_norm": 0.7496652603149414,
      "learning_rate": 7.80279153534444e-05,
      "loss": 2.0628,
      "step": 508
    },
    {
      "epoch": 0.6809364548494983,
      "grad_norm": 0.7691856622695923,
      "learning_rate": 7.798289058982441e-05,
      "loss": 2.1493,
      "step": 509
    },
    {
      "epoch": 0.6822742474916388,
      "grad_norm": 0.7692395448684692,
      "learning_rate": 7.793786582620442e-05,
      "loss": 1.9799,
      "step": 510
    },
    {
      "epoch": 0.6836120401337793,
      "grad_norm": 0.5502052903175354,
      "learning_rate": 7.789284106258442e-05,
      "loss": 2.0543,
      "step": 511
    },
    {
      "epoch": 0.6849498327759197,
      "grad_norm": 0.6818161010742188,
      "learning_rate": 7.784781629896443e-05,
      "loss": 1.943,
      "step": 512
    },
    {
      "epoch": 0.6862876254180602,
      "grad_norm": 0.7240835428237915,
      "learning_rate": 7.780279153534444e-05,
      "loss": 1.5104,
      "step": 513
    },
    {
      "epoch": 0.6876254180602007,
      "grad_norm": 0.7366359829902649,
      "learning_rate": 7.775776677172445e-05,
      "loss": 2.0822,
      "step": 514
    },
    {
      "epoch": 0.6889632107023411,
      "grad_norm": 1.0837323665618896,
      "learning_rate": 7.771274200810447e-05,
      "loss": 2.289,
      "step": 515
    },
    {
      "epoch": 0.6903010033444816,
      "grad_norm": 0.651714026927948,
      "learning_rate": 7.766771724448447e-05,
      "loss": 1.8688,
      "step": 516
    },
    {
      "epoch": 0.691638795986622,
      "grad_norm": 0.6614827513694763,
      "learning_rate": 7.762269248086448e-05,
      "loss": 1.6443,
      "step": 517
    },
    {
      "epoch": 0.6929765886287625,
      "grad_norm": 0.7544852495193481,
      "learning_rate": 7.757766771724448e-05,
      "loss": 1.6606,
      "step": 518
    },
    {
      "epoch": 0.6943143812709031,
      "grad_norm": 0.6697096824645996,
      "learning_rate": 7.753264295362451e-05,
      "loss": 1.5946,
      "step": 519
    },
    {
      "epoch": 0.6956521739130435,
      "grad_norm": 0.7795844674110413,
      "learning_rate": 7.74876181900045e-05,
      "loss": 2.059,
      "step": 520
    },
    {
      "epoch": 0.696989966555184,
      "grad_norm": 0.6018288731575012,
      "learning_rate": 7.744259342638451e-05,
      "loss": 2.1916,
      "step": 521
    },
    {
      "epoch": 0.6983277591973244,
      "grad_norm": 0.7487697601318359,
      "learning_rate": 7.739756866276452e-05,
      "loss": 2.1951,
      "step": 522
    },
    {
      "epoch": 0.6996655518394649,
      "grad_norm": 1.5919585227966309,
      "learning_rate": 7.735254389914454e-05,
      "loss": 2.2205,
      "step": 523
    },
    {
      "epoch": 0.7010033444816054,
      "grad_norm": 0.6275078058242798,
      "learning_rate": 7.730751913552454e-05,
      "loss": 2.1698,
      "step": 524
    },
    {
      "epoch": 0.7023411371237458,
      "grad_norm": 0.7217388153076172,
      "learning_rate": 7.726249437190454e-05,
      "loss": 1.9864,
      "step": 525
    },
    {
      "epoch": 0.7036789297658863,
      "grad_norm": 1.1377520561218262,
      "learning_rate": 7.721746960828457e-05,
      "loss": 2.1668,
      "step": 526
    },
    {
      "epoch": 0.7050167224080267,
      "grad_norm": 0.718572735786438,
      "learning_rate": 7.717244484466458e-05,
      "loss": 1.928,
      "step": 527
    },
    {
      "epoch": 0.7063545150501672,
      "grad_norm": 0.765325665473938,
      "learning_rate": 7.712742008104457e-05,
      "loss": 2.1302,
      "step": 528
    },
    {
      "epoch": 0.7076923076923077,
      "grad_norm": 1.549810528755188,
      "learning_rate": 7.708239531742458e-05,
      "loss": 2.189,
      "step": 529
    },
    {
      "epoch": 0.7090301003344481,
      "grad_norm": 2.3518106937408447,
      "learning_rate": 7.70373705538046e-05,
      "loss": 1.9373,
      "step": 530
    },
    {
      "epoch": 0.7103678929765886,
      "grad_norm": 1.188962697982788,
      "learning_rate": 7.699234579018461e-05,
      "loss": 1.6998,
      "step": 531
    },
    {
      "epoch": 0.711705685618729,
      "grad_norm": 0.9281237721443176,
      "learning_rate": 7.69473210265646e-05,
      "loss": 2.1245,
      "step": 532
    },
    {
      "epoch": 0.7130434782608696,
      "grad_norm": 0.6713751554489136,
      "learning_rate": 7.690229626294463e-05,
      "loss": 2.3883,
      "step": 533
    },
    {
      "epoch": 0.7143812709030101,
      "grad_norm": 1.8563570976257324,
      "learning_rate": 7.685727149932464e-05,
      "loss": 2.4456,
      "step": 534
    },
    {
      "epoch": 0.7157190635451505,
      "grad_norm": 0.6634758710861206,
      "learning_rate": 7.681224673570464e-05,
      "loss": 2.2008,
      "step": 535
    },
    {
      "epoch": 0.717056856187291,
      "grad_norm": 0.5978294014930725,
      "learning_rate": 7.676722197208465e-05,
      "loss": 2.0657,
      "step": 536
    },
    {
      "epoch": 0.7183946488294315,
      "grad_norm": 0.6472600698471069,
      "learning_rate": 7.672219720846466e-05,
      "loss": 2.1728,
      "step": 537
    },
    {
      "epoch": 0.7197324414715719,
      "grad_norm": 0.6482618451118469,
      "learning_rate": 7.667717244484467e-05,
      "loss": 2.446,
      "step": 538
    },
    {
      "epoch": 0.7210702341137124,
      "grad_norm": 0.6306278705596924,
      "learning_rate": 7.663214768122468e-05,
      "loss": 1.5909,
      "step": 539
    },
    {
      "epoch": 0.7224080267558528,
      "grad_norm": 0.6908814907073975,
      "learning_rate": 7.658712291760469e-05,
      "loss": 2.1809,
      "step": 540
    },
    {
      "epoch": 0.7237458193979933,
      "grad_norm": 0.8022807240486145,
      "learning_rate": 7.65420981539847e-05,
      "loss": 2.227,
      "step": 541
    },
    {
      "epoch": 0.7250836120401338,
      "grad_norm": 0.7036264538764954,
      "learning_rate": 7.64970733903647e-05,
      "loss": 1.8489,
      "step": 542
    },
    {
      "epoch": 0.7264214046822742,
      "grad_norm": 0.5983272194862366,
      "learning_rate": 7.645204862674471e-05,
      "loss": 2.0517,
      "step": 543
    },
    {
      "epoch": 0.7277591973244147,
      "grad_norm": 0.6938000917434692,
      "learning_rate": 7.640702386312472e-05,
      "loss": 2.0043,
      "step": 544
    },
    {
      "epoch": 0.7290969899665551,
      "grad_norm": 1.180989384651184,
      "learning_rate": 7.636199909950473e-05,
      "loss": 1.6897,
      "step": 545
    },
    {
      "epoch": 0.7304347826086957,
      "grad_norm": 1.2202868461608887,
      "learning_rate": 7.631697433588474e-05,
      "loss": 2.2268,
      "step": 546
    },
    {
      "epoch": 0.7317725752508362,
      "grad_norm": 0.6938366889953613,
      "learning_rate": 7.627194957226475e-05,
      "loss": 1.9375,
      "step": 547
    },
    {
      "epoch": 0.7331103678929766,
      "grad_norm": 0.8012475371360779,
      "learning_rate": 7.622692480864476e-05,
      "loss": 2.5131,
      "step": 548
    },
    {
      "epoch": 0.7344481605351171,
      "grad_norm": 0.7444472908973694,
      "learning_rate": 7.618190004502476e-05,
      "loss": 1.9647,
      "step": 549
    },
    {
      "epoch": 0.7357859531772575,
      "grad_norm": 0.6282299160957336,
      "learning_rate": 7.613687528140477e-05,
      "loss": 2.2758,
      "step": 550
    },
    {
      "epoch": 0.737123745819398,
      "grad_norm": 0.6200249195098877,
      "learning_rate": 7.60918505177848e-05,
      "loss": 2.1907,
      "step": 551
    },
    {
      "epoch": 0.7384615384615385,
      "grad_norm": 0.7774984836578369,
      "learning_rate": 7.604682575416479e-05,
      "loss": 2.2005,
      "step": 552
    },
    {
      "epoch": 0.7397993311036789,
      "grad_norm": 0.6397766470909119,
      "learning_rate": 7.60018009905448e-05,
      "loss": 1.0321,
      "step": 553
    },
    {
      "epoch": 0.7411371237458194,
      "grad_norm": 0.7659112811088562,
      "learning_rate": 7.595677622692481e-05,
      "loss": 1.5511,
      "step": 554
    },
    {
      "epoch": 0.7424749163879598,
      "grad_norm": 0.540540874004364,
      "learning_rate": 7.591175146330483e-05,
      "loss": 2.1449,
      "step": 555
    },
    {
      "epoch": 0.7438127090301003,
      "grad_norm": 0.6674241423606873,
      "learning_rate": 7.586672669968482e-05,
      "loss": 2.1938,
      "step": 556
    },
    {
      "epoch": 0.7451505016722408,
      "grad_norm": 0.651637077331543,
      "learning_rate": 7.582170193606483e-05,
      "loss": 1.995,
      "step": 557
    },
    {
      "epoch": 0.7464882943143812,
      "grad_norm": 1.0496723651885986,
      "learning_rate": 7.577667717244484e-05,
      "loss": 1.5,
      "step": 558
    },
    {
      "epoch": 0.7478260869565218,
      "grad_norm": 0.6831291317939758,
      "learning_rate": 7.573165240882486e-05,
      "loss": 1.8952,
      "step": 559
    },
    {
      "epoch": 0.7491638795986622,
      "grad_norm": 0.6953374147415161,
      "learning_rate": 7.568662764520487e-05,
      "loss": 2.3361,
      "step": 560
    },
    {
      "epoch": 0.7505016722408027,
      "grad_norm": 0.8107741475105286,
      "learning_rate": 7.564160288158487e-05,
      "loss": 1.7492,
      "step": 561
    },
    {
      "epoch": 0.7518394648829432,
      "grad_norm": 0.6309297680854797,
      "learning_rate": 7.559657811796489e-05,
      "loss": 1.9916,
      "step": 562
    },
    {
      "epoch": 0.7531772575250836,
      "grad_norm": 1.0587294101715088,
      "learning_rate": 7.55515533543449e-05,
      "loss": 2.224,
      "step": 563
    },
    {
      "epoch": 0.7545150501672241,
      "grad_norm": 0.7029111981391907,
      "learning_rate": 7.550652859072491e-05,
      "loss": 1.9918,
      "step": 564
    },
    {
      "epoch": 0.7558528428093646,
      "grad_norm": 0.6372597217559814,
      "learning_rate": 7.54615038271049e-05,
      "loss": 1.6158,
      "step": 565
    },
    {
      "epoch": 0.757190635451505,
      "grad_norm": 0.7088492512702942,
      "learning_rate": 7.541647906348492e-05,
      "loss": 2.2011,
      "step": 566
    },
    {
      "epoch": 0.7585284280936455,
      "grad_norm": 0.6593747735023499,
      "learning_rate": 7.537145429986493e-05,
      "loss": 1.5873,
      "step": 567
    },
    {
      "epoch": 0.7598662207357859,
      "grad_norm": 0.7396959066390991,
      "learning_rate": 7.532642953624494e-05,
      "loss": 1.7784,
      "step": 568
    },
    {
      "epoch": 0.7612040133779264,
      "grad_norm": 0.9161505103111267,
      "learning_rate": 7.528140477262495e-05,
      "loss": 1.7815,
      "step": 569
    },
    {
      "epoch": 0.7625418060200669,
      "grad_norm": 0.6812069416046143,
      "learning_rate": 7.523638000900496e-05,
      "loss": 2.1424,
      "step": 570
    },
    {
      "epoch": 0.7638795986622073,
      "grad_norm": 0.6250039339065552,
      "learning_rate": 7.519135524538497e-05,
      "loss": 1.8774,
      "step": 571
    },
    {
      "epoch": 0.7652173913043478,
      "grad_norm": 0.7947536706924438,
      "learning_rate": 7.514633048176498e-05,
      "loss": 2.2003,
      "step": 572
    },
    {
      "epoch": 0.7665551839464882,
      "grad_norm": 0.6720924377441406,
      "learning_rate": 7.510130571814498e-05,
      "loss": 2.2568,
      "step": 573
    },
    {
      "epoch": 0.7678929765886288,
      "grad_norm": 0.8429411053657532,
      "learning_rate": 7.505628095452499e-05,
      "loss": 1.8639,
      "step": 574
    },
    {
      "epoch": 0.7692307692307693,
      "grad_norm": 0.7361516952514648,
      "learning_rate": 7.5011256190905e-05,
      "loss": 1.7005,
      "step": 575
    },
    {
      "epoch": 0.7705685618729097,
      "grad_norm": 1.273979663848877,
      "learning_rate": 7.496623142728501e-05,
      "loss": 2.4655,
      "step": 576
    },
    {
      "epoch": 0.7719063545150502,
      "grad_norm": 1.8873357772827148,
      "learning_rate": 7.492120666366502e-05,
      "loss": 2.338,
      "step": 577
    },
    {
      "epoch": 0.7732441471571906,
      "grad_norm": 0.571962296962738,
      "learning_rate": 7.487618190004503e-05,
      "loss": 2.0726,
      "step": 578
    },
    {
      "epoch": 0.7745819397993311,
      "grad_norm": 0.6177148818969727,
      "learning_rate": 7.483115713642504e-05,
      "loss": 2.1553,
      "step": 579
    },
    {
      "epoch": 0.7759197324414716,
      "grad_norm": 0.8624194860458374,
      "learning_rate": 7.478613237280504e-05,
      "loss": 2.4043,
      "step": 580
    },
    {
      "epoch": 0.777257525083612,
      "grad_norm": 0.5737015604972839,
      "learning_rate": 7.474110760918505e-05,
      "loss": 1.6189,
      "step": 581
    },
    {
      "epoch": 0.7785953177257525,
      "grad_norm": 0.6640171408653259,
      "learning_rate": 7.469608284556506e-05,
      "loss": 2.3091,
      "step": 582
    },
    {
      "epoch": 0.7799331103678929,
      "grad_norm": 1.1256778240203857,
      "learning_rate": 7.465105808194507e-05,
      "loss": 2.1997,
      "step": 583
    },
    {
      "epoch": 0.7812709030100334,
      "grad_norm": 0.6835268139839172,
      "learning_rate": 7.460603331832509e-05,
      "loss": 2.019,
      "step": 584
    },
    {
      "epoch": 0.782608695652174,
      "grad_norm": 0.6177977919578552,
      "learning_rate": 7.456100855470509e-05,
      "loss": 2.3367,
      "step": 585
    },
    {
      "epoch": 0.7839464882943143,
      "grad_norm": 0.7783511877059937,
      "learning_rate": 7.45159837910851e-05,
      "loss": 1.9634,
      "step": 586
    },
    {
      "epoch": 0.7852842809364549,
      "grad_norm": 0.503692626953125,
      "learning_rate": 7.447095902746512e-05,
      "loss": 2.1588,
      "step": 587
    },
    {
      "epoch": 0.7866220735785954,
      "grad_norm": 0.8090397119522095,
      "learning_rate": 7.442593426384513e-05,
      "loss": 1.6442,
      "step": 588
    },
    {
      "epoch": 0.7879598662207358,
      "grad_norm": 0.6573639512062073,
      "learning_rate": 7.438090950022512e-05,
      "loss": 2.2257,
      "step": 589
    },
    {
      "epoch": 0.7892976588628763,
      "grad_norm": 0.7102932929992676,
      "learning_rate": 7.433588473660513e-05,
      "loss": 1.6468,
      "step": 590
    },
    {
      "epoch": 0.7906354515050167,
      "grad_norm": 0.7322196364402771,
      "learning_rate": 7.429085997298515e-05,
      "loss": 2.2053,
      "step": 591
    },
    {
      "epoch": 0.7919732441471572,
      "grad_norm": 0.7007878422737122,
      "learning_rate": 7.424583520936516e-05,
      "loss": 2.0766,
      "step": 592
    },
    {
      "epoch": 0.7933110367892977,
      "grad_norm": 0.6183981895446777,
      "learning_rate": 7.420081044574516e-05,
      "loss": 2.0603,
      "step": 593
    },
    {
      "epoch": 0.7946488294314381,
      "grad_norm": 0.5837265253067017,
      "learning_rate": 7.415578568212516e-05,
      "loss": 2.3202,
      "step": 594
    },
    {
      "epoch": 0.7959866220735786,
      "grad_norm": 1.104719638824463,
      "learning_rate": 7.411076091850519e-05,
      "loss": 1.3282,
      "step": 595
    },
    {
      "epoch": 0.797324414715719,
      "grad_norm": 0.6628216505050659,
      "learning_rate": 7.40657361548852e-05,
      "loss": 2.3223,
      "step": 596
    },
    {
      "epoch": 0.7986622073578595,
      "grad_norm": 0.5468476414680481,
      "learning_rate": 7.402071139126519e-05,
      "loss": 1.7866,
      "step": 597
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.8073409795761108,
      "learning_rate": 7.397568662764521e-05,
      "loss": 1.7743,
      "step": 598
    },
    {
      "epoch": 0.8013377926421404,
      "grad_norm": 1.0640450716018677,
      "learning_rate": 7.393066186402522e-05,
      "loss": 2.1797,
      "step": 599
    },
    {
      "epoch": 0.802675585284281,
      "grad_norm": 0.7137813568115234,
      "learning_rate": 7.388563710040523e-05,
      "loss": 2.0888,
      "step": 600
    },
    {
      "epoch": 0.8040133779264214,
      "grad_norm": 0.5866499543190002,
      "learning_rate": 7.384061233678522e-05,
      "loss": 1.668,
      "step": 601
    },
    {
      "epoch": 0.8053511705685619,
      "grad_norm": 0.7044905424118042,
      "learning_rate": 7.379558757316525e-05,
      "loss": 1.9982,
      "step": 602
    },
    {
      "epoch": 0.8066889632107024,
      "grad_norm": 0.5737768411636353,
      "learning_rate": 7.375056280954526e-05,
      "loss": 2.0302,
      "step": 603
    },
    {
      "epoch": 0.8080267558528428,
      "grad_norm": 0.5827451348304749,
      "learning_rate": 7.370553804592526e-05,
      "loss": 2.3226,
      "step": 604
    },
    {
      "epoch": 0.8093645484949833,
      "grad_norm": 0.7087153792381287,
      "learning_rate": 7.366051328230527e-05,
      "loss": 2.0301,
      "step": 605
    },
    {
      "epoch": 0.8107023411371237,
      "grad_norm": 0.6226567029953003,
      "learning_rate": 7.361548851868528e-05,
      "loss": 1.7037,
      "step": 606
    },
    {
      "epoch": 0.8120401337792642,
      "grad_norm": 0.7785759568214417,
      "learning_rate": 7.357046375506529e-05,
      "loss": 1.7154,
      "step": 607
    },
    {
      "epoch": 0.8133779264214047,
      "grad_norm": 0.777509868144989,
      "learning_rate": 7.35254389914453e-05,
      "loss": 2.0251,
      "step": 608
    },
    {
      "epoch": 0.8147157190635451,
      "grad_norm": 0.6344072222709656,
      "learning_rate": 7.348041422782531e-05,
      "loss": 2.2202,
      "step": 609
    },
    {
      "epoch": 0.8160535117056856,
      "grad_norm": 0.7604480385780334,
      "learning_rate": 7.343538946420532e-05,
      "loss": 1.9614,
      "step": 610
    },
    {
      "epoch": 0.8173913043478261,
      "grad_norm": 0.6319754719734192,
      "learning_rate": 7.339036470058532e-05,
      "loss": 1.6291,
      "step": 611
    },
    {
      "epoch": 0.8187290969899665,
      "grad_norm": 0.9950410723686218,
      "learning_rate": 7.334533993696533e-05,
      "loss": 1.8727,
      "step": 612
    },
    {
      "epoch": 0.820066889632107,
      "grad_norm": 0.6576692461967468,
      "learning_rate": 7.330031517334534e-05,
      "loss": 1.92,
      "step": 613
    },
    {
      "epoch": 0.8214046822742475,
      "grad_norm": 0.5095275640487671,
      "learning_rate": 7.325529040972535e-05,
      "loss": 1.9408,
      "step": 614
    },
    {
      "epoch": 0.822742474916388,
      "grad_norm": 1.188237190246582,
      "learning_rate": 7.321026564610536e-05,
      "loss": 1.9193,
      "step": 615
    },
    {
      "epoch": 0.8240802675585285,
      "grad_norm": 0.7145055532455444,
      "learning_rate": 7.316524088248538e-05,
      "loss": 2.2026,
      "step": 616
    },
    {
      "epoch": 0.8254180602006689,
      "grad_norm": 0.6884480714797974,
      "learning_rate": 7.312021611886538e-05,
      "loss": 2.3884,
      "step": 617
    },
    {
      "epoch": 0.8267558528428094,
      "grad_norm": 0.6090062856674194,
      "learning_rate": 7.307519135524538e-05,
      "loss": 2.1149,
      "step": 618
    },
    {
      "epoch": 0.8280936454849498,
      "grad_norm": 1.0304170846939087,
      "learning_rate": 7.303016659162539e-05,
      "loss": 1.4435,
      "step": 619
    },
    {
      "epoch": 0.8294314381270903,
      "grad_norm": 0.6356250047683716,
      "learning_rate": 7.298514182800542e-05,
      "loss": 1.8632,
      "step": 620
    },
    {
      "epoch": 0.8307692307692308,
      "grad_norm": 0.6120384931564331,
      "learning_rate": 7.294011706438541e-05,
      "loss": 2.0522,
      "step": 621
    },
    {
      "epoch": 0.8321070234113712,
      "grad_norm": 0.6501190662384033,
      "learning_rate": 7.289509230076542e-05,
      "loss": 1.8754,
      "step": 622
    },
    {
      "epoch": 0.8334448160535117,
      "grad_norm": 1.3946788311004639,
      "learning_rate": 7.285006753714544e-05,
      "loss": 2.0818,
      "step": 623
    },
    {
      "epoch": 0.8347826086956521,
      "grad_norm": 0.6905943751335144,
      "learning_rate": 7.280504277352545e-05,
      "loss": 2.4726,
      "step": 624
    },
    {
      "epoch": 0.8361204013377926,
      "grad_norm": 1.378237247467041,
      "learning_rate": 7.276001800990544e-05,
      "loss": 2.0315,
      "step": 625
    },
    {
      "epoch": 0.8374581939799332,
      "grad_norm": 0.9301738142967224,
      "learning_rate": 7.271499324628545e-05,
      "loss": 1.027,
      "step": 626
    },
    {
      "epoch": 0.8387959866220736,
      "grad_norm": 0.9967605471611023,
      "learning_rate": 7.266996848266548e-05,
      "loss": 2.1218,
      "step": 627
    },
    {
      "epoch": 0.8401337792642141,
      "grad_norm": 1.6421177387237549,
      "learning_rate": 7.262494371904548e-05,
      "loss": 1.5805,
      "step": 628
    },
    {
      "epoch": 0.8414715719063545,
      "grad_norm": 1.0547858476638794,
      "learning_rate": 7.257991895542548e-05,
      "loss": 2.3767,
      "step": 629
    },
    {
      "epoch": 0.842809364548495,
      "grad_norm": 0.9341603517532349,
      "learning_rate": 7.253489419180549e-05,
      "loss": 1.8782,
      "step": 630
    },
    {
      "epoch": 0.8441471571906355,
      "grad_norm": 0.8098312616348267,
      "learning_rate": 7.248986942818551e-05,
      "loss": 1.8284,
      "step": 631
    },
    {
      "epoch": 0.8454849498327759,
      "grad_norm": 1.4215975999832153,
      "learning_rate": 7.244484466456552e-05,
      "loss": 1.6072,
      "step": 632
    },
    {
      "epoch": 0.8468227424749164,
      "grad_norm": 0.6881840229034424,
      "learning_rate": 7.239981990094553e-05,
      "loss": 1.9254,
      "step": 633
    },
    {
      "epoch": 0.8481605351170568,
      "grad_norm": 0.7351963520050049,
      "learning_rate": 7.235479513732554e-05,
      "loss": 1.9477,
      "step": 634
    },
    {
      "epoch": 0.8494983277591973,
      "grad_norm": 0.7751598954200745,
      "learning_rate": 7.230977037370554e-05,
      "loss": 2.0508,
      "step": 635
    },
    {
      "epoch": 0.8508361204013378,
      "grad_norm": 0.7352134585380554,
      "learning_rate": 7.226474561008555e-05,
      "loss": 1.9826,
      "step": 636
    },
    {
      "epoch": 0.8521739130434782,
      "grad_norm": 1.0797083377838135,
      "learning_rate": 7.221972084646556e-05,
      "loss": 1.8754,
      "step": 637
    },
    {
      "epoch": 0.8535117056856187,
      "grad_norm": 0.6877202391624451,
      "learning_rate": 7.217469608284557e-05,
      "loss": 2.1698,
      "step": 638
    },
    {
      "epoch": 0.8548494983277592,
      "grad_norm": 0.5914489030838013,
      "learning_rate": 7.212967131922558e-05,
      "loss": 2.1072,
      "step": 639
    },
    {
      "epoch": 0.8561872909698997,
      "grad_norm": 1.0543609857559204,
      "learning_rate": 7.208464655560559e-05,
      "loss": 1.9583,
      "step": 640
    },
    {
      "epoch": 0.8575250836120402,
      "grad_norm": 0.6509416699409485,
      "learning_rate": 7.20396217919856e-05,
      "loss": 2.0584,
      "step": 641
    },
    {
      "epoch": 0.8588628762541806,
      "grad_norm": 0.6920568943023682,
      "learning_rate": 7.19945970283656e-05,
      "loss": 2.0691,
      "step": 642
    },
    {
      "epoch": 0.8602006688963211,
      "grad_norm": 0.5659633874893188,
      "learning_rate": 7.194957226474561e-05,
      "loss": 2.3212,
      "step": 643
    },
    {
      "epoch": 0.8615384615384616,
      "grad_norm": 0.8833758234977722,
      "learning_rate": 7.190454750112562e-05,
      "loss": 1.9745,
      "step": 644
    },
    {
      "epoch": 0.862876254180602,
      "grad_norm": 1.0847338438034058,
      "learning_rate": 7.185952273750563e-05,
      "loss": 2.17,
      "step": 645
    },
    {
      "epoch": 0.8642140468227425,
      "grad_norm": 0.7263421416282654,
      "learning_rate": 7.181449797388564e-05,
      "loss": 1.5802,
      "step": 646
    },
    {
      "epoch": 0.8655518394648829,
      "grad_norm": 1.1162469387054443,
      "learning_rate": 7.176947321026565e-05,
      "loss": 2.0613,
      "step": 647
    },
    {
      "epoch": 0.8668896321070234,
      "grad_norm": 0.73091721534729,
      "learning_rate": 7.172444844664566e-05,
      "loss": 1.8372,
      "step": 648
    },
    {
      "epoch": 0.8682274247491639,
      "grad_norm": 0.6744244694709778,
      "learning_rate": 7.167942368302566e-05,
      "loss": 2.0031,
      "step": 649
    },
    {
      "epoch": 0.8695652173913043,
      "grad_norm": 0.7309931516647339,
      "learning_rate": 7.163439891940567e-05,
      "loss": 1.5731,
      "step": 650
    },
    {
      "epoch": 0.8709030100334448,
      "grad_norm": 0.6159527897834778,
      "learning_rate": 7.158937415578568e-05,
      "loss": 2.2952,
      "step": 651
    },
    {
      "epoch": 0.8722408026755852,
      "grad_norm": 0.8553061485290527,
      "learning_rate": 7.15443493921657e-05,
      "loss": 1.8854,
      "step": 652
    },
    {
      "epoch": 0.8735785953177257,
      "grad_norm": 0.6783524751663208,
      "learning_rate": 7.149932462854571e-05,
      "loss": 1.13,
      "step": 653
    },
    {
      "epoch": 0.8749163879598663,
      "grad_norm": 1.1411305665969849,
      "learning_rate": 7.145429986492571e-05,
      "loss": 1.9138,
      "step": 654
    },
    {
      "epoch": 0.8762541806020067,
      "grad_norm": 0.7974840998649597,
      "learning_rate": 7.140927510130572e-05,
      "loss": 1.3333,
      "step": 655
    },
    {
      "epoch": 0.8775919732441472,
      "grad_norm": 0.8446215391159058,
      "learning_rate": 7.136425033768574e-05,
      "loss": 2.2973,
      "step": 656
    },
    {
      "epoch": 0.8789297658862876,
      "grad_norm": 0.7412329912185669,
      "learning_rate": 7.131922557406575e-05,
      "loss": 1.7935,
      "step": 657
    },
    {
      "epoch": 0.8802675585284281,
      "grad_norm": 1.7755156755447388,
      "learning_rate": 7.127420081044574e-05,
      "loss": 2.0122,
      "step": 658
    },
    {
      "epoch": 0.8816053511705686,
      "grad_norm": 1.0286380052566528,
      "learning_rate": 7.122917604682576e-05,
      "loss": 1.9987,
      "step": 659
    },
    {
      "epoch": 0.882943143812709,
      "grad_norm": 0.6704300045967102,
      "learning_rate": 7.118415128320577e-05,
      "loss": 1.8126,
      "step": 660
    },
    {
      "epoch": 0.8842809364548495,
      "grad_norm": 0.6966663002967834,
      "learning_rate": 7.113912651958578e-05,
      "loss": 1.993,
      "step": 661
    },
    {
      "epoch": 0.88561872909699,
      "grad_norm": 0.6740978360176086,
      "learning_rate": 7.109410175596578e-05,
      "loss": 1.9854,
      "step": 662
    },
    {
      "epoch": 0.8869565217391304,
      "grad_norm": 0.7223276495933533,
      "learning_rate": 7.10490769923458e-05,
      "loss": 2.0942,
      "step": 663
    },
    {
      "epoch": 0.8882943143812709,
      "grad_norm": 0.9881120324134827,
      "learning_rate": 7.10040522287258e-05,
      "loss": 1.9087,
      "step": 664
    },
    {
      "epoch": 0.8896321070234113,
      "grad_norm": 0.7445497512817383,
      "learning_rate": 7.095902746510582e-05,
      "loss": 2.058,
      "step": 665
    },
    {
      "epoch": 0.8909698996655518,
      "grad_norm": 1.0675506591796875,
      "learning_rate": 7.091400270148581e-05,
      "loss": 1.9869,
      "step": 666
    },
    {
      "epoch": 0.8923076923076924,
      "grad_norm": 1.2335395812988281,
      "learning_rate": 7.086897793786583e-05,
      "loss": 2.1621,
      "step": 667
    },
    {
      "epoch": 0.8936454849498328,
      "grad_norm": 0.6886176466941833,
      "learning_rate": 7.082395317424584e-05,
      "loss": 1.5078,
      "step": 668
    },
    {
      "epoch": 0.8949832775919733,
      "grad_norm": 0.71738600730896,
      "learning_rate": 7.077892841062585e-05,
      "loss": 2.3195,
      "step": 669
    },
    {
      "epoch": 0.8963210702341137,
      "grad_norm": 0.708991527557373,
      "learning_rate": 7.073390364700586e-05,
      "loss": 2.059,
      "step": 670
    },
    {
      "epoch": 0.8976588628762542,
      "grad_norm": 0.6323236227035522,
      "learning_rate": 7.068887888338587e-05,
      "loss": 1.9016,
      "step": 671
    },
    {
      "epoch": 0.8989966555183947,
      "grad_norm": 0.6371700763702393,
      "learning_rate": 7.064385411976588e-05,
      "loss": 2.1694,
      "step": 672
    },
    {
      "epoch": 0.9003344481605351,
      "grad_norm": 0.7130315899848938,
      "learning_rate": 7.059882935614588e-05,
      "loss": 1.8981,
      "step": 673
    },
    {
      "epoch": 0.9016722408026756,
      "grad_norm": 0.8963072299957275,
      "learning_rate": 7.055380459252589e-05,
      "loss": 2.056,
      "step": 674
    },
    {
      "epoch": 0.903010033444816,
      "grad_norm": 0.6768928170204163,
      "learning_rate": 7.05087798289059e-05,
      "loss": 1.9999,
      "step": 675
    },
    {
      "epoch": 0.9043478260869565,
      "grad_norm": 0.7496529221534729,
      "learning_rate": 7.046375506528591e-05,
      "loss": 2.2434,
      "step": 676
    },
    {
      "epoch": 0.905685618729097,
      "grad_norm": 0.5942232608795166,
      "learning_rate": 7.041873030166593e-05,
      "loss": 2.3195,
      "step": 677
    },
    {
      "epoch": 0.9070234113712374,
      "grad_norm": 0.6655867695808411,
      "learning_rate": 7.037370553804593e-05,
      "loss": 2.0573,
      "step": 678
    },
    {
      "epoch": 0.9083612040133779,
      "grad_norm": 0.5738136172294617,
      "learning_rate": 7.032868077442594e-05,
      "loss": 1.2416,
      "step": 679
    },
    {
      "epoch": 0.9096989966555183,
      "grad_norm": 0.8400653004646301,
      "learning_rate": 7.028365601080594e-05,
      "loss": 1.8816,
      "step": 680
    },
    {
      "epoch": 0.9110367892976589,
      "grad_norm": 0.6366567015647888,
      "learning_rate": 7.023863124718597e-05,
      "loss": 1.9616,
      "step": 681
    },
    {
      "epoch": 0.9123745819397994,
      "grad_norm": 0.7917408347129822,
      "learning_rate": 7.019360648356596e-05,
      "loss": 2.0806,
      "step": 682
    },
    {
      "epoch": 0.9137123745819398,
      "grad_norm": 0.669217586517334,
      "learning_rate": 7.014858171994597e-05,
      "loss": 2.0481,
      "step": 683
    },
    {
      "epoch": 0.9150501672240803,
      "grad_norm": 1.5556551218032837,
      "learning_rate": 7.010355695632598e-05,
      "loss": 1.5202,
      "step": 684
    },
    {
      "epoch": 0.9163879598662207,
      "grad_norm": 0.9249728918075562,
      "learning_rate": 7.0058532192706e-05,
      "loss": 2.0939,
      "step": 685
    },
    {
      "epoch": 0.9177257525083612,
      "grad_norm": 0.5922223329544067,
      "learning_rate": 7.0013507429086e-05,
      "loss": 2.058,
      "step": 686
    },
    {
      "epoch": 0.9190635451505017,
      "grad_norm": 0.5889205932617188,
      "learning_rate": 6.9968482665466e-05,
      "loss": 1.6894,
      "step": 687
    },
    {
      "epoch": 0.9204013377926421,
      "grad_norm": 0.7091755270957947,
      "learning_rate": 6.992345790184603e-05,
      "loss": 1.9788,
      "step": 688
    },
    {
      "epoch": 0.9217391304347826,
      "grad_norm": 0.5706994533538818,
      "learning_rate": 6.987843313822603e-05,
      "loss": 1.7282,
      "step": 689
    },
    {
      "epoch": 0.9230769230769231,
      "grad_norm": 0.6377424597740173,
      "learning_rate": 6.983340837460603e-05,
      "loss": 2.0029,
      "step": 690
    },
    {
      "epoch": 0.9244147157190635,
      "grad_norm": 1.2235517501831055,
      "learning_rate": 6.978838361098604e-05,
      "loss": 1.6211,
      "step": 691
    },
    {
      "epoch": 0.925752508361204,
      "grad_norm": 0.7518335580825806,
      "learning_rate": 6.974335884736606e-05,
      "loss": 2.0324,
      "step": 692
    },
    {
      "epoch": 0.9270903010033444,
      "grad_norm": 0.8257300853729248,
      "learning_rate": 6.969833408374607e-05,
      "loss": 1.8237,
      "step": 693
    },
    {
      "epoch": 0.928428093645485,
      "grad_norm": 0.7107146382331848,
      "learning_rate": 6.965330932012606e-05,
      "loss": 2.4846,
      "step": 694
    },
    {
      "epoch": 0.9297658862876255,
      "grad_norm": 0.954113245010376,
      "learning_rate": 6.960828455650609e-05,
      "loss": 2.0246,
      "step": 695
    },
    {
      "epoch": 0.9311036789297659,
      "grad_norm": 0.6087449789047241,
      "learning_rate": 6.95632597928861e-05,
      "loss": 1.9656,
      "step": 696
    },
    {
      "epoch": 0.9324414715719064,
      "grad_norm": 0.6922059059143066,
      "learning_rate": 6.95182350292661e-05,
      "loss": 2.3123,
      "step": 697
    },
    {
      "epoch": 0.9337792642140468,
      "grad_norm": 0.7577765583992004,
      "learning_rate": 6.94732102656461e-05,
      "loss": 1.5737,
      "step": 698
    },
    {
      "epoch": 0.9351170568561873,
      "grad_norm": 0.7660884857177734,
      "learning_rate": 6.942818550202612e-05,
      "loss": 2.2649,
      "step": 699
    },
    {
      "epoch": 0.9364548494983278,
      "grad_norm": 1.2487679719924927,
      "learning_rate": 6.938316073840613e-05,
      "loss": 2.0546,
      "step": 700
    },
    {
      "epoch": 0.9377926421404682,
      "grad_norm": 0.8137415647506714,
      "learning_rate": 6.933813597478614e-05,
      "loss": 2.1294,
      "step": 701
    },
    {
      "epoch": 0.9391304347826087,
      "grad_norm": 0.8820820450782776,
      "learning_rate": 6.929311121116615e-05,
      "loss": 2.3508,
      "step": 702
    },
    {
      "epoch": 0.9404682274247491,
      "grad_norm": 0.5844343900680542,
      "learning_rate": 6.924808644754615e-05,
      "loss": 2.024,
      "step": 703
    },
    {
      "epoch": 0.9418060200668896,
      "grad_norm": 0.9741575121879578,
      "learning_rate": 6.920306168392616e-05,
      "loss": 1.9389,
      "step": 704
    },
    {
      "epoch": 0.9431438127090301,
      "grad_norm": 0.5414264798164368,
      "learning_rate": 6.915803692030617e-05,
      "loss": 2.3536,
      "step": 705
    },
    {
      "epoch": 0.9444816053511705,
      "grad_norm": 0.9424920678138733,
      "learning_rate": 6.911301215668618e-05,
      "loss": 1.7482,
      "step": 706
    },
    {
      "epoch": 0.945819397993311,
      "grad_norm": 1.173538327217102,
      "learning_rate": 6.906798739306619e-05,
      "loss": 2.0628,
      "step": 707
    },
    {
      "epoch": 0.9471571906354515,
      "grad_norm": 0.808866024017334,
      "learning_rate": 6.90229626294462e-05,
      "loss": 2.0757,
      "step": 708
    },
    {
      "epoch": 0.948494983277592,
      "grad_norm": 0.8062437772750854,
      "learning_rate": 6.89779378658262e-05,
      "loss": 1.4797,
      "step": 709
    },
    {
      "epoch": 0.9498327759197325,
      "grad_norm": 0.6568036079406738,
      "learning_rate": 6.893291310220622e-05,
      "loss": 2.1079,
      "step": 710
    },
    {
      "epoch": 0.9511705685618729,
      "grad_norm": 1.0303947925567627,
      "learning_rate": 6.888788833858622e-05,
      "loss": 2.1879,
      "step": 711
    },
    {
      "epoch": 0.9525083612040134,
      "grad_norm": 1.439601182937622,
      "learning_rate": 6.884286357496623e-05,
      "loss": 1.5109,
      "step": 712
    },
    {
      "epoch": 0.9538461538461539,
      "grad_norm": 1.9849913120269775,
      "learning_rate": 6.879783881134625e-05,
      "loss": 1.929,
      "step": 713
    },
    {
      "epoch": 0.9551839464882943,
      "grad_norm": 0.6140236258506775,
      "learning_rate": 6.875281404772625e-05,
      "loss": 2.2385,
      "step": 714
    },
    {
      "epoch": 0.9565217391304348,
      "grad_norm": 0.7368261218070984,
      "learning_rate": 6.870778928410626e-05,
      "loss": 1.3549,
      "step": 715
    },
    {
      "epoch": 0.9578595317725752,
      "grad_norm": 0.8577214479446411,
      "learning_rate": 6.866276452048627e-05,
      "loss": 2.1128,
      "step": 716
    },
    {
      "epoch": 0.9591973244147157,
      "grad_norm": 0.7355601191520691,
      "learning_rate": 6.861773975686629e-05,
      "loss": 1.9433,
      "step": 717
    },
    {
      "epoch": 0.9605351170568562,
      "grad_norm": 1.2294349670410156,
      "learning_rate": 6.857271499324628e-05,
      "loss": 2.2753,
      "step": 718
    },
    {
      "epoch": 0.9618729096989966,
      "grad_norm": 0.574518084526062,
      "learning_rate": 6.852769022962629e-05,
      "loss": 2.0646,
      "step": 719
    },
    {
      "epoch": 0.9632107023411371,
      "grad_norm": 0.715045154094696,
      "learning_rate": 6.84826654660063e-05,
      "loss": 2.0687,
      "step": 720
    },
    {
      "epoch": 0.9645484949832775,
      "grad_norm": 0.8325941562652588,
      "learning_rate": 6.843764070238632e-05,
      "loss": 1.85,
      "step": 721
    },
    {
      "epoch": 0.9658862876254181,
      "grad_norm": 0.6069225668907166,
      "learning_rate": 6.839261593876632e-05,
      "loss": 2.1878,
      "step": 722
    },
    {
      "epoch": 0.9672240802675586,
      "grad_norm": 0.705606997013092,
      "learning_rate": 6.834759117514633e-05,
      "loss": 2.2931,
      "step": 723
    },
    {
      "epoch": 0.968561872909699,
      "grad_norm": 0.6511003375053406,
      "learning_rate": 6.830256641152635e-05,
      "loss": 1.9663,
      "step": 724
    },
    {
      "epoch": 0.9698996655518395,
      "grad_norm": 0.6704803109169006,
      "learning_rate": 6.825754164790636e-05,
      "loss": 1.5408,
      "step": 725
    },
    {
      "epoch": 0.9712374581939799,
      "grad_norm": 1.087570309638977,
      "learning_rate": 6.821251688428637e-05,
      "loss": 2.3817,
      "step": 726
    },
    {
      "epoch": 0.9725752508361204,
      "grad_norm": 0.7136285901069641,
      "learning_rate": 6.816749212066636e-05,
      "loss": 2.0755,
      "step": 727
    },
    {
      "epoch": 0.9739130434782609,
      "grad_norm": 0.7055271863937378,
      "learning_rate": 6.812246735704638e-05,
      "loss": 2.1872,
      "step": 728
    },
    {
      "epoch": 0.9752508361204013,
      "grad_norm": 2.6934823989868164,
      "learning_rate": 6.807744259342639e-05,
      "loss": 1.5151,
      "step": 729
    },
    {
      "epoch": 0.9765886287625418,
      "grad_norm": 0.7742595672607422,
      "learning_rate": 6.80324178298064e-05,
      "loss": 2.255,
      "step": 730
    },
    {
      "epoch": 0.9779264214046822,
      "grad_norm": 0.7820495963096619,
      "learning_rate": 6.798739306618641e-05,
      "loss": 2.2767,
      "step": 731
    },
    {
      "epoch": 0.9792642140468227,
      "grad_norm": 0.8606453537940979,
      "learning_rate": 6.794236830256642e-05,
      "loss": 1.7905,
      "step": 732
    },
    {
      "epoch": 0.9806020066889632,
      "grad_norm": 0.9574686884880066,
      "learning_rate": 6.789734353894643e-05,
      "loss": 2.1741,
      "step": 733
    },
    {
      "epoch": 0.9819397993311036,
      "grad_norm": 1.447963833808899,
      "learning_rate": 6.785231877532643e-05,
      "loss": 1.8766,
      "step": 734
    },
    {
      "epoch": 0.9832775919732442,
      "grad_norm": 3.8072526454925537,
      "learning_rate": 6.780729401170644e-05,
      "loss": 2.0198,
      "step": 735
    },
    {
      "epoch": 0.9846153846153847,
      "grad_norm": 0.5300580263137817,
      "learning_rate": 6.776226924808645e-05,
      "loss": 1.6352,
      "step": 736
    },
    {
      "epoch": 0.9859531772575251,
      "grad_norm": 0.9182571172714233,
      "learning_rate": 6.771724448446646e-05,
      "loss": 2.3257,
      "step": 737
    },
    {
      "epoch": 0.9872909698996656,
      "grad_norm": 1.5257279872894287,
      "learning_rate": 6.767221972084647e-05,
      "loss": 2.3333,
      "step": 738
    },
    {
      "epoch": 0.988628762541806,
      "grad_norm": 0.793891191482544,
      "learning_rate": 6.762719495722648e-05,
      "loss": 1.776,
      "step": 739
    },
    {
      "epoch": 0.9899665551839465,
      "grad_norm": 0.6769539713859558,
      "learning_rate": 6.758217019360649e-05,
      "loss": 2.0308,
      "step": 740
    },
    {
      "epoch": 0.991304347826087,
      "grad_norm": 0.7656939029693604,
      "learning_rate": 6.75371454299865e-05,
      "loss": 2.0104,
      "step": 741
    },
    {
      "epoch": 0.9926421404682274,
      "grad_norm": 1.0917755365371704,
      "learning_rate": 6.74921206663665e-05,
      "loss": 2.0117,
      "step": 742
    },
    {
      "epoch": 0.9939799331103679,
      "grad_norm": 0.5329183340072632,
      "learning_rate": 6.744709590274651e-05,
      "loss": 1.6329,
      "step": 743
    },
    {
      "epoch": 0.9953177257525083,
      "grad_norm": 0.612562358379364,
      "learning_rate": 6.740207113912652e-05,
      "loss": 2.2978,
      "step": 744
    },
    {
      "epoch": 0.9966555183946488,
      "grad_norm": 0.9256593585014343,
      "learning_rate": 6.735704637550653e-05,
      "loss": 2.2044,
      "step": 745
    },
    {
      "epoch": 0.9979933110367893,
      "grad_norm": 0.6008355617523193,
      "learning_rate": 6.731202161188654e-05,
      "loss": 2.3045,
      "step": 746
    },
    {
      "epoch": 0.9993311036789297,
      "grad_norm": 0.7152503728866577,
      "learning_rate": 6.726699684826655e-05,
      "loss": 2.0138,
      "step": 747
    },
    {
      "epoch": 1.0006688963210701,
      "grad_norm": 0.6082767844200134,
      "learning_rate": 6.722197208464655e-05,
      "loss": 1.6355,
      "step": 748
    },
    {
      "epoch": 1.0020066889632107,
      "grad_norm": 0.8323435187339783,
      "learning_rate": 6.717694732102658e-05,
      "loss": 1.5551,
      "step": 749
    },
    {
      "epoch": 1.0033444816053512,
      "grad_norm": 0.777487576007843,
      "learning_rate": 6.713192255740659e-05,
      "loss": 2.0041,
      "step": 750
    },
    {
      "epoch": 1.0046822742474917,
      "grad_norm": 1.172985315322876,
      "learning_rate": 6.708689779378658e-05,
      "loss": 0.7001,
      "step": 751
    },
    {
      "epoch": 1.0060200668896322,
      "grad_norm": 0.7470185160636902,
      "learning_rate": 6.704187303016659e-05,
      "loss": 1.7772,
      "step": 752
    },
    {
      "epoch": 1.0073578595317725,
      "grad_norm": 1.2879897356033325,
      "learning_rate": 6.699684826654661e-05,
      "loss": 1.5489,
      "step": 753
    },
    {
      "epoch": 1.008695652173913,
      "grad_norm": 1.5060155391693115,
      "learning_rate": 6.695182350292662e-05,
      "loss": 1.5242,
      "step": 754
    },
    {
      "epoch": 1.0100334448160535,
      "grad_norm": 0.8411580324172974,
      "learning_rate": 6.690679873930662e-05,
      "loss": 1.6664,
      "step": 755
    },
    {
      "epoch": 1.011371237458194,
      "grad_norm": 0.9809452295303345,
      "learning_rate": 6.686177397568662e-05,
      "loss": 1.8596,
      "step": 756
    },
    {
      "epoch": 1.0127090301003345,
      "grad_norm": 1.2678918838500977,
      "learning_rate": 6.681674921206665e-05,
      "loss": 1.6882,
      "step": 757
    },
    {
      "epoch": 1.0140468227424748,
      "grad_norm": 0.9003990888595581,
      "learning_rate": 6.677172444844665e-05,
      "loss": 1.8339,
      "step": 758
    },
    {
      "epoch": 1.0153846153846153,
      "grad_norm": 0.9132852554321289,
      "learning_rate": 6.672669968482665e-05,
      "loss": 1.6924,
      "step": 759
    },
    {
      "epoch": 1.0167224080267558,
      "grad_norm": 0.8782814741134644,
      "learning_rate": 6.668167492120667e-05,
      "loss": 1.2357,
      "step": 760
    },
    {
      "epoch": 1.0180602006688964,
      "grad_norm": 1.0629113912582397,
      "learning_rate": 6.663665015758668e-05,
      "loss": 1.4501,
      "step": 761
    },
    {
      "epoch": 1.0193979933110369,
      "grad_norm": 0.6594319343566895,
      "learning_rate": 6.659162539396669e-05,
      "loss": 1.9228,
      "step": 762
    },
    {
      "epoch": 1.0207357859531772,
      "grad_norm": 1.1182608604431152,
      "learning_rate": 6.654660063034668e-05,
      "loss": 1.562,
      "step": 763
    },
    {
      "epoch": 1.0220735785953177,
      "grad_norm": 0.6796782612800598,
      "learning_rate": 6.65015758667267e-05,
      "loss": 1.8346,
      "step": 764
    },
    {
      "epoch": 1.0234113712374582,
      "grad_norm": 1.0874909162521362,
      "learning_rate": 6.645655110310671e-05,
      "loss": 1.2697,
      "step": 765
    },
    {
      "epoch": 1.0247491638795987,
      "grad_norm": 1.1255723237991333,
      "learning_rate": 6.641152633948672e-05,
      "loss": 1.1482,
      "step": 766
    },
    {
      "epoch": 1.0260869565217392,
      "grad_norm": 0.7903443574905396,
      "learning_rate": 6.636650157586673e-05,
      "loss": 1.5898,
      "step": 767
    },
    {
      "epoch": 1.0274247491638795,
      "grad_norm": 0.8102263808250427,
      "learning_rate": 6.632147681224674e-05,
      "loss": 1.5937,
      "step": 768
    },
    {
      "epoch": 1.02876254180602,
      "grad_norm": 0.7469982504844666,
      "learning_rate": 6.627645204862675e-05,
      "loss": 1.542,
      "step": 769
    },
    {
      "epoch": 1.0301003344481605,
      "grad_norm": 0.8481405377388,
      "learning_rate": 6.623142728500676e-05,
      "loss": 1.0759,
      "step": 770
    },
    {
      "epoch": 1.031438127090301,
      "grad_norm": 0.8804197907447815,
      "learning_rate": 6.618640252138677e-05,
      "loss": 1.4243,
      "step": 771
    },
    {
      "epoch": 1.0327759197324415,
      "grad_norm": 0.8232119083404541,
      "learning_rate": 6.614137775776677e-05,
      "loss": 1.767,
      "step": 772
    },
    {
      "epoch": 1.034113712374582,
      "grad_norm": 1.0143418312072754,
      "learning_rate": 6.609635299414678e-05,
      "loss": 1.6867,
      "step": 773
    },
    {
      "epoch": 1.0354515050167223,
      "grad_norm": 0.8692298531532288,
      "learning_rate": 6.605132823052679e-05,
      "loss": 1.5972,
      "step": 774
    },
    {
      "epoch": 1.0367892976588629,
      "grad_norm": 0.8228359222412109,
      "learning_rate": 6.60063034669068e-05,
      "loss": 1.6464,
      "step": 775
    },
    {
      "epoch": 1.0381270903010034,
      "grad_norm": 0.9182949662208557,
      "learning_rate": 6.596127870328681e-05,
      "loss": 1.2447,
      "step": 776
    },
    {
      "epoch": 1.0394648829431439,
      "grad_norm": 0.9668188095092773,
      "learning_rate": 6.591625393966682e-05,
      "loss": 1.5519,
      "step": 777
    },
    {
      "epoch": 1.0408026755852844,
      "grad_norm": 0.8035921454429626,
      "learning_rate": 6.587122917604684e-05,
      "loss": 1.6115,
      "step": 778
    },
    {
      "epoch": 1.0421404682274247,
      "grad_norm": 0.7637916803359985,
      "learning_rate": 6.582620441242683e-05,
      "loss": 1.5989,
      "step": 779
    },
    {
      "epoch": 1.0434782608695652,
      "grad_norm": 0.6795792579650879,
      "learning_rate": 6.578117964880684e-05,
      "loss": 1.8817,
      "step": 780
    },
    {
      "epoch": 1.0448160535117057,
      "grad_norm": 0.7362264394760132,
      "learning_rate": 6.573615488518685e-05,
      "loss": 1.6863,
      "step": 781
    },
    {
      "epoch": 1.0461538461538462,
      "grad_norm": 0.7315294742584229,
      "learning_rate": 6.569113012156687e-05,
      "loss": 1.8927,
      "step": 782
    },
    {
      "epoch": 1.0474916387959867,
      "grad_norm": 0.9518543481826782,
      "learning_rate": 6.564610535794687e-05,
      "loss": 1.3103,
      "step": 783
    },
    {
      "epoch": 1.048829431438127,
      "grad_norm": 0.8763629794120789,
      "learning_rate": 6.560108059432688e-05,
      "loss": 1.9248,
      "step": 784
    },
    {
      "epoch": 1.0501672240802675,
      "grad_norm": 0.8956972360610962,
      "learning_rate": 6.55560558307069e-05,
      "loss": 2.4056,
      "step": 785
    },
    {
      "epoch": 1.051505016722408,
      "grad_norm": 1.1922390460968018,
      "learning_rate": 6.551103106708691e-05,
      "loss": 1.3751,
      "step": 786
    },
    {
      "epoch": 1.0528428093645485,
      "grad_norm": 0.8947060108184814,
      "learning_rate": 6.54660063034669e-05,
      "loss": 1.7996,
      "step": 787
    },
    {
      "epoch": 1.054180602006689,
      "grad_norm": 0.7719162702560425,
      "learning_rate": 6.542098153984691e-05,
      "loss": 1.7215,
      "step": 788
    },
    {
      "epoch": 1.0555183946488294,
      "grad_norm": 0.8137053847312927,
      "learning_rate": 6.537595677622693e-05,
      "loss": 1.6754,
      "step": 789
    },
    {
      "epoch": 1.0568561872909699,
      "grad_norm": 0.8411099910736084,
      "learning_rate": 6.533093201260694e-05,
      "loss": 1.7742,
      "step": 790
    },
    {
      "epoch": 1.0581939799331104,
      "grad_norm": 0.9793359041213989,
      "learning_rate": 6.528590724898694e-05,
      "loss": 1.5735,
      "step": 791
    },
    {
      "epoch": 1.0595317725752509,
      "grad_norm": 0.9073910117149353,
      "learning_rate": 6.524088248536695e-05,
      "loss": 2.2354,
      "step": 792
    },
    {
      "epoch": 1.0608695652173914,
      "grad_norm": 0.7567430138587952,
      "learning_rate": 6.519585772174697e-05,
      "loss": 1.5626,
      "step": 793
    },
    {
      "epoch": 1.0622073578595317,
      "grad_norm": 1.06880521774292,
      "learning_rate": 6.515083295812698e-05,
      "loss": 1.2051,
      "step": 794
    },
    {
      "epoch": 1.0635451505016722,
      "grad_norm": 0.8187976479530334,
      "learning_rate": 6.510580819450697e-05,
      "loss": 1.631,
      "step": 795
    },
    {
      "epoch": 1.0648829431438127,
      "grad_norm": 0.7056008577346802,
      "learning_rate": 6.5060783430887e-05,
      "loss": 1.5343,
      "step": 796
    },
    {
      "epoch": 1.0662207357859532,
      "grad_norm": 0.9352847337722778,
      "learning_rate": 6.5015758667267e-05,
      "loss": 1.6003,
      "step": 797
    },
    {
      "epoch": 1.0675585284280937,
      "grad_norm": 1.8865182399749756,
      "learning_rate": 6.497073390364701e-05,
      "loss": 1.7215,
      "step": 798
    },
    {
      "epoch": 1.068896321070234,
      "grad_norm": 1.0380245447158813,
      "learning_rate": 6.492570914002702e-05,
      "loss": 1.6079,
      "step": 799
    },
    {
      "epoch": 1.0702341137123745,
      "grad_norm": 1.001865029335022,
      "learning_rate": 6.488068437640703e-05,
      "loss": 1.0845,
      "step": 800
    },
    {
      "epoch": 1.071571906354515,
      "grad_norm": 1.0492470264434814,
      "learning_rate": 6.483565961278704e-05,
      "loss": 1.4899,
      "step": 801
    },
    {
      "epoch": 1.0729096989966556,
      "grad_norm": 1.1831380128860474,
      "learning_rate": 6.479063484916705e-05,
      "loss": 0.7326,
      "step": 802
    },
    {
      "epoch": 1.074247491638796,
      "grad_norm": 1.1278901100158691,
      "learning_rate": 6.474561008554705e-05,
      "loss": 1.5677,
      "step": 803
    },
    {
      "epoch": 1.0755852842809364,
      "grad_norm": 0.8802807927131653,
      "learning_rate": 6.470058532192706e-05,
      "loss": 1.7562,
      "step": 804
    },
    {
      "epoch": 1.0769230769230769,
      "grad_norm": 1.0205730199813843,
      "learning_rate": 6.465556055830707e-05,
      "loss": 0.9363,
      "step": 805
    },
    {
      "epoch": 1.0782608695652174,
      "grad_norm": 1.9332672357559204,
      "learning_rate": 6.461053579468708e-05,
      "loss": 1.2796,
      "step": 806
    },
    {
      "epoch": 1.079598662207358,
      "grad_norm": 1.5062698125839233,
      "learning_rate": 6.456551103106709e-05,
      "loss": 1.3188,
      "step": 807
    },
    {
      "epoch": 1.0809364548494984,
      "grad_norm": 0.7134535312652588,
      "learning_rate": 6.45204862674471e-05,
      "loss": 1.928,
      "step": 808
    },
    {
      "epoch": 1.0822742474916387,
      "grad_norm": 1.903796911239624,
      "learning_rate": 6.44754615038271e-05,
      "loss": 1.1583,
      "step": 809
    },
    {
      "epoch": 1.0836120401337792,
      "grad_norm": 0.7796040773391724,
      "learning_rate": 6.443043674020711e-05,
      "loss": 1.9661,
      "step": 810
    },
    {
      "epoch": 1.0849498327759197,
      "grad_norm": 0.7826983332633972,
      "learning_rate": 6.438541197658712e-05,
      "loss": 1.8202,
      "step": 811
    },
    {
      "epoch": 1.0862876254180602,
      "grad_norm": 0.7324210405349731,
      "learning_rate": 6.434038721296713e-05,
      "loss": 1.5366,
      "step": 812
    },
    {
      "epoch": 1.0876254180602007,
      "grad_norm": 0.7018693685531616,
      "learning_rate": 6.429536244934714e-05,
      "loss": 1.8534,
      "step": 813
    },
    {
      "epoch": 1.088963210702341,
      "grad_norm": 0.876075029373169,
      "learning_rate": 6.425033768572716e-05,
      "loss": 1.2189,
      "step": 814
    },
    {
      "epoch": 1.0903010033444815,
      "grad_norm": 1.021077275276184,
      "learning_rate": 6.420531292210716e-05,
      "loss": 1.4601,
      "step": 815
    },
    {
      "epoch": 1.091638795986622,
      "grad_norm": 1.157410740852356,
      "learning_rate": 6.416028815848717e-05,
      "loss": 0.9841,
      "step": 816
    },
    {
      "epoch": 1.0929765886287626,
      "grad_norm": 0.6732658743858337,
      "learning_rate": 6.411526339486717e-05,
      "loss": 1.9742,
      "step": 817
    },
    {
      "epoch": 1.094314381270903,
      "grad_norm": 0.6634247899055481,
      "learning_rate": 6.40702386312472e-05,
      "loss": 1.5411,
      "step": 818
    },
    {
      "epoch": 1.0956521739130434,
      "grad_norm": 0.7590674161911011,
      "learning_rate": 6.402521386762719e-05,
      "loss": 1.028,
      "step": 819
    },
    {
      "epoch": 1.0969899665551839,
      "grad_norm": 0.8274590969085693,
      "learning_rate": 6.39801891040072e-05,
      "loss": 1.6494,
      "step": 820
    },
    {
      "epoch": 1.0983277591973244,
      "grad_norm": 0.8210983276367188,
      "learning_rate": 6.393516434038722e-05,
      "loss": 1.5706,
      "step": 821
    },
    {
      "epoch": 1.099665551839465,
      "grad_norm": 0.9263409376144409,
      "learning_rate": 6.389013957676723e-05,
      "loss": 1.6716,
      "step": 822
    },
    {
      "epoch": 1.1010033444816054,
      "grad_norm": 0.8053122162818909,
      "learning_rate": 6.384511481314724e-05,
      "loss": 1.5703,
      "step": 823
    },
    {
      "epoch": 1.1023411371237457,
      "grad_norm": 1.2410510778427124,
      "learning_rate": 6.380009004952723e-05,
      "loss": 1.6517,
      "step": 824
    },
    {
      "epoch": 1.1036789297658862,
      "grad_norm": 0.909188449382782,
      "learning_rate": 6.375506528590726e-05,
      "loss": 1.177,
      "step": 825
    },
    {
      "epoch": 1.1050167224080267,
      "grad_norm": 0.673254132270813,
      "learning_rate": 6.371004052228727e-05,
      "loss": 1.5251,
      "step": 826
    },
    {
      "epoch": 1.1063545150501672,
      "grad_norm": 0.9418015480041504,
      "learning_rate": 6.366501575866727e-05,
      "loss": 1.2767,
      "step": 827
    },
    {
      "epoch": 1.1076923076923078,
      "grad_norm": 0.7347880601882935,
      "learning_rate": 6.361999099504727e-05,
      "loss": 1.5897,
      "step": 828
    },
    {
      "epoch": 1.109030100334448,
      "grad_norm": 0.9692310094833374,
      "learning_rate": 6.357496623142729e-05,
      "loss": 1.909,
      "step": 829
    },
    {
      "epoch": 1.1103678929765886,
      "grad_norm": 0.8513740301132202,
      "learning_rate": 6.35299414678073e-05,
      "loss": 1.2062,
      "step": 830
    },
    {
      "epoch": 1.111705685618729,
      "grad_norm": 1.5622446537017822,
      "learning_rate": 6.348491670418731e-05,
      "loss": 0.8553,
      "step": 831
    },
    {
      "epoch": 1.1130434782608696,
      "grad_norm": 0.8817566633224487,
      "learning_rate": 6.343989194056732e-05,
      "loss": 1.1544,
      "step": 832
    },
    {
      "epoch": 1.11438127090301,
      "grad_norm": 0.8441960215568542,
      "learning_rate": 6.339486717694733e-05,
      "loss": 1.2105,
      "step": 833
    },
    {
      "epoch": 1.1157190635451506,
      "grad_norm": 1.094710350036621,
      "learning_rate": 6.334984241332733e-05,
      "loss": 1.3695,
      "step": 834
    },
    {
      "epoch": 1.117056856187291,
      "grad_norm": 1.8443351984024048,
      "learning_rate": 6.330481764970734e-05,
      "loss": 2.2446,
      "step": 835
    },
    {
      "epoch": 1.1183946488294314,
      "grad_norm": 1.042940616607666,
      "learning_rate": 6.325979288608735e-05,
      "loss": 1.9706,
      "step": 836
    },
    {
      "epoch": 1.119732441471572,
      "grad_norm": 0.7949560880661011,
      "learning_rate": 6.321476812246736e-05,
      "loss": 1.7105,
      "step": 837
    },
    {
      "epoch": 1.1210702341137124,
      "grad_norm": 0.8403007388114929,
      "learning_rate": 6.316974335884737e-05,
      "loss": 1.6519,
      "step": 838
    },
    {
      "epoch": 1.122408026755853,
      "grad_norm": 1.5427777767181396,
      "learning_rate": 6.312471859522738e-05,
      "loss": 1.136,
      "step": 839
    },
    {
      "epoch": 1.1237458193979932,
      "grad_norm": 0.619282603263855,
      "learning_rate": 6.307969383160739e-05,
      "loss": 1.308,
      "step": 840
    },
    {
      "epoch": 1.1250836120401337,
      "grad_norm": 0.8113716244697571,
      "learning_rate": 6.30346690679874e-05,
      "loss": 1.9532,
      "step": 841
    },
    {
      "epoch": 1.1264214046822743,
      "grad_norm": 0.6016692519187927,
      "learning_rate": 6.29896443043674e-05,
      "loss": 1.3914,
      "step": 842
    },
    {
      "epoch": 1.1277591973244148,
      "grad_norm": 0.9051966071128845,
      "learning_rate": 6.294461954074741e-05,
      "loss": 1.6739,
      "step": 843
    },
    {
      "epoch": 1.1290969899665553,
      "grad_norm": 0.8936472535133362,
      "learning_rate": 6.289959477712742e-05,
      "loss": 1.5069,
      "step": 844
    },
    {
      "epoch": 1.1304347826086956,
      "grad_norm": 0.8642289042472839,
      "learning_rate": 6.285457001350743e-05,
      "loss": 1.6832,
      "step": 845
    },
    {
      "epoch": 1.131772575250836,
      "grad_norm": 1.186710238456726,
      "learning_rate": 6.280954524988744e-05,
      "loss": 1.6192,
      "step": 846
    },
    {
      "epoch": 1.1331103678929766,
      "grad_norm": 0.723852813243866,
      "learning_rate": 6.276452048626746e-05,
      "loss": 1.8716,
      "step": 847
    },
    {
      "epoch": 1.134448160535117,
      "grad_norm": 0.9492783546447754,
      "learning_rate": 6.271949572264745e-05,
      "loss": 1.6641,
      "step": 848
    },
    {
      "epoch": 1.1357859531772576,
      "grad_norm": 1.0893625020980835,
      "learning_rate": 6.267447095902746e-05,
      "loss": 1.5037,
      "step": 849
    },
    {
      "epoch": 1.137123745819398,
      "grad_norm": 1.28972589969635,
      "learning_rate": 6.262944619540749e-05,
      "loss": 1.0721,
      "step": 850
    },
    {
      "epoch": 1.1384615384615384,
      "grad_norm": 1.0664037466049194,
      "learning_rate": 6.25844214317875e-05,
      "loss": 1.6358,
      "step": 851
    },
    {
      "epoch": 1.139799331103679,
      "grad_norm": 0.8718981146812439,
      "learning_rate": 6.253939666816749e-05,
      "loss": 1.7783,
      "step": 852
    },
    {
      "epoch": 1.1411371237458194,
      "grad_norm": 0.7964752912521362,
      "learning_rate": 6.24943719045475e-05,
      "loss": 1.8482,
      "step": 853
    },
    {
      "epoch": 1.14247491638796,
      "grad_norm": 1.1163746118545532,
      "learning_rate": 6.244934714092752e-05,
      "loss": 1.5371,
      "step": 854
    },
    {
      "epoch": 1.1438127090301002,
      "grad_norm": 0.8743355870246887,
      "learning_rate": 6.240432237730753e-05,
      "loss": 1.5555,
      "step": 855
    },
    {
      "epoch": 1.1451505016722408,
      "grad_norm": 1.3836225271224976,
      "learning_rate": 6.235929761368752e-05,
      "loss": 1.138,
      "step": 856
    },
    {
      "epoch": 1.1464882943143813,
      "grad_norm": 0.9284037351608276,
      "learning_rate": 6.231427285006755e-05,
      "loss": 1.8,
      "step": 857
    },
    {
      "epoch": 1.1478260869565218,
      "grad_norm": 2.1215121746063232,
      "learning_rate": 6.226924808644755e-05,
      "loss": 1.886,
      "step": 858
    },
    {
      "epoch": 1.1491638795986623,
      "grad_norm": 0.8018773198127747,
      "learning_rate": 6.222422332282756e-05,
      "loss": 1.4954,
      "step": 859
    },
    {
      "epoch": 1.1505016722408028,
      "grad_norm": 0.7150079607963562,
      "learning_rate": 6.217919855920756e-05,
      "loss": 1.2708,
      "step": 860
    },
    {
      "epoch": 1.151839464882943,
      "grad_norm": 3.145190954208374,
      "learning_rate": 6.213417379558758e-05,
      "loss": 1.2389,
      "step": 861
    },
    {
      "epoch": 1.1531772575250836,
      "grad_norm": 0.9456586837768555,
      "learning_rate": 6.208914903196759e-05,
      "loss": 2.124,
      "step": 862
    },
    {
      "epoch": 1.154515050167224,
      "grad_norm": 1.0844643115997314,
      "learning_rate": 6.20441242683476e-05,
      "loss": 1.4282,
      "step": 863
    },
    {
      "epoch": 1.1558528428093646,
      "grad_norm": 0.8809278607368469,
      "learning_rate": 6.199909950472759e-05,
      "loss": 1.8585,
      "step": 864
    },
    {
      "epoch": 1.1571906354515051,
      "grad_norm": 0.9258586764335632,
      "learning_rate": 6.195407474110761e-05,
      "loss": 1.1901,
      "step": 865
    },
    {
      "epoch": 1.1585284280936454,
      "grad_norm": 0.8438937664031982,
      "learning_rate": 6.190904997748762e-05,
      "loss": 1.6645,
      "step": 866
    },
    {
      "epoch": 1.159866220735786,
      "grad_norm": 0.8730605244636536,
      "learning_rate": 6.186402521386763e-05,
      "loss": 1.5863,
      "step": 867
    },
    {
      "epoch": 1.1612040133779264,
      "grad_norm": 0.8524886965751648,
      "learning_rate": 6.181900045024764e-05,
      "loss": 1.2708,
      "step": 868
    },
    {
      "epoch": 1.162541806020067,
      "grad_norm": 0.6810708045959473,
      "learning_rate": 6.177397568662765e-05,
      "loss": 1.3026,
      "step": 869
    },
    {
      "epoch": 1.1638795986622075,
      "grad_norm": 1.1528704166412354,
      "learning_rate": 6.172895092300766e-05,
      "loss": 1.7219,
      "step": 870
    },
    {
      "epoch": 1.1652173913043478,
      "grad_norm": 2.7838613986968994,
      "learning_rate": 6.168392615938767e-05,
      "loss": 1.4851,
      "step": 871
    },
    {
      "epoch": 1.1665551839464883,
      "grad_norm": 0.8822845816612244,
      "learning_rate": 6.163890139576767e-05,
      "loss": 1.652,
      "step": 872
    },
    {
      "epoch": 1.1678929765886288,
      "grad_norm": 0.6918888688087463,
      "learning_rate": 6.159387663214768e-05,
      "loss": 1.2737,
      "step": 873
    },
    {
      "epoch": 1.1692307692307693,
      "grad_norm": 1.1061296463012695,
      "learning_rate": 6.154885186852769e-05,
      "loss": 1.4387,
      "step": 874
    },
    {
      "epoch": 1.1705685618729098,
      "grad_norm": 0.6670346856117249,
      "learning_rate": 6.150382710490771e-05,
      "loss": 1.9764,
      "step": 875
    },
    {
      "epoch": 1.17190635451505,
      "grad_norm": 0.8202233910560608,
      "learning_rate": 6.145880234128771e-05,
      "loss": 1.6687,
      "step": 876
    },
    {
      "epoch": 1.1732441471571906,
      "grad_norm": 0.7462705373764038,
      "learning_rate": 6.141377757766772e-05,
      "loss": 1.7787,
      "step": 877
    },
    {
      "epoch": 1.1745819397993311,
      "grad_norm": 0.8808805346488953,
      "learning_rate": 6.136875281404773e-05,
      "loss": 1.2746,
      "step": 878
    },
    {
      "epoch": 1.1759197324414716,
      "grad_norm": 0.7754594087600708,
      "learning_rate": 6.132372805042775e-05,
      "loss": 1.9316,
      "step": 879
    },
    {
      "epoch": 1.1772575250836121,
      "grad_norm": 1.5388227701187134,
      "learning_rate": 6.127870328680774e-05,
      "loss": 1.6537,
      "step": 880
    },
    {
      "epoch": 1.1785953177257524,
      "grad_norm": 2.2294013500213623,
      "learning_rate": 6.123367852318775e-05,
      "loss": 1.3761,
      "step": 881
    },
    {
      "epoch": 1.179933110367893,
      "grad_norm": 0.7639999389648438,
      "learning_rate": 6.118865375956776e-05,
      "loss": 1.522,
      "step": 882
    },
    {
      "epoch": 1.1812709030100335,
      "grad_norm": 1.0136057138442993,
      "learning_rate": 6.114362899594778e-05,
      "loss": 1.7424,
      "step": 883
    },
    {
      "epoch": 1.182608695652174,
      "grad_norm": 0.8844318985939026,
      "learning_rate": 6.109860423232778e-05,
      "loss": 1.1706,
      "step": 884
    },
    {
      "epoch": 1.1839464882943145,
      "grad_norm": 0.8913040161132812,
      "learning_rate": 6.105357946870779e-05,
      "loss": 1.7598,
      "step": 885
    },
    {
      "epoch": 1.1852842809364548,
      "grad_norm": 1.07845938205719,
      "learning_rate": 6.100855470508781e-05,
      "loss": 1.0585,
      "step": 886
    },
    {
      "epoch": 1.1866220735785953,
      "grad_norm": 0.9585360288619995,
      "learning_rate": 6.096352994146781e-05,
      "loss": 1.0684,
      "step": 887
    },
    {
      "epoch": 1.1879598662207358,
      "grad_norm": 1.028442621231079,
      "learning_rate": 6.091850517784782e-05,
      "loss": 1.8733,
      "step": 888
    },
    {
      "epoch": 1.1892976588628763,
      "grad_norm": 0.9173264503479004,
      "learning_rate": 6.087348041422783e-05,
      "loss": 1.7119,
      "step": 889
    },
    {
      "epoch": 1.1906354515050168,
      "grad_norm": 1.4423737525939941,
      "learning_rate": 6.082845565060784e-05,
      "loss": 1.963,
      "step": 890
    },
    {
      "epoch": 1.191973244147157,
      "grad_norm": 0.9499053955078125,
      "learning_rate": 6.0783430886987844e-05,
      "loss": 1.4102,
      "step": 891
    },
    {
      "epoch": 1.1933110367892976,
      "grad_norm": 0.7227963805198669,
      "learning_rate": 6.073840612336785e-05,
      "loss": 1.8352,
      "step": 892
    },
    {
      "epoch": 1.1946488294314381,
      "grad_norm": 0.8120591044425964,
      "learning_rate": 6.069338135974787e-05,
      "loss": 1.6841,
      "step": 893
    },
    {
      "epoch": 1.1959866220735786,
      "grad_norm": 0.7347455024719238,
      "learning_rate": 6.0648356596127876e-05,
      "loss": 1.8131,
      "step": 894
    },
    {
      "epoch": 1.1973244147157192,
      "grad_norm": 0.9052204489707947,
      "learning_rate": 6.060333183250788e-05,
      "loss": 1.4982,
      "step": 895
    },
    {
      "epoch": 1.1986622073578594,
      "grad_norm": 0.8606255650520325,
      "learning_rate": 6.055830706888789e-05,
      "loss": 1.1158,
      "step": 896
    },
    {
      "epoch": 1.2,
      "grad_norm": 1.5991266965866089,
      "learning_rate": 6.05132823052679e-05,
      "loss": 1.4721,
      "step": 897
    },
    {
      "epoch": 1.2013377926421405,
      "grad_norm": 0.9641832113265991,
      "learning_rate": 6.046825754164791e-05,
      "loss": 1.9599,
      "step": 898
    },
    {
      "epoch": 1.202675585284281,
      "grad_norm": 0.9988728165626526,
      "learning_rate": 6.042323277802792e-05,
      "loss": 1.4815,
      "step": 899
    },
    {
      "epoch": 1.2040133779264215,
      "grad_norm": 0.8313688635826111,
      "learning_rate": 6.037820801440792e-05,
      "loss": 1.0745,
      "step": 900
    },
    {
      "epoch": 1.2053511705685618,
      "grad_norm": 0.6813498735427856,
      "learning_rate": 6.033318325078794e-05,
      "loss": 1.3578,
      "step": 901
    },
    {
      "epoch": 1.2066889632107023,
      "grad_norm": 1.4857420921325684,
      "learning_rate": 6.0288158487167945e-05,
      "loss": 1.5083,
      "step": 902
    },
    {
      "epoch": 1.2080267558528428,
      "grad_norm": 0.7191517353057861,
      "learning_rate": 6.0243133723547954e-05,
      "loss": 1.8211,
      "step": 903
    },
    {
      "epoch": 1.2093645484949833,
      "grad_norm": 0.6989690065383911,
      "learning_rate": 6.019810895992797e-05,
      "loss": 1.8502,
      "step": 904
    },
    {
      "epoch": 1.2107023411371238,
      "grad_norm": 1.0212314128875732,
      "learning_rate": 6.015308419630797e-05,
      "loss": 1.9849,
      "step": 905
    },
    {
      "epoch": 1.2120401337792641,
      "grad_norm": 2.5734026432037354,
      "learning_rate": 6.010805943268798e-05,
      "loss": 1.2135,
      "step": 906
    },
    {
      "epoch": 1.2133779264214046,
      "grad_norm": 0.762624979019165,
      "learning_rate": 6.006303466906799e-05,
      "loss": 1.6969,
      "step": 907
    },
    {
      "epoch": 1.2147157190635451,
      "grad_norm": 0.9556645154953003,
      "learning_rate": 6.0018009905448004e-05,
      "loss": 1.4878,
      "step": 908
    },
    {
      "epoch": 1.2160535117056857,
      "grad_norm": 0.6057252883911133,
      "learning_rate": 5.9972985141828005e-05,
      "loss": 1.4873,
      "step": 909
    },
    {
      "epoch": 1.2173913043478262,
      "grad_norm": 1.1988812685012817,
      "learning_rate": 5.9927960378208014e-05,
      "loss": 1.4444,
      "step": 910
    },
    {
      "epoch": 1.2187290969899665,
      "grad_norm": 0.646827757358551,
      "learning_rate": 5.988293561458803e-05,
      "loss": 1.1479,
      "step": 911
    },
    {
      "epoch": 1.220066889632107,
      "grad_norm": 0.762706458568573,
      "learning_rate": 5.983791085096804e-05,
      "loss": 1.5255,
      "step": 912
    },
    {
      "epoch": 1.2214046822742475,
      "grad_norm": 0.9942954778671265,
      "learning_rate": 5.9792886087348046e-05,
      "loss": 1.252,
      "step": 913
    },
    {
      "epoch": 1.222742474916388,
      "grad_norm": 1.1604199409484863,
      "learning_rate": 5.974786132372805e-05,
      "loss": 1.7593,
      "step": 914
    },
    {
      "epoch": 1.2240802675585285,
      "grad_norm": 1.2815771102905273,
      "learning_rate": 5.9702836560108064e-05,
      "loss": 1.7823,
      "step": 915
    },
    {
      "epoch": 1.2254180602006688,
      "grad_norm": 0.7978178858757019,
      "learning_rate": 5.965781179648807e-05,
      "loss": 2.0054,
      "step": 916
    },
    {
      "epoch": 1.2267558528428093,
      "grad_norm": 0.6075098514556885,
      "learning_rate": 5.961278703286808e-05,
      "loss": 2.0769,
      "step": 917
    },
    {
      "epoch": 1.2280936454849498,
      "grad_norm": 0.7329334020614624,
      "learning_rate": 5.956776226924808e-05,
      "loss": 1.7853,
      "step": 918
    },
    {
      "epoch": 1.2294314381270903,
      "grad_norm": 0.689140260219574,
      "learning_rate": 5.95227375056281e-05,
      "loss": 1.9128,
      "step": 919
    },
    {
      "epoch": 1.2307692307692308,
      "grad_norm": 0.7186599969863892,
      "learning_rate": 5.9477712742008107e-05,
      "loss": 1.8959,
      "step": 920
    },
    {
      "epoch": 1.2321070234113711,
      "grad_norm": 1.0237096548080444,
      "learning_rate": 5.9432687978388115e-05,
      "loss": 2.0922,
      "step": 921
    },
    {
      "epoch": 1.2334448160535116,
      "grad_norm": 1.1331475973129272,
      "learning_rate": 5.938766321476813e-05,
      "loss": 1.5555,
      "step": 922
    },
    {
      "epoch": 1.2347826086956522,
      "grad_norm": 0.5831222534179688,
      "learning_rate": 5.934263845114814e-05,
      "loss": 1.3537,
      "step": 923
    },
    {
      "epoch": 1.2361204013377927,
      "grad_norm": 0.8829436302185059,
      "learning_rate": 5.929761368752814e-05,
      "loss": 1.56,
      "step": 924
    },
    {
      "epoch": 1.2374581939799332,
      "grad_norm": 0.6521071791648865,
      "learning_rate": 5.925258892390815e-05,
      "loss": 1.2735,
      "step": 925
    },
    {
      "epoch": 1.2387959866220735,
      "grad_norm": 0.8493558168411255,
      "learning_rate": 5.9207564160288165e-05,
      "loss": 1.8301,
      "step": 926
    },
    {
      "epoch": 1.240133779264214,
      "grad_norm": 0.889323890209198,
      "learning_rate": 5.9162539396668173e-05,
      "loss": 1.47,
      "step": 927
    },
    {
      "epoch": 1.2414715719063545,
      "grad_norm": 1.5412036180496216,
      "learning_rate": 5.9117514633048175e-05,
      "loss": 1.8443,
      "step": 928
    },
    {
      "epoch": 1.242809364548495,
      "grad_norm": 0.8223730325698853,
      "learning_rate": 5.907248986942819e-05,
      "loss": 1.4285,
      "step": 929
    },
    {
      "epoch": 1.2441471571906355,
      "grad_norm": 0.9287546277046204,
      "learning_rate": 5.90274651058082e-05,
      "loss": 1.5424,
      "step": 930
    },
    {
      "epoch": 1.2454849498327758,
      "grad_norm": 1.4095571041107178,
      "learning_rate": 5.898244034218821e-05,
      "loss": 1.1844,
      "step": 931
    },
    {
      "epoch": 1.2468227424749163,
      "grad_norm": 0.9519053101539612,
      "learning_rate": 5.893741557856821e-05,
      "loss": 1.5734,
      "step": 932
    },
    {
      "epoch": 1.2481605351170568,
      "grad_norm": 1.132487416267395,
      "learning_rate": 5.889239081494823e-05,
      "loss": 1.6002,
      "step": 933
    },
    {
      "epoch": 1.2494983277591973,
      "grad_norm": 0.6693798899650574,
      "learning_rate": 5.8847366051328234e-05,
      "loss": 1.8101,
      "step": 934
    },
    {
      "epoch": 1.2508361204013378,
      "grad_norm": 1.3155882358551025,
      "learning_rate": 5.880234128770824e-05,
      "loss": 1.6621,
      "step": 935
    },
    {
      "epoch": 1.2521739130434781,
      "grad_norm": 2.0933713912963867,
      "learning_rate": 5.8757316524088244e-05,
      "loss": 1.4421,
      "step": 936
    },
    {
      "epoch": 1.2535117056856186,
      "grad_norm": 0.9255110025405884,
      "learning_rate": 5.8712291760468266e-05,
      "loss": 2.0006,
      "step": 937
    },
    {
      "epoch": 1.2548494983277592,
      "grad_norm": 0.7264999151229858,
      "learning_rate": 5.866726699684827e-05,
      "loss": 1.9256,
      "step": 938
    },
    {
      "epoch": 1.2561872909698997,
      "grad_norm": 0.767331600189209,
      "learning_rate": 5.8622242233228277e-05,
      "loss": 1.8369,
      "step": 939
    },
    {
      "epoch": 1.2575250836120402,
      "grad_norm": 1.104103684425354,
      "learning_rate": 5.857721746960829e-05,
      "loss": 1.576,
      "step": 940
    },
    {
      "epoch": 1.2588628762541805,
      "grad_norm": 0.7990514636039734,
      "learning_rate": 5.85321927059883e-05,
      "loss": 1.8798,
      "step": 941
    },
    {
      "epoch": 1.2602006688963212,
      "grad_norm": 0.6627078056335449,
      "learning_rate": 5.84871679423683e-05,
      "loss": 1.8592,
      "step": 942
    },
    {
      "epoch": 1.2615384615384615,
      "grad_norm": 0.730588436126709,
      "learning_rate": 5.844214317874831e-05,
      "loss": 0.9581,
      "step": 943
    },
    {
      "epoch": 1.262876254180602,
      "grad_norm": 0.5782043933868408,
      "learning_rate": 5.8397118415128326e-05,
      "loss": 1.5693,
      "step": 944
    },
    {
      "epoch": 1.2642140468227425,
      "grad_norm": 1.0330079793930054,
      "learning_rate": 5.8352093651508335e-05,
      "loss": 1.3259,
      "step": 945
    },
    {
      "epoch": 1.2655518394648828,
      "grad_norm": 0.7264488935470581,
      "learning_rate": 5.830706888788834e-05,
      "loss": 1.8417,
      "step": 946
    },
    {
      "epoch": 1.2668896321070235,
      "grad_norm": 0.8809522986412048,
      "learning_rate": 5.826204412426836e-05,
      "loss": 1.3526,
      "step": 947
    },
    {
      "epoch": 1.2682274247491638,
      "grad_norm": 0.9869521856307983,
      "learning_rate": 5.821701936064836e-05,
      "loss": 1.3958,
      "step": 948
    },
    {
      "epoch": 1.2695652173913043,
      "grad_norm": 0.7876471281051636,
      "learning_rate": 5.817199459702837e-05,
      "loss": 1.5426,
      "step": 949
    },
    {
      "epoch": 1.2709030100334449,
      "grad_norm": 0.836608350276947,
      "learning_rate": 5.812696983340837e-05,
      "loss": 1.8164,
      "step": 950
    },
    {
      "epoch": 1.2722408026755851,
      "grad_norm": 0.7498127222061157,
      "learning_rate": 5.808194506978839e-05,
      "loss": 1.7363,
      "step": 951
    },
    {
      "epoch": 1.2735785953177259,
      "grad_norm": 0.9935570955276489,
      "learning_rate": 5.8036920306168395e-05,
      "loss": 1.5462,
      "step": 952
    },
    {
      "epoch": 1.2749163879598662,
      "grad_norm": 1.1275084018707275,
      "learning_rate": 5.7991895542548404e-05,
      "loss": 1.5394,
      "step": 953
    },
    {
      "epoch": 1.2762541806020067,
      "grad_norm": 0.996572732925415,
      "learning_rate": 5.7946870778928405e-05,
      "loss": 1.5601,
      "step": 954
    },
    {
      "epoch": 1.2775919732441472,
      "grad_norm": 1.216365933418274,
      "learning_rate": 5.790184601530843e-05,
      "loss": 1.4987,
      "step": 955
    },
    {
      "epoch": 1.2789297658862877,
      "grad_norm": 1.7709954977035522,
      "learning_rate": 5.785682125168843e-05,
      "loss": 1.3511,
      "step": 956
    },
    {
      "epoch": 1.2802675585284282,
      "grad_norm": 0.7698276042938232,
      "learning_rate": 5.781179648806844e-05,
      "loss": 1.8515,
      "step": 957
    },
    {
      "epoch": 1.2816053511705685,
      "grad_norm": 0.8181136846542358,
      "learning_rate": 5.776677172444845e-05,
      "loss": 1.5038,
      "step": 958
    },
    {
      "epoch": 1.282943143812709,
      "grad_norm": 0.9560226798057556,
      "learning_rate": 5.772174696082846e-05,
      "loss": 1.8003,
      "step": 959
    },
    {
      "epoch": 1.2842809364548495,
      "grad_norm": 0.9249825477600098,
      "learning_rate": 5.7676722197208464e-05,
      "loss": 1.8096,
      "step": 960
    },
    {
      "epoch": 1.28561872909699,
      "grad_norm": 0.7355000376701355,
      "learning_rate": 5.763169743358847e-05,
      "loss": 1.9474,
      "step": 961
    },
    {
      "epoch": 1.2869565217391306,
      "grad_norm": 0.7304123640060425,
      "learning_rate": 5.758667266996849e-05,
      "loss": 2.0113,
      "step": 962
    },
    {
      "epoch": 1.2882943143812708,
      "grad_norm": 1.1533622741699219,
      "learning_rate": 5.7541647906348496e-05,
      "loss": 1.0177,
      "step": 963
    },
    {
      "epoch": 1.2896321070234114,
      "grad_norm": 0.8715507984161377,
      "learning_rate": 5.74966231427285e-05,
      "loss": 1.4832,
      "step": 964
    },
    {
      "epoch": 1.2909698996655519,
      "grad_norm": 1.0615633726119995,
      "learning_rate": 5.745159837910852e-05,
      "loss": 1.5385,
      "step": 965
    },
    {
      "epoch": 1.2923076923076924,
      "grad_norm": 1.2473442554473877,
      "learning_rate": 5.740657361548852e-05,
      "loss": 1.3534,
      "step": 966
    },
    {
      "epoch": 1.293645484949833,
      "grad_norm": 1.1255221366882324,
      "learning_rate": 5.736154885186853e-05,
      "loss": 1.666,
      "step": 967
    },
    {
      "epoch": 1.2949832775919732,
      "grad_norm": 1.0651443004608154,
      "learning_rate": 5.731652408824853e-05,
      "loss": 1.7063,
      "step": 968
    },
    {
      "epoch": 1.2963210702341137,
      "grad_norm": 0.7863209247589111,
      "learning_rate": 5.7271499324628554e-05,
      "loss": 1.8771,
      "step": 969
    },
    {
      "epoch": 1.2976588628762542,
      "grad_norm": 1.0300050973892212,
      "learning_rate": 5.7226474561008556e-05,
      "loss": 1.6595,
      "step": 970
    },
    {
      "epoch": 1.2989966555183947,
      "grad_norm": 1.2955358028411865,
      "learning_rate": 5.7181449797388565e-05,
      "loss": 1.1926,
      "step": 971
    },
    {
      "epoch": 1.3003344481605352,
      "grad_norm": 1.4787100553512573,
      "learning_rate": 5.7136425033768573e-05,
      "loss": 1.6715,
      "step": 972
    },
    {
      "epoch": 1.3016722408026755,
      "grad_norm": 1.446374773979187,
      "learning_rate": 5.709140027014859e-05,
      "loss": 1.3466,
      "step": 973
    },
    {
      "epoch": 1.303010033444816,
      "grad_norm": 0.8684519529342651,
      "learning_rate": 5.704637550652859e-05,
      "loss": 1.3811,
      "step": 974
    },
    {
      "epoch": 1.3043478260869565,
      "grad_norm": 4.5600361824035645,
      "learning_rate": 5.70013507429086e-05,
      "loss": 1.3829,
      "step": 975
    },
    {
      "epoch": 1.305685618729097,
      "grad_norm": 1.1232842206954956,
      "learning_rate": 5.6956325979288615e-05,
      "loss": 1.7418,
      "step": 976
    },
    {
      "epoch": 1.3070234113712376,
      "grad_norm": 0.8644034266471863,
      "learning_rate": 5.691130121566862e-05,
      "loss": 1.3921,
      "step": 977
    },
    {
      "epoch": 1.3083612040133779,
      "grad_norm": 1.126865029335022,
      "learning_rate": 5.6866276452048625e-05,
      "loss": 1.2891,
      "step": 978
    },
    {
      "epoch": 1.3096989966555184,
      "grad_norm": 0.7839686870574951,
      "learning_rate": 5.6821251688428634e-05,
      "loss": 1.6133,
      "step": 979
    },
    {
      "epoch": 1.3110367892976589,
      "grad_norm": 0.9094774127006531,
      "learning_rate": 5.677622692480865e-05,
      "loss": 1.518,
      "step": 980
    },
    {
      "epoch": 1.3123745819397994,
      "grad_norm": 1.0354312658309937,
      "learning_rate": 5.673120216118866e-05,
      "loss": 1.1613,
      "step": 981
    },
    {
      "epoch": 1.31371237458194,
      "grad_norm": 1.0073140859603882,
      "learning_rate": 5.6686177397568666e-05,
      "loss": 2.0232,
      "step": 982
    },
    {
      "epoch": 1.3150501672240802,
      "grad_norm": 0.9858584403991699,
      "learning_rate": 5.664115263394868e-05,
      "loss": 0.9684,
      "step": 983
    },
    {
      "epoch": 1.3163879598662207,
      "grad_norm": 0.9003674387931824,
      "learning_rate": 5.659612787032868e-05,
      "loss": 1.4406,
      "step": 984
    },
    {
      "epoch": 1.3177257525083612,
      "grad_norm": 0.771503210067749,
      "learning_rate": 5.655110310670869e-05,
      "loss": 1.6696,
      "step": 985
    },
    {
      "epoch": 1.3190635451505017,
      "grad_norm": 2.726419687271118,
      "learning_rate": 5.65060783430887e-05,
      "loss": 1.1897,
      "step": 986
    },
    {
      "epoch": 1.3204013377926422,
      "grad_norm": 2.211148738861084,
      "learning_rate": 5.6461053579468716e-05,
      "loss": 2.0176,
      "step": 987
    },
    {
      "epoch": 1.3217391304347825,
      "grad_norm": 0.803394615650177,
      "learning_rate": 5.641602881584872e-05,
      "loss": 1.6801,
      "step": 988
    },
    {
      "epoch": 1.323076923076923,
      "grad_norm": 0.9389605522155762,
      "learning_rate": 5.6371004052228726e-05,
      "loss": 1.8548,
      "step": 989
    },
    {
      "epoch": 1.3244147157190636,
      "grad_norm": 1.031009316444397,
      "learning_rate": 5.6325979288608735e-05,
      "loss": 1.5096,
      "step": 990
    },
    {
      "epoch": 1.325752508361204,
      "grad_norm": 1.1561170816421509,
      "learning_rate": 5.628095452498875e-05,
      "loss": 1.7523,
      "step": 991
    },
    {
      "epoch": 1.3270903010033446,
      "grad_norm": 0.7659590840339661,
      "learning_rate": 5.623592976136875e-05,
      "loss": 1.0119,
      "step": 992
    },
    {
      "epoch": 1.3284280936454849,
      "grad_norm": 0.6387937664985657,
      "learning_rate": 5.619090499774876e-05,
      "loss": 2.0261,
      "step": 993
    },
    {
      "epoch": 1.3297658862876254,
      "grad_norm": 0.7081699967384338,
      "learning_rate": 5.6145880234128776e-05,
      "loss": 1.602,
      "step": 994
    },
    {
      "epoch": 1.3311036789297659,
      "grad_norm": 0.6930742263793945,
      "learning_rate": 5.6100855470508785e-05,
      "loss": 1.7137,
      "step": 995
    },
    {
      "epoch": 1.3324414715719064,
      "grad_norm": 0.7945417165756226,
      "learning_rate": 5.605583070688879e-05,
      "loss": 1.384,
      "step": 996
    },
    {
      "epoch": 1.333779264214047,
      "grad_norm": 0.736883282661438,
      "learning_rate": 5.6010805943268795e-05,
      "loss": 1.7751,
      "step": 997
    },
    {
      "epoch": 1.3351170568561872,
      "grad_norm": 0.7430248260498047,
      "learning_rate": 5.596578117964881e-05,
      "loss": 1.4957,
      "step": 998
    },
    {
      "epoch": 1.3364548494983277,
      "grad_norm": 0.7859342694282532,
      "learning_rate": 5.592075641602882e-05,
      "loss": 1.6672,
      "step": 999
    },
    {
      "epoch": 1.3377926421404682,
      "grad_norm": 1.0065340995788574,
      "learning_rate": 5.587573165240883e-05,
      "loss": 1.4852,
      "step": 1000
    },
    {
      "epoch": 1.3391304347826087,
      "grad_norm": 0.8629280924797058,
      "learning_rate": 5.583070688878884e-05,
      "loss": 1.6693,
      "step": 1001
    },
    {
      "epoch": 1.3404682274247492,
      "grad_norm": 0.7851661443710327,
      "learning_rate": 5.5785682125168845e-05,
      "loss": 1.319,
      "step": 1002
    },
    {
      "epoch": 1.3418060200668895,
      "grad_norm": 1.035677433013916,
      "learning_rate": 5.574065736154885e-05,
      "loss": 1.6294,
      "step": 1003
    },
    {
      "epoch": 1.34314381270903,
      "grad_norm": 1.3206753730773926,
      "learning_rate": 5.569563259792886e-05,
      "loss": 0.8565,
      "step": 1004
    },
    {
      "epoch": 1.3444816053511706,
      "grad_norm": 0.796152651309967,
      "learning_rate": 5.565060783430888e-05,
      "loss": 1.9462,
      "step": 1005
    },
    {
      "epoch": 1.345819397993311,
      "grad_norm": 1.0995591878890991,
      "learning_rate": 5.5605583070688886e-05,
      "loss": 1.8064,
      "step": 1006
    },
    {
      "epoch": 1.3471571906354516,
      "grad_norm": 0.8635403513908386,
      "learning_rate": 5.556055830706889e-05,
      "loss": 1.411,
      "step": 1007
    },
    {
      "epoch": 1.3484949832775919,
      "grad_norm": 1.2002208232879639,
      "learning_rate": 5.5515533543448896e-05,
      "loss": 1.4774,
      "step": 1008
    },
    {
      "epoch": 1.3498327759197324,
      "grad_norm": 0.9561333656311035,
      "learning_rate": 5.547050877982891e-05,
      "loss": 1.4763,
      "step": 1009
    },
    {
      "epoch": 1.351170568561873,
      "grad_norm": 0.8307086229324341,
      "learning_rate": 5.542548401620892e-05,
      "loss": 1.7697,
      "step": 1010
    },
    {
      "epoch": 1.3525083612040134,
      "grad_norm": 0.797065794467926,
      "learning_rate": 5.538045925258892e-05,
      "loss": 1.6327,
      "step": 1011
    },
    {
      "epoch": 1.353846153846154,
      "grad_norm": 1.3037409782409668,
      "learning_rate": 5.533543448896894e-05,
      "loss": 0.8671,
      "step": 1012
    },
    {
      "epoch": 1.3551839464882942,
      "grad_norm": 0.8096974492073059,
      "learning_rate": 5.5290409725348946e-05,
      "loss": 1.7238,
      "step": 1013
    },
    {
      "epoch": 1.3565217391304347,
      "grad_norm": 0.9276493191719055,
      "learning_rate": 5.5245384961728954e-05,
      "loss": 1.3982,
      "step": 1014
    },
    {
      "epoch": 1.3578595317725752,
      "grad_norm": 0.7602790594100952,
      "learning_rate": 5.5200360198108956e-05,
      "loss": 1.818,
      "step": 1015
    },
    {
      "epoch": 1.3591973244147157,
      "grad_norm": 0.7128927111625671,
      "learning_rate": 5.515533543448897e-05,
      "loss": 1.8052,
      "step": 1016
    },
    {
      "epoch": 1.3605351170568563,
      "grad_norm": 1.1914445161819458,
      "learning_rate": 5.511031067086898e-05,
      "loss": 1.6337,
      "step": 1017
    },
    {
      "epoch": 1.3618729096989965,
      "grad_norm": 2.203735828399658,
      "learning_rate": 5.506528590724899e-05,
      "loss": 1.4609,
      "step": 1018
    },
    {
      "epoch": 1.363210702341137,
      "grad_norm": 1.0320219993591309,
      "learning_rate": 5.5020261143629004e-05,
      "loss": 1.2802,
      "step": 1019
    },
    {
      "epoch": 1.3645484949832776,
      "grad_norm": 0.956654965877533,
      "learning_rate": 5.497523638000901e-05,
      "loss": 1.5686,
      "step": 1020
    },
    {
      "epoch": 1.365886287625418,
      "grad_norm": 0.8791170716285706,
      "learning_rate": 5.4930211616389015e-05,
      "loss": 1.4996,
      "step": 1021
    },
    {
      "epoch": 1.3672240802675586,
      "grad_norm": 0.8812573552131653,
      "learning_rate": 5.488518685276902e-05,
      "loss": 1.5253,
      "step": 1022
    },
    {
      "epoch": 1.3685618729096989,
      "grad_norm": 1.1747952699661255,
      "learning_rate": 5.484016208914904e-05,
      "loss": 1.0431,
      "step": 1023
    },
    {
      "epoch": 1.3698996655518394,
      "grad_norm": 0.892593502998352,
      "learning_rate": 5.479513732552905e-05,
      "loss": 1.8528,
      "step": 1024
    },
    {
      "epoch": 1.37123745819398,
      "grad_norm": 0.8653114438056946,
      "learning_rate": 5.475011256190905e-05,
      "loss": 1.6524,
      "step": 1025
    },
    {
      "epoch": 1.3725752508361204,
      "grad_norm": 1.1998345851898193,
      "learning_rate": 5.470508779828906e-05,
      "loss": 1.6844,
      "step": 1026
    },
    {
      "epoch": 1.373913043478261,
      "grad_norm": 0.8225812315940857,
      "learning_rate": 5.466006303466907e-05,
      "loss": 1.5627,
      "step": 1027
    },
    {
      "epoch": 1.3752508361204012,
      "grad_norm": 2.0456786155700684,
      "learning_rate": 5.461503827104908e-05,
      "loss": 1.7507,
      "step": 1028
    },
    {
      "epoch": 1.3765886287625417,
      "grad_norm": 0.9611921906471252,
      "learning_rate": 5.457001350742908e-05,
      "loss": 1.4718,
      "step": 1029
    },
    {
      "epoch": 1.3779264214046822,
      "grad_norm": 0.7811959385871887,
      "learning_rate": 5.4524988743809105e-05,
      "loss": 1.7484,
      "step": 1030
    },
    {
      "epoch": 1.3792642140468228,
      "grad_norm": 0.9279390573501587,
      "learning_rate": 5.447996398018911e-05,
      "loss": 1.7394,
      "step": 1031
    },
    {
      "epoch": 1.3806020066889633,
      "grad_norm": 0.9735180139541626,
      "learning_rate": 5.4434939216569116e-05,
      "loss": 1.8804,
      "step": 1032
    },
    {
      "epoch": 1.3819397993311036,
      "grad_norm": 0.8117311000823975,
      "learning_rate": 5.438991445294912e-05,
      "loss": 1.2928,
      "step": 1033
    },
    {
      "epoch": 1.383277591973244,
      "grad_norm": 0.8646334409713745,
      "learning_rate": 5.434488968932914e-05,
      "loss": 1.8303,
      "step": 1034
    },
    {
      "epoch": 1.3846153846153846,
      "grad_norm": 0.9675334692001343,
      "learning_rate": 5.429986492570914e-05,
      "loss": 1.8803,
      "step": 1035
    },
    {
      "epoch": 1.385953177257525,
      "grad_norm": 0.8046660423278809,
      "learning_rate": 5.425484016208915e-05,
      "loss": 1.7378,
      "step": 1036
    },
    {
      "epoch": 1.3872909698996656,
      "grad_norm": 0.6853929758071899,
      "learning_rate": 5.4209815398469166e-05,
      "loss": 1.8967,
      "step": 1037
    },
    {
      "epoch": 1.388628762541806,
      "grad_norm": 0.9253784418106079,
      "learning_rate": 5.4164790634849174e-05,
      "loss": 1.3781,
      "step": 1038
    },
    {
      "epoch": 1.3899665551839464,
      "grad_norm": 0.8796266317367554,
      "learning_rate": 5.4119765871229176e-05,
      "loss": 1.478,
      "step": 1039
    },
    {
      "epoch": 1.391304347826087,
      "grad_norm": 1.2055280208587646,
      "learning_rate": 5.4074741107609185e-05,
      "loss": 1.6066,
      "step": 1040
    },
    {
      "epoch": 1.3926421404682274,
      "grad_norm": 0.8891392946243286,
      "learning_rate": 5.40297163439892e-05,
      "loss": 1.4703,
      "step": 1041
    },
    {
      "epoch": 1.393979933110368,
      "grad_norm": 1.0517340898513794,
      "learning_rate": 5.398469158036921e-05,
      "loss": 1.7708,
      "step": 1042
    },
    {
      "epoch": 1.3953177257525082,
      "grad_norm": 0.7426386475563049,
      "learning_rate": 5.393966681674921e-05,
      "loss": 1.6997,
      "step": 1043
    },
    {
      "epoch": 1.396655518394649,
      "grad_norm": 0.7888435125350952,
      "learning_rate": 5.389464205312922e-05,
      "loss": 2.0114,
      "step": 1044
    },
    {
      "epoch": 1.3979933110367893,
      "grad_norm": 0.8119488954544067,
      "learning_rate": 5.3849617289509234e-05,
      "loss": 1.7638,
      "step": 1045
    },
    {
      "epoch": 1.3993311036789298,
      "grad_norm": 0.8995359539985657,
      "learning_rate": 5.380459252588924e-05,
      "loss": 1.9673,
      "step": 1046
    },
    {
      "epoch": 1.4006688963210703,
      "grad_norm": 0.903998851776123,
      "learning_rate": 5.3759567762269245e-05,
      "loss": 1.664,
      "step": 1047
    },
    {
      "epoch": 1.4020066889632106,
      "grad_norm": 0.7715796828269958,
      "learning_rate": 5.371454299864927e-05,
      "loss": 1.9947,
      "step": 1048
    },
    {
      "epoch": 1.4033444816053513,
      "grad_norm": 1.005541443824768,
      "learning_rate": 5.366951823502927e-05,
      "loss": 0.9912,
      "step": 1049
    },
    {
      "epoch": 1.4046822742474916,
      "grad_norm": 1.4997984170913696,
      "learning_rate": 5.362449347140928e-05,
      "loss": 1.5869,
      "step": 1050
    },
    {
      "epoch": 1.406020066889632,
      "grad_norm": 0.7256458401679993,
      "learning_rate": 5.357946870778928e-05,
      "loss": 1.7334,
      "step": 1051
    },
    {
      "epoch": 1.4073578595317726,
      "grad_norm": 0.9447379112243652,
      "learning_rate": 5.35344439441693e-05,
      "loss": 1.5052,
      "step": 1052
    },
    {
      "epoch": 1.4086956521739131,
      "grad_norm": 0.7913347482681274,
      "learning_rate": 5.34894191805493e-05,
      "loss": 2.0669,
      "step": 1053
    },
    {
      "epoch": 1.4100334448160536,
      "grad_norm": 1.1335333585739136,
      "learning_rate": 5.344439441692931e-05,
      "loss": 1.8613,
      "step": 1054
    },
    {
      "epoch": 1.411371237458194,
      "grad_norm": 0.7583142518997192,
      "learning_rate": 5.339936965330933e-05,
      "loss": 2.0301,
      "step": 1055
    },
    {
      "epoch": 1.4127090301003344,
      "grad_norm": 1.1084086894989014,
      "learning_rate": 5.3354344889689335e-05,
      "loss": 1.6319,
      "step": 1056
    },
    {
      "epoch": 1.414046822742475,
      "grad_norm": 1.1328237056732178,
      "learning_rate": 5.330932012606934e-05,
      "loss": 1.2222,
      "step": 1057
    },
    {
      "epoch": 1.4153846153846155,
      "grad_norm": 1.1727817058563232,
      "learning_rate": 5.3264295362449346e-05,
      "loss": 1.4558,
      "step": 1058
    },
    {
      "epoch": 1.416722408026756,
      "grad_norm": 1.4107069969177246,
      "learning_rate": 5.321927059882936e-05,
      "loss": 1.0919,
      "step": 1059
    },
    {
      "epoch": 1.4180602006688963,
      "grad_norm": 0.8056756854057312,
      "learning_rate": 5.317424583520937e-05,
      "loss": 1.9293,
      "step": 1060
    },
    {
      "epoch": 1.4193979933110368,
      "grad_norm": 0.814609169960022,
      "learning_rate": 5.312922107158937e-05,
      "loss": 1.9909,
      "step": 1061
    },
    {
      "epoch": 1.4207357859531773,
      "grad_norm": 0.8857481479644775,
      "learning_rate": 5.308419630796938e-05,
      "loss": 1.7429,
      "step": 1062
    },
    {
      "epoch": 1.4220735785953178,
      "grad_norm": 0.7829902172088623,
      "learning_rate": 5.3039171544349396e-05,
      "loss": 1.7188,
      "step": 1063
    },
    {
      "epoch": 1.4234113712374583,
      "grad_norm": 0.8884121179580688,
      "learning_rate": 5.2994146780729404e-05,
      "loss": 1.4074,
      "step": 1064
    },
    {
      "epoch": 1.4247491638795986,
      "grad_norm": 1.5147448778152466,
      "learning_rate": 5.2949122017109406e-05,
      "loss": 1.2415,
      "step": 1065
    },
    {
      "epoch": 1.4260869565217391,
      "grad_norm": 1.034368634223938,
      "learning_rate": 5.290409725348943e-05,
      "loss": 1.3288,
      "step": 1066
    },
    {
      "epoch": 1.4274247491638796,
      "grad_norm": 0.6370885372161865,
      "learning_rate": 5.285907248986943e-05,
      "loss": 1.3238,
      "step": 1067
    },
    {
      "epoch": 1.4287625418060201,
      "grad_norm": 0.7260492444038391,
      "learning_rate": 5.281404772624944e-05,
      "loss": 1.2832,
      "step": 1068
    },
    {
      "epoch": 1.4301003344481606,
      "grad_norm": 0.7294813990592957,
      "learning_rate": 5.276902296262945e-05,
      "loss": 1.652,
      "step": 1069
    },
    {
      "epoch": 1.431438127090301,
      "grad_norm": 0.8732017278671265,
      "learning_rate": 5.272399819900946e-05,
      "loss": 1.5793,
      "step": 1070
    },
    {
      "epoch": 1.4327759197324414,
      "grad_norm": 0.734139084815979,
      "learning_rate": 5.2678973435389464e-05,
      "loss": 1.6961,
      "step": 1071
    },
    {
      "epoch": 1.434113712374582,
      "grad_norm": 0.9525502920150757,
      "learning_rate": 5.263394867176947e-05,
      "loss": 1.6207,
      "step": 1072
    },
    {
      "epoch": 1.4354515050167225,
      "grad_norm": 1.0080657005310059,
      "learning_rate": 5.258892390814949e-05,
      "loss": 1.4763,
      "step": 1073
    },
    {
      "epoch": 1.436789297658863,
      "grad_norm": 0.7712482810020447,
      "learning_rate": 5.25438991445295e-05,
      "loss": 1.5778,
      "step": 1074
    },
    {
      "epoch": 1.4381270903010033,
      "grad_norm": 1.0089412927627563,
      "learning_rate": 5.24988743809095e-05,
      "loss": 1.7167,
      "step": 1075
    },
    {
      "epoch": 1.4394648829431438,
      "grad_norm": 0.7854228019714355,
      "learning_rate": 5.245384961728951e-05,
      "loss": 1.0463,
      "step": 1076
    },
    {
      "epoch": 1.4408026755852843,
      "grad_norm": 1.0776352882385254,
      "learning_rate": 5.240882485366952e-05,
      "loss": 1.074,
      "step": 1077
    },
    {
      "epoch": 1.4421404682274248,
      "grad_norm": 1.3480685949325562,
      "learning_rate": 5.236380009004953e-05,
      "loss": 1.6723,
      "step": 1078
    },
    {
      "epoch": 1.4434782608695653,
      "grad_norm": 0.8922518491744995,
      "learning_rate": 5.231877532642954e-05,
      "loss": 1.4558,
      "step": 1079
    },
    {
      "epoch": 1.4448160535117056,
      "grad_norm": 0.832686185836792,
      "learning_rate": 5.227375056280954e-05,
      "loss": 1.4743,
      "step": 1080
    },
    {
      "epoch": 1.4461538461538461,
      "grad_norm": 0.9173155426979065,
      "learning_rate": 5.222872579918956e-05,
      "loss": 1.6476,
      "step": 1081
    },
    {
      "epoch": 1.4474916387959866,
      "grad_norm": 1.1362848281860352,
      "learning_rate": 5.2183701035569566e-05,
      "loss": 1.2387,
      "step": 1082
    },
    {
      "epoch": 1.4488294314381271,
      "grad_norm": 0.9539154171943665,
      "learning_rate": 5.2138676271949574e-05,
      "loss": 1.6121,
      "step": 1083
    },
    {
      "epoch": 1.4501672240802677,
      "grad_norm": 1.0462729930877686,
      "learning_rate": 5.209365150832959e-05,
      "loss": 1.3236,
      "step": 1084
    },
    {
      "epoch": 1.451505016722408,
      "grad_norm": 0.9178979992866516,
      "learning_rate": 5.204862674470959e-05,
      "loss": 1.6152,
      "step": 1085
    },
    {
      "epoch": 1.4528428093645485,
      "grad_norm": 0.9143950939178467,
      "learning_rate": 5.20036019810896e-05,
      "loss": 1.7637,
      "step": 1086
    },
    {
      "epoch": 1.454180602006689,
      "grad_norm": 0.9704089164733887,
      "learning_rate": 5.195857721746961e-05,
      "loss": 0.9608,
      "step": 1087
    },
    {
      "epoch": 1.4555183946488295,
      "grad_norm": 0.6342671513557434,
      "learning_rate": 5.1913552453849624e-05,
      "loss": 1.4244,
      "step": 1088
    },
    {
      "epoch": 1.45685618729097,
      "grad_norm": 1.4497864246368408,
      "learning_rate": 5.186852769022963e-05,
      "loss": 1.86,
      "step": 1089
    },
    {
      "epoch": 1.4581939799331103,
      "grad_norm": 0.6924653649330139,
      "learning_rate": 5.1823502926609634e-05,
      "loss": 1.7662,
      "step": 1090
    },
    {
      "epoch": 1.4595317725752508,
      "grad_norm": 0.9326013922691345,
      "learning_rate": 5.177847816298965e-05,
      "loss": 1.2249,
      "step": 1091
    },
    {
      "epoch": 1.4608695652173913,
      "grad_norm": 0.8724308013916016,
      "learning_rate": 5.173345339936966e-05,
      "loss": 1.2829,
      "step": 1092
    },
    {
      "epoch": 1.4622073578595318,
      "grad_norm": 1.2138267755508423,
      "learning_rate": 5.168842863574967e-05,
      "loss": 1.025,
      "step": 1093
    },
    {
      "epoch": 1.4635451505016723,
      "grad_norm": 0.9376569390296936,
      "learning_rate": 5.164340387212967e-05,
      "loss": 1.5824,
      "step": 1094
    },
    {
      "epoch": 1.4648829431438126,
      "grad_norm": 0.81294184923172,
      "learning_rate": 5.1598379108509684e-05,
      "loss": 1.2309,
      "step": 1095
    },
    {
      "epoch": 1.4662207357859531,
      "grad_norm": 1.194340467453003,
      "learning_rate": 5.155335434488969e-05,
      "loss": 1.1125,
      "step": 1096
    },
    {
      "epoch": 1.4675585284280936,
      "grad_norm": 0.8782414793968201,
      "learning_rate": 5.15083295812697e-05,
      "loss": 1.6568,
      "step": 1097
    },
    {
      "epoch": 1.4688963210702342,
      "grad_norm": 0.653789222240448,
      "learning_rate": 5.14633048176497e-05,
      "loss": 1.3559,
      "step": 1098
    },
    {
      "epoch": 1.4702341137123747,
      "grad_norm": 0.9649755358695984,
      "learning_rate": 5.141828005402972e-05,
      "loss": 1.3056,
      "step": 1099
    },
    {
      "epoch": 1.471571906354515,
      "grad_norm": 0.8330010175704956,
      "learning_rate": 5.137325529040973e-05,
      "loss": 1.5524,
      "step": 1100
    },
    {
      "epoch": 1.4729096989966555,
      "grad_norm": 1.0811066627502441,
      "learning_rate": 5.1328230526789736e-05,
      "loss": 1.4753,
      "step": 1101
    },
    {
      "epoch": 1.474247491638796,
      "grad_norm": 0.8166443705558777,
      "learning_rate": 5.128320576316975e-05,
      "loss": 1.7568,
      "step": 1102
    },
    {
      "epoch": 1.4755852842809365,
      "grad_norm": 1.201744794845581,
      "learning_rate": 5.123818099954976e-05,
      "loss": 1.2685,
      "step": 1103
    },
    {
      "epoch": 1.476923076923077,
      "grad_norm": 0.8646392226219177,
      "learning_rate": 5.119315623592976e-05,
      "loss": 1.5086,
      "step": 1104
    },
    {
      "epoch": 1.4782608695652173,
      "grad_norm": 0.9522072076797485,
      "learning_rate": 5.114813147230977e-05,
      "loss": 1.8555,
      "step": 1105
    },
    {
      "epoch": 1.4795986622073578,
      "grad_norm": 0.8802261352539062,
      "learning_rate": 5.1103106708689785e-05,
      "loss": 1.6987,
      "step": 1106
    },
    {
      "epoch": 1.4809364548494983,
      "grad_norm": 0.8111658096313477,
      "learning_rate": 5.1058081945069794e-05,
      "loss": 1.6291,
      "step": 1107
    },
    {
      "epoch": 1.4822742474916388,
      "grad_norm": 1.1969858407974243,
      "learning_rate": 5.1013057181449796e-05,
      "loss": 1.332,
      "step": 1108
    },
    {
      "epoch": 1.4836120401337793,
      "grad_norm": 0.8505260944366455,
      "learning_rate": 5.096803241782981e-05,
      "loss": 1.81,
      "step": 1109
    },
    {
      "epoch": 1.4849498327759196,
      "grad_norm": 0.7508300542831421,
      "learning_rate": 5.092300765420982e-05,
      "loss": 1.7325,
      "step": 1110
    },
    {
      "epoch": 1.4862876254180601,
      "grad_norm": 0.8055757284164429,
      "learning_rate": 5.087798289058983e-05,
      "loss": 1.8016,
      "step": 1111
    },
    {
      "epoch": 1.4876254180602007,
      "grad_norm": 0.8028207421302795,
      "learning_rate": 5.083295812696983e-05,
      "loss": 2.0108,
      "step": 1112
    },
    {
      "epoch": 1.4889632107023412,
      "grad_norm": 0.8982310891151428,
      "learning_rate": 5.078793336334985e-05,
      "loss": 1.8517,
      "step": 1113
    },
    {
      "epoch": 1.4903010033444817,
      "grad_norm": 1.2262216806411743,
      "learning_rate": 5.0742908599729854e-05,
      "loss": 1.5475,
      "step": 1114
    },
    {
      "epoch": 1.491638795986622,
      "grad_norm": 0.692025899887085,
      "learning_rate": 5.069788383610986e-05,
      "loss": 1.2832,
      "step": 1115
    },
    {
      "epoch": 1.4929765886287625,
      "grad_norm": 1.0912442207336426,
      "learning_rate": 5.0652859072489864e-05,
      "loss": 1.778,
      "step": 1116
    },
    {
      "epoch": 1.494314381270903,
      "grad_norm": 0.8672052025794983,
      "learning_rate": 5.0607834308869886e-05,
      "loss": 1.6913,
      "step": 1117
    },
    {
      "epoch": 1.4956521739130435,
      "grad_norm": 0.9315889477729797,
      "learning_rate": 5.056280954524989e-05,
      "loss": 1.6336,
      "step": 1118
    },
    {
      "epoch": 1.496989966555184,
      "grad_norm": 0.8745747208595276,
      "learning_rate": 5.05177847816299e-05,
      "loss": 1.5267,
      "step": 1119
    },
    {
      "epoch": 1.4983277591973243,
      "grad_norm": 1.3531934022903442,
      "learning_rate": 5.047276001800991e-05,
      "loss": 1.3997,
      "step": 1120
    },
    {
      "epoch": 1.4996655518394648,
      "grad_norm": 0.9693718552589417,
      "learning_rate": 5.042773525438992e-05,
      "loss": 1.5723,
      "step": 1121
    },
    {
      "epoch": 1.5010033444816053,
      "grad_norm": 0.8846158385276794,
      "learning_rate": 5.038271049076992e-05,
      "loss": 1.1203,
      "step": 1122
    },
    {
      "epoch": 1.5023411371237458,
      "grad_norm": 1.0722146034240723,
      "learning_rate": 5.033768572714993e-05,
      "loss": 1.3621,
      "step": 1123
    },
    {
      "epoch": 1.5036789297658864,
      "grad_norm": 0.9507117867469788,
      "learning_rate": 5.0292660963529947e-05,
      "loss": 1.768,
      "step": 1124
    },
    {
      "epoch": 1.5050167224080266,
      "grad_norm": 0.9553461074829102,
      "learning_rate": 5.0247636199909955e-05,
      "loss": 2.001,
      "step": 1125
    },
    {
      "epoch": 1.5063545150501674,
      "grad_norm": 1.0646768808364868,
      "learning_rate": 5.020261143628996e-05,
      "loss": 1.7424,
      "step": 1126
    },
    {
      "epoch": 1.5076923076923077,
      "grad_norm": 0.8669482469558716,
      "learning_rate": 5.015758667266998e-05,
      "loss": 1.067,
      "step": 1127
    },
    {
      "epoch": 1.5090301003344482,
      "grad_norm": 1.4234883785247803,
      "learning_rate": 5.011256190904998e-05,
      "loss": 1.1849,
      "step": 1128
    },
    {
      "epoch": 1.5103678929765887,
      "grad_norm": 1.3873578310012817,
      "learning_rate": 5.006753714542999e-05,
      "loss": 1.8282,
      "step": 1129
    },
    {
      "epoch": 1.511705685618729,
      "grad_norm": 0.8728304505348206,
      "learning_rate": 5.002251238180999e-05,
      "loss": 1.8036,
      "step": 1130
    },
    {
      "epoch": 1.5130434782608697,
      "grad_norm": 0.86270672082901,
      "learning_rate": 4.997748761819001e-05,
      "loss": 1.6129,
      "step": 1131
    },
    {
      "epoch": 1.51438127090301,
      "grad_norm": 1.422959804534912,
      "learning_rate": 4.9932462854570015e-05,
      "loss": 1.6149,
      "step": 1132
    },
    {
      "epoch": 1.5157190635451505,
      "grad_norm": 0.7661011815071106,
      "learning_rate": 4.9887438090950024e-05,
      "loss": 2.2417,
      "step": 1133
    },
    {
      "epoch": 1.517056856187291,
      "grad_norm": 0.9919681549072266,
      "learning_rate": 4.984241332733003e-05,
      "loss": 1.4351,
      "step": 1134
    },
    {
      "epoch": 1.5183946488294313,
      "grad_norm": 0.8810451030731201,
      "learning_rate": 4.979738856371005e-05,
      "loss": 1.5639,
      "step": 1135
    },
    {
      "epoch": 1.519732441471572,
      "grad_norm": 1.5315157175064087,
      "learning_rate": 4.975236380009005e-05,
      "loss": 1.5218,
      "step": 1136
    },
    {
      "epoch": 1.5210702341137123,
      "grad_norm": 1.3847192525863647,
      "learning_rate": 4.9707339036470065e-05,
      "loss": 1.18,
      "step": 1137
    },
    {
      "epoch": 1.5224080267558529,
      "grad_norm": 1.2832859754562378,
      "learning_rate": 4.966231427285007e-05,
      "loss": 1.4576,
      "step": 1138
    },
    {
      "epoch": 1.5237458193979934,
      "grad_norm": 0.918787956237793,
      "learning_rate": 4.961728950923008e-05,
      "loss": 1.6287,
      "step": 1139
    },
    {
      "epoch": 1.5250836120401337,
      "grad_norm": 1.324271559715271,
      "learning_rate": 4.9572264745610084e-05,
      "loss": 1.658,
      "step": 1140
    },
    {
      "epoch": 1.5264214046822744,
      "grad_norm": 1.5736923217773438,
      "learning_rate": 4.95272399819901e-05,
      "loss": 1.5404,
      "step": 1141
    },
    {
      "epoch": 1.5277591973244147,
      "grad_norm": 0.7107148766517639,
      "learning_rate": 4.94822152183701e-05,
      "loss": 2.0322,
      "step": 1142
    },
    {
      "epoch": 1.5290969899665552,
      "grad_norm": 0.8476628065109253,
      "learning_rate": 4.9437190454750117e-05,
      "loss": 1.3441,
      "step": 1143
    },
    {
      "epoch": 1.5304347826086957,
      "grad_norm": 0.6295139193534851,
      "learning_rate": 4.9392165691130125e-05,
      "loss": 1.3322,
      "step": 1144
    },
    {
      "epoch": 1.531772575250836,
      "grad_norm": 0.8430982232093811,
      "learning_rate": 4.9347140927510134e-05,
      "loss": 1.0624,
      "step": 1145
    },
    {
      "epoch": 1.5331103678929767,
      "grad_norm": 1.353960394859314,
      "learning_rate": 4.930211616389014e-05,
      "loss": 1.0658,
      "step": 1146
    },
    {
      "epoch": 1.534448160535117,
      "grad_norm": 1.780779480934143,
      "learning_rate": 4.925709140027015e-05,
      "loss": 1.8983,
      "step": 1147
    },
    {
      "epoch": 1.5357859531772575,
      "grad_norm": 0.9181867241859436,
      "learning_rate": 4.921206663665016e-05,
      "loss": 1.3606,
      "step": 1148
    },
    {
      "epoch": 1.537123745819398,
      "grad_norm": 1.472495675086975,
      "learning_rate": 4.916704187303017e-05,
      "loss": 1.5477,
      "step": 1149
    },
    {
      "epoch": 1.5384615384615383,
      "grad_norm": 1.0969715118408203,
      "learning_rate": 4.912201710941018e-05,
      "loss": 1.5731,
      "step": 1150
    },
    {
      "epoch": 1.539799331103679,
      "grad_norm": 0.8129613995552063,
      "learning_rate": 4.9076992345790185e-05,
      "loss": 1.6769,
      "step": 1151
    },
    {
      "epoch": 1.5411371237458193,
      "grad_norm": 0.8746471405029297,
      "learning_rate": 4.9031967582170194e-05,
      "loss": 1.129,
      "step": 1152
    },
    {
      "epoch": 1.5424749163879599,
      "grad_norm": 1.206492304801941,
      "learning_rate": 4.898694281855021e-05,
      "loss": 1.3039,
      "step": 1153
    },
    {
      "epoch": 1.5438127090301004,
      "grad_norm": 1.562818169593811,
      "learning_rate": 4.894191805493021e-05,
      "loss": 1.3924,
      "step": 1154
    },
    {
      "epoch": 1.5451505016722407,
      "grad_norm": 2.0223031044006348,
      "learning_rate": 4.8896893291310226e-05,
      "loss": 1.346,
      "step": 1155
    },
    {
      "epoch": 1.5464882943143814,
      "grad_norm": 1.0795880556106567,
      "learning_rate": 4.885186852769023e-05,
      "loss": 1.4994,
      "step": 1156
    },
    {
      "epoch": 1.5478260869565217,
      "grad_norm": 1.5896492004394531,
      "learning_rate": 4.8806843764070244e-05,
      "loss": 1.2325,
      "step": 1157
    },
    {
      "epoch": 1.5491638795986622,
      "grad_norm": 0.875981867313385,
      "learning_rate": 4.8761819000450245e-05,
      "loss": 2.1033,
      "step": 1158
    },
    {
      "epoch": 1.5505016722408027,
      "grad_norm": 0.7678603529930115,
      "learning_rate": 4.871679423683026e-05,
      "loss": 1.7353,
      "step": 1159
    },
    {
      "epoch": 1.551839464882943,
      "grad_norm": 1.0514904260635376,
      "learning_rate": 4.867176947321026e-05,
      "loss": 1.5402,
      "step": 1160
    },
    {
      "epoch": 1.5531772575250837,
      "grad_norm": 0.9771605134010315,
      "learning_rate": 4.862674470959028e-05,
      "loss": 1.1979,
      "step": 1161
    },
    {
      "epoch": 1.554515050167224,
      "grad_norm": 0.8701416850090027,
      "learning_rate": 4.8581719945970286e-05,
      "loss": 1.6865,
      "step": 1162
    },
    {
      "epoch": 1.5558528428093645,
      "grad_norm": 1.5773226022720337,
      "learning_rate": 4.8536695182350295e-05,
      "loss": 1.3537,
      "step": 1163
    },
    {
      "epoch": 1.557190635451505,
      "grad_norm": 0.8707898855209351,
      "learning_rate": 4.8491670418730304e-05,
      "loss": 1.6703,
      "step": 1164
    },
    {
      "epoch": 1.5585284280936453,
      "grad_norm": 0.8050997257232666,
      "learning_rate": 4.844664565511031e-05,
      "loss": 1.4691,
      "step": 1165
    },
    {
      "epoch": 1.559866220735786,
      "grad_norm": 0.828519880771637,
      "learning_rate": 4.840162089149032e-05,
      "loss": 0.9642,
      "step": 1166
    },
    {
      "epoch": 1.5612040133779264,
      "grad_norm": 0.8314371109008789,
      "learning_rate": 4.835659612787033e-05,
      "loss": 1.7049,
      "step": 1167
    },
    {
      "epoch": 1.5625418060200669,
      "grad_norm": 1.1783218383789062,
      "learning_rate": 4.831157136425034e-05,
      "loss": 1.3039,
      "step": 1168
    },
    {
      "epoch": 1.5638795986622074,
      "grad_norm": 0.7392192482948303,
      "learning_rate": 4.8266546600630347e-05,
      "loss": 2.0674,
      "step": 1169
    },
    {
      "epoch": 1.5652173913043477,
      "grad_norm": 0.969502866268158,
      "learning_rate": 4.8221521837010355e-05,
      "loss": 1.7466,
      "step": 1170
    },
    {
      "epoch": 1.5665551839464884,
      "grad_norm": 0.8645799160003662,
      "learning_rate": 4.817649707339037e-05,
      "loss": 1.6764,
      "step": 1171
    },
    {
      "epoch": 1.5678929765886287,
      "grad_norm": 1.0396491289138794,
      "learning_rate": 4.813147230977037e-05,
      "loss": 1.7447,
      "step": 1172
    },
    {
      "epoch": 1.5692307692307692,
      "grad_norm": 0.8425093293190002,
      "learning_rate": 4.808644754615039e-05,
      "loss": 1.9199,
      "step": 1173
    },
    {
      "epoch": 1.5705685618729097,
      "grad_norm": 1.2044014930725098,
      "learning_rate": 4.8041422782530396e-05,
      "loss": 1.6451,
      "step": 1174
    },
    {
      "epoch": 1.57190635451505,
      "grad_norm": 0.7370679378509521,
      "learning_rate": 4.7996398018910405e-05,
      "loss": 1.9203,
      "step": 1175
    },
    {
      "epoch": 1.5732441471571907,
      "grad_norm": 0.7956492900848389,
      "learning_rate": 4.7951373255290413e-05,
      "loss": 1.6613,
      "step": 1176
    },
    {
      "epoch": 1.574581939799331,
      "grad_norm": 0.969325065612793,
      "learning_rate": 4.790634849167042e-05,
      "loss": 1.775,
      "step": 1177
    },
    {
      "epoch": 1.5759197324414715,
      "grad_norm": 0.5993706583976746,
      "learning_rate": 4.786132372805043e-05,
      "loss": 1.3278,
      "step": 1178
    },
    {
      "epoch": 1.577257525083612,
      "grad_norm": 1.4955739974975586,
      "learning_rate": 4.781629896443044e-05,
      "loss": 1.8495,
      "step": 1179
    },
    {
      "epoch": 1.5785953177257523,
      "grad_norm": 0.760959804058075,
      "learning_rate": 4.777127420081045e-05,
      "loss": 1.1063,
      "step": 1180
    },
    {
      "epoch": 1.579933110367893,
      "grad_norm": 0.7534853219985962,
      "learning_rate": 4.7726249437190456e-05,
      "loss": 1.1722,
      "step": 1181
    },
    {
      "epoch": 1.5812709030100334,
      "grad_norm": 0.8445656299591064,
      "learning_rate": 4.7681224673570465e-05,
      "loss": 1.8389,
      "step": 1182
    },
    {
      "epoch": 1.5826086956521739,
      "grad_norm": 0.8543136119842529,
      "learning_rate": 4.7636199909950474e-05,
      "loss": 1.841,
      "step": 1183
    },
    {
      "epoch": 1.5839464882943144,
      "grad_norm": 1.0799471139907837,
      "learning_rate": 4.759117514633049e-05,
      "loss": 0.7011,
      "step": 1184
    },
    {
      "epoch": 1.585284280936455,
      "grad_norm": 0.9755014777183533,
      "learning_rate": 4.754615038271049e-05,
      "loss": 1.7702,
      "step": 1185
    },
    {
      "epoch": 1.5866220735785954,
      "grad_norm": 0.6892296671867371,
      "learning_rate": 4.7501125619090506e-05,
      "loss": 1.4784,
      "step": 1186
    },
    {
      "epoch": 1.5879598662207357,
      "grad_norm": 0.9510697722434998,
      "learning_rate": 4.745610085547051e-05,
      "loss": 1.5627,
      "step": 1187
    },
    {
      "epoch": 1.5892976588628762,
      "grad_norm": 1.0505142211914062,
      "learning_rate": 4.741107609185052e-05,
      "loss": 1.5736,
      "step": 1188
    },
    {
      "epoch": 1.5906354515050167,
      "grad_norm": 1.007228970527649,
      "learning_rate": 4.736605132823053e-05,
      "loss": 1.0688,
      "step": 1189
    },
    {
      "epoch": 1.5919732441471572,
      "grad_norm": 1.0539839267730713,
      "learning_rate": 4.732102656461054e-05,
      "loss": 1.6078,
      "step": 1190
    },
    {
      "epoch": 1.5933110367892978,
      "grad_norm": 0.7513359189033508,
      "learning_rate": 4.727600180099055e-05,
      "loss": 1.9948,
      "step": 1191
    },
    {
      "epoch": 1.594648829431438,
      "grad_norm": 0.9461736679077148,
      "learning_rate": 4.723097703737056e-05,
      "loss": 1.516,
      "step": 1192
    },
    {
      "epoch": 1.5959866220735786,
      "grad_norm": 0.9117209315299988,
      "learning_rate": 4.7185952273750566e-05,
      "loss": 1.6311,
      "step": 1193
    },
    {
      "epoch": 1.597324414715719,
      "grad_norm": 0.7895516157150269,
      "learning_rate": 4.7140927510130575e-05,
      "loss": 1.808,
      "step": 1194
    },
    {
      "epoch": 1.5986622073578596,
      "grad_norm": 1.0118502378463745,
      "learning_rate": 4.7095902746510583e-05,
      "loss": 1.6352,
      "step": 1195
    },
    {
      "epoch": 1.6,
      "grad_norm": 1.148454189300537,
      "learning_rate": 4.705087798289059e-05,
      "loss": 1.9035,
      "step": 1196
    },
    {
      "epoch": 1.6013377926421404,
      "grad_norm": 1.7625706195831299,
      "learning_rate": 4.70058532192706e-05,
      "loss": 1.3833,
      "step": 1197
    },
    {
      "epoch": 1.602675585284281,
      "grad_norm": 0.8092257976531982,
      "learning_rate": 4.6960828455650616e-05,
      "loss": 1.6618,
      "step": 1198
    },
    {
      "epoch": 1.6040133779264214,
      "grad_norm": 0.810875654220581,
      "learning_rate": 4.691580369203062e-05,
      "loss": 1.3098,
      "step": 1199
    },
    {
      "epoch": 1.605351170568562,
      "grad_norm": 0.8476257920265198,
      "learning_rate": 4.687077892841063e-05,
      "loss": 1.0308,
      "step": 1200
    },
    {
      "epoch": 1.6066889632107024,
      "grad_norm": 1.0738558769226074,
      "learning_rate": 4.6825754164790635e-05,
      "loss": 1.3836,
      "step": 1201
    },
    {
      "epoch": 1.6080267558528427,
      "grad_norm": 0.9387577772140503,
      "learning_rate": 4.678072940117065e-05,
      "loss": 1.3271,
      "step": 1202
    },
    {
      "epoch": 1.6093645484949832,
      "grad_norm": 1.344914436340332,
      "learning_rate": 4.673570463755065e-05,
      "loss": 2.1565,
      "step": 1203
    },
    {
      "epoch": 1.6107023411371237,
      "grad_norm": 0.9820061922073364,
      "learning_rate": 4.669067987393067e-05,
      "loss": 1.6031,
      "step": 1204
    },
    {
      "epoch": 1.6120401337792643,
      "grad_norm": 0.7256208062171936,
      "learning_rate": 4.664565511031067e-05,
      "loss": 1.9967,
      "step": 1205
    },
    {
      "epoch": 1.6133779264214048,
      "grad_norm": 0.8569954633712769,
      "learning_rate": 4.6600630346690685e-05,
      "loss": 1.6004,
      "step": 1206
    },
    {
      "epoch": 1.614715719063545,
      "grad_norm": 1.1218992471694946,
      "learning_rate": 4.655560558307069e-05,
      "loss": 1.5533,
      "step": 1207
    },
    {
      "epoch": 1.6160535117056856,
      "grad_norm": 1.0124077796936035,
      "learning_rate": 4.65105808194507e-05,
      "loss": 1.7343,
      "step": 1208
    },
    {
      "epoch": 1.617391304347826,
      "grad_norm": 0.9381147623062134,
      "learning_rate": 4.646555605583071e-05,
      "loss": 1.2428,
      "step": 1209
    },
    {
      "epoch": 1.6187290969899666,
      "grad_norm": 1.1799885034561157,
      "learning_rate": 4.642053129221072e-05,
      "loss": 1.6861,
      "step": 1210
    },
    {
      "epoch": 1.620066889632107,
      "grad_norm": 0.9148082733154297,
      "learning_rate": 4.637550652859073e-05,
      "loss": 1.3491,
      "step": 1211
    },
    {
      "epoch": 1.6214046822742474,
      "grad_norm": 1.0270072221755981,
      "learning_rate": 4.6330481764970736e-05,
      "loss": 1.4965,
      "step": 1212
    },
    {
      "epoch": 1.6227424749163881,
      "grad_norm": 0.8440571427345276,
      "learning_rate": 4.6285457001350745e-05,
      "loss": 1.9027,
      "step": 1213
    },
    {
      "epoch": 1.6240802675585284,
      "grad_norm": 0.8079622387886047,
      "learning_rate": 4.624043223773075e-05,
      "loss": 1.718,
      "step": 1214
    },
    {
      "epoch": 1.625418060200669,
      "grad_norm": 0.6909911036491394,
      "learning_rate": 4.619540747411076e-05,
      "loss": 1.9922,
      "step": 1215
    },
    {
      "epoch": 1.6267558528428094,
      "grad_norm": 1.2815625667572021,
      "learning_rate": 4.615038271049078e-05,
      "loss": 1.1817,
      "step": 1216
    },
    {
      "epoch": 1.6280936454849497,
      "grad_norm": 0.7101358771324158,
      "learning_rate": 4.610535794687078e-05,
      "loss": 1.6939,
      "step": 1217
    },
    {
      "epoch": 1.6294314381270905,
      "grad_norm": 0.7537413835525513,
      "learning_rate": 4.6060333183250794e-05,
      "loss": 1.9988,
      "step": 1218
    },
    {
      "epoch": 1.6307692307692307,
      "grad_norm": 0.7323551177978516,
      "learning_rate": 4.6015308419630796e-05,
      "loss": 1.9491,
      "step": 1219
    },
    {
      "epoch": 1.6321070234113713,
      "grad_norm": 1.0128388404846191,
      "learning_rate": 4.597028365601081e-05,
      "loss": 1.4372,
      "step": 1220
    },
    {
      "epoch": 1.6334448160535118,
      "grad_norm": 1.1328787803649902,
      "learning_rate": 4.5925258892390813e-05,
      "loss": 2.1449,
      "step": 1221
    },
    {
      "epoch": 1.634782608695652,
      "grad_norm": 0.9315971732139587,
      "learning_rate": 4.588023412877083e-05,
      "loss": 1.7581,
      "step": 1222
    },
    {
      "epoch": 1.6361204013377928,
      "grad_norm": 0.8285197019577026,
      "learning_rate": 4.583520936515083e-05,
      "loss": 1.6367,
      "step": 1223
    },
    {
      "epoch": 1.637458193979933,
      "grad_norm": 0.6926225423812866,
      "learning_rate": 4.5790184601530846e-05,
      "loss": 1.654,
      "step": 1224
    },
    {
      "epoch": 1.6387959866220736,
      "grad_norm": 0.824735164642334,
      "learning_rate": 4.5745159837910855e-05,
      "loss": 1.7624,
      "step": 1225
    },
    {
      "epoch": 1.640133779264214,
      "grad_norm": 0.9015515446662903,
      "learning_rate": 4.570013507429086e-05,
      "loss": 1.2503,
      "step": 1226
    },
    {
      "epoch": 1.6414715719063544,
      "grad_norm": 1.440433382987976,
      "learning_rate": 4.565511031067087e-05,
      "loss": 1.2357,
      "step": 1227
    },
    {
      "epoch": 1.6428093645484951,
      "grad_norm": 0.8556510806083679,
      "learning_rate": 4.561008554705088e-05,
      "loss": 1.3501,
      "step": 1228
    },
    {
      "epoch": 1.6441471571906354,
      "grad_norm": 0.7644262313842773,
      "learning_rate": 4.556506078343089e-05,
      "loss": 1.8273,
      "step": 1229
    },
    {
      "epoch": 1.645484949832776,
      "grad_norm": 2.027939558029175,
      "learning_rate": 4.55200360198109e-05,
      "loss": 1.1069,
      "step": 1230
    },
    {
      "epoch": 1.6468227424749164,
      "grad_norm": 1.0590338706970215,
      "learning_rate": 4.5475011256190906e-05,
      "loss": 1.3507,
      "step": 1231
    },
    {
      "epoch": 1.6481605351170567,
      "grad_norm": 1.007128119468689,
      "learning_rate": 4.5429986492570915e-05,
      "loss": 1.5589,
      "step": 1232
    },
    {
      "epoch": 1.6494983277591975,
      "grad_norm": 0.8251166343688965,
      "learning_rate": 4.538496172895092e-05,
      "loss": 1.7861,
      "step": 1233
    },
    {
      "epoch": 1.6508361204013378,
      "grad_norm": 0.9109119176864624,
      "learning_rate": 4.533993696533094e-05,
      "loss": 1.6144,
      "step": 1234
    },
    {
      "epoch": 1.6521739130434783,
      "grad_norm": 0.7894558310508728,
      "learning_rate": 4.529491220171094e-05,
      "loss": 1.5139,
      "step": 1235
    },
    {
      "epoch": 1.6535117056856188,
      "grad_norm": 0.7755162715911865,
      "learning_rate": 4.5249887438090956e-05,
      "loss": 1.7346,
      "step": 1236
    },
    {
      "epoch": 1.654849498327759,
      "grad_norm": 2.059093952178955,
      "learning_rate": 4.520486267447096e-05,
      "loss": 1.7854,
      "step": 1237
    },
    {
      "epoch": 1.6561872909698998,
      "grad_norm": 0.908669650554657,
      "learning_rate": 4.515983791085097e-05,
      "loss": 1.6148,
      "step": 1238
    },
    {
      "epoch": 1.65752508361204,
      "grad_norm": 0.776725709438324,
      "learning_rate": 4.5114813147230975e-05,
      "loss": 1.6544,
      "step": 1239
    },
    {
      "epoch": 1.6588628762541806,
      "grad_norm": 0.6670528650283813,
      "learning_rate": 4.506978838361099e-05,
      "loss": 1.3139,
      "step": 1240
    },
    {
      "epoch": 1.6602006688963211,
      "grad_norm": 1.0721896886825562,
      "learning_rate": 4.502476361999099e-05,
      "loss": 1.5505,
      "step": 1241
    },
    {
      "epoch": 1.6615384615384614,
      "grad_norm": 1.018912434577942,
      "learning_rate": 4.497973885637101e-05,
      "loss": 1.9343,
      "step": 1242
    },
    {
      "epoch": 1.6628762541806021,
      "grad_norm": 0.8976886868476868,
      "learning_rate": 4.4934714092751016e-05,
      "loss": 1.7946,
      "step": 1243
    },
    {
      "epoch": 1.6642140468227424,
      "grad_norm": 0.9715926647186279,
      "learning_rate": 4.4889689329131025e-05,
      "loss": 1.7167,
      "step": 1244
    },
    {
      "epoch": 1.665551839464883,
      "grad_norm": 0.8015096783638,
      "learning_rate": 4.484466456551103e-05,
      "loss": 1.5664,
      "step": 1245
    },
    {
      "epoch": 1.6668896321070235,
      "grad_norm": 0.968388020992279,
      "learning_rate": 4.479963980189104e-05,
      "loss": 1.4593,
      "step": 1246
    },
    {
      "epoch": 1.6682274247491637,
      "grad_norm": 0.811292290687561,
      "learning_rate": 4.475461503827105e-05,
      "loss": 1.7979,
      "step": 1247
    },
    {
      "epoch": 1.6695652173913045,
      "grad_norm": 0.7946311235427856,
      "learning_rate": 4.470959027465106e-05,
      "loss": 1.8932,
      "step": 1248
    },
    {
      "epoch": 1.6709030100334448,
      "grad_norm": 0.7939193248748779,
      "learning_rate": 4.466456551103107e-05,
      "loss": 1.4773,
      "step": 1249
    },
    {
      "epoch": 1.6722408026755853,
      "grad_norm": 1.309445858001709,
      "learning_rate": 4.4619540747411076e-05,
      "loss": 1.8627,
      "step": 1250
    },
    {
      "epoch": 1.6735785953177258,
      "grad_norm": 0.7015190720558167,
      "learning_rate": 4.4574515983791085e-05,
      "loss": 1.4652,
      "step": 1251
    },
    {
      "epoch": 1.674916387959866,
      "grad_norm": 1.03852117061615,
      "learning_rate": 4.45294912201711e-05,
      "loss": 1.8289,
      "step": 1252
    },
    {
      "epoch": 1.6762541806020068,
      "grad_norm": 0.6862776279449463,
      "learning_rate": 4.44844664565511e-05,
      "loss": 1.3505,
      "step": 1253
    },
    {
      "epoch": 1.677591973244147,
      "grad_norm": 0.7088447213172913,
      "learning_rate": 4.443944169293112e-05,
      "loss": 1.3436,
      "step": 1254
    },
    {
      "epoch": 1.6789297658862876,
      "grad_norm": 0.6423237919807434,
      "learning_rate": 4.439441692931112e-05,
      "loss": 2.086,
      "step": 1255
    },
    {
      "epoch": 1.6802675585284281,
      "grad_norm": 0.926794171333313,
      "learning_rate": 4.4349392165691134e-05,
      "loss": 1.7186,
      "step": 1256
    },
    {
      "epoch": 1.6816053511705684,
      "grad_norm": 0.6868398189544678,
      "learning_rate": 4.430436740207114e-05,
      "loss": 1.7386,
      "step": 1257
    },
    {
      "epoch": 1.6829431438127092,
      "grad_norm": 1.3709686994552612,
      "learning_rate": 4.425934263845115e-05,
      "loss": 1.5422,
      "step": 1258
    },
    {
      "epoch": 1.6842809364548494,
      "grad_norm": 1.1822702884674072,
      "learning_rate": 4.421431787483116e-05,
      "loss": 1.6106,
      "step": 1259
    },
    {
      "epoch": 1.68561872909699,
      "grad_norm": 0.7609877586364746,
      "learning_rate": 4.416929311121117e-05,
      "loss": 0.7967,
      "step": 1260
    },
    {
      "epoch": 1.6869565217391305,
      "grad_norm": 0.9301615357398987,
      "learning_rate": 4.412426834759118e-05,
      "loss": 1.7581,
      "step": 1261
    },
    {
      "epoch": 1.6882943143812708,
      "grad_norm": 0.9290110468864441,
      "learning_rate": 4.4079243583971186e-05,
      "loss": 1.6569,
      "step": 1262
    },
    {
      "epoch": 1.6896321070234115,
      "grad_norm": 0.904699981212616,
      "learning_rate": 4.4034218820351194e-05,
      "loss": 1.3858,
      "step": 1263
    },
    {
      "epoch": 1.6909698996655518,
      "grad_norm": 0.8743433356285095,
      "learning_rate": 4.39891940567312e-05,
      "loss": 0.8902,
      "step": 1264
    },
    {
      "epoch": 1.6923076923076923,
      "grad_norm": 0.5912821292877197,
      "learning_rate": 4.394416929311121e-05,
      "loss": 1.6238,
      "step": 1265
    },
    {
      "epoch": 1.6936454849498328,
      "grad_norm": 2.2498979568481445,
      "learning_rate": 4.389914452949122e-05,
      "loss": 1.8626,
      "step": 1266
    },
    {
      "epoch": 1.694983277591973,
      "grad_norm": 0.8445337414741516,
      "learning_rate": 4.385411976587123e-05,
      "loss": 1.8312,
      "step": 1267
    },
    {
      "epoch": 1.6963210702341138,
      "grad_norm": 0.9713656902313232,
      "learning_rate": 4.380909500225124e-05,
      "loss": 1.3857,
      "step": 1268
    },
    {
      "epoch": 1.6976588628762541,
      "grad_norm": 0.6458280086517334,
      "learning_rate": 4.376407023863125e-05,
      "loss": 1.2622,
      "step": 1269
    },
    {
      "epoch": 1.6989966555183946,
      "grad_norm": 0.8908253908157349,
      "learning_rate": 4.371904547501126e-05,
      "loss": 1.798,
      "step": 1270
    },
    {
      "epoch": 1.7003344481605351,
      "grad_norm": 0.9305428862571716,
      "learning_rate": 4.367402071139127e-05,
      "loss": 1.7508,
      "step": 1271
    },
    {
      "epoch": 1.7016722408026754,
      "grad_norm": 0.8213643431663513,
      "learning_rate": 4.362899594777128e-05,
      "loss": 1.3151,
      "step": 1272
    },
    {
      "epoch": 1.7030100334448162,
      "grad_norm": 1.0373735427856445,
      "learning_rate": 4.358397118415129e-05,
      "loss": 1.4885,
      "step": 1273
    },
    {
      "epoch": 1.7043478260869565,
      "grad_norm": 0.676937460899353,
      "learning_rate": 4.3538946420531296e-05,
      "loss": 1.963,
      "step": 1274
    },
    {
      "epoch": 1.705685618729097,
      "grad_norm": 0.9667220115661621,
      "learning_rate": 4.3493921656911304e-05,
      "loss": 1.647,
      "step": 1275
    },
    {
      "epoch": 1.7070234113712375,
      "grad_norm": 1.1586287021636963,
      "learning_rate": 4.344889689329131e-05,
      "loss": 1.6821,
      "step": 1276
    },
    {
      "epoch": 1.7083612040133778,
      "grad_norm": 1.7140305042266846,
      "learning_rate": 4.340387212967132e-05,
      "loss": 1.6735,
      "step": 1277
    },
    {
      "epoch": 1.7096989966555185,
      "grad_norm": 1.2710798978805542,
      "learning_rate": 4.335884736605133e-05,
      "loss": 1.3605,
      "step": 1278
    },
    {
      "epoch": 1.7110367892976588,
      "grad_norm": 1.4276046752929688,
      "learning_rate": 4.331382260243134e-05,
      "loss": 1.2628,
      "step": 1279
    },
    {
      "epoch": 1.7123745819397993,
      "grad_norm": 0.8968702554702759,
      "learning_rate": 4.326879783881135e-05,
      "loss": 2.0376,
      "step": 1280
    },
    {
      "epoch": 1.7137123745819398,
      "grad_norm": 0.9779416918754578,
      "learning_rate": 4.322377307519136e-05,
      "loss": 1.9223,
      "step": 1281
    },
    {
      "epoch": 1.71505016722408,
      "grad_norm": 1.2591009140014648,
      "learning_rate": 4.3178748311571364e-05,
      "loss": 1.3221,
      "step": 1282
    },
    {
      "epoch": 1.7163879598662208,
      "grad_norm": 1.216707468032837,
      "learning_rate": 4.313372354795138e-05,
      "loss": 1.0194,
      "step": 1283
    },
    {
      "epoch": 1.7177257525083611,
      "grad_norm": 0.8018258213996887,
      "learning_rate": 4.308869878433138e-05,
      "loss": 2.0731,
      "step": 1284
    },
    {
      "epoch": 1.7190635451505016,
      "grad_norm": 0.8288760781288147,
      "learning_rate": 4.30436740207114e-05,
      "loss": 1.9664,
      "step": 1285
    },
    {
      "epoch": 1.7204013377926421,
      "grad_norm": 2.0245273113250732,
      "learning_rate": 4.29986492570914e-05,
      "loss": 1.4066,
      "step": 1286
    },
    {
      "epoch": 1.7217391304347827,
      "grad_norm": 1.0415517091751099,
      "learning_rate": 4.2953624493471414e-05,
      "loss": 1.6966,
      "step": 1287
    },
    {
      "epoch": 1.7230769230769232,
      "grad_norm": 1.4432035684585571,
      "learning_rate": 4.290859972985142e-05,
      "loss": 0.825,
      "step": 1288
    },
    {
      "epoch": 1.7244147157190635,
      "grad_norm": 0.7454544305801392,
      "learning_rate": 4.286357496623143e-05,
      "loss": 1.8722,
      "step": 1289
    },
    {
      "epoch": 1.725752508361204,
      "grad_norm": 1.177752137184143,
      "learning_rate": 4.281855020261144e-05,
      "loss": 1.4524,
      "step": 1290
    },
    {
      "epoch": 1.7270903010033445,
      "grad_norm": 0.9258737564086914,
      "learning_rate": 4.277352543899145e-05,
      "loss": 1.7513,
      "step": 1291
    },
    {
      "epoch": 1.728428093645485,
      "grad_norm": 0.832943856716156,
      "learning_rate": 4.272850067537146e-05,
      "loss": 1.5639,
      "step": 1292
    },
    {
      "epoch": 1.7297658862876255,
      "grad_norm": 1.6282193660736084,
      "learning_rate": 4.2683475911751466e-05,
      "loss": 1.3246,
      "step": 1293
    },
    {
      "epoch": 1.7311036789297658,
      "grad_norm": 0.7570889592170715,
      "learning_rate": 4.2638451148131474e-05,
      "loss": 1.3322,
      "step": 1294
    },
    {
      "epoch": 1.7324414715719063,
      "grad_norm": 0.8353405594825745,
      "learning_rate": 4.259342638451148e-05,
      "loss": 1.2958,
      "step": 1295
    },
    {
      "epoch": 1.7337792642140468,
      "grad_norm": 1.2555522918701172,
      "learning_rate": 4.254840162089149e-05,
      "loss": 1.4349,
      "step": 1296
    },
    {
      "epoch": 1.7351170568561873,
      "grad_norm": 1.1173676252365112,
      "learning_rate": 4.250337685727151e-05,
      "loss": 1.4082,
      "step": 1297
    },
    {
      "epoch": 1.7364548494983278,
      "grad_norm": 0.9005771279335022,
      "learning_rate": 4.245835209365151e-05,
      "loss": 1.9892,
      "step": 1298
    },
    {
      "epoch": 1.7377926421404681,
      "grad_norm": 0.9720417261123657,
      "learning_rate": 4.2413327330031524e-05,
      "loss": 1.5311,
      "step": 1299
    },
    {
      "epoch": 1.7391304347826086,
      "grad_norm": 0.9608173370361328,
      "learning_rate": 4.2368302566411526e-05,
      "loss": 1.4644,
      "step": 1300
    },
    {
      "epoch": 1.7404682274247492,
      "grad_norm": 1.2789133787155151,
      "learning_rate": 4.232327780279154e-05,
      "loss": 2.0233,
      "step": 1301
    },
    {
      "epoch": 1.7418060200668897,
      "grad_norm": 0.9710900783538818,
      "learning_rate": 4.227825303917154e-05,
      "loss": 1.1806,
      "step": 1302
    },
    {
      "epoch": 1.7431438127090302,
      "grad_norm": 0.8562449812889099,
      "learning_rate": 4.223322827555156e-05,
      "loss": 1.6512,
      "step": 1303
    },
    {
      "epoch": 1.7444816053511705,
      "grad_norm": 0.9922163486480713,
      "learning_rate": 4.218820351193156e-05,
      "loss": 1.7093,
      "step": 1304
    },
    {
      "epoch": 1.745819397993311,
      "grad_norm": 0.8094335794448853,
      "learning_rate": 4.2143178748311576e-05,
      "loss": 1.6508,
      "step": 1305
    },
    {
      "epoch": 1.7471571906354515,
      "grad_norm": 0.8714074492454529,
      "learning_rate": 4.2098153984691584e-05,
      "loss": 1.6098,
      "step": 1306
    },
    {
      "epoch": 1.748494983277592,
      "grad_norm": 1.1597920656204224,
      "learning_rate": 4.205312922107159e-05,
      "loss": 1.2336,
      "step": 1307
    },
    {
      "epoch": 1.7498327759197325,
      "grad_norm": 1.0653190612792969,
      "learning_rate": 4.20081044574516e-05,
      "loss": 1.1729,
      "step": 1308
    },
    {
      "epoch": 1.7511705685618728,
      "grad_norm": 0.941969096660614,
      "learning_rate": 4.196307969383161e-05,
      "loss": 1.808,
      "step": 1309
    },
    {
      "epoch": 1.7525083612040135,
      "grad_norm": 1.3117737770080566,
      "learning_rate": 4.191805493021162e-05,
      "loss": 1.0482,
      "step": 1310
    },
    {
      "epoch": 1.7538461538461538,
      "grad_norm": 0.8123752474784851,
      "learning_rate": 4.187303016659163e-05,
      "loss": 1.8081,
      "step": 1311
    },
    {
      "epoch": 1.7551839464882943,
      "grad_norm": 1.3279463052749634,
      "learning_rate": 4.1828005402971636e-05,
      "loss": 1.431,
      "step": 1312
    },
    {
      "epoch": 1.7565217391304349,
      "grad_norm": 1.044995665550232,
      "learning_rate": 4.1782980639351644e-05,
      "loss": 1.5174,
      "step": 1313
    },
    {
      "epoch": 1.7578595317725751,
      "grad_norm": 0.918773889541626,
      "learning_rate": 4.173795587573165e-05,
      "loss": 1.1729,
      "step": 1314
    },
    {
      "epoch": 1.7591973244147159,
      "grad_norm": 1.465911626815796,
      "learning_rate": 4.169293111211167e-05,
      "loss": 1.4036,
      "step": 1315
    },
    {
      "epoch": 1.7605351170568562,
      "grad_norm": 0.9047989845275879,
      "learning_rate": 4.164790634849167e-05,
      "loss": 1.0355,
      "step": 1316
    },
    {
      "epoch": 1.7618729096989967,
      "grad_norm": 1.1229565143585205,
      "learning_rate": 4.1602881584871685e-05,
      "loss": 1.3043,
      "step": 1317
    },
    {
      "epoch": 1.7632107023411372,
      "grad_norm": 0.7313392162322998,
      "learning_rate": 4.155785682125169e-05,
      "loss": 1.5846,
      "step": 1318
    },
    {
      "epoch": 1.7645484949832775,
      "grad_norm": 0.8520455360412598,
      "learning_rate": 4.15128320576317e-05,
      "loss": 1.4211,
      "step": 1319
    },
    {
      "epoch": 1.7658862876254182,
      "grad_norm": 0.8551596999168396,
      "learning_rate": 4.1467807294011704e-05,
      "loss": 1.6073,
      "step": 1320
    },
    {
      "epoch": 1.7672240802675585,
      "grad_norm": 0.9196562767028809,
      "learning_rate": 4.142278253039172e-05,
      "loss": 1.5378,
      "step": 1321
    },
    {
      "epoch": 1.768561872909699,
      "grad_norm": 1.1131138801574707,
      "learning_rate": 4.137775776677172e-05,
      "loss": 1.0312,
      "step": 1322
    },
    {
      "epoch": 1.7698996655518395,
      "grad_norm": 0.8004671335220337,
      "learning_rate": 4.133273300315174e-05,
      "loss": 2.0511,
      "step": 1323
    },
    {
      "epoch": 1.7712374581939798,
      "grad_norm": 1.1453050374984741,
      "learning_rate": 4.1287708239531745e-05,
      "loss": 0.7885,
      "step": 1324
    },
    {
      "epoch": 1.7725752508361206,
      "grad_norm": 1.051133394241333,
      "learning_rate": 4.1242683475911754e-05,
      "loss": 1.4642,
      "step": 1325
    },
    {
      "epoch": 1.7739130434782608,
      "grad_norm": 0.8506733775138855,
      "learning_rate": 4.119765871229176e-05,
      "loss": 1.396,
      "step": 1326
    },
    {
      "epoch": 1.7752508361204014,
      "grad_norm": 0.800424337387085,
      "learning_rate": 4.115263394867177e-05,
      "loss": 1.5075,
      "step": 1327
    },
    {
      "epoch": 1.7765886287625419,
      "grad_norm": 1.6729928255081177,
      "learning_rate": 4.110760918505178e-05,
      "loss": 0.978,
      "step": 1328
    },
    {
      "epoch": 1.7779264214046822,
      "grad_norm": 0.8187434077262878,
      "learning_rate": 4.106258442143179e-05,
      "loss": 1.6308,
      "step": 1329
    },
    {
      "epoch": 1.779264214046823,
      "grad_norm": 1.0891520977020264,
      "learning_rate": 4.10175596578118e-05,
      "loss": 1.4337,
      "step": 1330
    },
    {
      "epoch": 1.7806020066889632,
      "grad_norm": 0.9008198976516724,
      "learning_rate": 4.0972534894191806e-05,
      "loss": 1.476,
      "step": 1331
    },
    {
      "epoch": 1.7819397993311037,
      "grad_norm": 1.036777377128601,
      "learning_rate": 4.0927510130571814e-05,
      "loss": 1.4582,
      "step": 1332
    },
    {
      "epoch": 1.7832775919732442,
      "grad_norm": 1.3103868961334229,
      "learning_rate": 4.088248536695183e-05,
      "loss": 1.6545,
      "step": 1333
    },
    {
      "epoch": 1.7846153846153845,
      "grad_norm": 1.3254574537277222,
      "learning_rate": 4.083746060333183e-05,
      "loss": 1.5745,
      "step": 1334
    },
    {
      "epoch": 1.7859531772575252,
      "grad_norm": 0.771115243434906,
      "learning_rate": 4.079243583971185e-05,
      "loss": 1.9747,
      "step": 1335
    },
    {
      "epoch": 1.7872909698996655,
      "grad_norm": 0.7478201985359192,
      "learning_rate": 4.074741107609185e-05,
      "loss": 1.819,
      "step": 1336
    },
    {
      "epoch": 1.788628762541806,
      "grad_norm": 0.6987447142601013,
      "learning_rate": 4.0702386312471864e-05,
      "loss": 1.9189,
      "step": 1337
    },
    {
      "epoch": 1.7899665551839465,
      "grad_norm": 0.6925516724586487,
      "learning_rate": 4.0657361548851866e-05,
      "loss": 1.8319,
      "step": 1338
    },
    {
      "epoch": 1.7913043478260868,
      "grad_norm": 0.7896952033042908,
      "learning_rate": 4.061233678523188e-05,
      "loss": 0.9144,
      "step": 1339
    },
    {
      "epoch": 1.7926421404682276,
      "grad_norm": 0.847746729850769,
      "learning_rate": 4.056731202161189e-05,
      "loss": 1.4009,
      "step": 1340
    },
    {
      "epoch": 1.7939799331103679,
      "grad_norm": 0.8629260063171387,
      "learning_rate": 4.05222872579919e-05,
      "loss": 1.3139,
      "step": 1341
    },
    {
      "epoch": 1.7953177257525084,
      "grad_norm": 0.894623875617981,
      "learning_rate": 4.047726249437191e-05,
      "loss": 1.2251,
      "step": 1342
    },
    {
      "epoch": 1.7966555183946489,
      "grad_norm": 1.4640523195266724,
      "learning_rate": 4.0432237730751915e-05,
      "loss": 1.5554,
      "step": 1343
    },
    {
      "epoch": 1.7979933110367892,
      "grad_norm": 0.8280948996543884,
      "learning_rate": 4.0387212967131924e-05,
      "loss": 1.7077,
      "step": 1344
    },
    {
      "epoch": 1.79933110367893,
      "grad_norm": 1.387689471244812,
      "learning_rate": 4.034218820351193e-05,
      "loss": 1.4598,
      "step": 1345
    },
    {
      "epoch": 1.8006688963210702,
      "grad_norm": 0.7571278214454651,
      "learning_rate": 4.029716343989194e-05,
      "loss": 1.2455,
      "step": 1346
    },
    {
      "epoch": 1.8020066889632107,
      "grad_norm": 0.9558732509613037,
      "learning_rate": 4.025213867627195e-05,
      "loss": 0.9238,
      "step": 1347
    },
    {
      "epoch": 1.8033444816053512,
      "grad_norm": 1.0369471311569214,
      "learning_rate": 4.020711391265196e-05,
      "loss": 1.3457,
      "step": 1348
    },
    {
      "epoch": 1.8046822742474915,
      "grad_norm": 0.852948009967804,
      "learning_rate": 4.016208914903197e-05,
      "loss": 1.3754,
      "step": 1349
    },
    {
      "epoch": 1.8060200668896322,
      "grad_norm": 0.912213921546936,
      "learning_rate": 4.0117064385411976e-05,
      "loss": 1.8092,
      "step": 1350
    },
    {
      "epoch": 1.8073578595317725,
      "grad_norm": 0.8572192192077637,
      "learning_rate": 4.007203962179199e-05,
      "loss": 1.7593,
      "step": 1351
    },
    {
      "epoch": 1.808695652173913,
      "grad_norm": 1.1593003273010254,
      "learning_rate": 4.0027014858172e-05,
      "loss": 1.6094,
      "step": 1352
    },
    {
      "epoch": 1.8100334448160535,
      "grad_norm": 0.9357568621635437,
      "learning_rate": 3.998199009455201e-05,
      "loss": 0.9808,
      "step": 1353
    },
    {
      "epoch": 1.8113712374581938,
      "grad_norm": 0.8933264017105103,
      "learning_rate": 3.993696533093202e-05,
      "loss": 1.5225,
      "step": 1354
    },
    {
      "epoch": 1.8127090301003346,
      "grad_norm": 0.9033715724945068,
      "learning_rate": 3.9891940567312025e-05,
      "loss": 1.6487,
      "step": 1355
    },
    {
      "epoch": 1.8140468227424749,
      "grad_norm": 0.9341977834701538,
      "learning_rate": 3.9846915803692034e-05,
      "loss": 1.4483,
      "step": 1356
    },
    {
      "epoch": 1.8153846153846154,
      "grad_norm": 0.7608655095100403,
      "learning_rate": 3.980189104007204e-05,
      "loss": 2.0032,
      "step": 1357
    },
    {
      "epoch": 1.8167224080267559,
      "grad_norm": 1.3053033351898193,
      "learning_rate": 3.975686627645205e-05,
      "loss": 1.7181,
      "step": 1358
    },
    {
      "epoch": 1.8180602006688962,
      "grad_norm": 0.9982928037643433,
      "learning_rate": 3.971184151283206e-05,
      "loss": 1.5147,
      "step": 1359
    },
    {
      "epoch": 1.819397993311037,
      "grad_norm": 1.1024192571640015,
      "learning_rate": 3.966681674921207e-05,
      "loss": 1.561,
      "step": 1360
    },
    {
      "epoch": 1.8207357859531772,
      "grad_norm": 0.8846253752708435,
      "learning_rate": 3.962179198559208e-05,
      "loss": 1.4367,
      "step": 1361
    },
    {
      "epoch": 1.8220735785953177,
      "grad_norm": 0.8675643801689148,
      "learning_rate": 3.9576767221972085e-05,
      "loss": 1.8881,
      "step": 1362
    },
    {
      "epoch": 1.8234113712374582,
      "grad_norm": 0.8549279570579529,
      "learning_rate": 3.9531742458352094e-05,
      "loss": 1.6155,
      "step": 1363
    },
    {
      "epoch": 1.8247491638795985,
      "grad_norm": 0.8130283355712891,
      "learning_rate": 3.948671769473211e-05,
      "loss": 1.7252,
      "step": 1364
    },
    {
      "epoch": 1.8260869565217392,
      "grad_norm": 9.944559097290039,
      "learning_rate": 3.944169293111211e-05,
      "loss": 1.2356,
      "step": 1365
    },
    {
      "epoch": 1.8274247491638795,
      "grad_norm": 1.0192309617996216,
      "learning_rate": 3.9396668167492126e-05,
      "loss": 1.6147,
      "step": 1366
    },
    {
      "epoch": 1.82876254180602,
      "grad_norm": 1.1403286457061768,
      "learning_rate": 3.935164340387213e-05,
      "loss": 1.2417,
      "step": 1367
    },
    {
      "epoch": 1.8301003344481606,
      "grad_norm": 1.0655380487442017,
      "learning_rate": 3.9306618640252144e-05,
      "loss": 1.4588,
      "step": 1368
    },
    {
      "epoch": 1.8314381270903008,
      "grad_norm": 1.3363555669784546,
      "learning_rate": 3.926159387663215e-05,
      "loss": 1.5176,
      "step": 1369
    },
    {
      "epoch": 1.8327759197324416,
      "grad_norm": 0.7391565442085266,
      "learning_rate": 3.921656911301216e-05,
      "loss": 2.0288,
      "step": 1370
    },
    {
      "epoch": 1.8341137123745819,
      "grad_norm": 1.527122974395752,
      "learning_rate": 3.917154434939217e-05,
      "loss": 1.439,
      "step": 1371
    },
    {
      "epoch": 1.8354515050167224,
      "grad_norm": 1.5331617593765259,
      "learning_rate": 3.912651958577218e-05,
      "loss": 1.1387,
      "step": 1372
    },
    {
      "epoch": 1.836789297658863,
      "grad_norm": 0.7895675897598267,
      "learning_rate": 3.9081494822152187e-05,
      "loss": 1.7776,
      "step": 1373
    },
    {
      "epoch": 1.8381270903010032,
      "grad_norm": 1.14493727684021,
      "learning_rate": 3.9036470058532195e-05,
      "loss": 1.7798,
      "step": 1374
    },
    {
      "epoch": 1.839464882943144,
      "grad_norm": 1.087314248085022,
      "learning_rate": 3.8991445294912204e-05,
      "loss": 1.3838,
      "step": 1375
    },
    {
      "epoch": 1.8408026755852842,
      "grad_norm": 0.98976069688797,
      "learning_rate": 3.894642053129221e-05,
      "loss": 1.7391,
      "step": 1376
    },
    {
      "epoch": 1.8421404682274247,
      "grad_norm": 0.7153158783912659,
      "learning_rate": 3.890139576767222e-05,
      "loss": 2.0336,
      "step": 1377
    },
    {
      "epoch": 1.8434782608695652,
      "grad_norm": 1.0000951290130615,
      "learning_rate": 3.8856371004052236e-05,
      "loss": 1.6006,
      "step": 1378
    },
    {
      "epoch": 1.8448160535117055,
      "grad_norm": 0.8586902618408203,
      "learning_rate": 3.881134624043224e-05,
      "loss": 1.284,
      "step": 1379
    },
    {
      "epoch": 1.8461538461538463,
      "grad_norm": 0.8292785882949829,
      "learning_rate": 3.8766321476812253e-05,
      "loss": 1.5184,
      "step": 1380
    },
    {
      "epoch": 1.8474916387959865,
      "grad_norm": 0.7205365300178528,
      "learning_rate": 3.8721296713192255e-05,
      "loss": 1.8033,
      "step": 1381
    },
    {
      "epoch": 1.848829431438127,
      "grad_norm": 0.7597335577011108,
      "learning_rate": 3.867627194957227e-05,
      "loss": 1.4228,
      "step": 1382
    },
    {
      "epoch": 1.8501672240802676,
      "grad_norm": 0.9705151915550232,
      "learning_rate": 3.863124718595227e-05,
      "loss": 1.6322,
      "step": 1383
    },
    {
      "epoch": 1.851505016722408,
      "grad_norm": 0.9061914682388306,
      "learning_rate": 3.858622242233229e-05,
      "loss": 1.861,
      "step": 1384
    },
    {
      "epoch": 1.8528428093645486,
      "grad_norm": 0.9491883516311646,
      "learning_rate": 3.854119765871229e-05,
      "loss": 1.8632,
      "step": 1385
    },
    {
      "epoch": 1.8541806020066889,
      "grad_norm": 0.6613609790802002,
      "learning_rate": 3.8496172895092305e-05,
      "loss": 1.3203,
      "step": 1386
    },
    {
      "epoch": 1.8555183946488294,
      "grad_norm": 1.661139726638794,
      "learning_rate": 3.8451148131472314e-05,
      "loss": 0.9495,
      "step": 1387
    },
    {
      "epoch": 1.85685618729097,
      "grad_norm": 0.946655809879303,
      "learning_rate": 3.840612336785232e-05,
      "loss": 1.7735,
      "step": 1388
    },
    {
      "epoch": 1.8581939799331104,
      "grad_norm": 0.8011321425437927,
      "learning_rate": 3.836109860423233e-05,
      "loss": 1.0474,
      "step": 1389
    },
    {
      "epoch": 1.859531772575251,
      "grad_norm": 0.8001610040664673,
      "learning_rate": 3.831607384061234e-05,
      "loss": 1.6964,
      "step": 1390
    },
    {
      "epoch": 1.8608695652173912,
      "grad_norm": 1.0349432229995728,
      "learning_rate": 3.827104907699235e-05,
      "loss": 1.4787,
      "step": 1391
    },
    {
      "epoch": 1.8622073578595317,
      "grad_norm": 0.804012656211853,
      "learning_rate": 3.8226024313372357e-05,
      "loss": 1.9486,
      "step": 1392
    },
    {
      "epoch": 1.8635451505016722,
      "grad_norm": 0.936322033405304,
      "learning_rate": 3.8180999549752365e-05,
      "loss": 1.4181,
      "step": 1393
    },
    {
      "epoch": 1.8648829431438128,
      "grad_norm": 1.3157624006271362,
      "learning_rate": 3.8135974786132374e-05,
      "loss": 1.4829,
      "step": 1394
    },
    {
      "epoch": 1.8662207357859533,
      "grad_norm": 0.8876410722732544,
      "learning_rate": 3.809095002251238e-05,
      "loss": 1.6884,
      "step": 1395
    },
    {
      "epoch": 1.8675585284280936,
      "grad_norm": 0.6261321902275085,
      "learning_rate": 3.80459252588924e-05,
      "loss": 2.0704,
      "step": 1396
    },
    {
      "epoch": 1.868896321070234,
      "grad_norm": 1.0813729763031006,
      "learning_rate": 3.80009004952724e-05,
      "loss": 1.3557,
      "step": 1397
    },
    {
      "epoch": 1.8702341137123746,
      "grad_norm": 0.9512030482292175,
      "learning_rate": 3.7955875731652415e-05,
      "loss": 1.3012,
      "step": 1398
    },
    {
      "epoch": 1.871571906354515,
      "grad_norm": 0.8181292414665222,
      "learning_rate": 3.791085096803242e-05,
      "loss": 1.6619,
      "step": 1399
    },
    {
      "epoch": 1.8729096989966556,
      "grad_norm": 0.8774016499519348,
      "learning_rate": 3.786582620441243e-05,
      "loss": 1.2919,
      "step": 1400
    },
    {
      "epoch": 1.874247491638796,
      "grad_norm": 0.7443768978118896,
      "learning_rate": 3.7820801440792434e-05,
      "loss": 1.4092,
      "step": 1401
    },
    {
      "epoch": 1.8755852842809364,
      "grad_norm": 0.7041268348693848,
      "learning_rate": 3.777577667717245e-05,
      "loss": 1.0577,
      "step": 1402
    },
    {
      "epoch": 1.876923076923077,
      "grad_norm": 1.1805241107940674,
      "learning_rate": 3.773075191355245e-05,
      "loss": 1.4778,
      "step": 1403
    },
    {
      "epoch": 1.8782608695652174,
      "grad_norm": 1.1446001529693604,
      "learning_rate": 3.7685727149932466e-05,
      "loss": 1.3754,
      "step": 1404
    },
    {
      "epoch": 1.879598662207358,
      "grad_norm": 0.7583310604095459,
      "learning_rate": 3.7640702386312475e-05,
      "loss": 1.847,
      "step": 1405
    },
    {
      "epoch": 1.8809364548494982,
      "grad_norm": 1.094193696975708,
      "learning_rate": 3.7595677622692484e-05,
      "loss": 1.1759,
      "step": 1406
    },
    {
      "epoch": 1.8822742474916387,
      "grad_norm": 0.9978253841400146,
      "learning_rate": 3.755065285907249e-05,
      "loss": 1.2676,
      "step": 1407
    },
    {
      "epoch": 1.8836120401337793,
      "grad_norm": 0.7539755702018738,
      "learning_rate": 3.75056280954525e-05,
      "loss": 1.8526,
      "step": 1408
    },
    {
      "epoch": 1.8849498327759198,
      "grad_norm": 0.8578140735626221,
      "learning_rate": 3.746060333183251e-05,
      "loss": 1.4808,
      "step": 1409
    },
    {
      "epoch": 1.8862876254180603,
      "grad_norm": 0.8148081302642822,
      "learning_rate": 3.741557856821252e-05,
      "loss": 1.7241,
      "step": 1410
    },
    {
      "epoch": 1.8876254180602006,
      "grad_norm": 1.0469331741333008,
      "learning_rate": 3.7370553804592526e-05,
      "loss": 1.5271,
      "step": 1411
    },
    {
      "epoch": 1.8889632107023413,
      "grad_norm": 0.8617160320281982,
      "learning_rate": 3.7325529040972535e-05,
      "loss": 0.9646,
      "step": 1412
    },
    {
      "epoch": 1.8903010033444816,
      "grad_norm": 0.9113197326660156,
      "learning_rate": 3.7280504277352544e-05,
      "loss": 1.7127,
      "step": 1413
    },
    {
      "epoch": 1.891638795986622,
      "grad_norm": 0.8339391946792603,
      "learning_rate": 3.723547951373256e-05,
      "loss": 1.2338,
      "step": 1414
    },
    {
      "epoch": 1.8929765886287626,
      "grad_norm": 0.8415791988372803,
      "learning_rate": 3.719045475011256e-05,
      "loss": 1.7855,
      "step": 1415
    },
    {
      "epoch": 1.894314381270903,
      "grad_norm": 1.0455056428909302,
      "learning_rate": 3.7145429986492576e-05,
      "loss": 1.5371,
      "step": 1416
    },
    {
      "epoch": 1.8956521739130436,
      "grad_norm": 0.9208756685256958,
      "learning_rate": 3.710040522287258e-05,
      "loss": 1.5899,
      "step": 1417
    },
    {
      "epoch": 1.896989966555184,
      "grad_norm": 0.8601641654968262,
      "learning_rate": 3.705538045925259e-05,
      "loss": 1.5976,
      "step": 1418
    },
    {
      "epoch": 1.8983277591973244,
      "grad_norm": 1.0065354108810425,
      "learning_rate": 3.7010355695632595e-05,
      "loss": 1.9271,
      "step": 1419
    },
    {
      "epoch": 1.899665551839465,
      "grad_norm": 1.224617838859558,
      "learning_rate": 3.696533093201261e-05,
      "loss": 1.1049,
      "step": 1420
    },
    {
      "epoch": 1.9010033444816052,
      "grad_norm": 1.2477177381515503,
      "learning_rate": 3.692030616839261e-05,
      "loss": 1.1976,
      "step": 1421
    },
    {
      "epoch": 1.902341137123746,
      "grad_norm": 0.8188490867614746,
      "learning_rate": 3.687528140477263e-05,
      "loss": 1.2126,
      "step": 1422
    },
    {
      "epoch": 1.9036789297658863,
      "grad_norm": 0.927735447883606,
      "learning_rate": 3.6830256641152636e-05,
      "loss": 1.6352,
      "step": 1423
    },
    {
      "epoch": 1.9050167224080268,
      "grad_norm": 0.7117443084716797,
      "learning_rate": 3.6785231877532645e-05,
      "loss": 1.1576,
      "step": 1424
    },
    {
      "epoch": 1.9063545150501673,
      "grad_norm": 0.9709138870239258,
      "learning_rate": 3.6740207113912653e-05,
      "loss": 1.7208,
      "step": 1425
    },
    {
      "epoch": 1.9076923076923076,
      "grad_norm": 0.9803403615951538,
      "learning_rate": 3.669518235029266e-05,
      "loss": 1.6992,
      "step": 1426
    },
    {
      "epoch": 1.9090301003344483,
      "grad_norm": 0.9272592067718506,
      "learning_rate": 3.665015758667267e-05,
      "loss": 1.7691,
      "step": 1427
    },
    {
      "epoch": 1.9103678929765886,
      "grad_norm": 0.8970061540603638,
      "learning_rate": 3.660513282305268e-05,
      "loss": 1.6872,
      "step": 1428
    },
    {
      "epoch": 1.9117056856187291,
      "grad_norm": 1.3868622779846191,
      "learning_rate": 3.656010805943269e-05,
      "loss": 1.9104,
      "step": 1429
    },
    {
      "epoch": 1.9130434782608696,
      "grad_norm": 0.9223452806472778,
      "learning_rate": 3.6515083295812696e-05,
      "loss": 1.5378,
      "step": 1430
    },
    {
      "epoch": 1.91438127090301,
      "grad_norm": 0.8893844485282898,
      "learning_rate": 3.6470058532192705e-05,
      "loss": 1.5726,
      "step": 1431
    },
    {
      "epoch": 1.9157190635451506,
      "grad_norm": 0.8713217377662659,
      "learning_rate": 3.642503376857272e-05,
      "loss": 1.3582,
      "step": 1432
    },
    {
      "epoch": 1.917056856187291,
      "grad_norm": 0.9675815105438232,
      "learning_rate": 3.638000900495272e-05,
      "loss": 1.521,
      "step": 1433
    },
    {
      "epoch": 1.9183946488294314,
      "grad_norm": 1.0128233432769775,
      "learning_rate": 3.633498424133274e-05,
      "loss": 1.9584,
      "step": 1434
    },
    {
      "epoch": 1.919732441471572,
      "grad_norm": 0.7074108719825745,
      "learning_rate": 3.628995947771274e-05,
      "loss": 1.9104,
      "step": 1435
    },
    {
      "epoch": 1.9210702341137122,
      "grad_norm": 0.6000598669052124,
      "learning_rate": 3.6244934714092755e-05,
      "loss": 2.0013,
      "step": 1436
    },
    {
      "epoch": 1.922408026755853,
      "grad_norm": 0.8514284491539001,
      "learning_rate": 3.619990995047276e-05,
      "loss": 1.716,
      "step": 1437
    },
    {
      "epoch": 1.9237458193979933,
      "grad_norm": 0.6201356649398804,
      "learning_rate": 3.615488518685277e-05,
      "loss": 1.4419,
      "step": 1438
    },
    {
      "epoch": 1.9250836120401338,
      "grad_norm": 0.9943525791168213,
      "learning_rate": 3.610986042323278e-05,
      "loss": 1.9337,
      "step": 1439
    },
    {
      "epoch": 1.9264214046822743,
      "grad_norm": 0.9183377623558044,
      "learning_rate": 3.606483565961279e-05,
      "loss": 1.5447,
      "step": 1440
    },
    {
      "epoch": 1.9277591973244146,
      "grad_norm": 0.8715210556983948,
      "learning_rate": 3.60198108959928e-05,
      "loss": 1.7298,
      "step": 1441
    },
    {
      "epoch": 1.9290969899665553,
      "grad_norm": 1.0677322149276733,
      "learning_rate": 3.5974786132372806e-05,
      "loss": 1.1434,
      "step": 1442
    },
    {
      "epoch": 1.9304347826086956,
      "grad_norm": 0.8093985915184021,
      "learning_rate": 3.5929761368752815e-05,
      "loss": 1.9209,
      "step": 1443
    },
    {
      "epoch": 1.9317725752508361,
      "grad_norm": 1.3754236698150635,
      "learning_rate": 3.5884736605132823e-05,
      "loss": 1.7104,
      "step": 1444
    },
    {
      "epoch": 1.9331103678929766,
      "grad_norm": 0.8817523121833801,
      "learning_rate": 3.583971184151283e-05,
      "loss": 1.5823,
      "step": 1445
    },
    {
      "epoch": 1.934448160535117,
      "grad_norm": 1.1452676057815552,
      "learning_rate": 3.579468707789284e-05,
      "loss": 1.2673,
      "step": 1446
    },
    {
      "epoch": 1.9357859531772577,
      "grad_norm": 1.0037072896957397,
      "learning_rate": 3.5749662314272856e-05,
      "loss": 1.8337,
      "step": 1447
    },
    {
      "epoch": 1.937123745819398,
      "grad_norm": 1.1124719381332397,
      "learning_rate": 3.570463755065286e-05,
      "loss": 1.5884,
      "step": 1448
    },
    {
      "epoch": 1.9384615384615385,
      "grad_norm": 0.6976829767227173,
      "learning_rate": 3.565961278703287e-05,
      "loss": 2.3163,
      "step": 1449
    },
    {
      "epoch": 1.939799331103679,
      "grad_norm": 0.9209871888160706,
      "learning_rate": 3.561458802341288e-05,
      "loss": 1.5843,
      "step": 1450
    },
    {
      "epoch": 1.9411371237458193,
      "grad_norm": 1.010508418083191,
      "learning_rate": 3.556956325979289e-05,
      "loss": 1.458,
      "step": 1451
    },
    {
      "epoch": 1.94247491638796,
      "grad_norm": 1.0236648321151733,
      "learning_rate": 3.55245384961729e-05,
      "loss": 1.6633,
      "step": 1452
    },
    {
      "epoch": 1.9438127090301003,
      "grad_norm": 0.8461986184120178,
      "learning_rate": 3.547951373255291e-05,
      "loss": 1.7553,
      "step": 1453
    },
    {
      "epoch": 1.9451505016722408,
      "grad_norm": 1.7666321992874146,
      "learning_rate": 3.5434488968932916e-05,
      "loss": 1.8603,
      "step": 1454
    },
    {
      "epoch": 1.9464882943143813,
      "grad_norm": 1.007495641708374,
      "learning_rate": 3.5389464205312925e-05,
      "loss": 1.4298,
      "step": 1455
    },
    {
      "epoch": 1.9478260869565216,
      "grad_norm": 1.0924739837646484,
      "learning_rate": 3.534443944169293e-05,
      "loss": 1.2259,
      "step": 1456
    },
    {
      "epoch": 1.9491638795986623,
      "grad_norm": 1.3044383525848389,
      "learning_rate": 3.529941467807294e-05,
      "loss": 1.0427,
      "step": 1457
    },
    {
      "epoch": 1.9505016722408026,
      "grad_norm": 1.1054308414459229,
      "learning_rate": 3.525438991445295e-05,
      "loss": 1.2018,
      "step": 1458
    },
    {
      "epoch": 1.9518394648829431,
      "grad_norm": 0.857029378414154,
      "learning_rate": 3.5209365150832966e-05,
      "loss": 1.6903,
      "step": 1459
    },
    {
      "epoch": 1.9531772575250836,
      "grad_norm": 0.8971465826034546,
      "learning_rate": 3.516434038721297e-05,
      "loss": 1.8465,
      "step": 1460
    },
    {
      "epoch": 1.954515050167224,
      "grad_norm": 0.7438530921936035,
      "learning_rate": 3.511931562359298e-05,
      "loss": 2.0821,
      "step": 1461
    },
    {
      "epoch": 1.9558528428093647,
      "grad_norm": 0.9445087909698486,
      "learning_rate": 3.5074290859972985e-05,
      "loss": 1.186,
      "step": 1462
    },
    {
      "epoch": 1.957190635451505,
      "grad_norm": 0.8798640370368958,
      "learning_rate": 3.5029266096353e-05,
      "loss": 1.2806,
      "step": 1463
    },
    {
      "epoch": 1.9585284280936455,
      "grad_norm": 0.7574049830436707,
      "learning_rate": 3.4984241332733e-05,
      "loss": 1.9215,
      "step": 1464
    },
    {
      "epoch": 1.959866220735786,
      "grad_norm": 0.7600085139274597,
      "learning_rate": 3.493921656911302e-05,
      "loss": 1.547,
      "step": 1465
    },
    {
      "epoch": 1.9612040133779263,
      "grad_norm": 1.3530699014663696,
      "learning_rate": 3.489419180549302e-05,
      "loss": 1.3317,
      "step": 1466
    },
    {
      "epoch": 1.962541806020067,
      "grad_norm": 0.9421098232269287,
      "learning_rate": 3.4849167041873035e-05,
      "loss": 1.496,
      "step": 1467
    },
    {
      "epoch": 1.9638795986622073,
      "grad_norm": 0.9440025091171265,
      "learning_rate": 3.480414227825304e-05,
      "loss": 1.418,
      "step": 1468
    },
    {
      "epoch": 1.9652173913043478,
      "grad_norm": 0.8065316081047058,
      "learning_rate": 3.475911751463305e-05,
      "loss": 1.5454,
      "step": 1469
    },
    {
      "epoch": 1.9665551839464883,
      "grad_norm": 1.2114293575286865,
      "learning_rate": 3.471409275101306e-05,
      "loss": 1.214,
      "step": 1470
    },
    {
      "epoch": 1.9678929765886286,
      "grad_norm": 0.8474501371383667,
      "learning_rate": 3.466906798739307e-05,
      "loss": 1.4953,
      "step": 1471
    },
    {
      "epoch": 1.9692307692307693,
      "grad_norm": 1.0701537132263184,
      "learning_rate": 3.462404322377308e-05,
      "loss": 1.488,
      "step": 1472
    },
    {
      "epoch": 1.9705685618729096,
      "grad_norm": 0.6921608448028564,
      "learning_rate": 3.4579018460153086e-05,
      "loss": 1.8654,
      "step": 1473
    },
    {
      "epoch": 1.9719063545150501,
      "grad_norm": 1.0466644763946533,
      "learning_rate": 3.4533993696533095e-05,
      "loss": 1.7775,
      "step": 1474
    },
    {
      "epoch": 1.9732441471571907,
      "grad_norm": 0.8933880925178528,
      "learning_rate": 3.44889689329131e-05,
      "loss": 1.5537,
      "step": 1475
    },
    {
      "epoch": 1.974581939799331,
      "grad_norm": 0.8331461548805237,
      "learning_rate": 3.444394416929311e-05,
      "loss": 1.5999,
      "step": 1476
    },
    {
      "epoch": 1.9759197324414717,
      "grad_norm": 1.0497558116912842,
      "learning_rate": 3.439891940567313e-05,
      "loss": 1.2683,
      "step": 1477
    },
    {
      "epoch": 1.977257525083612,
      "grad_norm": 0.812122642993927,
      "learning_rate": 3.435389464205313e-05,
      "loss": 1.3971,
      "step": 1478
    },
    {
      "epoch": 1.9785953177257525,
      "grad_norm": 0.7708330154418945,
      "learning_rate": 3.4308869878433144e-05,
      "loss": 1.4712,
      "step": 1479
    },
    {
      "epoch": 1.979933110367893,
      "grad_norm": 0.8662110567092896,
      "learning_rate": 3.4263845114813146e-05,
      "loss": 1.7005,
      "step": 1480
    },
    {
      "epoch": 1.9812709030100333,
      "grad_norm": 0.7595421671867371,
      "learning_rate": 3.421882035119316e-05,
      "loss": 1.2517,
      "step": 1481
    },
    {
      "epoch": 1.982608695652174,
      "grad_norm": 0.6963253021240234,
      "learning_rate": 3.417379558757316e-05,
      "loss": 1.6449,
      "step": 1482
    },
    {
      "epoch": 1.9839464882943143,
      "grad_norm": 1.4446080923080444,
      "learning_rate": 3.412877082395318e-05,
      "loss": 1.8439,
      "step": 1483
    },
    {
      "epoch": 1.9852842809364548,
      "grad_norm": 0.791908323764801,
      "learning_rate": 3.408374606033318e-05,
      "loss": 1.7487,
      "step": 1484
    },
    {
      "epoch": 1.9866220735785953,
      "grad_norm": 0.8825159072875977,
      "learning_rate": 3.4038721296713196e-05,
      "loss": 1.7007,
      "step": 1485
    },
    {
      "epoch": 1.9879598662207358,
      "grad_norm": 1.163138747215271,
      "learning_rate": 3.3993696533093204e-05,
      "loss": 1.7588,
      "step": 1486
    },
    {
      "epoch": 1.9892976588628764,
      "grad_norm": 0.8407654166221619,
      "learning_rate": 3.394867176947321e-05,
      "loss": 1.7077,
      "step": 1487
    },
    {
      "epoch": 1.9906354515050166,
      "grad_norm": 1.0220328569412231,
      "learning_rate": 3.390364700585322e-05,
      "loss": 1.804,
      "step": 1488
    },
    {
      "epoch": 1.9919732441471572,
      "grad_norm": 1.0959333181381226,
      "learning_rate": 3.385862224223323e-05,
      "loss": 1.3809,
      "step": 1489
    },
    {
      "epoch": 1.9933110367892977,
      "grad_norm": 1.8417096138000488,
      "learning_rate": 3.381359747861324e-05,
      "loss": 1.5266,
      "step": 1490
    },
    {
      "epoch": 1.9946488294314382,
      "grad_norm": 0.7798132300376892,
      "learning_rate": 3.376857271499325e-05,
      "loss": 1.6998,
      "step": 1491
    },
    {
      "epoch": 1.9959866220735787,
      "grad_norm": 0.9236243963241577,
      "learning_rate": 3.3723547951373256e-05,
      "loss": 1.7454,
      "step": 1492
    },
    {
      "epoch": 1.997324414715719,
      "grad_norm": 0.82822185754776,
      "learning_rate": 3.3678523187753265e-05,
      "loss": 1.6078,
      "step": 1493
    },
    {
      "epoch": 1.9986622073578595,
      "grad_norm": 1.1015570163726807,
      "learning_rate": 3.363349842413327e-05,
      "loss": 1.2679,
      "step": 1494
    },
    {
      "epoch": 2.0,
      "grad_norm": 0.8461845517158508,
      "learning_rate": 3.358847366051329e-05,
      "loss": 1.6352,
      "step": 1495
    },
    {
      "epoch": 2.0013377926421403,
      "grad_norm": 1.279486894607544,
      "learning_rate": 3.354344889689329e-05,
      "loss": 1.1376,
      "step": 1496
    },
    {
      "epoch": 2.002675585284281,
      "grad_norm": 0.776619017124176,
      "learning_rate": 3.3498424133273306e-05,
      "loss": 0.4116,
      "step": 1497
    },
    {
      "epoch": 2.0040133779264213,
      "grad_norm": 1.0830862522125244,
      "learning_rate": 3.345339936965331e-05,
      "loss": 0.6109,
      "step": 1498
    },
    {
      "epoch": 2.005351170568562,
      "grad_norm": 1.01801598072052,
      "learning_rate": 3.340837460603332e-05,
      "loss": 0.8726,
      "step": 1499
    },
    {
      "epoch": 2.0066889632107023,
      "grad_norm": 1.0562922954559326,
      "learning_rate": 3.3363349842413325e-05,
      "loss": 1.1031,
      "step": 1500
    },
    {
      "epoch": 2.0080267558528426,
      "grad_norm": 2.5661964416503906,
      "learning_rate": 3.331832507879334e-05,
      "loss": 0.888,
      "step": 1501
    },
    {
      "epoch": 2.0093645484949834,
      "grad_norm": 0.9264085292816162,
      "learning_rate": 3.327330031517334e-05,
      "loss": 0.6023,
      "step": 1502
    },
    {
      "epoch": 2.0107023411371236,
      "grad_norm": 1.4456504583358765,
      "learning_rate": 3.322827555155336e-05,
      "loss": 0.7641,
      "step": 1503
    },
    {
      "epoch": 2.0120401337792644,
      "grad_norm": 1.4428790807724,
      "learning_rate": 3.3183250787933366e-05,
      "loss": 1.1088,
      "step": 1504
    },
    {
      "epoch": 2.0133779264214047,
      "grad_norm": 1.8238784074783325,
      "learning_rate": 3.3138226024313374e-05,
      "loss": 1.26,
      "step": 1505
    },
    {
      "epoch": 2.014715719063545,
      "grad_norm": 1.0849965810775757,
      "learning_rate": 3.309320126069338e-05,
      "loss": 1.3907,
      "step": 1506
    },
    {
      "epoch": 2.0160535117056857,
      "grad_norm": 1.50315260887146,
      "learning_rate": 3.304817649707339e-05,
      "loss": 0.9065,
      "step": 1507
    },
    {
      "epoch": 2.017391304347826,
      "grad_norm": 1.1935383081436157,
      "learning_rate": 3.30031517334534e-05,
      "loss": 1.0666,
      "step": 1508
    },
    {
      "epoch": 2.0187290969899667,
      "grad_norm": 1.2051082849502563,
      "learning_rate": 3.295812696983341e-05,
      "loss": 1.0776,
      "step": 1509
    },
    {
      "epoch": 2.020066889632107,
      "grad_norm": 0.8955470323562622,
      "learning_rate": 3.291310220621342e-05,
      "loss": 1.1224,
      "step": 1510
    },
    {
      "epoch": 2.0214046822742473,
      "grad_norm": 1.3287490606307983,
      "learning_rate": 3.2868077442593426e-05,
      "loss": 0.9316,
      "step": 1511
    },
    {
      "epoch": 2.022742474916388,
      "grad_norm": 0.9810749292373657,
      "learning_rate": 3.2823052678973435e-05,
      "loss": 1.0389,
      "step": 1512
    },
    {
      "epoch": 2.0240802675585283,
      "grad_norm": 1.1453120708465576,
      "learning_rate": 3.277802791535345e-05,
      "loss": 0.862,
      "step": 1513
    },
    {
      "epoch": 2.025418060200669,
      "grad_norm": 1.0180597305297852,
      "learning_rate": 3.273300315173345e-05,
      "loss": 0.9974,
      "step": 1514
    },
    {
      "epoch": 2.0267558528428093,
      "grad_norm": 0.8884549140930176,
      "learning_rate": 3.268797838811347e-05,
      "loss": 1.3196,
      "step": 1515
    },
    {
      "epoch": 2.0280936454849496,
      "grad_norm": 0.8313940167427063,
      "learning_rate": 3.264295362449347e-05,
      "loss": 1.6277,
      "step": 1516
    },
    {
      "epoch": 2.0294314381270904,
      "grad_norm": 1.1740089654922485,
      "learning_rate": 3.2597928860873484e-05,
      "loss": 1.0187,
      "step": 1517
    },
    {
      "epoch": 2.0307692307692307,
      "grad_norm": 1.3176401853561401,
      "learning_rate": 3.2552904097253486e-05,
      "loss": 1.2444,
      "step": 1518
    },
    {
      "epoch": 2.0321070234113714,
      "grad_norm": 0.8785193562507629,
      "learning_rate": 3.25078793336335e-05,
      "loss": 1.7614,
      "step": 1519
    },
    {
      "epoch": 2.0334448160535117,
      "grad_norm": 1.3594410419464111,
      "learning_rate": 3.246285457001351e-05,
      "loss": 1.0677,
      "step": 1520
    },
    {
      "epoch": 2.034782608695652,
      "grad_norm": 1.3269301652908325,
      "learning_rate": 3.241782980639352e-05,
      "loss": 0.4644,
      "step": 1521
    },
    {
      "epoch": 2.0361204013377927,
      "grad_norm": 1.262554407119751,
      "learning_rate": 3.237280504277353e-05,
      "loss": 1.0349,
      "step": 1522
    },
    {
      "epoch": 2.037458193979933,
      "grad_norm": 1.3310198783874512,
      "learning_rate": 3.2327780279153536e-05,
      "loss": 0.6012,
      "step": 1523
    },
    {
      "epoch": 2.0387959866220737,
      "grad_norm": 1.1281541585922241,
      "learning_rate": 3.2282755515533544e-05,
      "loss": 1.2607,
      "step": 1524
    },
    {
      "epoch": 2.040133779264214,
      "grad_norm": 1.0073496103286743,
      "learning_rate": 3.223773075191355e-05,
      "loss": 1.2157,
      "step": 1525
    },
    {
      "epoch": 2.0414715719063543,
      "grad_norm": 1.030488133430481,
      "learning_rate": 3.219270598829356e-05,
      "loss": 0.808,
      "step": 1526
    },
    {
      "epoch": 2.042809364548495,
      "grad_norm": 1.070237159729004,
      "learning_rate": 3.214768122467357e-05,
      "loss": 1.4373,
      "step": 1527
    },
    {
      "epoch": 2.0441471571906353,
      "grad_norm": 1.4718862771987915,
      "learning_rate": 3.210265646105358e-05,
      "loss": 1.0276,
      "step": 1528
    },
    {
      "epoch": 2.045484949832776,
      "grad_norm": 1.0908019542694092,
      "learning_rate": 3.205763169743359e-05,
      "loss": 0.3747,
      "step": 1529
    },
    {
      "epoch": 2.0468227424749164,
      "grad_norm": 1.0104992389678955,
      "learning_rate": 3.2012606933813596e-05,
      "loss": 1.295,
      "step": 1530
    },
    {
      "epoch": 2.0481605351170566,
      "grad_norm": 1.1235511302947998,
      "learning_rate": 3.196758217019361e-05,
      "loss": 1.3199,
      "step": 1531
    },
    {
      "epoch": 2.0494983277591974,
      "grad_norm": 0.7247384190559387,
      "learning_rate": 3.192255740657362e-05,
      "loss": 1.1936,
      "step": 1532
    },
    {
      "epoch": 2.0508361204013377,
      "grad_norm": 1.0165960788726807,
      "learning_rate": 3.187753264295363e-05,
      "loss": 0.6051,
      "step": 1533
    },
    {
      "epoch": 2.0521739130434784,
      "grad_norm": 1.3204349279403687,
      "learning_rate": 3.183250787933364e-05,
      "loss": 0.7344,
      "step": 1534
    },
    {
      "epoch": 2.0535117056856187,
      "grad_norm": 1.2122710943222046,
      "learning_rate": 3.1787483115713646e-05,
      "loss": 0.898,
      "step": 1535
    },
    {
      "epoch": 2.054849498327759,
      "grad_norm": 1.2342621088027954,
      "learning_rate": 3.1742458352093654e-05,
      "loss": 1.1225,
      "step": 1536
    },
    {
      "epoch": 2.0561872909698997,
      "grad_norm": 0.9262746572494507,
      "learning_rate": 3.169743358847366e-05,
      "loss": 1.527,
      "step": 1537
    },
    {
      "epoch": 2.05752508361204,
      "grad_norm": 1.1269612312316895,
      "learning_rate": 3.165240882485367e-05,
      "loss": 1.2711,
      "step": 1538
    },
    {
      "epoch": 2.0588628762541807,
      "grad_norm": 0.9769323468208313,
      "learning_rate": 3.160738406123368e-05,
      "loss": 1.3137,
      "step": 1539
    },
    {
      "epoch": 2.060200668896321,
      "grad_norm": 1.2974296808242798,
      "learning_rate": 3.156235929761369e-05,
      "loss": 0.9984,
      "step": 1540
    },
    {
      "epoch": 2.0615384615384613,
      "grad_norm": 0.9156148433685303,
      "learning_rate": 3.15173345339937e-05,
      "loss": 1.0303,
      "step": 1541
    },
    {
      "epoch": 2.062876254180602,
      "grad_norm": 1.0938788652420044,
      "learning_rate": 3.1472309770373706e-05,
      "loss": 0.6634,
      "step": 1542
    },
    {
      "epoch": 2.0642140468227423,
      "grad_norm": 1.010788917541504,
      "learning_rate": 3.1427285006753714e-05,
      "loss": 0.662,
      "step": 1543
    },
    {
      "epoch": 2.065551839464883,
      "grad_norm": 1.0869221687316895,
      "learning_rate": 3.138226024313373e-05,
      "loss": 1.3342,
      "step": 1544
    },
    {
      "epoch": 2.0668896321070234,
      "grad_norm": 1.017852783203125,
      "learning_rate": 3.133723547951373e-05,
      "loss": 0.608,
      "step": 1545
    },
    {
      "epoch": 2.068227424749164,
      "grad_norm": 1.0046050548553467,
      "learning_rate": 3.129221071589375e-05,
      "loss": 1.1453,
      "step": 1546
    },
    {
      "epoch": 2.0695652173913044,
      "grad_norm": 1.4137753248214722,
      "learning_rate": 3.124718595227375e-05,
      "loss": 0.8363,
      "step": 1547
    },
    {
      "epoch": 2.0709030100334447,
      "grad_norm": 0.852415919303894,
      "learning_rate": 3.1202161188653764e-05,
      "loss": 1.4471,
      "step": 1548
    },
    {
      "epoch": 2.0722408026755854,
      "grad_norm": 1.1125855445861816,
      "learning_rate": 3.115713642503377e-05,
      "loss": 0.9525,
      "step": 1549
    },
    {
      "epoch": 2.0735785953177257,
      "grad_norm": 1.400482177734375,
      "learning_rate": 3.111211166141378e-05,
      "loss": 1.3223,
      "step": 1550
    },
    {
      "epoch": 2.074916387959866,
      "grad_norm": 1.349930763244629,
      "learning_rate": 3.106708689779379e-05,
      "loss": 1.1897,
      "step": 1551
    },
    {
      "epoch": 2.0762541806020067,
      "grad_norm": 1.397952914237976,
      "learning_rate": 3.10220621341738e-05,
      "loss": 0.7269,
      "step": 1552
    },
    {
      "epoch": 2.077591973244147,
      "grad_norm": 1.566179871559143,
      "learning_rate": 3.097703737055381e-05,
      "loss": 0.5265,
      "step": 1553
    },
    {
      "epoch": 2.0789297658862878,
      "grad_norm": 1.1740292310714722,
      "learning_rate": 3.0932012606933816e-05,
      "loss": 0.574,
      "step": 1554
    },
    {
      "epoch": 2.080267558528428,
      "grad_norm": 1.0403320789337158,
      "learning_rate": 3.0886987843313824e-05,
      "loss": 0.9537,
      "step": 1555
    },
    {
      "epoch": 2.0816053511705688,
      "grad_norm": 0.9628156423568726,
      "learning_rate": 3.084196307969383e-05,
      "loss": 0.6196,
      "step": 1556
    },
    {
      "epoch": 2.082943143812709,
      "grad_norm": 1.2764846086502075,
      "learning_rate": 3.079693831607384e-05,
      "loss": 0.9719,
      "step": 1557
    },
    {
      "epoch": 2.0842809364548494,
      "grad_norm": 1.7418354749679565,
      "learning_rate": 3.075191355245386e-05,
      "loss": 0.2701,
      "step": 1558
    },
    {
      "epoch": 2.08561872909699,
      "grad_norm": 1.3813650608062744,
      "learning_rate": 3.070688878883386e-05,
      "loss": 1.0376,
      "step": 1559
    },
    {
      "epoch": 2.0869565217391304,
      "grad_norm": 1.4286768436431885,
      "learning_rate": 3.0661864025213874e-05,
      "loss": 0.4777,
      "step": 1560
    },
    {
      "epoch": 2.088294314381271,
      "grad_norm": 0.9821745157241821,
      "learning_rate": 3.0616839261593876e-05,
      "loss": 1.1567,
      "step": 1561
    },
    {
      "epoch": 2.0896321070234114,
      "grad_norm": 1.05975341796875,
      "learning_rate": 3.057181449797389e-05,
      "loss": 0.9883,
      "step": 1562
    },
    {
      "epoch": 2.0909698996655517,
      "grad_norm": 1.042338490486145,
      "learning_rate": 3.052678973435389e-05,
      "loss": 0.8155,
      "step": 1563
    },
    {
      "epoch": 2.0923076923076924,
      "grad_norm": 0.7823744416236877,
      "learning_rate": 3.0481764970733905e-05,
      "loss": 1.2554,
      "step": 1564
    },
    {
      "epoch": 2.0936454849498327,
      "grad_norm": 1.068821907043457,
      "learning_rate": 3.0436740207113913e-05,
      "loss": 1.263,
      "step": 1565
    },
    {
      "epoch": 2.0949832775919734,
      "grad_norm": 1.022455096244812,
      "learning_rate": 3.0391715443493922e-05,
      "loss": 1.1237,
      "step": 1566
    },
    {
      "epoch": 2.0963210702341137,
      "grad_norm": 1.326230525970459,
      "learning_rate": 3.0346690679873934e-05,
      "loss": 1.033,
      "step": 1567
    },
    {
      "epoch": 2.097658862876254,
      "grad_norm": 0.8274069428443909,
      "learning_rate": 3.030166591625394e-05,
      "loss": 1.544,
      "step": 1568
    },
    {
      "epoch": 2.0989966555183948,
      "grad_norm": 1.4057950973510742,
      "learning_rate": 3.025664115263395e-05,
      "loss": 0.8866,
      "step": 1569
    },
    {
      "epoch": 2.100334448160535,
      "grad_norm": 1.241181492805481,
      "learning_rate": 3.021161638901396e-05,
      "loss": 1.037,
      "step": 1570
    },
    {
      "epoch": 2.101672240802676,
      "grad_norm": 1.1959797143936157,
      "learning_rate": 3.016659162539397e-05,
      "loss": 0.8406,
      "step": 1571
    },
    {
      "epoch": 2.103010033444816,
      "grad_norm": 1.3782483339309692,
      "learning_rate": 3.0121566861773977e-05,
      "loss": 1.0415,
      "step": 1572
    },
    {
      "epoch": 2.1043478260869564,
      "grad_norm": 0.8823965787887573,
      "learning_rate": 3.0076542098153985e-05,
      "loss": 0.612,
      "step": 1573
    },
    {
      "epoch": 2.105685618729097,
      "grad_norm": 1.0144659280776978,
      "learning_rate": 3.0031517334533994e-05,
      "loss": 1.4259,
      "step": 1574
    },
    {
      "epoch": 2.1070234113712374,
      "grad_norm": 1.094332218170166,
      "learning_rate": 2.9986492570914003e-05,
      "loss": 1.2305,
      "step": 1575
    },
    {
      "epoch": 2.108361204013378,
      "grad_norm": 1.0350522994995117,
      "learning_rate": 2.9941467807294015e-05,
      "loss": 0.6495,
      "step": 1576
    },
    {
      "epoch": 2.1096989966555184,
      "grad_norm": 1.2167094945907593,
      "learning_rate": 2.9896443043674023e-05,
      "loss": 0.5618,
      "step": 1577
    },
    {
      "epoch": 2.1110367892976587,
      "grad_norm": 0.9851863384246826,
      "learning_rate": 2.9851418280054032e-05,
      "loss": 1.4429,
      "step": 1578
    },
    {
      "epoch": 2.1123745819397994,
      "grad_norm": 1.2461366653442383,
      "learning_rate": 2.980639351643404e-05,
      "loss": 0.6586,
      "step": 1579
    },
    {
      "epoch": 2.1137123745819397,
      "grad_norm": 1.1684248447418213,
      "learning_rate": 2.976136875281405e-05,
      "loss": 1.109,
      "step": 1580
    },
    {
      "epoch": 2.1150501672240805,
      "grad_norm": 1.1089441776275635,
      "learning_rate": 2.9716343989194058e-05,
      "loss": 1.4378,
      "step": 1581
    },
    {
      "epoch": 2.1163879598662207,
      "grad_norm": 1.082276463508606,
      "learning_rate": 2.967131922557407e-05,
      "loss": 1.0216,
      "step": 1582
    },
    {
      "epoch": 2.117725752508361,
      "grad_norm": 1.1534171104431152,
      "learning_rate": 2.9626294461954075e-05,
      "loss": 0.798,
      "step": 1583
    },
    {
      "epoch": 2.1190635451505018,
      "grad_norm": 1.5649821758270264,
      "learning_rate": 2.9581269698334087e-05,
      "loss": 0.4634,
      "step": 1584
    },
    {
      "epoch": 2.120401337792642,
      "grad_norm": 0.8794542551040649,
      "learning_rate": 2.9536244934714095e-05,
      "loss": 0.8743,
      "step": 1585
    },
    {
      "epoch": 2.121739130434783,
      "grad_norm": 0.8998076319694519,
      "learning_rate": 2.9491220171094104e-05,
      "loss": 1.8453,
      "step": 1586
    },
    {
      "epoch": 2.123076923076923,
      "grad_norm": 0.9526141285896301,
      "learning_rate": 2.9446195407474116e-05,
      "loss": 1.0177,
      "step": 1587
    },
    {
      "epoch": 2.1244147157190634,
      "grad_norm": 1.1450635194778442,
      "learning_rate": 2.940117064385412e-05,
      "loss": 1.2151,
      "step": 1588
    },
    {
      "epoch": 2.125752508361204,
      "grad_norm": 1.4250377416610718,
      "learning_rate": 2.9356145880234133e-05,
      "loss": 0.7914,
      "step": 1589
    },
    {
      "epoch": 2.1270903010033444,
      "grad_norm": 1.0612053871154785,
      "learning_rate": 2.9311121116614138e-05,
      "loss": 1.3046,
      "step": 1590
    },
    {
      "epoch": 2.128428093645485,
      "grad_norm": 1.2039984464645386,
      "learning_rate": 2.926609635299415e-05,
      "loss": 0.5904,
      "step": 1591
    },
    {
      "epoch": 2.1297658862876254,
      "grad_norm": 1.2266685962677002,
      "learning_rate": 2.9221071589374155e-05,
      "loss": 1.2272,
      "step": 1592
    },
    {
      "epoch": 2.1311036789297657,
      "grad_norm": 1.5131837129592896,
      "learning_rate": 2.9176046825754167e-05,
      "loss": 1.1346,
      "step": 1593
    },
    {
      "epoch": 2.1324414715719064,
      "grad_norm": 1.3815799951553345,
      "learning_rate": 2.913102206213418e-05,
      "loss": 0.5999,
      "step": 1594
    },
    {
      "epoch": 2.1337792642140467,
      "grad_norm": 0.8422326445579529,
      "learning_rate": 2.9085997298514185e-05,
      "loss": 0.8242,
      "step": 1595
    },
    {
      "epoch": 2.1351170568561875,
      "grad_norm": 0.8974445462226868,
      "learning_rate": 2.9040972534894197e-05,
      "loss": 1.3765,
      "step": 1596
    },
    {
      "epoch": 2.1364548494983278,
      "grad_norm": 1.0997154712677002,
      "learning_rate": 2.8995947771274202e-05,
      "loss": 0.8954,
      "step": 1597
    },
    {
      "epoch": 2.137792642140468,
      "grad_norm": 1.0048385858535767,
      "learning_rate": 2.8950923007654214e-05,
      "loss": 1.3485,
      "step": 1598
    },
    {
      "epoch": 2.139130434782609,
      "grad_norm": 0.8595561981201172,
      "learning_rate": 2.890589824403422e-05,
      "loss": 0.7611,
      "step": 1599
    },
    {
      "epoch": 2.140468227424749,
      "grad_norm": 1.1249717473983765,
      "learning_rate": 2.886087348041423e-05,
      "loss": 1.0467,
      "step": 1600
    },
    {
      "epoch": 2.14180602006689,
      "grad_norm": 1.151814341545105,
      "learning_rate": 2.8815848716794236e-05,
      "loss": 1.0023,
      "step": 1601
    },
    {
      "epoch": 2.14314381270903,
      "grad_norm": 0.9440100789070129,
      "learning_rate": 2.8770823953174248e-05,
      "loss": 1.3554,
      "step": 1602
    },
    {
      "epoch": 2.1444816053511704,
      "grad_norm": 1.1240310668945312,
      "learning_rate": 2.872579918955426e-05,
      "loss": 1.126,
      "step": 1603
    },
    {
      "epoch": 2.145819397993311,
      "grad_norm": 1.3048230409622192,
      "learning_rate": 2.8680774425934265e-05,
      "loss": 1.0317,
      "step": 1604
    },
    {
      "epoch": 2.1471571906354514,
      "grad_norm": 1.1779913902282715,
      "learning_rate": 2.8635749662314277e-05,
      "loss": 1.0175,
      "step": 1605
    },
    {
      "epoch": 2.148494983277592,
      "grad_norm": 1.0159926414489746,
      "learning_rate": 2.8590724898694282e-05,
      "loss": 0.9013,
      "step": 1606
    },
    {
      "epoch": 2.1498327759197324,
      "grad_norm": 1.1253591775894165,
      "learning_rate": 2.8545700135074294e-05,
      "loss": 1.0637,
      "step": 1607
    },
    {
      "epoch": 2.1511705685618727,
      "grad_norm": 1.393731713294983,
      "learning_rate": 2.85006753714543e-05,
      "loss": 1.4256,
      "step": 1608
    },
    {
      "epoch": 2.1525083612040135,
      "grad_norm": 1.1206722259521484,
      "learning_rate": 2.845565060783431e-05,
      "loss": 1.1309,
      "step": 1609
    },
    {
      "epoch": 2.1538461538461537,
      "grad_norm": 0.8800863027572632,
      "learning_rate": 2.8410625844214317e-05,
      "loss": 0.8318,
      "step": 1610
    },
    {
      "epoch": 2.1551839464882945,
      "grad_norm": 1.0208646059036255,
      "learning_rate": 2.836560108059433e-05,
      "loss": 1.379,
      "step": 1611
    },
    {
      "epoch": 2.1565217391304348,
      "grad_norm": 1.1892729997634888,
      "learning_rate": 2.832057631697434e-05,
      "loss": 0.773,
      "step": 1612
    },
    {
      "epoch": 2.157859531772575,
      "grad_norm": 1.4659501314163208,
      "learning_rate": 2.8275551553354346e-05,
      "loss": 0.659,
      "step": 1613
    },
    {
      "epoch": 2.159197324414716,
      "grad_norm": 1.2769265174865723,
      "learning_rate": 2.8230526789734358e-05,
      "loss": 0.9489,
      "step": 1614
    },
    {
      "epoch": 2.160535117056856,
      "grad_norm": 1.3630527257919312,
      "learning_rate": 2.8185502026114363e-05,
      "loss": 0.8874,
      "step": 1615
    },
    {
      "epoch": 2.161872909698997,
      "grad_norm": 0.9882266521453857,
      "learning_rate": 2.8140477262494375e-05,
      "loss": 1.2838,
      "step": 1616
    },
    {
      "epoch": 2.163210702341137,
      "grad_norm": 0.9982869625091553,
      "learning_rate": 2.809545249887438e-05,
      "loss": 1.5081,
      "step": 1617
    },
    {
      "epoch": 2.1645484949832774,
      "grad_norm": 1.4063819646835327,
      "learning_rate": 2.8050427735254392e-05,
      "loss": 0.9401,
      "step": 1618
    },
    {
      "epoch": 2.165886287625418,
      "grad_norm": 1.2039049863815308,
      "learning_rate": 2.8005402971634397e-05,
      "loss": 1.1665,
      "step": 1619
    },
    {
      "epoch": 2.1672240802675584,
      "grad_norm": 1.0905070304870605,
      "learning_rate": 2.796037820801441e-05,
      "loss": 0.9254,
      "step": 1620
    },
    {
      "epoch": 2.168561872909699,
      "grad_norm": 1.2284868955612183,
      "learning_rate": 2.791535344439442e-05,
      "loss": 0.6904,
      "step": 1621
    },
    {
      "epoch": 2.1698996655518394,
      "grad_norm": 0.9538512825965881,
      "learning_rate": 2.7870328680774427e-05,
      "loss": 0.8834,
      "step": 1622
    },
    {
      "epoch": 2.1712374581939797,
      "grad_norm": 1.2037982940673828,
      "learning_rate": 2.782530391715444e-05,
      "loss": 1.1294,
      "step": 1623
    },
    {
      "epoch": 2.1725752508361205,
      "grad_norm": 0.953117847442627,
      "learning_rate": 2.7780279153534444e-05,
      "loss": 1.4922,
      "step": 1624
    },
    {
      "epoch": 2.1739130434782608,
      "grad_norm": 0.8379032015800476,
      "learning_rate": 2.7735254389914456e-05,
      "loss": 0.9568,
      "step": 1625
    },
    {
      "epoch": 2.1752508361204015,
      "grad_norm": 1.0359519720077515,
      "learning_rate": 2.769022962629446e-05,
      "loss": 0.4698,
      "step": 1626
    },
    {
      "epoch": 2.1765886287625418,
      "grad_norm": 0.8869040012359619,
      "learning_rate": 2.7645204862674473e-05,
      "loss": 1.0776,
      "step": 1627
    },
    {
      "epoch": 2.177926421404682,
      "grad_norm": 1.293930172920227,
      "learning_rate": 2.7600180099054478e-05,
      "loss": 0.8307,
      "step": 1628
    },
    {
      "epoch": 2.179264214046823,
      "grad_norm": 1.2503477334976196,
      "learning_rate": 2.755515533543449e-05,
      "loss": 1.0004,
      "step": 1629
    },
    {
      "epoch": 2.180602006688963,
      "grad_norm": 1.387004017829895,
      "learning_rate": 2.7510130571814502e-05,
      "loss": 1.2621,
      "step": 1630
    },
    {
      "epoch": 2.181939799331104,
      "grad_norm": 1.4255646467208862,
      "learning_rate": 2.7465105808194507e-05,
      "loss": 0.679,
      "step": 1631
    },
    {
      "epoch": 2.183277591973244,
      "grad_norm": 1.2277337312698364,
      "learning_rate": 2.742008104457452e-05,
      "loss": 0.8446,
      "step": 1632
    },
    {
      "epoch": 2.184615384615385,
      "grad_norm": 1.272117257118225,
      "learning_rate": 2.7375056280954524e-05,
      "loss": 1.309,
      "step": 1633
    },
    {
      "epoch": 2.185953177257525,
      "grad_norm": 0.906496524810791,
      "learning_rate": 2.7330031517334536e-05,
      "loss": 0.3541,
      "step": 1634
    },
    {
      "epoch": 2.1872909698996654,
      "grad_norm": 1.2406556606292725,
      "learning_rate": 2.728500675371454e-05,
      "loss": 0.6467,
      "step": 1635
    },
    {
      "epoch": 2.188628762541806,
      "grad_norm": 1.1095494031906128,
      "learning_rate": 2.7239981990094554e-05,
      "loss": 0.9538,
      "step": 1636
    },
    {
      "epoch": 2.1899665551839465,
      "grad_norm": 1.2391284704208374,
      "learning_rate": 2.719495722647456e-05,
      "loss": 1.0218,
      "step": 1637
    },
    {
      "epoch": 2.1913043478260867,
      "grad_norm": 1.4398272037506104,
      "learning_rate": 2.714993246285457e-05,
      "loss": 0.9383,
      "step": 1638
    },
    {
      "epoch": 2.1926421404682275,
      "grad_norm": 1.0370733737945557,
      "learning_rate": 2.7104907699234583e-05,
      "loss": 0.5855,
      "step": 1639
    },
    {
      "epoch": 2.1939799331103678,
      "grad_norm": 1.026199221611023,
      "learning_rate": 2.7059882935614588e-05,
      "loss": 1.1646,
      "step": 1640
    },
    {
      "epoch": 2.1953177257525085,
      "grad_norm": 0.794501781463623,
      "learning_rate": 2.70148581719946e-05,
      "loss": 1.6246,
      "step": 1641
    },
    {
      "epoch": 2.196655518394649,
      "grad_norm": 1.6995491981506348,
      "learning_rate": 2.6969833408374605e-05,
      "loss": 0.8617,
      "step": 1642
    },
    {
      "epoch": 2.1979933110367895,
      "grad_norm": 0.9656935930252075,
      "learning_rate": 2.6924808644754617e-05,
      "loss": 0.6808,
      "step": 1643
    },
    {
      "epoch": 2.19933110367893,
      "grad_norm": 1.2793020009994507,
      "learning_rate": 2.6879783881134622e-05,
      "loss": 0.7165,
      "step": 1644
    },
    {
      "epoch": 2.20066889632107,
      "grad_norm": 1.0213698148727417,
      "learning_rate": 2.6834759117514634e-05,
      "loss": 0.5796,
      "step": 1645
    },
    {
      "epoch": 2.202006688963211,
      "grad_norm": 0.9900079965591431,
      "learning_rate": 2.678973435389464e-05,
      "loss": 0.3005,
      "step": 1646
    },
    {
      "epoch": 2.203344481605351,
      "grad_norm": 1.314415693283081,
      "learning_rate": 2.674470959027465e-05,
      "loss": 0.8062,
      "step": 1647
    },
    {
      "epoch": 2.2046822742474914,
      "grad_norm": 1.058727502822876,
      "learning_rate": 2.6699684826654663e-05,
      "loss": 1.0138,
      "step": 1648
    },
    {
      "epoch": 2.206020066889632,
      "grad_norm": 1.2562659978866577,
      "learning_rate": 2.665466006303467e-05,
      "loss": 0.5364,
      "step": 1649
    },
    {
      "epoch": 2.2073578595317724,
      "grad_norm": 1.1769195795059204,
      "learning_rate": 2.660963529941468e-05,
      "loss": 0.8234,
      "step": 1650
    },
    {
      "epoch": 2.208695652173913,
      "grad_norm": 1.1072698831558228,
      "learning_rate": 2.6564610535794686e-05,
      "loss": 1.0157,
      "step": 1651
    },
    {
      "epoch": 2.2100334448160535,
      "grad_norm": 1.242130160331726,
      "learning_rate": 2.6519585772174698e-05,
      "loss": 0.772,
      "step": 1652
    },
    {
      "epoch": 2.211371237458194,
      "grad_norm": 1.0669097900390625,
      "learning_rate": 2.6474561008554703e-05,
      "loss": 1.1534,
      "step": 1653
    },
    {
      "epoch": 2.2127090301003345,
      "grad_norm": 1.0712904930114746,
      "learning_rate": 2.6429536244934715e-05,
      "loss": 1.1747,
      "step": 1654
    },
    {
      "epoch": 2.2140468227424748,
      "grad_norm": 1.408160924911499,
      "learning_rate": 2.6384511481314724e-05,
      "loss": 1.3222,
      "step": 1655
    },
    {
      "epoch": 2.2153846153846155,
      "grad_norm": 1.0598413944244385,
      "learning_rate": 2.6339486717694732e-05,
      "loss": 0.8916,
      "step": 1656
    },
    {
      "epoch": 2.216722408026756,
      "grad_norm": 1.279381513595581,
      "learning_rate": 2.6294461954074744e-05,
      "loss": 0.7631,
      "step": 1657
    },
    {
      "epoch": 2.218060200668896,
      "grad_norm": 1.2336814403533936,
      "learning_rate": 2.624943719045475e-05,
      "loss": 1.2253,
      "step": 1658
    },
    {
      "epoch": 2.219397993311037,
      "grad_norm": 1.1152136325836182,
      "learning_rate": 2.620441242683476e-05,
      "loss": 1.0768,
      "step": 1659
    },
    {
      "epoch": 2.220735785953177,
      "grad_norm": 1.0855528116226196,
      "learning_rate": 2.615938766321477e-05,
      "loss": 1.3688,
      "step": 1660
    },
    {
      "epoch": 2.222073578595318,
      "grad_norm": 1.0505130290985107,
      "learning_rate": 2.611436289959478e-05,
      "loss": 1.5493,
      "step": 1661
    },
    {
      "epoch": 2.223411371237458,
      "grad_norm": 1.069624423980713,
      "learning_rate": 2.6069338135974787e-05,
      "loss": 1.0145,
      "step": 1662
    },
    {
      "epoch": 2.224749163879599,
      "grad_norm": 1.2100669145584106,
      "learning_rate": 2.6024313372354796e-05,
      "loss": 0.5034,
      "step": 1663
    },
    {
      "epoch": 2.226086956521739,
      "grad_norm": 0.8809670209884644,
      "learning_rate": 2.5979288608734804e-05,
      "loss": 1.7303,
      "step": 1664
    },
    {
      "epoch": 2.2274247491638794,
      "grad_norm": 1.0502550601959229,
      "learning_rate": 2.5934263845114816e-05,
      "loss": 0.9524,
      "step": 1665
    },
    {
      "epoch": 2.22876254180602,
      "grad_norm": 1.357090711593628,
      "learning_rate": 2.5889239081494825e-05,
      "loss": 0.6862,
      "step": 1666
    },
    {
      "epoch": 2.2301003344481605,
      "grad_norm": 1.0590691566467285,
      "learning_rate": 2.5844214317874833e-05,
      "loss": 1.1155,
      "step": 1667
    },
    {
      "epoch": 2.231438127090301,
      "grad_norm": 1.186330795288086,
      "learning_rate": 2.5799189554254842e-05,
      "loss": 0.6492,
      "step": 1668
    },
    {
      "epoch": 2.2327759197324415,
      "grad_norm": 1.2332799434661865,
      "learning_rate": 2.575416479063485e-05,
      "loss": 1.2474,
      "step": 1669
    },
    {
      "epoch": 2.234113712374582,
      "grad_norm": 1.0606275796890259,
      "learning_rate": 2.570914002701486e-05,
      "loss": 0.8702,
      "step": 1670
    },
    {
      "epoch": 2.2354515050167225,
      "grad_norm": 1.2487508058547974,
      "learning_rate": 2.5664115263394868e-05,
      "loss": 0.7746,
      "step": 1671
    },
    {
      "epoch": 2.236789297658863,
      "grad_norm": 1.1616050004959106,
      "learning_rate": 2.561909049977488e-05,
      "loss": 0.7112,
      "step": 1672
    },
    {
      "epoch": 2.2381270903010035,
      "grad_norm": 1.1081942319869995,
      "learning_rate": 2.5574065736154885e-05,
      "loss": 1.0841,
      "step": 1673
    },
    {
      "epoch": 2.239464882943144,
      "grad_norm": 1.123223900794983,
      "learning_rate": 2.5529040972534897e-05,
      "loss": 0.9884,
      "step": 1674
    },
    {
      "epoch": 2.240802675585284,
      "grad_norm": 1.2441766262054443,
      "learning_rate": 2.5484016208914905e-05,
      "loss": 0.9708,
      "step": 1675
    },
    {
      "epoch": 2.242140468227425,
      "grad_norm": 1.051093578338623,
      "learning_rate": 2.5438991445294914e-05,
      "loss": 0.7997,
      "step": 1676
    },
    {
      "epoch": 2.243478260869565,
      "grad_norm": 1.2590465545654297,
      "learning_rate": 2.5393966681674926e-05,
      "loss": 1.4614,
      "step": 1677
    },
    {
      "epoch": 2.244816053511706,
      "grad_norm": 0.8667411804199219,
      "learning_rate": 2.534894191805493e-05,
      "loss": 1.6479,
      "step": 1678
    },
    {
      "epoch": 2.246153846153846,
      "grad_norm": 1.396817922592163,
      "learning_rate": 2.5303917154434943e-05,
      "loss": 0.7776,
      "step": 1679
    },
    {
      "epoch": 2.2474916387959865,
      "grad_norm": 1.1650742292404175,
      "learning_rate": 2.525889239081495e-05,
      "loss": 0.6929,
      "step": 1680
    },
    {
      "epoch": 2.248829431438127,
      "grad_norm": 0.9426375031471252,
      "learning_rate": 2.521386762719496e-05,
      "loss": 0.9831,
      "step": 1681
    },
    {
      "epoch": 2.2501672240802675,
      "grad_norm": 0.9585577845573425,
      "learning_rate": 2.5168842863574966e-05,
      "loss": 1.1836,
      "step": 1682
    },
    {
      "epoch": 2.251505016722408,
      "grad_norm": 1.1615204811096191,
      "learning_rate": 2.5123818099954978e-05,
      "loss": 0.4986,
      "step": 1683
    },
    {
      "epoch": 2.2528428093645485,
      "grad_norm": 1.2810487747192383,
      "learning_rate": 2.507879333633499e-05,
      "loss": 0.6468,
      "step": 1684
    },
    {
      "epoch": 2.254180602006689,
      "grad_norm": 1.1308361291885376,
      "learning_rate": 2.5033768572714995e-05,
      "loss": 0.7336,
      "step": 1685
    },
    {
      "epoch": 2.2555183946488295,
      "grad_norm": 0.7869702577590942,
      "learning_rate": 2.4988743809095003e-05,
      "loss": 1.0697,
      "step": 1686
    },
    {
      "epoch": 2.25685618729097,
      "grad_norm": 1.1998578310012817,
      "learning_rate": 2.4943719045475012e-05,
      "loss": 1.2017,
      "step": 1687
    },
    {
      "epoch": 2.2581939799331106,
      "grad_norm": 1.0956956148147583,
      "learning_rate": 2.4898694281855024e-05,
      "loss": 0.3821,
      "step": 1688
    },
    {
      "epoch": 2.259531772575251,
      "grad_norm": 1.1351637840270996,
      "learning_rate": 2.4853669518235032e-05,
      "loss": 1.1934,
      "step": 1689
    },
    {
      "epoch": 2.260869565217391,
      "grad_norm": 0.9958216547966003,
      "learning_rate": 2.480864475461504e-05,
      "loss": 1.2529,
      "step": 1690
    },
    {
      "epoch": 2.262207357859532,
      "grad_norm": 1.2681056261062622,
      "learning_rate": 2.476361999099505e-05,
      "loss": 1.0012,
      "step": 1691
    },
    {
      "epoch": 2.263545150501672,
      "grad_norm": 0.9700149893760681,
      "learning_rate": 2.4718595227375058e-05,
      "loss": 0.473,
      "step": 1692
    },
    {
      "epoch": 2.264882943143813,
      "grad_norm": 1.3021841049194336,
      "learning_rate": 2.4673570463755067e-05,
      "loss": 1.1692,
      "step": 1693
    },
    {
      "epoch": 2.266220735785953,
      "grad_norm": 1.036901831626892,
      "learning_rate": 2.4628545700135075e-05,
      "loss": 1.1165,
      "step": 1694
    },
    {
      "epoch": 2.2675585284280935,
      "grad_norm": 1.0550732612609863,
      "learning_rate": 2.4583520936515084e-05,
      "loss": 0.7615,
      "step": 1695
    },
    {
      "epoch": 2.268896321070234,
      "grad_norm": 1.2030292749404907,
      "learning_rate": 2.4538496172895093e-05,
      "loss": 1.1545,
      "step": 1696
    },
    {
      "epoch": 2.2702341137123745,
      "grad_norm": 1.2884608507156372,
      "learning_rate": 2.4493471409275105e-05,
      "loss": 1.0744,
      "step": 1697
    },
    {
      "epoch": 2.2715719063545152,
      "grad_norm": 0.9818195700645447,
      "learning_rate": 2.4448446645655113e-05,
      "loss": 0.9492,
      "step": 1698
    },
    {
      "epoch": 2.2729096989966555,
      "grad_norm": 1.0629665851593018,
      "learning_rate": 2.4403421882035122e-05,
      "loss": 0.8669,
      "step": 1699
    },
    {
      "epoch": 2.274247491638796,
      "grad_norm": 1.0125162601470947,
      "learning_rate": 2.435839711841513e-05,
      "loss": 0.9677,
      "step": 1700
    },
    {
      "epoch": 2.2755852842809365,
      "grad_norm": 1.0470705032348633,
      "learning_rate": 2.431337235479514e-05,
      "loss": 1.0364,
      "step": 1701
    },
    {
      "epoch": 2.276923076923077,
      "grad_norm": 1.2241395711898804,
      "learning_rate": 2.4268347591175148e-05,
      "loss": 0.906,
      "step": 1702
    },
    {
      "epoch": 2.2782608695652176,
      "grad_norm": 1.0276167392730713,
      "learning_rate": 2.4223322827555156e-05,
      "loss": 0.8643,
      "step": 1703
    },
    {
      "epoch": 2.279598662207358,
      "grad_norm": 1.1805189847946167,
      "learning_rate": 2.4178298063935165e-05,
      "loss": 0.5515,
      "step": 1704
    },
    {
      "epoch": 2.280936454849498,
      "grad_norm": 1.0454705953598022,
      "learning_rate": 2.4133273300315173e-05,
      "loss": 1.3343,
      "step": 1705
    },
    {
      "epoch": 2.282274247491639,
      "grad_norm": 0.7291643619537354,
      "learning_rate": 2.4088248536695185e-05,
      "loss": 1.178,
      "step": 1706
    },
    {
      "epoch": 2.283612040133779,
      "grad_norm": 1.2851074934005737,
      "learning_rate": 2.4043223773075194e-05,
      "loss": 1.1367,
      "step": 1707
    },
    {
      "epoch": 2.28494983277592,
      "grad_norm": 1.2466861009597778,
      "learning_rate": 2.3998199009455202e-05,
      "loss": 0.6178,
      "step": 1708
    },
    {
      "epoch": 2.28628762541806,
      "grad_norm": 0.9368792176246643,
      "learning_rate": 2.395317424583521e-05,
      "loss": 0.9587,
      "step": 1709
    },
    {
      "epoch": 2.2876254180602005,
      "grad_norm": 1.3019531965255737,
      "learning_rate": 2.390814948221522e-05,
      "loss": 0.7194,
      "step": 1710
    },
    {
      "epoch": 2.288963210702341,
      "grad_norm": 1.5246587991714478,
      "learning_rate": 2.3863124718595228e-05,
      "loss": 0.6984,
      "step": 1711
    },
    {
      "epoch": 2.2903010033444815,
      "grad_norm": 1.6100928783416748,
      "learning_rate": 2.3818099954975237e-05,
      "loss": 0.8573,
      "step": 1712
    },
    {
      "epoch": 2.2916387959866222,
      "grad_norm": 1.0381380319595337,
      "learning_rate": 2.3773075191355245e-05,
      "loss": 0.6743,
      "step": 1713
    },
    {
      "epoch": 2.2929765886287625,
      "grad_norm": 1.2357443571090698,
      "learning_rate": 2.3728050427735254e-05,
      "loss": 1.0916,
      "step": 1714
    },
    {
      "epoch": 2.294314381270903,
      "grad_norm": 0.9141361713409424,
      "learning_rate": 2.3683025664115266e-05,
      "loss": 0.8996,
      "step": 1715
    },
    {
      "epoch": 2.2956521739130435,
      "grad_norm": 1.232163667678833,
      "learning_rate": 2.3638000900495275e-05,
      "loss": 0.6892,
      "step": 1716
    },
    {
      "epoch": 2.296989966555184,
      "grad_norm": 1.240925908088684,
      "learning_rate": 2.3592976136875283e-05,
      "loss": 1.1866,
      "step": 1717
    },
    {
      "epoch": 2.2983277591973246,
      "grad_norm": 1.143701195716858,
      "learning_rate": 2.3547951373255292e-05,
      "loss": 1.1714,
      "step": 1718
    },
    {
      "epoch": 2.299665551839465,
      "grad_norm": 1.2745702266693115,
      "learning_rate": 2.35029266096353e-05,
      "loss": 0.6173,
      "step": 1719
    },
    {
      "epoch": 2.3010033444816056,
      "grad_norm": 1.3332494497299194,
      "learning_rate": 2.345790184601531e-05,
      "loss": 0.6876,
      "step": 1720
    },
    {
      "epoch": 2.302341137123746,
      "grad_norm": 1.3572496175765991,
      "learning_rate": 2.3412877082395317e-05,
      "loss": 0.8975,
      "step": 1721
    },
    {
      "epoch": 2.303678929765886,
      "grad_norm": 1.0085145235061646,
      "learning_rate": 2.3367852318775326e-05,
      "loss": 1.3663,
      "step": 1722
    },
    {
      "epoch": 2.305016722408027,
      "grad_norm": 1.1505640745162964,
      "learning_rate": 2.3322827555155335e-05,
      "loss": 0.8133,
      "step": 1723
    },
    {
      "epoch": 2.306354515050167,
      "grad_norm": 1.2748997211456299,
      "learning_rate": 2.3277802791535347e-05,
      "loss": 0.8408,
      "step": 1724
    },
    {
      "epoch": 2.3076923076923075,
      "grad_norm": 0.9610079526901245,
      "learning_rate": 2.3232778027915355e-05,
      "loss": 0.6906,
      "step": 1725
    },
    {
      "epoch": 2.309030100334448,
      "grad_norm": 1.1950584650039673,
      "learning_rate": 2.3187753264295364e-05,
      "loss": 0.9894,
      "step": 1726
    },
    {
      "epoch": 2.3103678929765885,
      "grad_norm": 1.2355742454528809,
      "learning_rate": 2.3142728500675372e-05,
      "loss": 0.7915,
      "step": 1727
    },
    {
      "epoch": 2.3117056856187292,
      "grad_norm": 1.3379088640213013,
      "learning_rate": 2.309770373705538e-05,
      "loss": 0.9516,
      "step": 1728
    },
    {
      "epoch": 2.3130434782608695,
      "grad_norm": 1.3441375494003296,
      "learning_rate": 2.305267897343539e-05,
      "loss": 0.7547,
      "step": 1729
    },
    {
      "epoch": 2.3143812709030103,
      "grad_norm": 1.0970510244369507,
      "learning_rate": 2.3007654209815398e-05,
      "loss": 1.0687,
      "step": 1730
    },
    {
      "epoch": 2.3157190635451506,
      "grad_norm": 1.1877330541610718,
      "learning_rate": 2.2962629446195407e-05,
      "loss": 1.5353,
      "step": 1731
    },
    {
      "epoch": 2.317056856187291,
      "grad_norm": 1.1516923904418945,
      "learning_rate": 2.2917604682575415e-05,
      "loss": 0.4395,
      "step": 1732
    },
    {
      "epoch": 2.3183946488294316,
      "grad_norm": 0.7589366436004639,
      "learning_rate": 2.2872579918955427e-05,
      "loss": 0.6784,
      "step": 1733
    },
    {
      "epoch": 2.319732441471572,
      "grad_norm": 1.0079516172409058,
      "learning_rate": 2.2827555155335436e-05,
      "loss": 0.5076,
      "step": 1734
    },
    {
      "epoch": 2.321070234113712,
      "grad_norm": 1.5039020776748657,
      "learning_rate": 2.2782530391715444e-05,
      "loss": 0.9259,
      "step": 1735
    },
    {
      "epoch": 2.322408026755853,
      "grad_norm": 0.8978360295295715,
      "learning_rate": 2.2737505628095453e-05,
      "loss": 1.4465,
      "step": 1736
    },
    {
      "epoch": 2.323745819397993,
      "grad_norm": 1.3363885879516602,
      "learning_rate": 2.269248086447546e-05,
      "loss": 0.8954,
      "step": 1737
    },
    {
      "epoch": 2.325083612040134,
      "grad_norm": 1.2241244316101074,
      "learning_rate": 2.264745610085547e-05,
      "loss": 1.3225,
      "step": 1738
    },
    {
      "epoch": 2.326421404682274,
      "grad_norm": 1.3648251295089722,
      "learning_rate": 2.260243133723548e-05,
      "loss": 1.0267,
      "step": 1739
    },
    {
      "epoch": 2.327759197324415,
      "grad_norm": 1.5363645553588867,
      "learning_rate": 2.2557406573615487e-05,
      "loss": 0.9331,
      "step": 1740
    },
    {
      "epoch": 2.3290969899665552,
      "grad_norm": 1.1219274997711182,
      "learning_rate": 2.2512381809995496e-05,
      "loss": 0.9591,
      "step": 1741
    },
    {
      "epoch": 2.3304347826086955,
      "grad_norm": 1.2403483390808105,
      "learning_rate": 2.2467357046375508e-05,
      "loss": 0.8553,
      "step": 1742
    },
    {
      "epoch": 2.3317725752508363,
      "grad_norm": 1.2419347763061523,
      "learning_rate": 2.2422332282755517e-05,
      "loss": 0.9683,
      "step": 1743
    },
    {
      "epoch": 2.3331103678929765,
      "grad_norm": 1.2146024703979492,
      "learning_rate": 2.2377307519135525e-05,
      "loss": 1.1727,
      "step": 1744
    },
    {
      "epoch": 2.334448160535117,
      "grad_norm": 1.3546491861343384,
      "learning_rate": 2.2332282755515534e-05,
      "loss": 0.8247,
      "step": 1745
    },
    {
      "epoch": 2.3357859531772576,
      "grad_norm": 1.0356041193008423,
      "learning_rate": 2.2287257991895542e-05,
      "loss": 0.9872,
      "step": 1746
    },
    {
      "epoch": 2.337123745819398,
      "grad_norm": 1.2764581441879272,
      "learning_rate": 2.224223322827555e-05,
      "loss": 1.1876,
      "step": 1747
    },
    {
      "epoch": 2.3384615384615386,
      "grad_norm": 1.2266361713409424,
      "learning_rate": 2.219720846465556e-05,
      "loss": 0.8758,
      "step": 1748
    },
    {
      "epoch": 2.339799331103679,
      "grad_norm": 1.0712088346481323,
      "learning_rate": 2.215218370103557e-05,
      "loss": 1.3622,
      "step": 1749
    },
    {
      "epoch": 2.3411371237458196,
      "grad_norm": 1.194495677947998,
      "learning_rate": 2.210715893741558e-05,
      "loss": 0.7502,
      "step": 1750
    },
    {
      "epoch": 2.34247491638796,
      "grad_norm": 1.4769166707992554,
      "learning_rate": 2.206213417379559e-05,
      "loss": 1.2554,
      "step": 1751
    },
    {
      "epoch": 2.3438127090301,
      "grad_norm": 1.2552289962768555,
      "learning_rate": 2.2017109410175597e-05,
      "loss": 0.8602,
      "step": 1752
    },
    {
      "epoch": 2.345150501672241,
      "grad_norm": 0.9770278334617615,
      "learning_rate": 2.1972084646555606e-05,
      "loss": 1.5884,
      "step": 1753
    },
    {
      "epoch": 2.346488294314381,
      "grad_norm": 1.1476101875305176,
      "learning_rate": 2.1927059882935614e-05,
      "loss": 0.5061,
      "step": 1754
    },
    {
      "epoch": 2.3478260869565215,
      "grad_norm": 1.1554605960845947,
      "learning_rate": 2.1882035119315626e-05,
      "loss": 0.4689,
      "step": 1755
    },
    {
      "epoch": 2.3491638795986622,
      "grad_norm": 1.296264886856079,
      "learning_rate": 2.1837010355695635e-05,
      "loss": 0.873,
      "step": 1756
    },
    {
      "epoch": 2.3505016722408025,
      "grad_norm": 1.245503544807434,
      "learning_rate": 2.1791985592075644e-05,
      "loss": 1.2616,
      "step": 1757
    },
    {
      "epoch": 2.3518394648829433,
      "grad_norm": 1.1375371217727661,
      "learning_rate": 2.1746960828455652e-05,
      "loss": 1.1452,
      "step": 1758
    },
    {
      "epoch": 2.3531772575250836,
      "grad_norm": 0.955712080001831,
      "learning_rate": 2.170193606483566e-05,
      "loss": 1.4884,
      "step": 1759
    },
    {
      "epoch": 2.3545150501672243,
      "grad_norm": 1.0706889629364014,
      "learning_rate": 2.165691130121567e-05,
      "loss": 0.7082,
      "step": 1760
    },
    {
      "epoch": 2.3558528428093646,
      "grad_norm": 1.3834351301193237,
      "learning_rate": 2.161188653759568e-05,
      "loss": 0.5975,
      "step": 1761
    },
    {
      "epoch": 2.357190635451505,
      "grad_norm": 1.1080961227416992,
      "learning_rate": 2.156686177397569e-05,
      "loss": 0.9926,
      "step": 1762
    },
    {
      "epoch": 2.3585284280936456,
      "grad_norm": 1.2083733081817627,
      "learning_rate": 2.15218370103557e-05,
      "loss": 0.6541,
      "step": 1763
    },
    {
      "epoch": 2.359866220735786,
      "grad_norm": 1.4147807359695435,
      "learning_rate": 2.1476812246735707e-05,
      "loss": 1.1652,
      "step": 1764
    },
    {
      "epoch": 2.361204013377926,
      "grad_norm": 0.9736313819885254,
      "learning_rate": 2.1431787483115716e-05,
      "loss": 1.1962,
      "step": 1765
    },
    {
      "epoch": 2.362541806020067,
      "grad_norm": 1.0683282613754272,
      "learning_rate": 2.1386762719495724e-05,
      "loss": 0.8665,
      "step": 1766
    },
    {
      "epoch": 2.363879598662207,
      "grad_norm": 0.9101349115371704,
      "learning_rate": 2.1341737955875733e-05,
      "loss": 1.4562,
      "step": 1767
    },
    {
      "epoch": 2.365217391304348,
      "grad_norm": 0.8101057410240173,
      "learning_rate": 2.129671319225574e-05,
      "loss": 1.436,
      "step": 1768
    },
    {
      "epoch": 2.3665551839464882,
      "grad_norm": 1.0484764575958252,
      "learning_rate": 2.1251688428635753e-05,
      "loss": 1.1506,
      "step": 1769
    },
    {
      "epoch": 2.367892976588629,
      "grad_norm": 1.45047926902771,
      "learning_rate": 2.1206663665015762e-05,
      "loss": 0.3479,
      "step": 1770
    },
    {
      "epoch": 2.3692307692307693,
      "grad_norm": 1.1299214363098145,
      "learning_rate": 2.116163890139577e-05,
      "loss": 1.1567,
      "step": 1771
    },
    {
      "epoch": 2.3705685618729095,
      "grad_norm": 1.1455305814743042,
      "learning_rate": 2.111661413777578e-05,
      "loss": 0.758,
      "step": 1772
    },
    {
      "epoch": 2.3719063545150503,
      "grad_norm": 1.1516034603118896,
      "learning_rate": 2.1071589374155788e-05,
      "loss": 0.6784,
      "step": 1773
    },
    {
      "epoch": 2.3732441471571906,
      "grad_norm": 1.4549493789672852,
      "learning_rate": 2.1026564610535796e-05,
      "loss": 0.9763,
      "step": 1774
    },
    {
      "epoch": 2.374581939799331,
      "grad_norm": 1.4241305589675903,
      "learning_rate": 2.0981539846915805e-05,
      "loss": 0.6554,
      "step": 1775
    },
    {
      "epoch": 2.3759197324414716,
      "grad_norm": 1.1807152032852173,
      "learning_rate": 2.0936515083295814e-05,
      "loss": 1.0083,
      "step": 1776
    },
    {
      "epoch": 2.377257525083612,
      "grad_norm": 1.2309315204620361,
      "learning_rate": 2.0891490319675822e-05,
      "loss": 1.4019,
      "step": 1777
    },
    {
      "epoch": 2.3785953177257526,
      "grad_norm": 1.0805270671844482,
      "learning_rate": 2.0846465556055834e-05,
      "loss": 0.929,
      "step": 1778
    },
    {
      "epoch": 2.379933110367893,
      "grad_norm": 1.2308298349380493,
      "learning_rate": 2.0801440792435843e-05,
      "loss": 1.3442,
      "step": 1779
    },
    {
      "epoch": 2.3812709030100336,
      "grad_norm": 1.1787298917770386,
      "learning_rate": 2.075641602881585e-05,
      "loss": 1.087,
      "step": 1780
    },
    {
      "epoch": 2.382608695652174,
      "grad_norm": 1.1068018674850464,
      "learning_rate": 2.071139126519586e-05,
      "loss": 1.0062,
      "step": 1781
    },
    {
      "epoch": 2.383946488294314,
      "grad_norm": 1.109654426574707,
      "learning_rate": 2.066636650157587e-05,
      "loss": 0.664,
      "step": 1782
    },
    {
      "epoch": 2.385284280936455,
      "grad_norm": 1.4103230237960815,
      "learning_rate": 2.0621341737955877e-05,
      "loss": 0.9542,
      "step": 1783
    },
    {
      "epoch": 2.3866220735785952,
      "grad_norm": 1.1636139154434204,
      "learning_rate": 2.0576316974335886e-05,
      "loss": 0.713,
      "step": 1784
    },
    {
      "epoch": 2.387959866220736,
      "grad_norm": 1.1543084383010864,
      "learning_rate": 2.0531292210715894e-05,
      "loss": 1.1145,
      "step": 1785
    },
    {
      "epoch": 2.3892976588628763,
      "grad_norm": 1.088383674621582,
      "learning_rate": 2.0486267447095903e-05,
      "loss": 0.852,
      "step": 1786
    },
    {
      "epoch": 2.3906354515050166,
      "grad_norm": 1.3434123992919922,
      "learning_rate": 2.0441242683475915e-05,
      "loss": 0.7964,
      "step": 1787
    },
    {
      "epoch": 2.3919732441471573,
      "grad_norm": 1.1704776287078857,
      "learning_rate": 2.0396217919855923e-05,
      "loss": 1.026,
      "step": 1788
    },
    {
      "epoch": 2.3933110367892976,
      "grad_norm": 1.2548800706863403,
      "learning_rate": 2.0351193156235932e-05,
      "loss": 1.0465,
      "step": 1789
    },
    {
      "epoch": 2.3946488294314383,
      "grad_norm": 1.332874059677124,
      "learning_rate": 2.030616839261594e-05,
      "loss": 0.8655,
      "step": 1790
    },
    {
      "epoch": 2.3959866220735786,
      "grad_norm": 0.925167977809906,
      "learning_rate": 2.026114362899595e-05,
      "loss": 0.9586,
      "step": 1791
    },
    {
      "epoch": 2.397324414715719,
      "grad_norm": 1.582779884338379,
      "learning_rate": 2.0216118865375958e-05,
      "loss": 0.9598,
      "step": 1792
    },
    {
      "epoch": 2.3986622073578596,
      "grad_norm": 1.0422955751419067,
      "learning_rate": 2.0171094101755966e-05,
      "loss": 1.144,
      "step": 1793
    },
    {
      "epoch": 2.4,
      "grad_norm": 1.4884545803070068,
      "learning_rate": 2.0126069338135975e-05,
      "loss": 0.6358,
      "step": 1794
    },
    {
      "epoch": 2.4013377926421406,
      "grad_norm": 1.151254653930664,
      "learning_rate": 2.0081044574515983e-05,
      "loss": 1.5272,
      "step": 1795
    },
    {
      "epoch": 2.402675585284281,
      "grad_norm": 1.0272475481033325,
      "learning_rate": 2.0036019810895995e-05,
      "loss": 1.107,
      "step": 1796
    },
    {
      "epoch": 2.4040133779264212,
      "grad_norm": 1.1820266246795654,
      "learning_rate": 1.9990995047276004e-05,
      "loss": 0.812,
      "step": 1797
    },
    {
      "epoch": 2.405351170568562,
      "grad_norm": 1.1445200443267822,
      "learning_rate": 1.9945970283656013e-05,
      "loss": 1.0356,
      "step": 1798
    },
    {
      "epoch": 2.4066889632107022,
      "grad_norm": 0.9422075748443604,
      "learning_rate": 1.990094552003602e-05,
      "loss": 1.2716,
      "step": 1799
    },
    {
      "epoch": 2.408026755852843,
      "grad_norm": 1.2195419073104858,
      "learning_rate": 1.985592075641603e-05,
      "loss": 1.0983,
      "step": 1800
    },
    {
      "epoch": 2.4093645484949833,
      "grad_norm": 1.0389440059661865,
      "learning_rate": 1.981089599279604e-05,
      "loss": 1.2389,
      "step": 1801
    },
    {
      "epoch": 2.4107023411371236,
      "grad_norm": 0.952129602432251,
      "learning_rate": 1.9765871229176047e-05,
      "loss": 0.6066,
      "step": 1802
    },
    {
      "epoch": 2.4120401337792643,
      "grad_norm": 1.1310745477676392,
      "learning_rate": 1.9720846465556056e-05,
      "loss": 0.7812,
      "step": 1803
    },
    {
      "epoch": 2.4133779264214046,
      "grad_norm": 1.4534422159194946,
      "learning_rate": 1.9675821701936064e-05,
      "loss": 1.0928,
      "step": 1804
    },
    {
      "epoch": 2.4147157190635453,
      "grad_norm": 1.0552500486373901,
      "learning_rate": 1.9630796938316076e-05,
      "loss": 1.0877,
      "step": 1805
    },
    {
      "epoch": 2.4160535117056856,
      "grad_norm": 1.2675952911376953,
      "learning_rate": 1.9585772174696085e-05,
      "loss": 0.4539,
      "step": 1806
    },
    {
      "epoch": 2.417391304347826,
      "grad_norm": 1.1485333442687988,
      "learning_rate": 1.9540747411076093e-05,
      "loss": 0.993,
      "step": 1807
    },
    {
      "epoch": 2.4187290969899666,
      "grad_norm": 1.238770842552185,
      "learning_rate": 1.9495722647456102e-05,
      "loss": 0.8684,
      "step": 1808
    },
    {
      "epoch": 2.420066889632107,
      "grad_norm": 1.22825026512146,
      "learning_rate": 1.945069788383611e-05,
      "loss": 1.023,
      "step": 1809
    },
    {
      "epoch": 2.4214046822742477,
      "grad_norm": 1.2012650966644287,
      "learning_rate": 1.940567312021612e-05,
      "loss": 0.9327,
      "step": 1810
    },
    {
      "epoch": 2.422742474916388,
      "grad_norm": 1.0674593448638916,
      "learning_rate": 1.9360648356596128e-05,
      "loss": 1.2149,
      "step": 1811
    },
    {
      "epoch": 2.4240802675585282,
      "grad_norm": 1.0509954690933228,
      "learning_rate": 1.9315623592976136e-05,
      "loss": 1.4321,
      "step": 1812
    },
    {
      "epoch": 2.425418060200669,
      "grad_norm": 1.2722355127334595,
      "learning_rate": 1.9270598829356145e-05,
      "loss": 0.387,
      "step": 1813
    },
    {
      "epoch": 2.4267558528428093,
      "grad_norm": 1.2516536712646484,
      "learning_rate": 1.9225574065736157e-05,
      "loss": 0.8528,
      "step": 1814
    },
    {
      "epoch": 2.42809364548495,
      "grad_norm": 1.777623176574707,
      "learning_rate": 1.9180549302116165e-05,
      "loss": 0.9748,
      "step": 1815
    },
    {
      "epoch": 2.4294314381270903,
      "grad_norm": 1.517867922782898,
      "learning_rate": 1.9135524538496174e-05,
      "loss": 0.9008,
      "step": 1816
    },
    {
      "epoch": 2.430769230769231,
      "grad_norm": 1.2444127798080444,
      "learning_rate": 1.9090499774876183e-05,
      "loss": 1.0568,
      "step": 1817
    },
    {
      "epoch": 2.4321070234113713,
      "grad_norm": 1.157259225845337,
      "learning_rate": 1.904547501125619e-05,
      "loss": 0.8248,
      "step": 1818
    },
    {
      "epoch": 2.4334448160535116,
      "grad_norm": 1.0810209512710571,
      "learning_rate": 1.90004502476362e-05,
      "loss": 0.9286,
      "step": 1819
    },
    {
      "epoch": 2.4347826086956523,
      "grad_norm": 1.2783986330032349,
      "learning_rate": 1.895542548401621e-05,
      "loss": 1.191,
      "step": 1820
    },
    {
      "epoch": 2.4361204013377926,
      "grad_norm": 1.5304255485534668,
      "learning_rate": 1.8910400720396217e-05,
      "loss": 0.9669,
      "step": 1821
    },
    {
      "epoch": 2.437458193979933,
      "grad_norm": 0.8730368614196777,
      "learning_rate": 1.8865375956776226e-05,
      "loss": 1.8674,
      "step": 1822
    },
    {
      "epoch": 2.4387959866220736,
      "grad_norm": 0.7612476348876953,
      "learning_rate": 1.8820351193156237e-05,
      "loss": 1.1188,
      "step": 1823
    },
    {
      "epoch": 2.440133779264214,
      "grad_norm": 0.917701005935669,
      "learning_rate": 1.8775326429536246e-05,
      "loss": 1.5137,
      "step": 1824
    },
    {
      "epoch": 2.4414715719063547,
      "grad_norm": 1.051805853843689,
      "learning_rate": 1.8730301665916255e-05,
      "loss": 0.7611,
      "step": 1825
    },
    {
      "epoch": 2.442809364548495,
      "grad_norm": 1.0616856813430786,
      "learning_rate": 1.8685276902296263e-05,
      "loss": 1.0647,
      "step": 1826
    },
    {
      "epoch": 2.4441471571906357,
      "grad_norm": 1.0747015476226807,
      "learning_rate": 1.8640252138676272e-05,
      "loss": 0.8052,
      "step": 1827
    },
    {
      "epoch": 2.445484949832776,
      "grad_norm": 1.0665451288223267,
      "learning_rate": 1.859522737505628e-05,
      "loss": 1.3131,
      "step": 1828
    },
    {
      "epoch": 2.4468227424749163,
      "grad_norm": 1.2551919221878052,
      "learning_rate": 1.855020261143629e-05,
      "loss": 1.0341,
      "step": 1829
    },
    {
      "epoch": 2.448160535117057,
      "grad_norm": 1.1262199878692627,
      "learning_rate": 1.8505177847816298e-05,
      "loss": 1.4566,
      "step": 1830
    },
    {
      "epoch": 2.4494983277591973,
      "grad_norm": 1.1426798105239868,
      "learning_rate": 1.8460153084196306e-05,
      "loss": 0.4731,
      "step": 1831
    },
    {
      "epoch": 2.4508361204013376,
      "grad_norm": 0.8868839740753174,
      "learning_rate": 1.8415128320576318e-05,
      "loss": 1.5829,
      "step": 1832
    },
    {
      "epoch": 2.4521739130434783,
      "grad_norm": 1.0325320959091187,
      "learning_rate": 1.8370103556956327e-05,
      "loss": 0.707,
      "step": 1833
    },
    {
      "epoch": 2.4535117056856186,
      "grad_norm": 1.1204519271850586,
      "learning_rate": 1.8325078793336335e-05,
      "loss": 0.58,
      "step": 1834
    },
    {
      "epoch": 2.4548494983277593,
      "grad_norm": 1.3483015298843384,
      "learning_rate": 1.8280054029716344e-05,
      "loss": 0.8913,
      "step": 1835
    },
    {
      "epoch": 2.4561872909698996,
      "grad_norm": 1.328149676322937,
      "learning_rate": 1.8235029266096353e-05,
      "loss": 0.8025,
      "step": 1836
    },
    {
      "epoch": 2.4575250836120404,
      "grad_norm": 1.5574856996536255,
      "learning_rate": 1.819000450247636e-05,
      "loss": 0.856,
      "step": 1837
    },
    {
      "epoch": 2.4588628762541807,
      "grad_norm": 1.2833364009857178,
      "learning_rate": 1.814497973885637e-05,
      "loss": 1.1202,
      "step": 1838
    },
    {
      "epoch": 2.460200668896321,
      "grad_norm": 1.504071593284607,
      "learning_rate": 1.809995497523638e-05,
      "loss": 0.8259,
      "step": 1839
    },
    {
      "epoch": 2.4615384615384617,
      "grad_norm": 0.9362914562225342,
      "learning_rate": 1.805493021161639e-05,
      "loss": 1.1352,
      "step": 1840
    },
    {
      "epoch": 2.462876254180602,
      "grad_norm": 1.0429112911224365,
      "learning_rate": 1.80099054479964e-05,
      "loss": 0.6208,
      "step": 1841
    },
    {
      "epoch": 2.4642140468227423,
      "grad_norm": 0.8809786438941956,
      "learning_rate": 1.7964880684376407e-05,
      "loss": 0.7602,
      "step": 1842
    },
    {
      "epoch": 2.465551839464883,
      "grad_norm": 0.8987886309623718,
      "learning_rate": 1.7919855920756416e-05,
      "loss": 0.3419,
      "step": 1843
    },
    {
      "epoch": 2.4668896321070233,
      "grad_norm": 1.066552758216858,
      "learning_rate": 1.7874831157136428e-05,
      "loss": 1.1363,
      "step": 1844
    },
    {
      "epoch": 2.468227424749164,
      "grad_norm": 1.1129825115203857,
      "learning_rate": 1.7829806393516437e-05,
      "loss": 1.0859,
      "step": 1845
    },
    {
      "epoch": 2.4695652173913043,
      "grad_norm": 1.2088080644607544,
      "learning_rate": 1.7784781629896445e-05,
      "loss": 0.9777,
      "step": 1846
    },
    {
      "epoch": 2.470903010033445,
      "grad_norm": 1.3047457933425903,
      "learning_rate": 1.7739756866276454e-05,
      "loss": 0.7029,
      "step": 1847
    },
    {
      "epoch": 2.4722408026755853,
      "grad_norm": 1.3074170351028442,
      "learning_rate": 1.7694732102656462e-05,
      "loss": 1.1114,
      "step": 1848
    },
    {
      "epoch": 2.4735785953177256,
      "grad_norm": 1.1256179809570312,
      "learning_rate": 1.764970733903647e-05,
      "loss": 1.0348,
      "step": 1849
    },
    {
      "epoch": 2.4749163879598663,
      "grad_norm": 1.1836167573928833,
      "learning_rate": 1.7604682575416483e-05,
      "loss": 1.1177,
      "step": 1850
    },
    {
      "epoch": 2.4762541806020066,
      "grad_norm": 1.1081132888793945,
      "learning_rate": 1.755965781179649e-05,
      "loss": 0.9263,
      "step": 1851
    },
    {
      "epoch": 2.477591973244147,
      "grad_norm": 0.8676427602767944,
      "learning_rate": 1.75146330481765e-05,
      "loss": 1.5092,
      "step": 1852
    },
    {
      "epoch": 2.4789297658862877,
      "grad_norm": 1.0682950019836426,
      "learning_rate": 1.746960828455651e-05,
      "loss": 1.1037,
      "step": 1853
    },
    {
      "epoch": 2.480267558528428,
      "grad_norm": 0.8002800941467285,
      "learning_rate": 1.7424583520936517e-05,
      "loss": 1.4235,
      "step": 1854
    },
    {
      "epoch": 2.4816053511705687,
      "grad_norm": 3.852647304534912,
      "learning_rate": 1.7379558757316526e-05,
      "loss": 0.5508,
      "step": 1855
    },
    {
      "epoch": 2.482943143812709,
      "grad_norm": 0.9870190024375916,
      "learning_rate": 1.7334533993696534e-05,
      "loss": 1.0963,
      "step": 1856
    },
    {
      "epoch": 2.4842809364548497,
      "grad_norm": 1.1127872467041016,
      "learning_rate": 1.7289509230076543e-05,
      "loss": 0.9211,
      "step": 1857
    },
    {
      "epoch": 2.48561872909699,
      "grad_norm": 1.0517997741699219,
      "learning_rate": 1.724448446645655e-05,
      "loss": 0.9661,
      "step": 1858
    },
    {
      "epoch": 2.4869565217391303,
      "grad_norm": 1.0883008241653442,
      "learning_rate": 1.7199459702836564e-05,
      "loss": 1.0869,
      "step": 1859
    },
    {
      "epoch": 2.488294314381271,
      "grad_norm": 0.9822888374328613,
      "learning_rate": 1.7154434939216572e-05,
      "loss": 1.1655,
      "step": 1860
    },
    {
      "epoch": 2.4896321070234113,
      "grad_norm": 1.487812876701355,
      "learning_rate": 1.710941017559658e-05,
      "loss": 0.6279,
      "step": 1861
    },
    {
      "epoch": 2.4909698996655516,
      "grad_norm": 1.1666566133499146,
      "learning_rate": 1.706438541197659e-05,
      "loss": 1.049,
      "step": 1862
    },
    {
      "epoch": 2.4923076923076923,
      "grad_norm": 1.2897906303405762,
      "learning_rate": 1.7019360648356598e-05,
      "loss": 1.3856,
      "step": 1863
    },
    {
      "epoch": 2.4936454849498326,
      "grad_norm": 1.2394946813583374,
      "learning_rate": 1.6974335884736607e-05,
      "loss": 1.0223,
      "step": 1864
    },
    {
      "epoch": 2.4949832775919734,
      "grad_norm": 1.536291241645813,
      "learning_rate": 1.6929311121116615e-05,
      "loss": 0.7075,
      "step": 1865
    },
    {
      "epoch": 2.4963210702341136,
      "grad_norm": 0.7085421085357666,
      "learning_rate": 1.6884286357496624e-05,
      "loss": 0.626,
      "step": 1866
    },
    {
      "epoch": 2.4976588628762544,
      "grad_norm": 1.1903938055038452,
      "learning_rate": 1.6839261593876632e-05,
      "loss": 1.4943,
      "step": 1867
    },
    {
      "epoch": 2.4989966555183947,
      "grad_norm": 0.9671469330787659,
      "learning_rate": 1.6794236830256644e-05,
      "loss": 1.4292,
      "step": 1868
    },
    {
      "epoch": 2.500334448160535,
      "grad_norm": 1.2777341604232788,
      "learning_rate": 1.6749212066636653e-05,
      "loss": 1.371,
      "step": 1869
    },
    {
      "epoch": 2.5016722408026757,
      "grad_norm": 1.1267082691192627,
      "learning_rate": 1.670418730301666e-05,
      "loss": 1.0386,
      "step": 1870
    },
    {
      "epoch": 2.503010033444816,
      "grad_norm": 1.0690659284591675,
      "learning_rate": 1.665916253939667e-05,
      "loss": 0.9463,
      "step": 1871
    },
    {
      "epoch": 2.5043478260869563,
      "grad_norm": 1.0973694324493408,
      "learning_rate": 1.661413777577668e-05,
      "loss": 0.7422,
      "step": 1872
    },
    {
      "epoch": 2.505685618729097,
      "grad_norm": 1.0973224639892578,
      "learning_rate": 1.6569113012156687e-05,
      "loss": 0.6723,
      "step": 1873
    },
    {
      "epoch": 2.5070234113712373,
      "grad_norm": 1.0682529211044312,
      "learning_rate": 1.6524088248536696e-05,
      "loss": 1.0982,
      "step": 1874
    },
    {
      "epoch": 2.508361204013378,
      "grad_norm": 1.1073600053787231,
      "learning_rate": 1.6479063484916704e-05,
      "loss": 0.8828,
      "step": 1875
    },
    {
      "epoch": 2.5096989966555183,
      "grad_norm": 0.9082316160202026,
      "learning_rate": 1.6434038721296713e-05,
      "loss": 1.1856,
      "step": 1876
    },
    {
      "epoch": 2.511036789297659,
      "grad_norm": 0.7870885729789734,
      "learning_rate": 1.6389013957676725e-05,
      "loss": 1.2523,
      "step": 1877
    },
    {
      "epoch": 2.5123745819397993,
      "grad_norm": 1.105562448501587,
      "learning_rate": 1.6343989194056734e-05,
      "loss": 0.5557,
      "step": 1878
    },
    {
      "epoch": 2.5137123745819396,
      "grad_norm": 1.1683121919631958,
      "learning_rate": 1.6298964430436742e-05,
      "loss": 1.0643,
      "step": 1879
    },
    {
      "epoch": 2.5150501672240804,
      "grad_norm": 1.3365424871444702,
      "learning_rate": 1.625393966681675e-05,
      "loss": 1.1505,
      "step": 1880
    },
    {
      "epoch": 2.5163879598662207,
      "grad_norm": 1.2845288515090942,
      "learning_rate": 1.620891490319676e-05,
      "loss": 0.787,
      "step": 1881
    },
    {
      "epoch": 2.517725752508361,
      "grad_norm": 0.5601235628128052,
      "learning_rate": 1.6163890139576768e-05,
      "loss": 0.5122,
      "step": 1882
    },
    {
      "epoch": 2.5190635451505017,
      "grad_norm": 1.0477005243301392,
      "learning_rate": 1.6118865375956776e-05,
      "loss": 0.7563,
      "step": 1883
    },
    {
      "epoch": 2.5204013377926424,
      "grad_norm": 1.183910608291626,
      "learning_rate": 1.6073840612336785e-05,
      "loss": 0.8595,
      "step": 1884
    },
    {
      "epoch": 2.5217391304347827,
      "grad_norm": 1.1977518796920776,
      "learning_rate": 1.6028815848716794e-05,
      "loss": 0.5492,
      "step": 1885
    },
    {
      "epoch": 2.523076923076923,
      "grad_norm": 1.104203224182129,
      "learning_rate": 1.5983791085096806e-05,
      "loss": 1.2017,
      "step": 1886
    },
    {
      "epoch": 2.5244147157190637,
      "grad_norm": 1.2420964241027832,
      "learning_rate": 1.5938766321476814e-05,
      "loss": 1.0783,
      "step": 1887
    },
    {
      "epoch": 2.525752508361204,
      "grad_norm": 1.2682381868362427,
      "learning_rate": 1.5893741557856823e-05,
      "loss": 0.5764,
      "step": 1888
    },
    {
      "epoch": 2.5270903010033443,
      "grad_norm": 1.061891794204712,
      "learning_rate": 1.584871679423683e-05,
      "loss": 0.5618,
      "step": 1889
    },
    {
      "epoch": 2.528428093645485,
      "grad_norm": 1.5407414436340332,
      "learning_rate": 1.580369203061684e-05,
      "loss": 0.88,
      "step": 1890
    },
    {
      "epoch": 2.5297658862876253,
      "grad_norm": 1.427150845527649,
      "learning_rate": 1.575866726699685e-05,
      "loss": 0.8215,
      "step": 1891
    },
    {
      "epoch": 2.5311036789297656,
      "grad_norm": 1.1984659433364868,
      "learning_rate": 1.5713642503376857e-05,
      "loss": 0.9356,
      "step": 1892
    },
    {
      "epoch": 2.5324414715719064,
      "grad_norm": 1.241512656211853,
      "learning_rate": 1.5668617739756866e-05,
      "loss": 0.6812,
      "step": 1893
    },
    {
      "epoch": 2.533779264214047,
      "grad_norm": 1.1882518529891968,
      "learning_rate": 1.5623592976136874e-05,
      "loss": 0.8977,
      "step": 1894
    },
    {
      "epoch": 2.5351170568561874,
      "grad_norm": 1.09954833984375,
      "learning_rate": 1.5578568212516886e-05,
      "loss": 1.3995,
      "step": 1895
    },
    {
      "epoch": 2.5364548494983277,
      "grad_norm": 1.0280219316482544,
      "learning_rate": 1.5533543448896895e-05,
      "loss": 1.048,
      "step": 1896
    },
    {
      "epoch": 2.5377926421404684,
      "grad_norm": 1.1472655534744263,
      "learning_rate": 1.5488518685276903e-05,
      "loss": 0.8748,
      "step": 1897
    },
    {
      "epoch": 2.5391304347826087,
      "grad_norm": 1.2066268920898438,
      "learning_rate": 1.5443493921656912e-05,
      "loss": 1.1831,
      "step": 1898
    },
    {
      "epoch": 2.540468227424749,
      "grad_norm": 1.6384567022323608,
      "learning_rate": 1.539846915803692e-05,
      "loss": 0.5115,
      "step": 1899
    },
    {
      "epoch": 2.5418060200668897,
      "grad_norm": 1.2105473279953003,
      "learning_rate": 1.535344439441693e-05,
      "loss": 1.0732,
      "step": 1900
    },
    {
      "epoch": 2.54314381270903,
      "grad_norm": 0.9396874904632568,
      "learning_rate": 1.5308419630796938e-05,
      "loss": 1.1214,
      "step": 1901
    },
    {
      "epoch": 2.5444816053511703,
      "grad_norm": 1.2806360721588135,
      "learning_rate": 1.5263394867176946e-05,
      "loss": 0.673,
      "step": 1902
    },
    {
      "epoch": 2.545819397993311,
      "grad_norm": 1.1149977445602417,
      "learning_rate": 1.5218370103556957e-05,
      "loss": 0.7108,
      "step": 1903
    },
    {
      "epoch": 2.5471571906354518,
      "grad_norm": 1.0973268747329712,
      "learning_rate": 1.5173345339936967e-05,
      "loss": 1.085,
      "step": 1904
    },
    {
      "epoch": 2.548494983277592,
      "grad_norm": 1.1979267597198486,
      "learning_rate": 1.5128320576316976e-05,
      "loss": 0.7764,
      "step": 1905
    },
    {
      "epoch": 2.5498327759197323,
      "grad_norm": 1.1910693645477295,
      "learning_rate": 1.5083295812696984e-05,
      "loss": 0.8246,
      "step": 1906
    },
    {
      "epoch": 2.551170568561873,
      "grad_norm": 1.1525737047195435,
      "learning_rate": 1.5038271049076993e-05,
      "loss": 1.0309,
      "step": 1907
    },
    {
      "epoch": 2.5525083612040134,
      "grad_norm": 1.1554816961288452,
      "learning_rate": 1.4993246285457001e-05,
      "loss": 1.1261,
      "step": 1908
    },
    {
      "epoch": 2.5538461538461537,
      "grad_norm": 1.2663363218307495,
      "learning_rate": 1.4948221521837012e-05,
      "loss": 0.8414,
      "step": 1909
    },
    {
      "epoch": 2.5551839464882944,
      "grad_norm": 0.9890291094779968,
      "learning_rate": 1.490319675821702e-05,
      "loss": 1.7599,
      "step": 1910
    },
    {
      "epoch": 2.5565217391304347,
      "grad_norm": 1.3378098011016846,
      "learning_rate": 1.4858171994597029e-05,
      "loss": 1.187,
      "step": 1911
    },
    {
      "epoch": 2.5578595317725754,
      "grad_norm": 0.920430600643158,
      "learning_rate": 1.4813147230977037e-05,
      "loss": 0.6821,
      "step": 1912
    },
    {
      "epoch": 2.5591973244147157,
      "grad_norm": 1.643335223197937,
      "learning_rate": 1.4768122467357048e-05,
      "loss": 0.8469,
      "step": 1913
    },
    {
      "epoch": 2.5605351170568564,
      "grad_norm": 1.3408782482147217,
      "learning_rate": 1.4723097703737058e-05,
      "loss": 1.0189,
      "step": 1914
    },
    {
      "epoch": 2.5618729096989967,
      "grad_norm": 1.2192609310150146,
      "learning_rate": 1.4678072940117067e-05,
      "loss": 1.0288,
      "step": 1915
    },
    {
      "epoch": 2.563210702341137,
      "grad_norm": 0.9527949690818787,
      "learning_rate": 1.4633048176497075e-05,
      "loss": 1.3139,
      "step": 1916
    },
    {
      "epoch": 2.5645484949832777,
      "grad_norm": 1.1736693382263184,
      "learning_rate": 1.4588023412877084e-05,
      "loss": 0.9382,
      "step": 1917
    },
    {
      "epoch": 2.565886287625418,
      "grad_norm": 1.4159702062606812,
      "learning_rate": 1.4542998649257092e-05,
      "loss": 0.6853,
      "step": 1918
    },
    {
      "epoch": 2.5672240802675583,
      "grad_norm": 1.1225204467773438,
      "learning_rate": 1.4497973885637101e-05,
      "loss": 1.079,
      "step": 1919
    },
    {
      "epoch": 2.568561872909699,
      "grad_norm": 1.1435071229934692,
      "learning_rate": 1.445294912201711e-05,
      "loss": 1.2635,
      "step": 1920
    },
    {
      "epoch": 2.5698996655518394,
      "grad_norm": 1.0206356048583984,
      "learning_rate": 1.4407924358397118e-05,
      "loss": 1.4776,
      "step": 1921
    },
    {
      "epoch": 2.57123745819398,
      "grad_norm": 1.7108975648880005,
      "learning_rate": 1.436289959477713e-05,
      "loss": 0.9222,
      "step": 1922
    },
    {
      "epoch": 2.5725752508361204,
      "grad_norm": 1.0113279819488525,
      "learning_rate": 1.4317874831157139e-05,
      "loss": 0.9089,
      "step": 1923
    },
    {
      "epoch": 2.573913043478261,
      "grad_norm": 1.2199150323867798,
      "learning_rate": 1.4272850067537147e-05,
      "loss": 0.8124,
      "step": 1924
    },
    {
      "epoch": 2.5752508361204014,
      "grad_norm": 0.9723790287971497,
      "learning_rate": 1.4227825303917156e-05,
      "loss": 1.6872,
      "step": 1925
    },
    {
      "epoch": 2.5765886287625417,
      "grad_norm": 1.3184490203857422,
      "learning_rate": 1.4182800540297164e-05,
      "loss": 0.6796,
      "step": 1926
    },
    {
      "epoch": 2.5779264214046824,
      "grad_norm": 0.9623216986656189,
      "learning_rate": 1.4137775776677173e-05,
      "loss": 0.9223,
      "step": 1927
    },
    {
      "epoch": 2.5792642140468227,
      "grad_norm": 1.3507014513015747,
      "learning_rate": 1.4092751013057182e-05,
      "loss": 0.8685,
      "step": 1928
    },
    {
      "epoch": 2.580602006688963,
      "grad_norm": 1.5632047653198242,
      "learning_rate": 1.404772624943719e-05,
      "loss": 1.086,
      "step": 1929
    },
    {
      "epoch": 2.5819397993311037,
      "grad_norm": 1.377564787864685,
      "learning_rate": 1.4002701485817199e-05,
      "loss": 0.9658,
      "step": 1930
    },
    {
      "epoch": 2.583277591973244,
      "grad_norm": 1.2215312719345093,
      "learning_rate": 1.395767672219721e-05,
      "loss": 1.0862,
      "step": 1931
    },
    {
      "epoch": 2.5846153846153848,
      "grad_norm": 1.145735740661621,
      "learning_rate": 1.391265195857722e-05,
      "loss": 1.2792,
      "step": 1932
    },
    {
      "epoch": 2.585953177257525,
      "grad_norm": 1.2771897315979004,
      "learning_rate": 1.3867627194957228e-05,
      "loss": 1.1728,
      "step": 1933
    },
    {
      "epoch": 2.587290969899666,
      "grad_norm": 1.2495319843292236,
      "learning_rate": 1.3822602431337236e-05,
      "loss": 0.8183,
      "step": 1934
    },
    {
      "epoch": 2.588628762541806,
      "grad_norm": 1.5561614036560059,
      "learning_rate": 1.3777577667717245e-05,
      "loss": 0.9294,
      "step": 1935
    },
    {
      "epoch": 2.5899665551839464,
      "grad_norm": 1.4182900190353394,
      "learning_rate": 1.3732552904097254e-05,
      "loss": 1.146,
      "step": 1936
    },
    {
      "epoch": 2.591304347826087,
      "grad_norm": 1.2488436698913574,
      "learning_rate": 1.3687528140477262e-05,
      "loss": 0.6245,
      "step": 1937
    },
    {
      "epoch": 2.5926421404682274,
      "grad_norm": 1.273505687713623,
      "learning_rate": 1.364250337685727e-05,
      "loss": 0.8321,
      "step": 1938
    },
    {
      "epoch": 2.5939799331103677,
      "grad_norm": 1.0358428955078125,
      "learning_rate": 1.359747861323728e-05,
      "loss": 0.9381,
      "step": 1939
    },
    {
      "epoch": 2.5953177257525084,
      "grad_norm": 1.0949110984802246,
      "learning_rate": 1.3552453849617291e-05,
      "loss": 1.4436,
      "step": 1940
    },
    {
      "epoch": 2.5966555183946487,
      "grad_norm": 1.0827977657318115,
      "learning_rate": 1.35074290859973e-05,
      "loss": 1.3695,
      "step": 1941
    },
    {
      "epoch": 2.5979933110367894,
      "grad_norm": 1.3221254348754883,
      "learning_rate": 1.3462404322377309e-05,
      "loss": 0.883,
      "step": 1942
    },
    {
      "epoch": 2.5993311036789297,
      "grad_norm": 1.330696702003479,
      "learning_rate": 1.3417379558757317e-05,
      "loss": 0.9751,
      "step": 1943
    },
    {
      "epoch": 2.6006688963210705,
      "grad_norm": 0.9758976101875305,
      "learning_rate": 1.3372354795137326e-05,
      "loss": 1.3133,
      "step": 1944
    },
    {
      "epoch": 2.6020066889632107,
      "grad_norm": 0.8660216331481934,
      "learning_rate": 1.3327330031517334e-05,
      "loss": 1.1024,
      "step": 1945
    },
    {
      "epoch": 2.603344481605351,
      "grad_norm": 1.2315067052841187,
      "learning_rate": 1.3282305267897343e-05,
      "loss": 0.7998,
      "step": 1946
    },
    {
      "epoch": 2.6046822742474918,
      "grad_norm": 1.1562628746032715,
      "learning_rate": 1.3237280504277352e-05,
      "loss": 1.3548,
      "step": 1947
    },
    {
      "epoch": 2.606020066889632,
      "grad_norm": 1.4570300579071045,
      "learning_rate": 1.3192255740657362e-05,
      "loss": 1.0354,
      "step": 1948
    },
    {
      "epoch": 2.6073578595317723,
      "grad_norm": 0.8457752466201782,
      "learning_rate": 1.3147230977037372e-05,
      "loss": 0.7525,
      "step": 1949
    },
    {
      "epoch": 2.608695652173913,
      "grad_norm": 1.1738195419311523,
      "learning_rate": 1.310220621341738e-05,
      "loss": 1.3463,
      "step": 1950
    },
    {
      "epoch": 2.6100334448160534,
      "grad_norm": 1.1737287044525146,
      "learning_rate": 1.305718144979739e-05,
      "loss": 0.8589,
      "step": 1951
    },
    {
      "epoch": 2.611371237458194,
      "grad_norm": 1.2601444721221924,
      "learning_rate": 1.3012156686177398e-05,
      "loss": 0.8973,
      "step": 1952
    },
    {
      "epoch": 2.6127090301003344,
      "grad_norm": 0.9962211847305298,
      "learning_rate": 1.2967131922557408e-05,
      "loss": 1.1899,
      "step": 1953
    },
    {
      "epoch": 2.614046822742475,
      "grad_norm": 1.0661948919296265,
      "learning_rate": 1.2922107158937417e-05,
      "loss": 1.2756,
      "step": 1954
    },
    {
      "epoch": 2.6153846153846154,
      "grad_norm": 1.3405736684799194,
      "learning_rate": 1.2877082395317425e-05,
      "loss": 0.7138,
      "step": 1955
    },
    {
      "epoch": 2.6167224080267557,
      "grad_norm": 1.1791526079177856,
      "learning_rate": 1.2832057631697434e-05,
      "loss": 0.9403,
      "step": 1956
    },
    {
      "epoch": 2.6180602006688964,
      "grad_norm": 1.0099623203277588,
      "learning_rate": 1.2787032868077442e-05,
      "loss": 0.8853,
      "step": 1957
    },
    {
      "epoch": 2.6193979933110367,
      "grad_norm": 1.1736373901367188,
      "learning_rate": 1.2742008104457453e-05,
      "loss": 1.0082,
      "step": 1958
    },
    {
      "epoch": 2.620735785953177,
      "grad_norm": 1.2148722410202026,
      "learning_rate": 1.2696983340837463e-05,
      "loss": 1.2064,
      "step": 1959
    },
    {
      "epoch": 2.6220735785953178,
      "grad_norm": 1.094408392906189,
      "learning_rate": 1.2651958577217472e-05,
      "loss": 0.9928,
      "step": 1960
    },
    {
      "epoch": 2.623411371237458,
      "grad_norm": 1.1903018951416016,
      "learning_rate": 1.260693381359748e-05,
      "loss": 0.9445,
      "step": 1961
    },
    {
      "epoch": 2.624749163879599,
      "grad_norm": 1.101300597190857,
      "learning_rate": 1.2561909049977489e-05,
      "loss": 0.6501,
      "step": 1962
    },
    {
      "epoch": 2.626086956521739,
      "grad_norm": 1.079803466796875,
      "learning_rate": 1.2516884286357497e-05,
      "loss": 1.1167,
      "step": 1963
    },
    {
      "epoch": 2.62742474916388,
      "grad_norm": 0.9757705926895142,
      "learning_rate": 1.2471859522737506e-05,
      "loss": 0.7433,
      "step": 1964
    },
    {
      "epoch": 2.62876254180602,
      "grad_norm": 1.028402328491211,
      "learning_rate": 1.2426834759117516e-05,
      "loss": 1.6142,
      "step": 1965
    },
    {
      "epoch": 2.6301003344481604,
      "grad_norm": 0.6608441472053528,
      "learning_rate": 1.2381809995497525e-05,
      "loss": 0.5881,
      "step": 1966
    },
    {
      "epoch": 2.631438127090301,
      "grad_norm": 1.1534054279327393,
      "learning_rate": 1.2336785231877533e-05,
      "loss": 1.2169,
      "step": 1967
    },
    {
      "epoch": 2.6327759197324414,
      "grad_norm": 1.0863515138626099,
      "learning_rate": 1.2291760468257542e-05,
      "loss": 1.288,
      "step": 1968
    },
    {
      "epoch": 2.6341137123745817,
      "grad_norm": 1.3784714937210083,
      "learning_rate": 1.2246735704637552e-05,
      "loss": 1.2264,
      "step": 1969
    },
    {
      "epoch": 2.6354515050167224,
      "grad_norm": 1.5014134645462036,
      "learning_rate": 1.2201710941017561e-05,
      "loss": 0.7316,
      "step": 1970
    },
    {
      "epoch": 2.6367892976588627,
      "grad_norm": 1.3054801225662231,
      "learning_rate": 1.215668617739757e-05,
      "loss": 0.412,
      "step": 1971
    },
    {
      "epoch": 2.6381270903010035,
      "grad_norm": 0.8251444101333618,
      "learning_rate": 1.2111661413777578e-05,
      "loss": 1.0635,
      "step": 1972
    },
    {
      "epoch": 2.6394648829431437,
      "grad_norm": 1.0586771965026855,
      "learning_rate": 1.2066636650157587e-05,
      "loss": 0.681,
      "step": 1973
    },
    {
      "epoch": 2.6408026755852845,
      "grad_norm": 1.3109058141708374,
      "learning_rate": 1.2021611886537597e-05,
      "loss": 0.9528,
      "step": 1974
    },
    {
      "epoch": 2.6421404682274248,
      "grad_norm": 1.1887587308883667,
      "learning_rate": 1.1976587122917606e-05,
      "loss": 0.8634,
      "step": 1975
    },
    {
      "epoch": 2.643478260869565,
      "grad_norm": 1.2000380754470825,
      "learning_rate": 1.1931562359297614e-05,
      "loss": 1.1268,
      "step": 1976
    },
    {
      "epoch": 2.644816053511706,
      "grad_norm": 1.0287446975708008,
      "learning_rate": 1.1886537595677623e-05,
      "loss": 0.678,
      "step": 1977
    },
    {
      "epoch": 2.646153846153846,
      "grad_norm": 1.051858901977539,
      "learning_rate": 1.1841512832057633e-05,
      "loss": 0.6412,
      "step": 1978
    },
    {
      "epoch": 2.6474916387959864,
      "grad_norm": 1.2014086246490479,
      "learning_rate": 1.1796488068437642e-05,
      "loss": 0.9895,
      "step": 1979
    },
    {
      "epoch": 2.648829431438127,
      "grad_norm": 0.8527210354804993,
      "learning_rate": 1.175146330481765e-05,
      "loss": 0.9464,
      "step": 1980
    },
    {
      "epoch": 2.650167224080268,
      "grad_norm": 1.1812256574630737,
      "learning_rate": 1.1706438541197659e-05,
      "loss": 0.7605,
      "step": 1981
    },
    {
      "epoch": 2.651505016722408,
      "grad_norm": 1.463782787322998,
      "learning_rate": 1.1661413777577667e-05,
      "loss": 1.1127,
      "step": 1982
    },
    {
      "epoch": 2.6528428093645484,
      "grad_norm": 1.2627677917480469,
      "learning_rate": 1.1616389013957678e-05,
      "loss": 0.5517,
      "step": 1983
    },
    {
      "epoch": 2.654180602006689,
      "grad_norm": 1.1560112237930298,
      "learning_rate": 1.1571364250337686e-05,
      "loss": 1.1212,
      "step": 1984
    },
    {
      "epoch": 2.6555183946488294,
      "grad_norm": 1.3743358850479126,
      "learning_rate": 1.1526339486717695e-05,
      "loss": 0.9603,
      "step": 1985
    },
    {
      "epoch": 2.6568561872909697,
      "grad_norm": 1.1926093101501465,
      "learning_rate": 1.1481314723097703e-05,
      "loss": 1.112,
      "step": 1986
    },
    {
      "epoch": 2.6581939799331105,
      "grad_norm": 1.288248896598816,
      "learning_rate": 1.1436289959477714e-05,
      "loss": 0.8998,
      "step": 1987
    },
    {
      "epoch": 2.6595317725752508,
      "grad_norm": 1.1038404703140259,
      "learning_rate": 1.1391265195857722e-05,
      "loss": 0.2956,
      "step": 1988
    },
    {
      "epoch": 2.660869565217391,
      "grad_norm": 1.2742241621017456,
      "learning_rate": 1.134624043223773e-05,
      "loss": 1.2681,
      "step": 1989
    },
    {
      "epoch": 2.6622073578595318,
      "grad_norm": 0.9164013862609863,
      "learning_rate": 1.130121566861774e-05,
      "loss": 0.8526,
      "step": 1990
    },
    {
      "epoch": 2.6635451505016725,
      "grad_norm": 1.274473786354065,
      "learning_rate": 1.1256190904997748e-05,
      "loss": 0.9956,
      "step": 1991
    },
    {
      "epoch": 2.664882943143813,
      "grad_norm": 1.241410255432129,
      "learning_rate": 1.1211166141377758e-05,
      "loss": 0.8181,
      "step": 1992
    },
    {
      "epoch": 2.666220735785953,
      "grad_norm": 1.138505458831787,
      "learning_rate": 1.1166141377757767e-05,
      "loss": 1.4564,
      "step": 1993
    },
    {
      "epoch": 2.667558528428094,
      "grad_norm": 1.180283784866333,
      "learning_rate": 1.1121116614137775e-05,
      "loss": 1.3439,
      "step": 1994
    },
    {
      "epoch": 2.668896321070234,
      "grad_norm": 0.9829315543174744,
      "learning_rate": 1.1076091850517786e-05,
      "loss": 1.4337,
      "step": 1995
    },
    {
      "epoch": 2.6702341137123744,
      "grad_norm": 1.160563588142395,
      "learning_rate": 1.1031067086897794e-05,
      "loss": 0.5688,
      "step": 1996
    },
    {
      "epoch": 2.671571906354515,
      "grad_norm": 1.1791166067123413,
      "learning_rate": 1.0986042323277803e-05,
      "loss": 0.4572,
      "step": 1997
    },
    {
      "epoch": 2.6729096989966554,
      "grad_norm": 1.2810428142547607,
      "learning_rate": 1.0941017559657813e-05,
      "loss": 0.5095,
      "step": 1998
    },
    {
      "epoch": 2.6742474916387957,
      "grad_norm": 1.0259909629821777,
      "learning_rate": 1.0895992796037822e-05,
      "loss": 1.3085,
      "step": 1999
    },
    {
      "epoch": 2.6755852842809364,
      "grad_norm": 0.9915300607681274,
      "learning_rate": 1.085096803241783e-05,
      "loss": 0.5626,
      "step": 2000
    }
  ],
  "logging_steps": 1,
  "max_steps": 2241,
  "num_input_tokens_seen": 0,
  "num_train_epochs": 3,
  "save_steps": 500,
  "stateful_callbacks": {
    "TrainerControl": {
      "args": {
        "should_epoch_stop": false,
        "should_evaluate": false,
        "should_log": false,
        "should_save": true,
        "should_training_stop": false
      },
      "attributes": {}
    }
  },
  "total_flos": 3.122019867483095e+18,
  "train_batch_size": 1,
  "trial_name": null,
  "trial_params": null
}