{
  "best_metric": null,
  "best_model_checkpoint": null,
  "epoch": 1.0,
  "eval_steps": 500,
  "global_step": 18025,
  "is_hyper_param_search": false,
  "is_local_process_zero": true,
  "is_world_process_zero": true,
  "log_history": [
    {
      "epoch": 0.0011095700416088765,
      "grad_norm": 0.6721351742744446,
      "learning_rate": 2.7739251040221912e-09,
      "loss": 1.1791,
      "step": 20
    },
    {
      "epoch": 0.002219140083217753,
      "grad_norm": 0.48403581976890564,
      "learning_rate": 5.5478502080443824e-09,
      "loss": 1.1797,
      "step": 40
    },
    {
      "epoch": 0.00332871012482663,
      "grad_norm": 0.328329473733902,
      "learning_rate": 8.321775312066573e-09,
      "loss": 1.1276,
      "step": 60
    },
    {
      "epoch": 0.004438280166435506,
      "grad_norm": 0.573009192943573,
      "learning_rate": 1.1095700416088765e-08,
      "loss": 1.1502,
      "step": 80
    },
    {
      "epoch": 0.005547850208044383,
      "grad_norm": 0.7344629764556885,
      "learning_rate": 1.3869625520110957e-08,
      "loss": 1.1745,
      "step": 100
    },
    {
      "epoch": 0.00665742024965326,
      "grad_norm": 0.4872736632823944,
      "learning_rate": 1.6643550624133146e-08,
      "loss": 1.1419,
      "step": 120
    },
    {
      "epoch": 0.007766990291262136,
      "grad_norm": 0.42412033677101135,
      "learning_rate": 1.9417475728155338e-08,
      "loss": 1.1833,
      "step": 140
    },
    {
      "epoch": 0.008876560332871012,
      "grad_norm": 0.41351547837257385,
      "learning_rate": 2.219140083217753e-08,
      "loss": 1.1966,
      "step": 160
    },
    {
      "epoch": 0.009986130374479889,
      "grad_norm": 0.5638367533683777,
      "learning_rate": 2.4965325936199722e-08,
      "loss": 1.1935,
      "step": 180
    },
    {
      "epoch": 0.011095700416088766,
      "grad_norm": 0.35621383786201477,
      "learning_rate": 2.7739251040221914e-08,
      "loss": 1.1403,
      "step": 200
    },
    {
      "epoch": 0.012205270457697643,
      "grad_norm": 0.519483208656311,
      "learning_rate": 3.0513176144244106e-08,
      "loss": 1.1716,
      "step": 220
    },
    {
      "epoch": 0.01331484049930652,
      "grad_norm": 0.655170202255249,
      "learning_rate": 3.328710124826629e-08,
      "loss": 1.187,
      "step": 240
    },
    {
      "epoch": 0.014424410540915394,
      "grad_norm": 0.30613973736763,
      "learning_rate": 3.606102635228848e-08,
      "loss": 1.1639,
      "step": 260
    },
    {
      "epoch": 0.015533980582524271,
      "grad_norm": 0.6322771310806274,
      "learning_rate": 3.8834951456310675e-08,
      "loss": 1.1547,
      "step": 280
    },
    {
      "epoch": 0.016643550624133148,
      "grad_norm": 0.622188925743103,
      "learning_rate": 4.1608876560332874e-08,
      "loss": 1.1878,
      "step": 300
    },
    {
      "epoch": 0.017753120665742025,
      "grad_norm": 0.636248767375946,
      "learning_rate": 4.438280166435506e-08,
      "loss": 1.1298,
      "step": 320
    },
    {
      "epoch": 0.0188626907073509,
      "grad_norm": 0.5820329189300537,
      "learning_rate": 4.715672676837725e-08,
      "loss": 1.1284,
      "step": 340
    },
    {
      "epoch": 0.019972260748959778,
      "grad_norm": 0.40758267045021057,
      "learning_rate": 4.9930651872399443e-08,
      "loss": 1.2361,
      "step": 360
    },
    {
      "epoch": 0.021081830790568655,
      "grad_norm": 0.4012329578399658,
      "learning_rate": 5.270457697642163e-08,
      "loss": 1.1518,
      "step": 380
    },
    {
      "epoch": 0.022191400832177532,
      "grad_norm": 0.5331623554229736,
      "learning_rate": 5.547850208044383e-08,
      "loss": 1.2203,
      "step": 400
    },
    {
      "epoch": 0.02330097087378641,
      "grad_norm": 0.5145316123962402,
      "learning_rate": 5.825242718446602e-08,
      "loss": 1.1253,
      "step": 420
    },
    {
      "epoch": 0.024410540915395285,
      "grad_norm": 0.36710885167121887,
      "learning_rate": 6.102635228848821e-08,
      "loss": 1.0685,
      "step": 440
    },
    {
      "epoch": 0.025520110957004162,
      "grad_norm": 0.3183213472366333,
      "learning_rate": 6.38002773925104e-08,
      "loss": 1.2097,
      "step": 460
    },
    {
      "epoch": 0.02662968099861304,
      "grad_norm": 0.3870026171207428,
      "learning_rate": 6.657420249653258e-08,
      "loss": 1.1605,
      "step": 480
    },
    {
      "epoch": 0.027739251040221916,
      "grad_norm": 0.5289874076843262,
      "learning_rate": 6.934812760055478e-08,
      "loss": 1.1163,
      "step": 500
    },
    {
      "epoch": 0.02884882108183079,
      "grad_norm": 0.6247478127479553,
      "learning_rate": 7.212205270457697e-08,
      "loss": 1.2463,
      "step": 520
    },
    {
      "epoch": 0.029958391123439666,
      "grad_norm": 0.5780977010726929,
      "learning_rate": 7.489597780859917e-08,
      "loss": 1.195,
      "step": 540
    },
    {
      "epoch": 0.031067961165048542,
      "grad_norm": 0.5350005626678467,
      "learning_rate": 7.766990291262135e-08,
      "loss": 1.1289,
      "step": 560
    },
    {
      "epoch": 0.03217753120665742,
      "grad_norm": 0.30264171957969666,
      "learning_rate": 8.044382801664355e-08,
      "loss": 1.0648,
      "step": 580
    },
    {
      "epoch": 0.033287101248266296,
      "grad_norm": 0.45016685128211975,
      "learning_rate": 8.321775312066575e-08,
      "loss": 1.153,
      "step": 600
    },
    {
      "epoch": 0.034396671289875176,
      "grad_norm": 0.5521411895751953,
      "learning_rate": 8.599167822468793e-08,
      "loss": 1.1257,
      "step": 620
    },
    {
      "epoch": 0.03550624133148405,
      "grad_norm": 0.4235968291759491,
      "learning_rate": 8.876560332871012e-08,
      "loss": 1.1198,
      "step": 640
    },
    {
      "epoch": 0.03661581137309293,
      "grad_norm": 0.5764958262443542,
      "learning_rate": 9.153952843273232e-08,
      "loss": 1.1015,
      "step": 660
    },
    {
      "epoch": 0.0377253814147018,
      "grad_norm": 0.5378324389457703,
      "learning_rate": 9.43134535367545e-08,
      "loss": 1.2342,
      "step": 680
    },
    {
      "epoch": 0.038834951456310676,
      "grad_norm": 0.3303524851799011,
      "learning_rate": 9.708737864077669e-08,
      "loss": 1.1156,
      "step": 700
    },
    {
      "epoch": 0.039944521497919556,
      "grad_norm": 0.5583963394165039,
      "learning_rate": 9.986130374479889e-08,
      "loss": 1.1579,
      "step": 720
    },
    {
      "epoch": 0.04105409153952843,
      "grad_norm": 0.813584566116333,
      "learning_rate": 1.0263522884882107e-07,
      "loss": 1.1749,
      "step": 740
    },
    {
      "epoch": 0.04216366158113731,
      "grad_norm": 0.6232322454452515,
      "learning_rate": 1.0540915395284326e-07,
      "loss": 1.195,
      "step": 760
    },
    {
      "epoch": 0.04327323162274618,
      "grad_norm": 0.5598679780960083,
      "learning_rate": 1.0818307905686546e-07,
      "loss": 1.1183,
      "step": 780
    },
    {
      "epoch": 0.044382801664355064,
      "grad_norm": 0.5374318361282349,
      "learning_rate": 1.1095700416088766e-07,
      "loss": 1.1944,
      "step": 800
    },
    {
      "epoch": 0.04549237170596394,
      "grad_norm": 0.19993217289447784,
      "learning_rate": 1.1373092926490985e-07,
      "loss": 1.0835,
      "step": 820
    },
    {
      "epoch": 0.04660194174757282,
      "grad_norm": 0.4393330514431,
      "learning_rate": 1.1650485436893204e-07,
      "loss": 1.1977,
      "step": 840
    },
    {
      "epoch": 0.04771151178918169,
      "grad_norm": 0.6222187876701355,
      "learning_rate": 1.1927877947295422e-07,
      "loss": 1.0523,
      "step": 860
    },
    {
      "epoch": 0.04882108183079057,
      "grad_norm": 0.319072425365448,
      "learning_rate": 1.2205270457697642e-07,
      "loss": 1.136,
      "step": 880
    },
    {
      "epoch": 0.049930651872399444,
      "grad_norm": 0.5406301617622375,
      "learning_rate": 1.248266296809986e-07,
      "loss": 1.2155,
      "step": 900
    },
    {
      "epoch": 0.051040221914008324,
      "grad_norm": 0.618211567401886,
      "learning_rate": 1.276005547850208e-07,
      "loss": 1.1562,
      "step": 920
    },
    {
      "epoch": 0.0521497919556172,
      "grad_norm": 0.3475450277328491,
      "learning_rate": 1.30374479889043e-07,
      "loss": 1.1551,
      "step": 940
    },
    {
      "epoch": 0.05325936199722608,
      "grad_norm": 0.4210108518600464,
      "learning_rate": 1.3314840499306516e-07,
      "loss": 1.1612,
      "step": 960
    },
    {
      "epoch": 0.05436893203883495,
      "grad_norm": 0.386306494474411,
      "learning_rate": 1.3592233009708736e-07,
      "loss": 1.2195,
      "step": 980
    },
    {
      "epoch": 0.05547850208044383,
      "grad_norm": 0.42582884430885315,
      "learning_rate": 1.3869625520110956e-07,
      "loss": 1.1139,
      "step": 1000
    },
    {
      "epoch": 0.056588072122052704,
      "grad_norm": 0.3357681930065155,
      "learning_rate": 1.4147018030513176e-07,
      "loss": 1.1757,
      "step": 1020
    },
    {
      "epoch": 0.05769764216366158,
      "grad_norm": 0.2897900938987732,
      "learning_rate": 1.4424410540915393e-07,
      "loss": 1.2098,
      "step": 1040
    },
    {
      "epoch": 0.05880721220527046,
      "grad_norm": 0.38025063276290894,
      "learning_rate": 1.4701803051317613e-07,
      "loss": 1.1234,
      "step": 1060
    },
    {
      "epoch": 0.05991678224687933,
      "grad_norm": 0.36913609504699707,
      "learning_rate": 1.4979195561719833e-07,
      "loss": 1.1552,
      "step": 1080
    },
    {
      "epoch": 0.06102635228848821,
      "grad_norm": 0.5480925440788269,
      "learning_rate": 1.525658807212205e-07,
      "loss": 1.135,
      "step": 1100
    },
    {
      "epoch": 0.062135922330097085,
      "grad_norm": 0.3799718916416168,
      "learning_rate": 1.553398058252427e-07,
      "loss": 1.1454,
      "step": 1120
    },
    {
      "epoch": 0.06324549237170596,
      "grad_norm": 0.3841489255428314,
      "learning_rate": 1.5811373092926493e-07,
      "loss": 1.1351,
      "step": 1140
    },
    {
      "epoch": 0.06435506241331485,
      "grad_norm": 0.29666101932525635,
      "learning_rate": 1.608876560332871e-07,
      "loss": 1.1616,
      "step": 1160
    },
    {
      "epoch": 0.06546463245492372,
      "grad_norm": 0.2889375686645508,
      "learning_rate": 1.636615811373093e-07,
      "loss": 1.1802,
      "step": 1180
    },
    {
      "epoch": 0.06657420249653259,
      "grad_norm": 0.5737839937210083,
      "learning_rate": 1.664355062413315e-07,
      "loss": 1.1667,
      "step": 1200
    },
    {
      "epoch": 0.06768377253814147,
      "grad_norm": 0.5003082156181335,
      "learning_rate": 1.6920943134535367e-07,
      "loss": 1.075,
      "step": 1220
    },
    {
      "epoch": 0.06879334257975035,
      "grad_norm": 0.46454185247421265,
      "learning_rate": 1.7198335644937587e-07,
      "loss": 1.1724,
      "step": 1240
    },
    {
      "epoch": 0.06990291262135923,
      "grad_norm": 0.32240554690361023,
      "learning_rate": 1.7475728155339807e-07,
      "loss": 1.1437,
      "step": 1260
    },
    {
      "epoch": 0.0710124826629681,
      "grad_norm": 0.42182767391204834,
      "learning_rate": 1.7753120665742024e-07,
      "loss": 0.9911,
      "step": 1280
    },
    {
      "epoch": 0.07212205270457697,
      "grad_norm": 0.4385708272457123,
      "learning_rate": 1.8030513176144244e-07,
      "loss": 1.0787,
      "step": 1300
    },
    {
      "epoch": 0.07323162274618586,
      "grad_norm": 0.3282943367958069,
      "learning_rate": 1.8307905686546463e-07,
      "loss": 1.1217,
      "step": 1320
    },
    {
      "epoch": 0.07434119278779473,
      "grad_norm": 0.7223221063613892,
      "learning_rate": 1.858529819694868e-07,
      "loss": 1.1546,
      "step": 1340
    },
    {
      "epoch": 0.0754507628294036,
      "grad_norm": 0.36656028032302856,
      "learning_rate": 1.88626907073509e-07,
      "loss": 1.163,
      "step": 1360
    },
    {
      "epoch": 0.07656033287101248,
      "grad_norm": 0.5122601389884949,
      "learning_rate": 1.914008321775312e-07,
      "loss": 1.053,
      "step": 1380
    },
    {
      "epoch": 0.07766990291262135,
      "grad_norm": 0.5633952021598816,
      "learning_rate": 1.9417475728155338e-07,
      "loss": 1.1306,
      "step": 1400
    },
    {
      "epoch": 0.07877947295423024,
      "grad_norm": 0.5684695243835449,
      "learning_rate": 1.9694868238557558e-07,
      "loss": 1.1589,
      "step": 1420
    },
    {
      "epoch": 0.07988904299583911,
      "grad_norm": 0.6536048054695129,
      "learning_rate": 1.9972260748959777e-07,
      "loss": 1.113,
      "step": 1440
    },
    {
      "epoch": 0.08099861303744799,
      "grad_norm": 0.4558582901954651,
      "learning_rate": 2.0249653259361995e-07,
      "loss": 1.0757,
      "step": 1460
    },
    {
      "epoch": 0.08210818307905686,
      "grad_norm": 0.6440749168395996,
      "learning_rate": 2.0527045769764214e-07,
      "loss": 1.1674,
      "step": 1480
    },
    {
      "epoch": 0.08321775312066575,
      "grad_norm": 0.41806939244270325,
      "learning_rate": 2.0804438280166434e-07,
      "loss": 1.0539,
      "step": 1500
    },
    {
      "epoch": 0.08432732316227462,
      "grad_norm": 0.5335156321525574,
      "learning_rate": 2.1081830790568652e-07,
      "loss": 1.1338,
      "step": 1520
    },
    {
      "epoch": 0.0854368932038835,
      "grad_norm": 0.35202693939208984,
      "learning_rate": 2.1359223300970871e-07,
      "loss": 1.1638,
      "step": 1540
    },
    {
      "epoch": 0.08654646324549237,
      "grad_norm": 0.45304250717163086,
      "learning_rate": 2.163661581137309e-07,
      "loss": 1.0912,
      "step": 1560
    },
    {
      "epoch": 0.08765603328710125,
      "grad_norm": 0.37071916460990906,
      "learning_rate": 2.191400832177531e-07,
      "loss": 1.0883,
      "step": 1580
    },
    {
      "epoch": 0.08876560332871013,
      "grad_norm": 0.5757469534873962,
      "learning_rate": 2.219140083217753e-07,
      "loss": 1.0081,
      "step": 1600
    },
    {
      "epoch": 0.089875173370319,
      "grad_norm": 0.4922785460948944,
      "learning_rate": 2.246879334257975e-07,
      "loss": 1.0775,
      "step": 1620
    },
    {
      "epoch": 0.09098474341192787,
      "grad_norm": 0.6358697414398193,
      "learning_rate": 2.274618585298197e-07,
      "loss": 1.1156,
      "step": 1640
    },
    {
      "epoch": 0.09209431345353676,
      "grad_norm": 0.33512547612190247,
      "learning_rate": 2.3023578363384188e-07,
      "loss": 1.049,
      "step": 1660
    },
    {
      "epoch": 0.09320388349514563,
      "grad_norm": 0.3588186502456665,
      "learning_rate": 2.3300970873786408e-07,
      "loss": 1.0028,
      "step": 1680
    },
    {
      "epoch": 0.09431345353675451,
      "grad_norm": 0.4455154836177826,
      "learning_rate": 2.3578363384188628e-07,
      "loss": 0.9894,
      "step": 1700
    },
    {
      "epoch": 0.09542302357836338,
      "grad_norm": 0.4005114734172821,
      "learning_rate": 2.3855755894590845e-07,
      "loss": 1.0441,
      "step": 1720
    },
    {
      "epoch": 0.09653259361997225,
      "grad_norm": 0.3630480170249939,
      "learning_rate": 2.413314840499306e-07,
      "loss": 1.1274,
      "step": 1740
    },
    {
      "epoch": 0.09764216366158114,
      "grad_norm": 0.38374799489974976,
      "learning_rate": 2.4410540915395285e-07,
      "loss": 1.0205,
      "step": 1760
    },
    {
      "epoch": 0.09875173370319001,
      "grad_norm": 0.240658700466156,
      "learning_rate": 2.46879334257975e-07,
      "loss": 0.9917,
      "step": 1780
    },
    {
      "epoch": 0.09986130374479889,
      "grad_norm": 0.32337549328804016,
      "learning_rate": 2.496532593619972e-07,
      "loss": 1.0758,
      "step": 1800
    },
    {
      "epoch": 0.10097087378640776,
      "grad_norm": 0.47185397148132324,
      "learning_rate": 2.524271844660194e-07,
      "loss": 1.1119,
      "step": 1820
    },
    {
      "epoch": 0.10208044382801665,
      "grad_norm": 0.5956501960754395,
      "learning_rate": 2.552011095700416e-07,
      "loss": 1.043,
      "step": 1840
    },
    {
      "epoch": 0.10319001386962552,
      "grad_norm": 0.36230626702308655,
      "learning_rate": 2.5797503467406376e-07,
      "loss": 1.0284,
      "step": 1860
    },
    {
      "epoch": 0.1042995839112344,
      "grad_norm": 0.4904063642024994,
      "learning_rate": 2.60748959778086e-07,
      "loss": 1.0458,
      "step": 1880
    },
    {
      "epoch": 0.10540915395284327,
      "grad_norm": 0.3035784661769867,
      "learning_rate": 2.6352288488210816e-07,
      "loss": 1.0256,
      "step": 1900
    },
    {
      "epoch": 0.10651872399445216,
      "grad_norm": 0.5605130791664124,
      "learning_rate": 2.6629680998613033e-07,
      "loss": 1.0233,
      "step": 1920
    },
    {
      "epoch": 0.10762829403606103,
      "grad_norm": 0.3493014872074127,
      "learning_rate": 2.6907073509015255e-07,
      "loss": 1.0023,
      "step": 1940
    },
    {
      "epoch": 0.1087378640776699,
      "grad_norm": 0.5957789421081543,
      "learning_rate": 2.7184466019417473e-07,
      "loss": 1.0315,
      "step": 1960
    },
    {
      "epoch": 0.10984743411927878,
      "grad_norm": 0.6720208525657654,
      "learning_rate": 2.746185852981969e-07,
      "loss": 1.0138,
      "step": 1980
    },
    {
      "epoch": 0.11095700416088766,
      "grad_norm": 0.43460479378700256,
      "learning_rate": 2.773925104022191e-07,
      "loss": 0.9863,
      "step": 2000
    },
    {
      "epoch": 0.11206657420249654,
      "grad_norm": 0.5312954783439636,
      "learning_rate": 2.801664355062413e-07,
      "loss": 1.0486,
      "step": 2020
    },
    {
      "epoch": 0.11317614424410541,
      "grad_norm": 0.6359843611717224,
      "learning_rate": 2.829403606102635e-07,
      "loss": 0.9457,
      "step": 2040
    },
    {
      "epoch": 0.11428571428571428,
      "grad_norm": 0.6254010796546936,
      "learning_rate": 2.857142857142857e-07,
      "loss": 1.1109,
      "step": 2060
    },
    {
      "epoch": 0.11539528432732316,
      "grad_norm": 0.36780887842178345,
      "learning_rate": 2.8848821081830787e-07,
      "loss": 1.0425,
      "step": 2080
    },
    {
      "epoch": 0.11650485436893204,
      "grad_norm": 0.46995213627815247,
      "learning_rate": 2.912621359223301e-07,
      "loss": 0.9666,
      "step": 2100
    },
    {
      "epoch": 0.11761442441054092,
      "grad_norm": 0.49816495180130005,
      "learning_rate": 2.9403606102635226e-07,
      "loss": 0.9935,
      "step": 2120
    },
    {
      "epoch": 0.11872399445214979,
      "grad_norm": 0.9470138549804688,
      "learning_rate": 2.9680998613037444e-07,
      "loss": 1.0937,
      "step": 2140
    },
    {
      "epoch": 0.11983356449375866,
      "grad_norm": 0.4583646357059479,
      "learning_rate": 2.9958391123439666e-07,
      "loss": 0.9972,
      "step": 2160
    },
    {
      "epoch": 0.12094313453536755,
      "grad_norm": 0.351114958524704,
      "learning_rate": 3.0235783633841883e-07,
      "loss": 1.0061,
      "step": 2180
    },
    {
      "epoch": 0.12205270457697642,
      "grad_norm": 0.4335211515426636,
      "learning_rate": 3.05131761442441e-07,
      "loss": 1.0091,
      "step": 2200
    },
    {
      "epoch": 0.1231622746185853,
      "grad_norm": 0.3483923673629761,
      "learning_rate": 3.0790568654646323e-07,
      "loss": 1.0092,
      "step": 2220
    },
    {
      "epoch": 0.12427184466019417,
      "grad_norm": 0.33233171701431274,
      "learning_rate": 3.106796116504854e-07,
      "loss": 1.0185,
      "step": 2240
    },
    {
      "epoch": 0.12538141470180306,
      "grad_norm": 0.34460940957069397,
      "learning_rate": 3.1345353675450763e-07,
      "loss": 0.9484,
      "step": 2260
    },
    {
      "epoch": 0.12649098474341192,
      "grad_norm": 0.42694029211997986,
      "learning_rate": 3.1622746185852985e-07,
      "loss": 0.9484,
      "step": 2280
    },
    {
      "epoch": 0.1276005547850208,
      "grad_norm": 0.5112186074256897,
      "learning_rate": 3.19001386962552e-07,
      "loss": 1.0039,
      "step": 2300
    },
    {
      "epoch": 0.1287101248266297,
      "grad_norm": 0.4560784101486206,
      "learning_rate": 3.217753120665742e-07,
      "loss": 0.9961,
      "step": 2320
    },
    {
      "epoch": 0.12981969486823855,
      "grad_norm": 0.5053315162658691,
      "learning_rate": 3.245492371705964e-07,
      "loss": 0.9542,
      "step": 2340
    },
    {
      "epoch": 0.13092926490984744,
      "grad_norm": 0.6796769499778748,
      "learning_rate": 3.273231622746186e-07,
      "loss": 0.9187,
      "step": 2360
    },
    {
      "epoch": 0.13203883495145632,
      "grad_norm": 0.7734983563423157,
      "learning_rate": 3.3009708737864077e-07,
      "loss": 0.9298,
      "step": 2380
    },
    {
      "epoch": 0.13314840499306518,
      "grad_norm": 0.4399431049823761,
      "learning_rate": 3.32871012482663e-07,
      "loss": 0.8959,
      "step": 2400
    },
    {
      "epoch": 0.13425797503467407,
      "grad_norm": 0.4783932864665985,
      "learning_rate": 3.3564493758668516e-07,
      "loss": 0.8539,
      "step": 2420
    },
    {
      "epoch": 0.13536754507628293,
      "grad_norm": 0.4672847092151642,
      "learning_rate": 3.3841886269070734e-07,
      "loss": 0.8839,
      "step": 2440
    },
    {
      "epoch": 0.13647711511789182,
      "grad_norm": 0.4219910204410553,
      "learning_rate": 3.4119278779472956e-07,
      "loss": 0.934,
      "step": 2460
    },
    {
      "epoch": 0.1375866851595007,
      "grad_norm": 0.3283788561820984,
      "learning_rate": 3.4396671289875173e-07,
      "loss": 0.8729,
      "step": 2480
    },
    {
      "epoch": 0.13869625520110956,
      "grad_norm": 0.6127363443374634,
      "learning_rate": 3.467406380027739e-07,
      "loss": 0.8355,
      "step": 2500
    },
    {
      "epoch": 0.13980582524271845,
      "grad_norm": 0.8837600350379944,
      "learning_rate": 3.4951456310679613e-07,
      "loss": 0.9256,
      "step": 2520
    },
    {
      "epoch": 0.1409153952843273,
      "grad_norm": 0.3368714153766632,
      "learning_rate": 3.522884882108183e-07,
      "loss": 0.8326,
      "step": 2540
    },
    {
      "epoch": 0.1420249653259362,
      "grad_norm": 0.6457244753837585,
      "learning_rate": 3.550624133148405e-07,
      "loss": 0.8638,
      "step": 2560
    },
    {
      "epoch": 0.14313453536754508,
      "grad_norm": 0.5497669577598572,
      "learning_rate": 3.578363384188627e-07,
      "loss": 0.8049,
      "step": 2580
    },
    {
      "epoch": 0.14424410540915394,
      "grad_norm": 0.5958977341651917,
      "learning_rate": 3.6061026352288487e-07,
      "loss": 0.8107,
      "step": 2600
    },
    {
      "epoch": 0.14535367545076283,
      "grad_norm": 0.5878711938858032,
      "learning_rate": 3.6338418862690704e-07,
      "loss": 0.8011,
      "step": 2620
    },
    {
      "epoch": 0.14646324549237172,
      "grad_norm": 0.4262014627456665,
      "learning_rate": 3.6615811373092927e-07,
      "loss": 0.898,
      "step": 2640
    },
    {
      "epoch": 0.14757281553398058,
      "grad_norm": 0.7306149005889893,
      "learning_rate": 3.6893203883495144e-07,
      "loss": 0.7726,
      "step": 2660
    },
    {
      "epoch": 0.14868238557558947,
      "grad_norm": 0.32822510600090027,
      "learning_rate": 3.717059639389736e-07,
      "loss": 0.775,
      "step": 2680
    },
    {
      "epoch": 0.14979195561719832,
      "grad_norm": 0.41548779606819153,
      "learning_rate": 3.7447988904299584e-07,
      "loss": 0.7613,
      "step": 2700
    },
    {
      "epoch": 0.1509015256588072,
      "grad_norm": 1.1839288473129272,
      "learning_rate": 3.77253814147018e-07,
      "loss": 0.8496,
      "step": 2720
    },
    {
      "epoch": 0.1520110957004161,
      "grad_norm": 0.5219757556915283,
      "learning_rate": 3.800277392510402e-07,
      "loss": 0.7402,
      "step": 2740
    },
    {
      "epoch": 0.15312066574202496,
      "grad_norm": 0.8173393607139587,
      "learning_rate": 3.828016643550624e-07,
      "loss": 0.7204,
      "step": 2760
    },
    {
      "epoch": 0.15423023578363385,
      "grad_norm": 0.49754881858825684,
      "learning_rate": 3.855755894590846e-07,
      "loss": 0.7716,
      "step": 2780
    },
    {
      "epoch": 0.1553398058252427,
      "grad_norm": 0.39697808027267456,
      "learning_rate": 3.8834951456310675e-07,
      "loss": 0.7791,
      "step": 2800
    },
    {
      "epoch": 0.1564493758668516,
      "grad_norm": 0.6214376091957092,
      "learning_rate": 3.91123439667129e-07,
      "loss": 0.6724,
      "step": 2820
    },
    {
      "epoch": 0.15755894590846048,
      "grad_norm": 0.6486151218414307,
      "learning_rate": 3.9389736477115115e-07,
      "loss": 0.8015,
      "step": 2840
    },
    {
      "epoch": 0.15866851595006934,
      "grad_norm": 0.5499553084373474,
      "learning_rate": 3.966712898751733e-07,
      "loss": 0.7871,
      "step": 2860
    },
    {
      "epoch": 0.15977808599167823,
      "grad_norm": 0.8797757029533386,
      "learning_rate": 3.9944521497919555e-07,
      "loss": 0.7183,
      "step": 2880
    },
    {
      "epoch": 0.1608876560332871,
      "grad_norm": 0.47135302424430847,
      "learning_rate": 4.022191400832177e-07,
      "loss": 0.7348,
      "step": 2900
    },
    {
      "epoch": 0.16199722607489597,
      "grad_norm": 0.8005576729774475,
      "learning_rate": 4.049930651872399e-07,
      "loss": 0.6212,
      "step": 2920
    },
    {
      "epoch": 0.16310679611650486,
      "grad_norm": 0.47837623953819275,
      "learning_rate": 4.077669902912621e-07,
      "loss": 0.6812,
      "step": 2940
    },
    {
      "epoch": 0.16421636615811372,
      "grad_norm": 0.36638781428337097,
      "learning_rate": 4.105409153952843e-07,
      "loss": 0.6925,
      "step": 2960
    },
    {
      "epoch": 0.1653259361997226,
      "grad_norm": 0.817538857460022,
      "learning_rate": 4.1331484049930646e-07,
      "loss": 0.6186,
      "step": 2980
    },
    {
      "epoch": 0.1664355062413315,
      "grad_norm": 0.5090010166168213,
      "learning_rate": 4.160887656033287e-07,
      "loss": 0.6844,
      "step": 3000
    },
    {
      "epoch": 0.16754507628294035,
      "grad_norm": 0.6102781295776367,
      "learning_rate": 4.1886269070735086e-07,
      "loss": 0.6943,
      "step": 3020
    },
    {
      "epoch": 0.16865464632454924,
      "grad_norm": 0.8231751918792725,
      "learning_rate": 4.2163661581137303e-07,
      "loss": 0.6391,
      "step": 3040
    },
    {
      "epoch": 0.16976421636615813,
      "grad_norm": 0.38910776376724243,
      "learning_rate": 4.2441054091539526e-07,
      "loss": 0.6937,
      "step": 3060
    },
    {
      "epoch": 0.170873786407767,
      "grad_norm": 0.5838291049003601,
      "learning_rate": 4.2718446601941743e-07,
      "loss": 0.5791,
      "step": 3080
    },
    {
      "epoch": 0.17198335644937587,
      "grad_norm": 0.519530177116394,
      "learning_rate": 4.299583911234396e-07,
      "loss": 0.7412,
      "step": 3100
    },
    {
      "epoch": 0.17309292649098473,
      "grad_norm": 0.45696595311164856,
      "learning_rate": 4.327323162274618e-07,
      "loss": 0.6469,
      "step": 3120
    },
    {
      "epoch": 0.17420249653259362,
      "grad_norm": 0.6771582961082458,
      "learning_rate": 4.35506241331484e-07,
      "loss": 0.6441,
      "step": 3140
    },
    {
      "epoch": 0.1753120665742025,
      "grad_norm": 0.559917151927948,
      "learning_rate": 4.382801664355062e-07,
      "loss": 0.5778,
      "step": 3160
    },
    {
      "epoch": 0.17642163661581137,
      "grad_norm": 0.9249961376190186,
      "learning_rate": 4.4105409153952845e-07,
      "loss": 0.6637,
      "step": 3180
    },
    {
      "epoch": 0.17753120665742025,
      "grad_norm": 0.5211077928543091,
      "learning_rate": 4.438280166435506e-07,
      "loss": 0.7047,
      "step": 3200
    },
    {
      "epoch": 0.1786407766990291,
      "grad_norm": 0.7488894462585449,
      "learning_rate": 4.4660194174757285e-07,
      "loss": 0.5802,
      "step": 3220
    },
    {
      "epoch": 0.179750346740638,
      "grad_norm": 0.6046866774559021,
      "learning_rate": 4.49375866851595e-07,
      "loss": 0.6601,
      "step": 3240
    },
    {
      "epoch": 0.1808599167822469,
      "grad_norm": 0.3715108036994934,
      "learning_rate": 4.521497919556172e-07,
      "loss": 0.6094,
      "step": 3260
    },
    {
      "epoch": 0.18196948682385575,
      "grad_norm": 0.5831759572029114,
      "learning_rate": 4.549237170596394e-07,
      "loss": 0.6086,
      "step": 3280
    },
    {
      "epoch": 0.18307905686546463,
      "grad_norm": 0.595746636390686,
      "learning_rate": 4.576976421636616e-07,
      "loss": 0.5916,
      "step": 3300
    },
    {
      "epoch": 0.18418862690707352,
      "grad_norm": 0.48339492082595825,
      "learning_rate": 4.6047156726768376e-07,
      "loss": 0.6127,
      "step": 3320
    },
    {
      "epoch": 0.18529819694868238,
      "grad_norm": 1.626143455505371,
      "learning_rate": 4.63245492371706e-07,
      "loss": 0.5437,
      "step": 3340
    },
    {
      "epoch": 0.18640776699029127,
      "grad_norm": 0.3789680004119873,
      "learning_rate": 4.6601941747572816e-07,
      "loss": 0.6499,
      "step": 3360
    },
    {
      "epoch": 0.18751733703190013,
      "grad_norm": 0.5178479552268982,
      "learning_rate": 4.6879334257975033e-07,
      "loss": 0.6509,
      "step": 3380
    },
    {
      "epoch": 0.18862690707350901,
      "grad_norm": 0.5709561109542847,
      "learning_rate": 4.7156726768377255e-07,
      "loss": 0.6429,
      "step": 3400
    },
    {
      "epoch": 0.1897364771151179,
      "grad_norm": 0.3643471896648407,
      "learning_rate": 4.743411927877947e-07,
      "loss": 0.6235,
      "step": 3420
    },
    {
      "epoch": 0.19084604715672676,
      "grad_norm": 0.5804113745689392,
      "learning_rate": 4.771151178918169e-07,
      "loss": 0.7341,
      "step": 3440
    },
    {
      "epoch": 0.19195561719833565,
      "grad_norm": 0.5089621543884277,
      "learning_rate": 4.798890429958391e-07,
      "loss": 0.6286,
      "step": 3460
    },
    {
      "epoch": 0.1930651872399445,
      "grad_norm": 0.4625658392906189,
      "learning_rate": 4.826629680998612e-07,
      "loss": 0.6128,
      "step": 3480
    },
    {
      "epoch": 0.1941747572815534,
      "grad_norm": 0.36961832642555237,
      "learning_rate": 4.854368932038835e-07,
      "loss": 0.6213,
      "step": 3500
    },
    {
      "epoch": 0.19528432732316228,
      "grad_norm": 0.4466856122016907,
      "learning_rate": 4.882108183079057e-07,
      "loss": 0.5383,
      "step": 3520
    },
    {
      "epoch": 0.19639389736477114,
      "grad_norm": 0.45287024974823,
      "learning_rate": 4.909847434119279e-07,
      "loss": 0.5064,
      "step": 3540
    },
    {
      "epoch": 0.19750346740638003,
      "grad_norm": 0.6351368427276611,
      "learning_rate": 4.9375866851595e-07,
      "loss": 0.5485,
      "step": 3560
    },
    {
      "epoch": 0.19861303744798892,
      "grad_norm": 0.46472978591918945,
      "learning_rate": 4.965325936199722e-07,
      "loss": 0.5392,
      "step": 3580
    },
    {
      "epoch": 0.19972260748959778,
      "grad_norm": 0.38963034749031067,
      "learning_rate": 4.993065187239944e-07,
      "loss": 0.5459,
      "step": 3600
    },
    {
      "epoch": 0.20083217753120666,
      "grad_norm": 2.1769580841064453,
      "learning_rate": 4.999986650611594e-07,
      "loss": 0.5158,
      "step": 3620
    },
    {
      "epoch": 0.20194174757281552,
      "grad_norm": 0.6485143899917603,
      "learning_rate": 4.999927320283929e-07,
      "loss": 0.5815,
      "step": 3640
    },
    {
      "epoch": 0.2030513176144244,
      "grad_norm": 0.37338986992836,
      "learning_rate": 4.999820526876891e-07,
      "loss": 0.5475,
      "step": 3660
    },
    {
      "epoch": 0.2041608876560333,
      "grad_norm": 0.4554106593132019,
      "learning_rate": 4.999666272418033e-07,
      "loss": 0.547,
      "step": 3680
    },
    {
      "epoch": 0.20527045769764216,
      "grad_norm": 0.3950905501842499,
      "learning_rate": 4.999464559835997e-07,
      "loss": 0.5561,
      "step": 3700
    },
    {
      "epoch": 0.20638002773925104,
      "grad_norm": 0.28335675597190857,
      "learning_rate": 4.999215392960455e-07,
      "loss": 0.6461,
      "step": 3720
    },
    {
      "epoch": 0.20748959778085993,
      "grad_norm": 0.3045244514942169,
      "learning_rate": 4.998918776522036e-07,
      "loss": 0.5206,
      "step": 3740
    },
    {
      "epoch": 0.2085991678224688,
      "grad_norm": 0.4339936375617981,
      "learning_rate": 4.998574716152234e-07,
      "loss": 0.4728,
      "step": 3760
    },
    {
      "epoch": 0.20970873786407768,
      "grad_norm": 0.39513078331947327,
      "learning_rate": 4.998183218383305e-07,
      "loss": 0.5485,
      "step": 3780
    },
    {
      "epoch": 0.21081830790568654,
      "grad_norm": 0.40521058440208435,
      "learning_rate": 4.997744290648143e-07,
      "loss": 0.6388,
      "step": 3800
    },
    {
      "epoch": 0.21192787794729542,
      "grad_norm": 0.3975263833999634,
      "learning_rate": 4.997257941280133e-07,
      "loss": 0.5521,
      "step": 3820
    },
    {
      "epoch": 0.2130374479889043,
      "grad_norm": 0.3691408634185791,
      "learning_rate": 4.996724179512999e-07,
      "loss": 0.5293,
      "step": 3840
    },
    {
      "epoch": 0.21414701803051317,
      "grad_norm": 0.30931538343429565,
      "learning_rate": 4.996143015480629e-07,
      "loss": 0.6779,
      "step": 3860
    },
    {
      "epoch": 0.21525658807212206,
      "grad_norm": 0.4070769250392914,
      "learning_rate": 4.995514460216873e-07,
      "loss": 0.4724,
      "step": 3880
    },
    {
      "epoch": 0.21636615811373092,
      "grad_norm": 0.34178927540779114,
      "learning_rate": 4.994838525655349e-07,
      "loss": 0.4932,
      "step": 3900
    },
    {
      "epoch": 0.2174757281553398,
      "grad_norm": 0.3776053190231323,
      "learning_rate": 4.994115224629204e-07,
      "loss": 0.513,
      "step": 3920
    },
    {
      "epoch": 0.2185852981969487,
      "grad_norm": 0.3009076416492462,
      "learning_rate": 4.993344570870874e-07,
      "loss": 0.4694,
      "step": 3940
    },
    {
      "epoch": 0.21969486823855755,
      "grad_norm": 0.24924315512180328,
      "learning_rate": 4.992526579011823e-07,
      "loss": 0.5135,
      "step": 3960
    },
    {
      "epoch": 0.22080443828016644,
      "grad_norm": 0.39235708117485046,
      "learning_rate": 4.991661264582271e-07,
      "loss": 0.5608,
      "step": 3980
    },
    {
      "epoch": 0.22191400832177532,
      "grad_norm": 0.34350383281707764,
      "learning_rate": 4.990748644010888e-07,
      "loss": 0.5201,
      "step": 4000
    },
    {
      "epoch": 0.22302357836338418,
      "grad_norm": 0.5332874059677124,
      "learning_rate": 4.989788734624492e-07,
      "loss": 0.5994,
      "step": 4020
    },
    {
      "epoch": 0.22413314840499307,
      "grad_norm": 0.32011643052101135,
      "learning_rate": 4.988781554647714e-07,
      "loss": 0.5103,
      "step": 4040
    },
    {
      "epoch": 0.22524271844660193,
      "grad_norm": 0.37372103333473206,
      "learning_rate": 4.987727123202655e-07,
      "loss": 0.5483,
      "step": 4060
    },
    {
      "epoch": 0.22635228848821082,
      "grad_norm": 0.3511541187763214,
      "learning_rate": 4.986625460308524e-07,
      "loss": 0.5508,
      "step": 4080
    },
    {
      "epoch": 0.2274618585298197,
      "grad_norm": 0.33943256735801697,
      "learning_rate": 4.985476586881254e-07,
      "loss": 0.5437,
      "step": 4100
    },
    {
      "epoch": 0.22857142857142856,
      "grad_norm": 0.5390433669090271,
      "learning_rate": 4.984280524733107e-07,
      "loss": 0.5161,
      "step": 4120
    },
    {
      "epoch": 0.22968099861303745,
      "grad_norm": 0.40753117203712463,
      "learning_rate": 4.983037296572259e-07,
      "loss": 0.4993,
      "step": 4140
    },
    {
      "epoch": 0.2307905686546463,
      "grad_norm": 0.34956789016723633,
      "learning_rate": 4.981746926002372e-07,
      "loss": 0.613,
      "step": 4160
    },
    {
      "epoch": 0.2319001386962552,
      "grad_norm": 0.3175369203090668,
      "learning_rate": 4.980409437522143e-07,
      "loss": 0.5396,
      "step": 4180
    },
    {
      "epoch": 0.23300970873786409,
      "grad_norm": 0.30835628509521484,
      "learning_rate": 4.979024856524839e-07,
      "loss": 0.5407,
      "step": 4200
    },
    {
      "epoch": 0.23411927877947294,
      "grad_norm": 0.3639371693134308,
      "learning_rate": 4.977593209297814e-07,
      "loss": 0.5457,
      "step": 4220
    },
    {
      "epoch": 0.23522884882108183,
      "grad_norm": 0.4240809381008148,
      "learning_rate": 4.976114523022015e-07,
      "loss": 0.5323,
      "step": 4240
    },
    {
      "epoch": 0.23633841886269072,
      "grad_norm": 0.509510338306427,
      "learning_rate": 4.974588825771457e-07,
      "loss": 0.5374,
      "step": 4260
    },
    {
      "epoch": 0.23744798890429958,
      "grad_norm": 0.3038425147533417,
      "learning_rate": 4.9730161465127e-07,
      "loss": 0.5041,
      "step": 4280
    },
    {
      "epoch": 0.23855755894590847,
      "grad_norm": 0.33369964361190796,
      "learning_rate": 4.971396515104292e-07,
      "loss": 0.601,
      "step": 4300
    },
    {
      "epoch": 0.23966712898751732,
      "grad_norm": 0.4764558970928192,
      "learning_rate": 4.969729962296203e-07,
      "loss": 0.6066,
      "step": 4320
    },
    {
      "epoch": 0.2407766990291262,
      "grad_norm": 0.4532679319381714,
      "learning_rate": 4.968016519729246e-07,
      "loss": 0.5999,
      "step": 4340
    },
    {
      "epoch": 0.2418862690707351,
      "grad_norm": 0.39469200372695923,
      "learning_rate": 4.966256219934471e-07,
      "loss": 0.498,
      "step": 4360
    },
    {
      "epoch": 0.24299583911234396,
      "grad_norm": 0.4191853702068329,
      "learning_rate": 4.964449096332547e-07,
      "loss": 0.5246,
      "step": 4380
    },
    {
      "epoch": 0.24410540915395285,
      "grad_norm": 0.3406555950641632,
      "learning_rate": 4.962595183233133e-07,
      "loss": 0.6331,
      "step": 4400
    },
    {
      "epoch": 0.24521497919556173,
      "grad_norm": 0.32338958978652954,
      "learning_rate": 4.960694515834224e-07,
      "loss": 0.5389,
      "step": 4420
    },
    {
      "epoch": 0.2463245492371706,
      "grad_norm": 0.8198554515838623,
      "learning_rate": 4.958747130221477e-07,
      "loss": 0.5678,
      "step": 4440
    },
    {
      "epoch": 0.24743411927877948,
      "grad_norm": 0.38297727704048157,
      "learning_rate": 4.956753063367537e-07,
      "loss": 0.4682,
      "step": 4460
    },
    {
      "epoch": 0.24854368932038834,
      "grad_norm": 0.36951327323913574,
      "learning_rate": 4.954712353131323e-07,
      "loss": 0.4903,
      "step": 4480
    },
    {
      "epoch": 0.24965325936199723,
      "grad_norm": 0.30048689246177673,
      "learning_rate": 4.952625038257321e-07,
      "loss": 0.6061,
      "step": 4500
    },
    {
      "epoch": 0.2507628294036061,
      "grad_norm": 0.2912724018096924,
      "learning_rate": 4.950491158374837e-07,
      "loss": 0.565,
      "step": 4520
    },
    {
      "epoch": 0.251872399445215,
      "grad_norm": 0.5042518377304077,
      "learning_rate": 4.948310753997254e-07,
      "loss": 0.5231,
      "step": 4540
    },
    {
      "epoch": 0.25298196948682383,
      "grad_norm": 0.31980255246162415,
      "learning_rate": 4.94608386652126e-07,
      "loss": 0.5077,
      "step": 4560
    },
    {
      "epoch": 0.2540915395284327,
      "grad_norm": 0.2944648861885071,
      "learning_rate": 4.943810538226056e-07,
      "loss": 0.4751,
      "step": 4580
    },
    {
      "epoch": 0.2552011095700416,
      "grad_norm": 0.24331466853618622,
      "learning_rate": 4.941490812272563e-07,
      "loss": 0.5061,
      "step": 4600
    },
    {
      "epoch": 0.2563106796116505,
      "grad_norm": 0.308380663394928,
      "learning_rate": 4.939124732702595e-07,
      "loss": 0.5207,
      "step": 4620
    },
    {
      "epoch": 0.2574202496532594,
      "grad_norm": 0.3826179802417755,
      "learning_rate": 4.936712344438028e-07,
      "loss": 0.5081,
      "step": 4640
    },
    {
      "epoch": 0.2585298196948682,
      "grad_norm": 0.41639629006385803,
      "learning_rate": 4.934253693279943e-07,
      "loss": 0.5334,
      "step": 4660
    },
    {
      "epoch": 0.2596393897364771,
      "grad_norm": 0.38280215859413147,
      "learning_rate": 4.931748825907759e-07,
      "loss": 0.5957,
      "step": 4680
    },
    {
      "epoch": 0.260748959778086,
      "grad_norm": 0.3500733971595764,
      "learning_rate": 4.929197789878347e-07,
      "loss": 0.5426,
      "step": 4700
    },
    {
      "epoch": 0.2618585298196949,
      "grad_norm": 0.4630734324455261,
      "learning_rate": 4.926600633625126e-07,
      "loss": 0.539,
      "step": 4720
    },
    {
      "epoch": 0.26296809986130376,
      "grad_norm": 0.3602588474750519,
      "learning_rate": 4.92395740645714e-07,
      "loss": 0.4219,
      "step": 4740
    },
    {
      "epoch": 0.26407766990291265,
      "grad_norm": 0.28921574354171753,
      "learning_rate": 4.92126815855813e-07,
      "loss": 0.5068,
      "step": 4760
    },
    {
      "epoch": 0.2651872399445215,
      "grad_norm": 0.298486590385437,
      "learning_rate": 4.918532940985576e-07,
      "loss": 0.5365,
      "step": 4780
    },
    {
      "epoch": 0.26629680998613037,
      "grad_norm": 0.32068467140197754,
      "learning_rate": 4.915751805669725e-07,
      "loss": 0.5623,
      "step": 4800
    },
    {
      "epoch": 0.26740638002773925,
      "grad_norm": 0.2494667023420334,
      "learning_rate": 4.912924805412613e-07,
      "loss": 0.5911,
      "step": 4820
    },
    {
      "epoch": 0.26851595006934814,
      "grad_norm": 0.3729526698589325,
      "learning_rate": 4.910051993887053e-07,
      "loss": 0.6284,
      "step": 4840
    },
    {
      "epoch": 0.26962552011095703,
      "grad_norm": 0.36454734206199646,
      "learning_rate": 4.907133425635625e-07,
      "loss": 0.5695,
      "step": 4860
    },
    {
      "epoch": 0.27073509015256586,
      "grad_norm": 0.3649137318134308,
      "learning_rate": 4.904169156069633e-07,
      "loss": 0.5287,
      "step": 4880
    },
    {
      "epoch": 0.27184466019417475,
      "grad_norm": 0.44556987285614014,
      "learning_rate": 4.90115924146806e-07,
      "loss": 0.5561,
      "step": 4900
    },
    {
      "epoch": 0.27295423023578363,
      "grad_norm": 0.3902558386325836,
      "learning_rate": 4.898103738976491e-07,
      "loss": 0.5358,
      "step": 4920
    },
    {
      "epoch": 0.2740638002773925,
      "grad_norm": 0.38199660181999207,
      "learning_rate": 4.895002706606037e-07,
      "loss": 0.5221,
      "step": 4940
    },
    {
      "epoch": 0.2751733703190014,
      "grad_norm": 0.7920161485671997,
      "learning_rate": 4.891856203232228e-07,
      "loss": 0.552,
      "step": 4960
    },
    {
      "epoch": 0.27628294036061024,
      "grad_norm": 0.40676286816596985,
      "learning_rate": 4.888664288593896e-07,
      "loss": 0.563,
      "step": 4980
    },
    {
      "epoch": 0.27739251040221913,
      "grad_norm": 0.2597495913505554,
      "learning_rate": 4.885427023292043e-07,
      "loss": 0.5276,
      "step": 5000
    },
    {
      "epoch": 0.278502080443828,
      "grad_norm": 0.408184289932251,
      "learning_rate": 4.882144468788685e-07,
      "loss": 0.4505,
      "step": 5020
    },
    {
      "epoch": 0.2796116504854369,
      "grad_norm": 0.31909486651420593,
      "learning_rate": 4.878816687405694e-07,
      "loss": 0.5883,
      "step": 5040
    },
    {
      "epoch": 0.2807212205270458,
      "grad_norm": 0.6439054012298584,
      "learning_rate": 4.875443742323607e-07,
      "loss": 0.5181,
      "step": 5060
    },
    {
      "epoch": 0.2818307905686546,
      "grad_norm": 0.3375921845436096,
      "learning_rate": 4.872025697580431e-07,
      "loss": 0.5607,
      "step": 5080
    },
    {
      "epoch": 0.2829403606102635,
      "grad_norm": 0.30383211374282837,
      "learning_rate": 4.868562618070422e-07,
      "loss": 0.517,
      "step": 5100
    },
    {
      "epoch": 0.2840499306518724,
      "grad_norm": 0.40573009848594666,
      "learning_rate": 4.865054569542859e-07,
      "loss": 0.5974,
      "step": 5120
    },
    {
      "epoch": 0.2851595006934813,
      "grad_norm": 0.2704174518585205,
      "learning_rate": 4.861501618600794e-07,
      "loss": 0.4676,
      "step": 5140
    },
    {
      "epoch": 0.28626907073509017,
      "grad_norm": 0.30422112345695496,
      "learning_rate": 4.857903832699784e-07,
      "loss": 0.5631,
      "step": 5160
    },
    {
      "epoch": 0.287378640776699,
      "grad_norm": 0.48548394441604614,
      "learning_rate": 4.854261280146615e-07,
      "loss": 0.6646,
      "step": 5180
    },
    {
      "epoch": 0.2884882108183079,
      "grad_norm": 0.49318018555641174,
      "learning_rate": 4.850574030097999e-07,
      "loss": 0.5939,
      "step": 5200
    },
    {
      "epoch": 0.2895977808599168,
      "grad_norm": 0.369650661945343,
      "learning_rate": 4.846842152559272e-07,
      "loss": 0.5602,
      "step": 5220
    },
    {
      "epoch": 0.29070735090152566,
      "grad_norm": 0.5085413455963135,
      "learning_rate": 4.843065718383051e-07,
      "loss": 0.5528,
      "step": 5240
    },
    {
      "epoch": 0.29181692094313455,
      "grad_norm": 0.30005863308906555,
      "learning_rate": 4.839244799267899e-07,
      "loss": 0.5668,
      "step": 5260
    },
    {
      "epoch": 0.29292649098474344,
      "grad_norm": 1.0485389232635498,
      "learning_rate": 4.83537946775696e-07,
      "loss": 0.5668,
      "step": 5280
    },
    {
      "epoch": 0.29403606102635227,
      "grad_norm": 0.38798120617866516,
      "learning_rate": 4.831469797236582e-07,
      "loss": 0.5526,
      "step": 5300
    },
    {
      "epoch": 0.29514563106796116,
      "grad_norm": 0.42610159516334534,
      "learning_rate": 4.827515861934924e-07,
      "loss": 0.5549,
      "step": 5320
    },
    {
      "epoch": 0.29625520110957004,
      "grad_norm": 0.2633446753025055,
      "learning_rate": 4.823517736920546e-07,
      "loss": 0.5283,
      "step": 5340
    },
    {
      "epoch": 0.29736477115117893,
      "grad_norm": 0.4579598903656006,
      "learning_rate": 4.819475498100985e-07,
      "loss": 0.5362,
      "step": 5360
    },
    {
      "epoch": 0.2984743411927878,
      "grad_norm": 0.39916613698005676,
      "learning_rate": 4.815389222221313e-07,
      "loss": 0.4562,
      "step": 5380
    },
    {
      "epoch": 0.29958391123439665,
      "grad_norm": 0.559020459651947,
      "learning_rate": 4.81125898686268e-07,
      "loss": 0.481,
      "step": 5400
    },
    {
      "epoch": 0.30069348127600554,
      "grad_norm": 0.40768253803253174,
      "learning_rate": 4.80708487044084e-07,
      "loss": 0.5346,
      "step": 5420
    },
    {
      "epoch": 0.3018030513176144,
      "grad_norm": 0.22490708529949188,
      "learning_rate": 4.802866952204667e-07,
      "loss": 0.5692,
      "step": 5440
    },
    {
      "epoch": 0.3029126213592233,
      "grad_norm": 0.3096957504749298,
      "learning_rate": 4.798605312234643e-07,
      "loss": 0.5559,
      "step": 5460
    },
    {
      "epoch": 0.3040221914008322,
      "grad_norm": 0.3028647303581238,
      "learning_rate": 4.794300031441342e-07,
      "loss": 0.5313,
      "step": 5480
    },
    {
      "epoch": 0.30513176144244103,
      "grad_norm": 0.7425801753997803,
      "learning_rate": 4.789951191563895e-07,
      "loss": 0.4875,
      "step": 5500
    },
    {
      "epoch": 0.3062413314840499,
      "grad_norm": 0.25657814741134644,
      "learning_rate": 4.785558875168434e-07,
      "loss": 0.4611,
      "step": 5520
    },
    {
      "epoch": 0.3073509015256588,
      "grad_norm": 0.374449759721756,
      "learning_rate": 4.781123165646529e-07,
      "loss": 0.5818,
      "step": 5540
    },
    {
      "epoch": 0.3084604715672677,
      "grad_norm": 0.3681221008300781,
      "learning_rate": 4.776644147213602e-07,
      "loss": 0.4757,
      "step": 5560
    },
    {
      "epoch": 0.3095700416088766,
      "grad_norm": 0.5280266404151917,
      "learning_rate": 4.772121904907328e-07,
      "loss": 0.4936,
      "step": 5580
    },
    {
      "epoch": 0.3106796116504854,
      "grad_norm": 0.447544664144516,
      "learning_rate": 4.7675565245860195e-07,
      "loss": 0.5231,
      "step": 5600
    },
    {
      "epoch": 0.3117891816920943,
      "grad_norm": 0.33150920271873474,
      "learning_rate": 4.7629480929270014e-07,
      "loss": 0.5644,
      "step": 5620
    },
    {
      "epoch": 0.3128987517337032,
      "grad_norm": 0.3428071141242981,
      "learning_rate": 4.7582966974249607e-07,
      "loss": 0.6091,
      "step": 5640
    },
    {
      "epoch": 0.31400832177531207,
      "grad_norm": 0.4063955545425415,
      "learning_rate": 4.753602426390285e-07,
      "loss": 0.5079,
      "step": 5660
    },
    {
      "epoch": 0.31511789181692096,
      "grad_norm": 0.25844889879226685,
      "learning_rate": 4.7488653689473903e-07,
      "loss": 0.6156,
      "step": 5680
    },
    {
      "epoch": 0.31622746185852985,
      "grad_norm": 0.2735048234462738,
      "learning_rate": 4.744085615033023e-07,
      "loss": 0.5386,
      "step": 5700
    },
    {
      "epoch": 0.3173370319001387,
      "grad_norm": 0.49888360500335693,
      "learning_rate": 4.739263255394559e-07,
      "loss": 0.5374,
      "step": 5720
    },
    {
      "epoch": 0.31844660194174756,
      "grad_norm": 0.26587429642677307,
      "learning_rate": 4.734398381588274e-07,
      "loss": 0.5424,
      "step": 5740
    },
    {
      "epoch": 0.31955617198335645,
      "grad_norm": 0.28447648882865906,
      "learning_rate": 4.7294910859776095e-07,
      "loss": 0.5161,
      "step": 5760
    },
    {
      "epoch": 0.32066574202496534,
      "grad_norm": 0.3417870104312897,
      "learning_rate": 4.7245414617314193e-07,
      "loss": 0.4308,
      "step": 5780
    },
    {
      "epoch": 0.3217753120665742,
      "grad_norm": 0.3430401086807251,
      "learning_rate": 4.719549602822199e-07,
      "loss": 0.5222,
      "step": 5800
    },
    {
      "epoch": 0.32288488210818306,
      "grad_norm": 0.31506893038749695,
      "learning_rate": 4.7145156040243017e-07,
      "loss": 0.4937,
      "step": 5820
    },
    {
      "epoch": 0.32399445214979194,
      "grad_norm": 0.327404648065567,
      "learning_rate": 4.709439560912139e-07,
      "loss": 0.5163,
      "step": 5840
    },
    {
      "epoch": 0.32510402219140083,
      "grad_norm": 0.291507750749588,
      "learning_rate": 4.704321569858368e-07,
      "loss": 0.4774,
      "step": 5860
    },
    {
      "epoch": 0.3262135922330097,
      "grad_norm": 0.2654714286327362,
      "learning_rate": 4.6991617280320614e-07,
      "loss": 0.4485,
      "step": 5880
    },
    {
      "epoch": 0.3273231622746186,
      "grad_norm": 0.38568347692489624,
      "learning_rate": 4.6939601333968583e-07,
      "loss": 0.5054,
      "step": 5900
    },
    {
      "epoch": 0.32843273231622744,
      "grad_norm": 0.28820493817329407,
      "learning_rate": 4.6887168847091085e-07,
      "loss": 0.5271,
      "step": 5920
    },
    {
      "epoch": 0.3295423023578363,
      "grad_norm": 0.2675837576389313,
      "learning_rate": 4.683432081516e-07,
      "loss": 0.4915,
      "step": 5940
    },
    {
      "epoch": 0.3306518723994452,
      "grad_norm": 0.3644208014011383,
      "learning_rate": 4.678105824153662e-07,
      "loss": 0.5216,
      "step": 5960
    },
    {
      "epoch": 0.3317614424410541,
      "grad_norm": 0.29700183868408203,
      "learning_rate": 4.6727382137452644e-07,
      "loss": 0.4904,
      "step": 5980
    },
    {
      "epoch": 0.332871012482663,
      "grad_norm": 0.25862252712249756,
      "learning_rate": 4.6673293521990966e-07,
      "loss": 0.516,
      "step": 6000
    },
    {
      "epoch": 0.3339805825242718,
      "grad_norm": 0.31422197818756104,
      "learning_rate": 4.661879342206636e-07,
      "loss": 0.4196,
      "step": 6020
    },
    {
      "epoch": 0.3350901525658807,
      "grad_norm": 0.3729874789714813,
      "learning_rate": 4.6563882872405924e-07,
      "loss": 0.5395,
      "step": 6040
    },
    {
      "epoch": 0.3361997226074896,
      "grad_norm": 0.7892670035362244,
      "learning_rate": 4.650856291552948e-07,
      "loss": 0.4989,
      "step": 6060
    },
    {
      "epoch": 0.3373092926490985,
      "grad_norm": 0.33878469467163086,
      "learning_rate": 4.645283460172976e-07,
      "loss": 0.5837,
      "step": 6080
    },
    {
      "epoch": 0.33841886269070737,
      "grad_norm": 0.37014704942703247,
      "learning_rate": 4.6396698989052473e-07,
      "loss": 0.4183,
      "step": 6100
    },
    {
      "epoch": 0.33952843273231625,
      "grad_norm": 0.3007761240005493,
      "learning_rate": 4.6340157143276233e-07,
      "loss": 0.4898,
      "step": 6120
    },
    {
      "epoch": 0.3406380027739251,
      "grad_norm": 1.216626524925232,
      "learning_rate": 4.628321013789228e-07,
      "loss": 0.5657,
      "step": 6140
    },
    {
      "epoch": 0.341747572815534,
      "grad_norm": 0.4057266414165497,
      "learning_rate": 4.622585905408414e-07,
      "loss": 0.5154,
      "step": 6160
    },
    {
      "epoch": 0.34285714285714286,
      "grad_norm": 0.2979249358177185,
      "learning_rate": 4.6168104980707103e-07,
      "loss": 0.5022,
      "step": 6180
    },
    {
      "epoch": 0.34396671289875175,
      "grad_norm": 0.4504467248916626,
      "learning_rate": 4.6109949014267494e-07,
      "loss": 0.4424,
      "step": 6200
    },
    {
      "epoch": 0.34507628294036063,
      "grad_norm": 0.32083943486213684,
      "learning_rate": 4.605139225890192e-07,
      "loss": 0.6416,
      "step": 6220
    },
    {
      "epoch": 0.34618585298196947,
      "grad_norm": 0.2717413604259491,
      "learning_rate": 4.5992435826356286e-07,
      "loss": 0.5305,
      "step": 6240
    },
    {
      "epoch": 0.34729542302357835,
      "grad_norm": 0.27425190806388855,
      "learning_rate": 4.593308083596464e-07,
      "loss": 0.5205,
      "step": 6260
    },
    {
      "epoch": 0.34840499306518724,
      "grad_norm": 0.3109380304813385,
      "learning_rate": 4.587332841462802e-07,
      "loss": 0.4805,
      "step": 6280
    },
    {
      "epoch": 0.34951456310679613,
      "grad_norm": 0.29059240221977234,
      "learning_rate": 4.581317969679296e-07,
      "loss": 0.5676,
      "step": 6300
    },
    {
      "epoch": 0.350624133148405,
      "grad_norm": 0.2471311241388321,
      "learning_rate": 4.575263582443e-07,
      "loss": 0.5058,
      "step": 6320
    },
    {
      "epoch": 0.35173370319001385,
      "grad_norm": 0.47161543369293213,
      "learning_rate": 4.5691697947012016e-07,
      "loss": 0.5187,
      "step": 6340
    },
    {
      "epoch": 0.35284327323162273,
      "grad_norm": 0.3279658257961273,
      "learning_rate": 4.563036722149236e-07,
      "loss": 0.504,
      "step": 6360
    },
    {
      "epoch": 0.3539528432732316,
      "grad_norm": 0.365337610244751,
      "learning_rate": 4.556864481228293e-07,
      "loss": 0.5314,
      "step": 6380
    },
    {
      "epoch": 0.3550624133148405,
      "grad_norm": 0.3830949664115906,
      "learning_rate": 4.5506531891232036e-07,
      "loss": 0.4771,
      "step": 6400
    },
    {
      "epoch": 0.3561719833564494,
      "grad_norm": 0.3073193430900574,
      "learning_rate": 4.5444029637602154e-07,
      "loss": 0.6175,
      "step": 6420
    },
    {
      "epoch": 0.3572815533980582,
      "grad_norm": 0.30972909927368164,
      "learning_rate": 4.5381139238047553e-07,
      "loss": 0.5965,
      "step": 6440
    },
    {
      "epoch": 0.3583911234396671,
      "grad_norm": 0.35211724042892456,
      "learning_rate": 4.531786188659177e-07,
      "loss": 0.4085,
      "step": 6460
    },
    {
      "epoch": 0.359500693481276,
      "grad_norm": 0.3817974030971527,
      "learning_rate": 4.525419878460489e-07,
      "loss": 0.4394,
      "step": 6480
    },
    {
      "epoch": 0.3606102635228849,
      "grad_norm": 0.26062262058258057,
      "learning_rate": 4.519015114078082e-07,
      "loss": 0.5138,
      "step": 6500
    },
    {
      "epoch": 0.3617198335644938,
      "grad_norm": 0.26502180099487305,
      "learning_rate": 4.5125720171114265e-07,
      "loss": 0.5103,
      "step": 6520
    },
    {
      "epoch": 0.36282940360610266,
      "grad_norm": 0.3360140025615692,
      "learning_rate": 4.506090709887767e-07,
      "loss": 0.5787,
      "step": 6540
    },
    {
      "epoch": 0.3639389736477115,
      "grad_norm": 0.4174387454986572,
      "learning_rate": 4.4995713154598014e-07,
      "loss": 0.5905,
      "step": 6560
    },
    {
      "epoch": 0.3650485436893204,
      "grad_norm": 0.30797651410102844,
      "learning_rate": 4.493013957603342e-07,
      "loss": 0.5341,
      "step": 6580
    },
    {
      "epoch": 0.36615811373092927,
      "grad_norm": 0.29396864771842957,
      "learning_rate": 4.4864187608149664e-07,
      "loss": 0.5531,
      "step": 6600
    },
    {
      "epoch": 0.36726768377253816,
      "grad_norm": 0.4340313971042633,
      "learning_rate": 4.4797858503096553e-07,
      "loss": 0.5408,
      "step": 6620
    },
    {
      "epoch": 0.36837725381414704,
      "grad_norm": 0.33175596594810486,
      "learning_rate": 4.473115352018412e-07,
      "loss": 0.5338,
      "step": 6640
    },
    {
      "epoch": 0.3694868238557559,
      "grad_norm": 0.2598438262939453,
      "learning_rate": 4.4664073925858737e-07,
      "loss": 0.4943,
      "step": 6660
    },
    {
      "epoch": 0.37059639389736476,
      "grad_norm": 0.42363619804382324,
      "learning_rate": 4.459662099367908e-07,
      "loss": 0.5188,
      "step": 6680
    },
    {
      "epoch": 0.37170596393897365,
      "grad_norm": 0.30839937925338745,
      "learning_rate": 4.4528796004291937e-07,
      "loss": 0.551,
      "step": 6700
    },
    {
      "epoch": 0.37281553398058254,
      "grad_norm": 0.39423060417175293,
      "learning_rate": 4.4460600245407876e-07,
      "loss": 0.5298,
      "step": 6720
    },
    {
      "epoch": 0.3739251040221914,
      "grad_norm": 0.2767972946166992,
      "learning_rate": 4.439203501177683e-07,
      "loss": 0.4744,
      "step": 6740
    },
    {
      "epoch": 0.37503467406380026,
      "grad_norm": 0.3349682092666626,
      "learning_rate": 4.432310160516348e-07,
      "loss": 0.6472,
      "step": 6760
    },
    {
      "epoch": 0.37614424410540914,
      "grad_norm": 0.2644417881965637,
      "learning_rate": 4.42538013343226e-07,
      "loss": 0.4105,
      "step": 6780
    },
    {
      "epoch": 0.37725381414701803,
      "grad_norm": 0.5239447951316833,
      "learning_rate": 4.4184135514974117e-07,
      "loss": 0.5414,
      "step": 6800
    },
    {
      "epoch": 0.3783633841886269,
      "grad_norm": 0.363673597574234,
      "learning_rate": 4.411410546977823e-07,
      "loss": 0.6091,
      "step": 6820
    },
    {
      "epoch": 0.3794729542302358,
      "grad_norm": 0.39377179741859436,
      "learning_rate": 4.4043712528310217e-07,
      "loss": 0.4794,
      "step": 6840
    },
    {
      "epoch": 0.38058252427184464,
      "grad_norm": 0.3427687883377075,
      "learning_rate": 4.397295802703523e-07,
      "loss": 0.532,
      "step": 6860
    },
    {
      "epoch": 0.3816920943134535,
      "grad_norm": 0.3745235502719879,
      "learning_rate": 4.390184330928295e-07,
      "loss": 0.5059,
      "step": 6880
    },
    {
      "epoch": 0.3828016643550624,
      "grad_norm": 0.38224560022354126,
      "learning_rate": 4.3830369725222017e-07,
      "loss": 0.6141,
      "step": 6900
    },
    {
      "epoch": 0.3839112343966713,
      "grad_norm": 0.4920729696750641,
      "learning_rate": 4.375853863183443e-07,
      "loss": 0.5215,
      "step": 6920
    },
    {
      "epoch": 0.3850208044382802,
      "grad_norm": 0.3441776633262634,
      "learning_rate": 4.3686351392889793e-07,
      "loss": 0.4538,
      "step": 6940
    },
    {
      "epoch": 0.386130374479889,
      "grad_norm": 0.3994586765766144,
      "learning_rate": 4.361380937891942e-07,
      "loss": 0.4517,
      "step": 6960
    },
    {
      "epoch": 0.3872399445214979,
      "grad_norm": 0.38988494873046875,
      "learning_rate": 4.3540913967190286e-07,
      "loss": 0.4544,
      "step": 6980
    },
    {
      "epoch": 0.3883495145631068,
      "grad_norm": 0.31946954131126404,
      "learning_rate": 4.346766654167893e-07,
      "loss": 0.4662,
      "step": 7000
    },
    {
      "epoch": 0.3894590846047157,
      "grad_norm": 0.336331844329834,
      "learning_rate": 4.33940684930451e-07,
      "loss": 0.4657,
      "step": 7020
    },
    {
      "epoch": 0.39056865464632456,
      "grad_norm": 0.35121986269950867,
      "learning_rate": 4.3320121218605454e-07,
      "loss": 0.4843,
      "step": 7040
    },
    {
      "epoch": 0.39167822468793345,
      "grad_norm": 0.32553473114967346,
      "learning_rate": 4.324582612230694e-07,
      "loss": 0.5287,
      "step": 7060
    },
    {
      "epoch": 0.3927877947295423,
      "grad_norm": 0.2604405879974365,
      "learning_rate": 4.3171184614700185e-07,
      "loss": 0.5274,
      "step": 7080
    },
    {
      "epoch": 0.39389736477115117,
      "grad_norm": 0.2987576723098755,
      "learning_rate": 4.309619811291271e-07,
      "loss": 0.4328,
      "step": 7100
    },
    {
      "epoch": 0.39500693481276006,
      "grad_norm": 0.3672533631324768,
      "learning_rate": 4.3020868040622023e-07,
      "loss": 0.5229,
      "step": 7120
    },
    {
      "epoch": 0.39611650485436894,
      "grad_norm": 0.46902307868003845,
      "learning_rate": 4.294519582802857e-07,
      "loss": 0.5167,
      "step": 7140
    },
    {
      "epoch": 0.39722607489597783,
      "grad_norm": 0.5896915793418884,
      "learning_rate": 4.2869182911828627e-07,
      "loss": 0.5236,
      "step": 7160
    },
    {
      "epoch": 0.39833564493758666,
      "grad_norm": 0.31331950426101685,
      "learning_rate": 4.2792830735186976e-07,
      "loss": 0.58,
      "step": 7180
    },
    {
      "epoch": 0.39944521497919555,
      "grad_norm": 0.4323517382144928,
      "learning_rate": 4.2716140747709516e-07,
      "loss": 0.4798,
      "step": 7200
    },
    {
      "epoch": 0.40055478502080444,
      "grad_norm": 0.24910062551498413,
      "learning_rate": 4.2639114405415777e-07,
      "loss": 0.5023,
      "step": 7220
    },
    {
      "epoch": 0.4016643550624133,
      "grad_norm": 0.3129631578922272,
      "learning_rate": 4.256175317071122e-07,
      "loss": 0.5323,
      "step": 7240
    },
    {
      "epoch": 0.4027739251040222,
      "grad_norm": 0.30831965804100037,
      "learning_rate": 4.248405851235952e-07,
      "loss": 0.6024,
      "step": 7260
    },
    {
      "epoch": 0.40388349514563104,
      "grad_norm": 0.2090596854686737,
      "learning_rate": 4.2406031905454664e-07,
      "loss": 0.5647,
      "step": 7280
    },
    {
      "epoch": 0.40499306518723993,
      "grad_norm": 0.3623282015323639,
      "learning_rate": 4.2327674831392923e-07,
      "loss": 0.4927,
      "step": 7300
    },
    {
      "epoch": 0.4061026352288488,
      "grad_norm": 0.2708721458911896,
      "learning_rate": 4.2248988777844756e-07,
      "loss": 0.4644,
      "step": 7320
    },
    {
      "epoch": 0.4072122052704577,
      "grad_norm": 0.3101958930492401,
      "learning_rate": 4.216997523872656e-07,
      "loss": 0.4676,
      "step": 7340
    },
    {
      "epoch": 0.4083217753120666,
      "grad_norm": 0.34548845887184143,
      "learning_rate": 4.2090635714172295e-07,
      "loss": 0.4972,
      "step": 7360
    },
    {
      "epoch": 0.4094313453536754,
      "grad_norm": 0.2833360433578491,
      "learning_rate": 4.2010971710505024e-07,
      "loss": 0.5035,
      "step": 7380
    },
    {
      "epoch": 0.4105409153952843,
      "grad_norm": 0.4793650805950165,
      "learning_rate": 4.1930984740208277e-07,
      "loss": 0.5244,
      "step": 7400
    },
    {
      "epoch": 0.4116504854368932,
      "grad_norm": 0.25977978110313416,
      "learning_rate": 4.185067632189737e-07,
      "loss": 0.4568,
      "step": 7420
    },
    {
      "epoch": 0.4127600554785021,
      "grad_norm": 0.6289175152778625,
      "learning_rate": 4.177004798029058e-07,
      "loss": 0.4981,
      "step": 7440
    },
    {
      "epoch": 0.413869625520111,
      "grad_norm": 0.33281320333480835,
      "learning_rate": 4.1689101246180134e-07,
      "loss": 0.5826,
      "step": 7460
    },
    {
      "epoch": 0.41497919556171986,
      "grad_norm": 0.39740175008773804,
      "learning_rate": 4.1607837656403245e-07,
      "loss": 0.4544,
      "step": 7480
    },
    {
      "epoch": 0.4160887656033287,
      "grad_norm": 0.24023501574993134,
      "learning_rate": 4.1526258753812833e-07,
      "loss": 0.5676,
      "step": 7500
    },
    {
      "epoch": 0.4171983356449376,
      "grad_norm": 0.3031497001647949,
      "learning_rate": 4.1444366087248304e-07,
      "loss": 0.4852,
      "step": 7520
    },
    {
      "epoch": 0.41830790568654647,
      "grad_norm": 1.512039303779602,
      "learning_rate": 4.136216121150611e-07,
      "loss": 0.6114,
      "step": 7540
    },
    {
      "epoch": 0.41941747572815535,
      "grad_norm": 0.3467373549938202,
      "learning_rate": 4.1279645687310245e-07,
      "loss": 0.4715,
      "step": 7560
    },
    {
      "epoch": 0.42052704576976424,
      "grad_norm": 0.2934076189994812,
      "learning_rate": 4.11968210812826e-07,
      "loss": 0.4977,
      "step": 7580
    },
    {
      "epoch": 0.42163661581137307,
      "grad_norm": 0.3921898305416107,
      "learning_rate": 4.111368896591323e-07,
      "loss": 0.5787,
      "step": 7600
    },
    {
      "epoch": 0.42274618585298196,
      "grad_norm": 1.2903082370758057,
      "learning_rate": 4.10302509195305e-07,
      "loss": 0.4826,
      "step": 7620
    },
    {
      "epoch": 0.42385575589459085,
      "grad_norm": 0.30528688430786133,
      "learning_rate": 4.0946508526271107e-07,
      "loss": 0.5653,
      "step": 7640
    },
    {
      "epoch": 0.42496532593619973,
      "grad_norm": 0.35863542556762695,
      "learning_rate": 4.086246337605002e-07,
      "loss": 0.4821,
      "step": 7660
    },
    {
      "epoch": 0.4260748959778086,
      "grad_norm": 0.28547871112823486,
      "learning_rate": 4.077811706453028e-07,
      "loss": 0.4127,
      "step": 7680
    },
    {
      "epoch": 0.42718446601941745,
      "grad_norm": 0.489197313785553,
      "learning_rate": 4.069347119309271e-07,
      "loss": 0.5363,
      "step": 7700
    },
    {
      "epoch": 0.42829403606102634,
      "grad_norm": 0.2887154221534729,
      "learning_rate": 4.060852736880553e-07,
      "loss": 0.5618,
      "step": 7720
    },
    {
      "epoch": 0.4294036061026352,
      "grad_norm": 0.2965123951435089,
      "learning_rate": 4.0523287204393795e-07,
      "loss": 0.4854,
      "step": 7740
    },
    {
      "epoch": 0.4305131761442441,
      "grad_norm": 0.326860636472702,
      "learning_rate": 4.0437752318208846e-07,
      "loss": 0.4852,
      "step": 7760
    },
    {
      "epoch": 0.431622746185853,
      "grad_norm": 0.33304309844970703,
      "learning_rate": 4.0351924334197516e-07,
      "loss": 0.4727,
      "step": 7780
    },
    {
      "epoch": 0.43273231622746183,
      "grad_norm": 0.26137974858283997,
      "learning_rate": 4.0265804881871366e-07,
      "loss": 0.536,
      "step": 7800
    },
    {
      "epoch": 0.4338418862690707,
      "grad_norm": 0.543980062007904,
      "learning_rate": 4.0179395596275665e-07,
      "loss": 0.5694,
      "step": 7820
    },
    {
      "epoch": 0.4349514563106796,
      "grad_norm": 0.3294641971588135,
      "learning_rate": 4.0092698117958447e-07,
      "loss": 0.5928,
      "step": 7840
    },
    {
      "epoch": 0.4360610263522885,
      "grad_norm": 0.3174699544906616,
      "learning_rate": 4.0005714092939255e-07,
      "loss": 0.5022,
      "step": 7860
    },
    {
      "epoch": 0.4371705963938974,
      "grad_norm": 0.3462190628051758,
      "learning_rate": 3.9918445172677995e-07,
      "loss": 0.4936,
      "step": 7880
    },
    {
      "epoch": 0.43828016643550627,
      "grad_norm": 0.2537282407283783,
      "learning_rate": 3.983089301404351e-07,
      "loss": 0.6196,
      "step": 7900
    },
    {
      "epoch": 0.4393897364771151,
      "grad_norm": 0.4872954189777374,
      "learning_rate": 3.9743059279282126e-07,
      "loss": 0.46,
      "step": 7920
    },
    {
      "epoch": 0.440499306518724,
      "grad_norm": 0.3006153702735901,
      "learning_rate": 3.9654945635986155e-07,
      "loss": 0.5201,
      "step": 7940
    },
    {
      "epoch": 0.4416088765603329,
      "grad_norm": 0.28118792176246643,
      "learning_rate": 3.9566553757062154e-07,
      "loss": 0.548,
      "step": 7960
    },
    {
      "epoch": 0.44271844660194176,
      "grad_norm": 0.4475691616535187,
      "learning_rate": 3.947788532069923e-07,
      "loss": 0.498,
      "step": 7980
    },
    {
      "epoch": 0.44382801664355065,
      "grad_norm": 0.39291325211524963,
      "learning_rate": 3.938894201033713e-07,
      "loss": 0.4702,
      "step": 8000
    },
    {
      "epoch": 0.4449375866851595,
      "grad_norm": 0.3244176506996155,
      "learning_rate": 3.929972551463431e-07,
      "loss": 0.4421,
      "step": 8020
    },
    {
      "epoch": 0.44604715672676837,
      "grad_norm": 0.3121548295021057,
      "learning_rate": 3.9210237527435864e-07,
      "loss": 0.5523,
      "step": 8040
    },
    {
      "epoch": 0.44715672676837726,
      "grad_norm": 0.33265426754951477,
      "learning_rate": 3.9120479747741344e-07,
      "loss": 0.5159,
      "step": 8060
    },
    {
      "epoch": 0.44826629680998614,
      "grad_norm": 0.298480361700058,
      "learning_rate": 3.903045387967256e-07,
      "loss": 0.4688,
      "step": 8080
    },
    {
      "epoch": 0.44937586685159503,
      "grad_norm": 0.2703563868999481,
      "learning_rate": 3.8940161632441157e-07,
      "loss": 0.4673,
      "step": 8100
    },
    {
      "epoch": 0.45048543689320386,
      "grad_norm": 0.4872739911079407,
      "learning_rate": 3.884960472031622e-07,
      "loss": 0.5151,
      "step": 8120
    },
    {
      "epoch": 0.45159500693481275,
      "grad_norm": 0.4316668212413788,
      "learning_rate": 3.87587848625917e-07,
      "loss": 0.5357,
      "step": 8140
    },
    {
      "epoch": 0.45270457697642164,
      "grad_norm": 0.2701937258243561,
      "learning_rate": 3.866770378355375e-07,
      "loss": 0.4859,
      "step": 8160
    },
    {
      "epoch": 0.4538141470180305,
      "grad_norm": 0.2920953035354614,
      "learning_rate": 3.8576363212448057e-07,
      "loss": 0.5778,
      "step": 8180
    },
    {
      "epoch": 0.4549237170596394,
      "grad_norm": 0.4294867217540741,
      "learning_rate": 3.8484764883446944e-07,
      "loss": 0.5387,
      "step": 8200
    },
    {
      "epoch": 0.45603328710124824,
      "grad_norm": 0.42127662897109985,
      "learning_rate": 3.8392910535616476e-07,
      "loss": 0.5998,
      "step": 8220
    },
    {
      "epoch": 0.45714285714285713,
      "grad_norm": 0.3392697870731354,
      "learning_rate": 3.8300801912883414e-07,
      "loss": 0.4547,
      "step": 8240
    },
    {
      "epoch": 0.458252427184466,
      "grad_norm": 0.2972329258918762,
      "learning_rate": 3.820844076400216e-07,
      "loss": 0.5605,
      "step": 8260
    },
    {
      "epoch": 0.4593619972260749,
      "grad_norm": 0.3029995560646057,
      "learning_rate": 3.8115828842521514e-07,
      "loss": 0.5124,
      "step": 8280
    },
    {
      "epoch": 0.4604715672676838,
      "grad_norm": 0.3053463101387024,
      "learning_rate": 3.802296790675137e-07,
      "loss": 0.5181,
      "step": 8300
    },
    {
      "epoch": 0.4615811373092926,
      "grad_norm": 0.4185832738876343,
      "learning_rate": 3.7929859719729394e-07,
      "loss": 0.5177,
      "step": 8320
    },
    {
      "epoch": 0.4626907073509015,
      "grad_norm": 0.3226168155670166,
      "learning_rate": 3.783650604918746e-07,
      "loss": 0.5008,
      "step": 8340
    },
    {
      "epoch": 0.4638002773925104,
      "grad_norm": 0.3534069359302521,
      "learning_rate": 3.7742908667518175e-07,
      "loss": 0.529,
      "step": 8360
    },
    {
      "epoch": 0.4649098474341193,
      "grad_norm": 0.37676820158958435,
      "learning_rate": 3.7649069351741185e-07,
      "loss": 0.5128,
      "step": 8380
    },
    {
      "epoch": 0.46601941747572817,
      "grad_norm": 0.30804529786109924,
      "learning_rate": 3.755498988346945e-07,
      "loss": 0.5262,
      "step": 8400
    },
    {
      "epoch": 0.46712898751733706,
      "grad_norm": 0.3896881639957428,
      "learning_rate": 3.746067204887538e-07,
      "loss": 0.4655,
      "step": 8420
    },
    {
      "epoch": 0.4682385575589459,
      "grad_norm": 0.3813321888446808,
      "learning_rate": 3.7366117638657e-07,
      "loss": 0.4867,
      "step": 8440
    },
    {
      "epoch": 0.4693481276005548,
      "grad_norm": 0.3187031149864197,
      "learning_rate": 3.72713284480039e-07,
      "loss": 0.4468,
      "step": 8460
    },
    {
      "epoch": 0.47045769764216366,
      "grad_norm": 0.3183457851409912,
      "learning_rate": 3.7176306276563126e-07,
      "loss": 0.491,
      "step": 8480
    },
    {
      "epoch": 0.47156726768377255,
      "grad_norm": 0.58977210521698,
      "learning_rate": 3.708105292840509e-07,
      "loss": 0.5106,
      "step": 8500
    },
    {
      "epoch": 0.47267683772538144,
      "grad_norm": 0.3110608756542206,
      "learning_rate": 3.698557021198925e-07,
      "loss": 0.4688,
      "step": 8520
    },
    {
      "epoch": 0.47378640776699027,
      "grad_norm": 0.23853054642677307,
      "learning_rate": 3.6889859940129814e-07,
      "loss": 0.5505,
      "step": 8540
    },
    {
      "epoch": 0.47489597780859916,
      "grad_norm": 0.24323715269565582,
      "learning_rate": 3.6793923929961296e-07,
      "loss": 0.4357,
      "step": 8560
    },
    {
      "epoch": 0.47600554785020804,
      "grad_norm": 0.39231613278388977,
      "learning_rate": 3.669776400290403e-07,
      "loss": 0.5529,
      "step": 8580
    },
    {
      "epoch": 0.47711511789181693,
      "grad_norm": 0.3708134889602661,
      "learning_rate": 3.66013819846296e-07,
      "loss": 0.6198,
      "step": 8600
    },
    {
      "epoch": 0.4782246879334258,
      "grad_norm": 0.2724001109600067,
      "learning_rate": 3.6504779705026156e-07,
      "loss": 0.4517,
      "step": 8620
    },
    {
      "epoch": 0.47933425797503465,
      "grad_norm": 0.3848133683204651,
      "learning_rate": 3.6407958998163687e-07,
      "loss": 0.5878,
      "step": 8640
    },
    {
      "epoch": 0.48044382801664354,
      "grad_norm": 0.3788512945175171,
      "learning_rate": 3.6310921702259184e-07,
      "loss": 0.4963,
      "step": 8660
    },
    {
      "epoch": 0.4815533980582524,
      "grad_norm": 0.45098355412483215,
      "learning_rate": 3.6213669659641757e-07,
      "loss": 0.4629,
      "step": 8680
    },
    {
      "epoch": 0.4826629680998613,
      "grad_norm": 0.5497767329216003,
      "learning_rate": 3.611620471671766e-07,
      "loss": 0.5813,
      "step": 8700
    },
    {
      "epoch": 0.4837725381414702,
      "grad_norm": 0.3716546297073364,
      "learning_rate": 3.6018528723935214e-07,
      "loss": 0.498,
      "step": 8720
    },
    {
      "epoch": 0.48488210818307903,
      "grad_norm": 0.323794960975647,
      "learning_rate": 3.5920643535749696e-07,
      "loss": 0.4899,
      "step": 8740
    },
    {
      "epoch": 0.4859916782246879,
      "grad_norm": 0.2931385338306427,
      "learning_rate": 3.582255101058811e-07,
      "loss": 0.56,
      "step": 8760
    },
    {
      "epoch": 0.4871012482662968,
      "grad_norm": 0.4833517372608185,
      "learning_rate": 3.572425301081392e-07,
      "loss": 0.5229,
      "step": 8780
    },
    {
      "epoch": 0.4882108183079057,
      "grad_norm": 0.3123915195465088,
      "learning_rate": 3.5625751402691693e-07,
      "loss": 0.5081,
      "step": 8800
    },
    {
      "epoch": 0.4893203883495146,
      "grad_norm": 0.24910438060760498,
      "learning_rate": 3.5527048056351654e-07,
      "loss": 0.5406,
      "step": 8820
    },
    {
      "epoch": 0.49042995839112347,
      "grad_norm": 0.4142923355102539,
      "learning_rate": 3.542814484575419e-07,
      "loss": 0.4703,
      "step": 8840
    },
    {
      "epoch": 0.4915395284327323,
      "grad_norm": 0.34447526931762695,
      "learning_rate": 3.532904364865426e-07,
      "loss": 0.4803,
      "step": 8860
    },
    {
      "epoch": 0.4926490984743412,
      "grad_norm": 0.8663418292999268,
      "learning_rate": 3.522974634656576e-07,
      "loss": 0.4718,
      "step": 8880
    },
    {
      "epoch": 0.49375866851595007,
      "grad_norm": 0.5423429012298584,
      "learning_rate": 3.5130254824725787e-07,
      "loss": 0.4895,
      "step": 8900
    },
    {
      "epoch": 0.49486823855755896,
      "grad_norm": 0.4453887641429901,
      "learning_rate": 3.503057097205885e-07,
      "loss": 0.5677,
      "step": 8920
    },
    {
      "epoch": 0.49597780859916785,
      "grad_norm": 0.41120442748069763,
      "learning_rate": 3.4930696681141034e-07,
      "loss": 0.5781,
      "step": 8940
    },
    {
      "epoch": 0.4970873786407767,
      "grad_norm": 0.38804373145103455,
      "learning_rate": 3.4830633848164006e-07,
      "loss": 0.5453,
      "step": 8960
    },
    {
      "epoch": 0.49819694868238557,
      "grad_norm": 0.4749200940132141,
      "learning_rate": 3.473038437289907e-07,
      "loss": 0.516,
      "step": 8980
    },
    {
      "epoch": 0.49930651872399445,
      "grad_norm": 0.34202054142951965,
      "learning_rate": 3.462995015866109e-07,
      "loss": 0.6462,
      "step": 9000
    },
    {
      "epoch": 0.5004160887656033,
      "grad_norm": 0.29288092255592346,
      "learning_rate": 3.452933311227232e-07,
      "loss": 0.4496,
      "step": 9020
    },
    {
      "epoch": 0.5015256588072122,
      "grad_norm": 0.4564000070095062,
      "learning_rate": 3.442853514402626e-07,
      "loss": 0.5254,
      "step": 9040
    },
    {
      "epoch": 0.5026352288488211,
      "grad_norm": 0.4368923604488373,
      "learning_rate": 3.432755816765131e-07,
      "loss": 0.3723,
      "step": 9060
    },
    {
      "epoch": 0.50374479889043,
      "grad_norm": 0.42233774065971375,
      "learning_rate": 3.422640410027451e-07,
      "loss": 0.4816,
      "step": 9080
    },
    {
      "epoch": 0.5048543689320388,
      "grad_norm": 0.5136800408363342,
      "learning_rate": 3.412507486238512e-07,
      "loss": 0.48,
      "step": 9100
    },
    {
      "epoch": 0.5059639389736477,
      "grad_norm": 0.6266002655029297,
      "learning_rate": 3.4023572377798116e-07,
      "loss": 0.4605,
      "step": 9120
    },
    {
      "epoch": 0.5070735090152566,
      "grad_norm": 0.38814643025398254,
      "learning_rate": 3.3921898573617715e-07,
      "loss": 0.4482,
      "step": 9140
    },
    {
      "epoch": 0.5081830790568654,
      "grad_norm": 0.3517705202102661,
      "learning_rate": 3.382005538020078e-07,
      "loss": 0.4514,
      "step": 9160
    },
    {
      "epoch": 0.5092926490984744,
      "grad_norm": 0.25241366028785706,
      "learning_rate": 3.371804473112014e-07,
      "loss": 0.5004,
      "step": 9180
    },
    {
      "epoch": 0.5104022191400832,
      "grad_norm": 0.5189043283462524,
      "learning_rate": 3.3615868563127937e-07,
      "loss": 0.5339,
      "step": 9200
    },
    {
      "epoch": 0.511511789181692,
      "grad_norm": 0.42029428482055664,
      "learning_rate": 3.3513528816118775e-07,
      "loss": 0.551,
      "step": 9220
    },
    {
      "epoch": 0.512621359223301,
      "grad_norm": 0.5070712566375732,
      "learning_rate": 3.341102743309296e-07,
      "loss": 0.5013,
      "step": 9240
    },
    {
      "epoch": 0.5137309292649098,
      "grad_norm": 0.2703372836112976,
      "learning_rate": 3.3308366360119584e-07,
      "loss": 0.5446,
      "step": 9260
    },
    {
      "epoch": 0.5148404993065188,
      "grad_norm": 0.39520278573036194,
      "learning_rate": 3.3205547546299575e-07,
      "loss": 0.5542,
      "step": 9280
    },
    {
      "epoch": 0.5159500693481276,
      "grad_norm": 0.4130886197090149,
      "learning_rate": 3.3102572943728673e-07,
      "loss": 0.4348,
      "step": 9300
    },
    {
      "epoch": 0.5170596393897364,
      "grad_norm": 0.30212274193763733,
      "learning_rate": 3.2999444507460437e-07,
      "loss": 0.4626,
      "step": 9320
    },
    {
      "epoch": 0.5181692094313454,
      "grad_norm": 0.24097274243831635,
      "learning_rate": 3.2896164195469033e-07,
      "loss": 0.5347,
      "step": 9340
    },
    {
      "epoch": 0.5192787794729542,
      "grad_norm": 0.2773560583591461,
      "learning_rate": 3.279273396861214e-07,
      "loss": 0.6156,
      "step": 9360
    },
    {
      "epoch": 0.5203883495145631,
      "grad_norm": 0.4777432382106781,
      "learning_rate": 3.268915579059366e-07,
      "loss": 0.4348,
      "step": 9380
    },
    {
      "epoch": 0.521497919556172,
      "grad_norm": 0.30696791410446167,
      "learning_rate": 3.2585431627926476e-07,
      "loss": 0.6035,
      "step": 9400
    },
    {
      "epoch": 0.5226074895977808,
      "grad_norm": 0.2942405343055725,
      "learning_rate": 3.248156344989512e-07,
      "loss": 0.4254,
      "step": 9420
    },
    {
      "epoch": 0.5237170596393897,
      "grad_norm": 0.498976469039917,
      "learning_rate": 3.237755322851834e-07,
      "loss": 0.5464,
      "step": 9440
    },
    {
      "epoch": 0.5248266296809986,
      "grad_norm": 0.5058267712593079,
      "learning_rate": 3.2273402938511706e-07,
      "loss": 0.5168,
      "step": 9460
    },
    {
      "epoch": 0.5259361997226075,
      "grad_norm": 0.164671391248703,
      "learning_rate": 3.2169114557250103e-07,
      "loss": 0.5291,
      "step": 9480
    },
    {
      "epoch": 0.5270457697642164,
      "grad_norm": 0.45560675859451294,
      "learning_rate": 3.206469006473017e-07,
      "loss": 0.5577,
      "step": 9500
    },
    {
      "epoch": 0.5281553398058253,
      "grad_norm": 0.4424368441104889,
      "learning_rate": 3.196013144353274e-07,
      "loss": 0.4828,
      "step": 9520
    },
    {
      "epoch": 0.5292649098474341,
      "grad_norm": 0.4512025713920593,
      "learning_rate": 3.185544067878518e-07,
      "loss": 0.5543,
      "step": 9540
    },
    {
      "epoch": 0.530374479889043,
      "grad_norm": 0.30722030997276306,
      "learning_rate": 3.175061975812371e-07,
      "loss": 0.5418,
      "step": 9560
    },
    {
      "epoch": 0.5314840499306519,
      "grad_norm": 0.38087835907936096,
      "learning_rate": 3.1645670671655645e-07,
      "loss": 0.4493,
      "step": 9580
    },
    {
      "epoch": 0.5325936199722607,
      "grad_norm": 0.3499601483345032,
      "learning_rate": 3.154059541192164e-07,
      "loss": 0.5878,
      "step": 9600
    },
    {
      "epoch": 0.5337031900138697,
      "grad_norm": 0.4532395601272583,
      "learning_rate": 3.1435395973857876e-07,
      "loss": 0.5829,
      "step": 9620
    },
    {
      "epoch": 0.5348127600554785,
      "grad_norm": 0.30889692902565,
      "learning_rate": 3.1330074354758094e-07,
      "loss": 0.5797,
      "step": 9640
    },
    {
      "epoch": 0.5359223300970873,
      "grad_norm": 0.4182281494140625,
      "learning_rate": 3.12246325542358e-07,
      "loss": 0.5118,
      "step": 9660
    },
    {
      "epoch": 0.5370319001386963,
      "grad_norm": 0.35436221957206726,
      "learning_rate": 3.11190725741862e-07,
      "loss": 0.6517,
      "step": 9680
    },
    {
      "epoch": 0.5381414701803051,
      "grad_norm": 0.33836600184440613,
      "learning_rate": 3.1013396418748234e-07,
      "loss": 0.4423,
      "step": 9700
    },
    {
      "epoch": 0.5392510402219141,
      "grad_norm": 0.34394633769989014,
      "learning_rate": 3.090760609426655e-07,
      "loss": 0.4322,
      "step": 9720
    },
    {
      "epoch": 0.5403606102635229,
      "grad_norm": 0.3953171670436859,
      "learning_rate": 3.080170360925336e-07,
      "loss": 0.443,
      "step": 9740
    },
    {
      "epoch": 0.5414701803051317,
      "grad_norm": 0.27416878938674927,
      "learning_rate": 3.069569097435033e-07,
      "loss": 0.4737,
      "step": 9760
    },
    {
      "epoch": 0.5425797503467407,
      "grad_norm": 0.325531005859375,
      "learning_rate": 3.0589570202290433e-07,
      "loss": 0.4698,
      "step": 9780
    },
    {
      "epoch": 0.5436893203883495,
      "grad_norm": 0.44285646080970764,
      "learning_rate": 3.0483343307859663e-07,
      "loss": 0.4886,
      "step": 9800
    },
    {
      "epoch": 0.5447988904299584,
      "grad_norm": 0.2852041721343994,
      "learning_rate": 3.0377012307858904e-07,
      "loss": 0.5289,
      "step": 9820
    },
    {
      "epoch": 0.5459084604715673,
      "grad_norm": 0.3136042356491089,
      "learning_rate": 3.027057922106549e-07,
      "loss": 0.494,
      "step": 9840
    },
    {
      "epoch": 0.5470180305131761,
      "grad_norm": 0.3801943361759186,
      "learning_rate": 3.0164046068195e-07,
      "loss": 0.4818,
      "step": 9860
    },
    {
      "epoch": 0.548127600554785,
      "grad_norm": 0.3817514479160309,
      "learning_rate": 3.0057414871862816e-07,
      "loss": 0.5448,
      "step": 9880
    },
    {
      "epoch": 0.5492371705963939,
      "grad_norm": 0.3009171485900879,
      "learning_rate": 2.9950687656545787e-07,
      "loss": 0.4765,
      "step": 9900
    },
    {
      "epoch": 0.5503467406380028,
      "grad_norm": 0.4833567440509796,
      "learning_rate": 2.9843866448543727e-07,
      "loss": 0.5342,
      "step": 9920
    },
    {
      "epoch": 0.5514563106796116,
      "grad_norm": 0.4005512297153473,
      "learning_rate": 2.973695327594099e-07,
      "loss": 0.4766,
      "step": 9940
    },
    {
      "epoch": 0.5525658807212205,
      "grad_norm": 0.33436936140060425,
      "learning_rate": 2.9629950168567954e-07,
      "loss": 0.4826,
      "step": 9960
    },
    {
      "epoch": 0.5536754507628294,
      "grad_norm": 0.33340537548065186,
      "learning_rate": 2.9522859157962454e-07,
      "loss": 0.5473,
      "step": 9980
    },
    {
      "epoch": 0.5547850208044383,
      "grad_norm": 0.33686235547065735,
      "learning_rate": 2.9415682277331265e-07,
      "loss": 0.5534,
      "step": 10000
    },
    {
      "epoch": 0.5558945908460472,
      "grad_norm": 0.3106544315814972,
      "learning_rate": 2.930842156151146e-07,
      "loss": 0.497,
      "step": 10020
    },
    {
      "epoch": 0.557004160887656,
      "grad_norm": 0.428392231464386,
      "learning_rate": 2.920107904693178e-07,
      "loss": 0.483,
      "step": 10040
    },
    {
      "epoch": 0.5581137309292649,
      "grad_norm": 0.3546343743801117,
      "learning_rate": 2.9093656771574006e-07,
      "loss": 0.4438,
      "step": 10060
    },
    {
      "epoch": 0.5592233009708738,
      "grad_norm": 0.5527733564376831,
      "learning_rate": 2.8986156774934204e-07,
      "loss": 0.5118,
      "step": 10080
    },
    {
      "epoch": 0.5603328710124826,
      "grad_norm": 0.33526676893234253,
      "learning_rate": 2.8878581097984075e-07,
      "loss": 0.565,
      "step": 10100
    },
    {
      "epoch": 0.5614424410540916,
      "grad_norm": 0.4417477548122406,
      "learning_rate": 2.877093178313214e-07,
      "loss": 0.4793,
      "step": 10120
    },
    {
      "epoch": 0.5625520110957004,
      "grad_norm": 0.2999446392059326,
      "learning_rate": 2.8663210874185013e-07,
      "loss": 0.5449,
      "step": 10140
    },
    {
      "epoch": 0.5636615811373092,
      "grad_norm": 0.38563647866249084,
      "learning_rate": 2.8555420416308573e-07,
      "loss": 0.5037,
      "step": 10160
    },
    {
      "epoch": 0.5647711511789182,
      "grad_norm": 0.8459754586219788,
      "learning_rate": 2.8447562455989134e-07,
      "loss": 0.5474,
      "step": 10180
    },
    {
      "epoch": 0.565880721220527,
      "grad_norm": 0.3799205720424652,
      "learning_rate": 2.8339639040994604e-07,
      "loss": 0.609,
      "step": 10200
    },
    {
      "epoch": 0.566990291262136,
      "grad_norm": 0.51907879114151,
      "learning_rate": 2.8231652220335603e-07,
      "loss": 0.4621,
      "step": 10220
    },
    {
      "epoch": 0.5680998613037448,
      "grad_norm": 0.3080946207046509,
      "learning_rate": 2.812360404422653e-07,
      "loss": 0.5304,
      "step": 10240
    },
    {
      "epoch": 0.5692094313453536,
      "grad_norm": 0.35099002718925476,
      "learning_rate": 2.80154965640467e-07,
      "loss": 0.5226,
      "step": 10260
    },
    {
      "epoch": 0.5703190013869626,
      "grad_norm": 0.3252660036087036,
      "learning_rate": 2.790733183230136e-07,
      "loss": 0.4481,
      "step": 10280
    },
    {
      "epoch": 0.5714285714285714,
      "grad_norm": 0.3965422511100769,
      "learning_rate": 2.7799111902582693e-07,
      "loss": 0.5616,
      "step": 10300
    },
    {
      "epoch": 0.5725381414701803,
      "grad_norm": 0.4213715195655823,
      "learning_rate": 2.7690838829530886e-07,
      "loss": 0.4895,
      "step": 10320
    },
    {
      "epoch": 0.5736477115117892,
      "grad_norm": 0.27551645040512085,
      "learning_rate": 2.758251466879508e-07,
      "loss": 0.532,
      "step": 10340
    },
    {
      "epoch": 0.574757281553398,
      "grad_norm": 0.6193405389785767,
      "learning_rate": 2.7474141476994366e-07,
      "loss": 0.5517,
      "step": 10360
    },
    {
      "epoch": 0.575866851595007,
      "grad_norm": 0.4643997251987457,
      "learning_rate": 2.736572131167872e-07,
      "loss": 0.6117,
      "step": 10380
    },
    {
      "epoch": 0.5769764216366158,
      "grad_norm": 0.3855043351650238,
      "learning_rate": 2.725725623128994e-07,
      "loss": 0.4358,
      "step": 10400
    },
    {
      "epoch": 0.5780859916782247,
      "grad_norm": 0.2621045410633087,
      "learning_rate": 2.71487482951226e-07,
      "loss": 0.5391,
      "step": 10420
    },
    {
      "epoch": 0.5791955617198336,
      "grad_norm": 0.4209457337856293,
      "learning_rate": 2.7040199563284894e-07,
      "loss": 0.5641,
      "step": 10440
    },
    {
      "epoch": 0.5803051317614425,
      "grad_norm": 0.3612224757671356,
      "learning_rate": 2.6931612096659566e-07,
      "loss": 0.5176,
      "step": 10460
    },
    {
      "epoch": 0.5814147018030513,
      "grad_norm": 0.30091843008995056,
      "learning_rate": 2.682298795686478e-07,
      "loss": 0.4728,
      "step": 10480
    },
    {
      "epoch": 0.5825242718446602,
      "grad_norm": 0.33328065276145935,
      "learning_rate": 2.671432920621495e-07,
      "loss": 0.5307,
      "step": 10500
    },
    {
      "epoch": 0.5836338418862691,
      "grad_norm": 0.2998218536376953,
      "learning_rate": 2.6605637907681613e-07,
      "loss": 0.5042,
      "step": 10520
    },
    {
      "epoch": 0.5847434119278779,
      "grad_norm": 0.576810896396637,
      "learning_rate": 2.6496916124854244e-07,
      "loss": 0.5064,
      "step": 10540
    },
    {
      "epoch": 0.5858529819694869,
      "grad_norm": 0.40196332335472107,
      "learning_rate": 2.638816592190112e-07,
      "loss": 0.5932,
      "step": 10560
    },
    {
      "epoch": 0.5869625520110957,
      "grad_norm": 0.524108350276947,
      "learning_rate": 2.627938936353006e-07,
      "loss": 0.5463,
      "step": 10580
    },
    {
      "epoch": 0.5880721220527045,
      "grad_norm": 0.45371681451797485,
      "learning_rate": 2.617058851494927e-07,
      "loss": 0.5356,
      "step": 10600
    },
    {
      "epoch": 0.5891816920943135,
      "grad_norm": 0.3213278651237488,
      "learning_rate": 2.606176544182813e-07,
      "loss": 0.5075,
      "step": 10620
    },
    {
      "epoch": 0.5902912621359223,
      "grad_norm": 0.34761932492256165,
      "learning_rate": 2.5952922210257964e-07,
      "loss": 0.5104,
      "step": 10640
    },
    {
      "epoch": 0.5914008321775313,
      "grad_norm": 0.32400938868522644,
      "learning_rate": 2.584406088671284e-07,
      "loss": 0.4889,
      "step": 10660
    },
    {
      "epoch": 0.5925104022191401,
      "grad_norm": 0.30970096588134766,
      "learning_rate": 2.573518353801028e-07,
      "loss": 0.6171,
      "step": 10680
    },
    {
      "epoch": 0.5936199722607489,
      "grad_norm": 0.3451375365257263,
      "learning_rate": 2.5626292231272086e-07,
      "loss": 0.4881,
      "step": 10700
    },
    {
      "epoch": 0.5947295423023579,
      "grad_norm": 0.41977164149284363,
      "learning_rate": 2.5517389033885056e-07,
      "loss": 0.5399,
      "step": 10720
    },
    {
      "epoch": 0.5958391123439667,
      "grad_norm": 0.3118828535079956,
      "learning_rate": 2.540847601346173e-07,
      "loss": 0.4543,
      "step": 10740
    },
    {
      "epoch": 0.5969486823855756,
      "grad_norm": 0.36801040172576904,
      "learning_rate": 2.5299555237801176e-07,
      "loss": 0.4706,
      "step": 10760
    },
    {
      "epoch": 0.5980582524271845,
      "grad_norm": 0.3566454350948334,
      "learning_rate": 2.5190628774849667e-07,
      "loss": 0.5271,
      "step": 10780
    },
    {
      "epoch": 0.5991678224687933,
      "grad_norm": 0.436729371547699,
      "learning_rate": 2.5081698692661475e-07,
      "loss": 0.5308,
      "step": 10800
    },
    {
      "epoch": 0.6002773925104022,
      "grad_norm": 0.39320242404937744,
      "learning_rate": 2.497276705935957e-07,
      "loss": 0.5804,
      "step": 10820
    },
    {
      "epoch": 0.6013869625520111,
      "grad_norm": 0.2793132960796356,
      "learning_rate": 2.4863835943096386e-07,
      "loss": 0.4361,
      "step": 10840
    },
    {
      "epoch": 0.60249653259362,
      "grad_norm": 0.31615859270095825,
      "learning_rate": 2.4754907412014526e-07,
      "loss": 0.468,
      "step": 10860
    },
    {
      "epoch": 0.6036061026352288,
      "grad_norm": 0.27115607261657715,
      "learning_rate": 2.464598353420754e-07,
      "loss": 0.4934,
      "step": 10880
    },
    {
      "epoch": 0.6047156726768377,
      "grad_norm": 0.41032761335372925,
      "learning_rate": 2.45370663776806e-07,
      "loss": 0.6101,
      "step": 10900
    },
    {
      "epoch": 0.6058252427184466,
      "grad_norm": 0.3607085943222046,
      "learning_rate": 2.442815801031128e-07,
      "loss": 0.4863,
      "step": 10920
    },
    {
      "epoch": 0.6069348127600555,
      "grad_norm": 0.3773590624332428,
      "learning_rate": 2.431926049981029e-07,
      "loss": 0.4498,
      "step": 10940
    },
    {
      "epoch": 0.6080443828016644,
      "grad_norm": 0.3329433500766754,
      "learning_rate": 2.4210375913682203e-07,
      "loss": 0.5377,
      "step": 10960
    },
    {
      "epoch": 0.6091539528432732,
      "grad_norm": 0.3469200134277344,
      "learning_rate": 2.4101506319186234e-07,
      "loss": 0.5213,
      "step": 10980
    },
    {
      "epoch": 0.6102635228848821,
      "grad_norm": 0.5309900045394897,
      "learning_rate": 2.399265378329694e-07,
      "loss": 0.5984,
      "step": 11000
    },
    {
      "epoch": 0.611373092926491,
      "grad_norm": 0.28028154373168945,
      "learning_rate": 2.388382037266504e-07,
      "loss": 0.5899,
      "step": 11020
    },
    {
      "epoch": 0.6124826629680998,
      "grad_norm": 0.3442562520503998,
      "learning_rate": 2.3775008153578108e-07,
      "loss": 0.4739,
      "step": 11040
    },
    {
      "epoch": 0.6135922330097088,
      "grad_norm": 0.38204333186149597,
      "learning_rate": 2.366621919192141e-07,
      "loss": 0.4846,
      "step": 11060
    },
    {
      "epoch": 0.6147018030513176,
      "grad_norm": 0.4290561378002167,
      "learning_rate": 2.3557455553138645e-07,
      "loss": 0.5242,
      "step": 11080
    },
    {
      "epoch": 0.6158113730929264,
      "grad_norm": 0.33580636978149414,
      "learning_rate": 2.3448719302192729e-07,
      "loss": 0.4827,
      "step": 11100
    },
    {
      "epoch": 0.6169209431345354,
      "grad_norm": 0.3730616569519043,
      "learning_rate": 2.3340012503526607e-07,
      "loss": 0.55,
      "step": 11120
    },
    {
      "epoch": 0.6180305131761442,
      "grad_norm": 0.5171761512756348,
      "learning_rate": 2.323133722102404e-07,
      "loss": 0.514,
      "step": 11140
    },
    {
      "epoch": 0.6191400832177532,
      "grad_norm": 0.328640878200531,
      "learning_rate": 2.3122695517970434e-07,
      "loss": 0.5019,
      "step": 11160
    },
    {
      "epoch": 0.620249653259362,
      "grad_norm": 0.34641680121421814,
      "learning_rate": 2.3014089457013675e-07,
      "loss": 0.5429,
      "step": 11180
    },
    {
      "epoch": 0.6213592233009708,
      "grad_norm": 0.3420324921607971,
      "learning_rate": 2.2905521100124935e-07,
      "loss": 0.4482,
      "step": 11200
    },
    {
      "epoch": 0.6224687933425798,
      "grad_norm": 0.44696226716041565,
      "learning_rate": 2.2796992508559563e-07,
      "loss": 0.5247,
      "step": 11220
    },
    {
      "epoch": 0.6235783633841886,
      "grad_norm": 0.3765786290168762,
      "learning_rate": 2.2688505742817916e-07,
      "loss": 0.5924,
      "step": 11240
    },
    {
      "epoch": 0.6246879334257975,
      "grad_norm": 0.32200539112091064,
      "learning_rate": 2.258006286260626e-07,
      "loss": 0.5605,
      "step": 11260
    },
    {
      "epoch": 0.6257975034674064,
      "grad_norm": 0.455307275056839,
      "learning_rate": 2.2471665926797676e-07,
      "loss": 0.5314,
      "step": 11280
    },
    {
      "epoch": 0.6269070735090153,
      "grad_norm": 0.25619956851005554,
      "learning_rate": 2.2363316993392932e-07,
      "loss": 0.551,
      "step": 11300
    },
    {
      "epoch": 0.6280166435506241,
      "grad_norm": 0.3347276747226715,
      "learning_rate": 2.225501811948145e-07,
      "loss": 0.4912,
      "step": 11320
    },
    {
      "epoch": 0.629126213592233,
      "grad_norm": 0.6607474088668823,
      "learning_rate": 2.2146771361202215e-07,
      "loss": 0.5319,
      "step": 11340
    },
    {
      "epoch": 0.6302357836338419,
      "grad_norm": 0.39210301637649536,
      "learning_rate": 2.203857877370477e-07,
      "loss": 0.4812,
      "step": 11360
    },
    {
      "epoch": 0.6313453536754507,
      "grad_norm": 0.36486566066741943,
      "learning_rate": 2.193044241111018e-07,
      "loss": 0.5066,
      "step": 11380
    },
    {
      "epoch": 0.6324549237170597,
      "grad_norm": 0.25275692343711853,
      "learning_rate": 2.182236432647204e-07,
      "loss": 0.4351,
      "step": 11400
    },
    {
      "epoch": 0.6335644937586685,
      "grad_norm": 0.32072117924690247,
      "learning_rate": 2.1714346571737485e-07,
      "loss": 0.4902,
      "step": 11420
    },
    {
      "epoch": 0.6346740638002774,
      "grad_norm": 0.3688344657421112,
      "learning_rate": 2.160639119770824e-07,
      "loss": 0.4802,
      "step": 11440
    },
    {
      "epoch": 0.6357836338418863,
      "grad_norm": 0.3499152958393097,
      "learning_rate": 2.1498500254001683e-07,
      "loss": 0.5426,
      "step": 11460
    },
    {
      "epoch": 0.6368932038834951,
      "grad_norm": 0.5433036088943481,
      "learning_rate": 2.1390675789011945e-07,
      "loss": 0.5413,
      "step": 11480
    },
    {
      "epoch": 0.6380027739251041,
      "grad_norm": 0.32321110367774963,
      "learning_rate": 2.128291984987099e-07,
      "loss": 0.5557,
      "step": 11500
    },
    {
      "epoch": 0.6391123439667129,
      "grad_norm": 0.3584541380405426,
      "learning_rate": 2.117523448240977e-07,
      "loss": 0.6089,
      "step": 11520
    },
    {
      "epoch": 0.6402219140083217,
      "grad_norm": 0.35222071409225464,
      "learning_rate": 2.1067621731119384e-07,
      "loss": 0.4796,
      "step": 11540
    },
    {
      "epoch": 0.6413314840499307,
      "grad_norm": 0.3017160892486572,
      "learning_rate": 2.0960083639112243e-07,
      "loss": 0.4427,
      "step": 11560
    },
    {
      "epoch": 0.6424410540915395,
      "grad_norm": 0.3295774459838867,
      "learning_rate": 2.0852622248083308e-07,
      "loss": 0.4628,
      "step": 11580
    },
    {
      "epoch": 0.6435506241331485,
      "grad_norm": 0.5670339465141296,
      "learning_rate": 2.0745239598271312e-07,
      "loss": 0.5061,
      "step": 11600
    },
    {
      "epoch": 0.6446601941747573,
      "grad_norm": 0.3149929344654083,
      "learning_rate": 2.0637937728420008e-07,
      "loss": 0.4442,
      "step": 11620
    },
    {
      "epoch": 0.6457697642163661,
      "grad_norm": 0.5195226669311523,
      "learning_rate": 2.0530718675739488e-07,
      "loss": 0.5651,
      "step": 11640
    },
    {
      "epoch": 0.6468793342579751,
      "grad_norm": 0.34720999002456665,
      "learning_rate": 2.0423584475867504e-07,
      "loss": 0.5341,
      "step": 11660
    },
    {
      "epoch": 0.6479889042995839,
      "grad_norm": 0.41216403245925903,
      "learning_rate": 2.0316537162830784e-07,
      "loss": 0.4756,
      "step": 11680
    },
    {
      "epoch": 0.6490984743411928,
      "grad_norm": 0.4366797208786011,
      "learning_rate": 2.020957876900648e-07,
      "loss": 0.5297,
      "step": 11700
    },
    {
      "epoch": 0.6502080443828017,
      "grad_norm": 0.3360641598701477,
      "learning_rate": 2.0102711325083513e-07,
      "loss": 0.6328,
      "step": 11720
    },
    {
      "epoch": 0.6513176144244105,
      "grad_norm": 0.37930089235305786,
      "learning_rate": 1.999593686002406e-07,
      "loss": 0.5153,
      "step": 11740
    },
    {
      "epoch": 0.6524271844660194,
      "grad_norm": 0.44946834444999695,
      "learning_rate": 1.9889257401025015e-07,
      "loss": 0.5232,
      "step": 11760
    },
    {
      "epoch": 0.6535367545076283,
      "grad_norm": 0.43730628490448,
      "learning_rate": 1.978267497347951e-07,
      "loss": 0.4573,
      "step": 11780
    },
    {
      "epoch": 0.6546463245492372,
      "grad_norm": 0.4298498332500458,
      "learning_rate": 1.9676191600938474e-07,
      "loss": 0.3999,
      "step": 11800
    },
    {
      "epoch": 0.655755894590846,
      "grad_norm": 0.3248540759086609,
      "learning_rate": 1.9569809305072177e-07,
      "loss": 0.5563,
      "step": 11820
    },
    {
      "epoch": 0.6568654646324549,
      "grad_norm": 0.3533152639865875,
      "learning_rate": 1.9463530105631877e-07,
      "loss": 0.5788,
      "step": 11840
    },
    {
      "epoch": 0.6579750346740638,
      "grad_norm": 0.4187324047088623,
      "learning_rate": 1.9357356020411475e-07,
      "loss": 0.4424,
      "step": 11860
    },
    {
      "epoch": 0.6590846047156727,
      "grad_norm": 0.39432525634765625,
      "learning_rate": 1.925128906520917e-07,
      "loss": 0.5991,
      "step": 11880
    },
    {
      "epoch": 0.6601941747572816,
      "grad_norm": 0.5896446108818054,
      "learning_rate": 1.9145331253789253e-07,
      "loss": 0.5596,
      "step": 11900
    },
    {
      "epoch": 0.6613037447988904,
      "grad_norm": 0.36868834495544434,
      "learning_rate": 1.90394845978438e-07,
      "loss": 0.5265,
      "step": 11920
    },
    {
      "epoch": 0.6624133148404993,
      "grad_norm": 0.386819452047348,
      "learning_rate": 1.8933751106954535e-07,
      "loss": 0.5376,
      "step": 11940
    },
    {
      "epoch": 0.6635228848821082,
      "grad_norm": 0.3883580267429352,
      "learning_rate": 1.8828132788554638e-07,
      "loss": 0.4808,
      "step": 11960
    },
    {
      "epoch": 0.664632454923717,
      "grad_norm": 0.6747732162475586,
      "learning_rate": 1.8722631647890657e-07,
      "loss": 0.4478,
      "step": 11980
    },
    {
      "epoch": 0.665742024965326,
      "grad_norm": 0.2752344012260437,
      "learning_rate": 1.8617249687984434e-07,
      "loss": 0.5043,
      "step": 12000
    },
    {
      "epoch": 0.6668515950069348,
      "grad_norm": 0.39787495136260986,
      "learning_rate": 1.8511988909595067e-07,
      "loss": 0.4899,
      "step": 12020
    },
    {
      "epoch": 0.6679611650485436,
      "grad_norm": 0.4791197180747986,
      "learning_rate": 1.8406851311180926e-07,
      "loss": 0.5389,
      "step": 12040
    },
    {
      "epoch": 0.6690707350901526,
      "grad_norm": 0.3360481858253479,
      "learning_rate": 1.8301838888861709e-07,
      "loss": 0.4804,
      "step": 12060
    },
    {
      "epoch": 0.6701803051317614,
      "grad_norm": 0.3347693681716919,
      "learning_rate": 1.819695363638055e-07,
      "loss": 0.558,
      "step": 12080
    },
    {
      "epoch": 0.6712898751733704,
      "grad_norm": 0.30149486660957336,
      "learning_rate": 1.809219754506618e-07,
      "loss": 0.4088,
      "step": 12100
    },
    {
      "epoch": 0.6723994452149792,
      "grad_norm": 0.30244728922843933,
      "learning_rate": 1.7987572603795078e-07,
      "loss": 0.5592,
      "step": 12120
    },
    {
      "epoch": 0.673509015256588,
      "grad_norm": 0.3256188929080963,
      "learning_rate": 1.7883080798953754e-07,
      "loss": 0.6117,
      "step": 12140
    },
    {
      "epoch": 0.674618585298197,
      "grad_norm": 0.36098477244377136,
      "learning_rate": 1.777872411440101e-07,
      "loss": 0.4261,
      "step": 12160
    },
    {
      "epoch": 0.6757281553398058,
      "grad_norm": 0.33403322100639343,
      "learning_rate": 1.767450453143029e-07,
      "loss": 0.503,
      "step": 12180
    },
    {
      "epoch": 0.6768377253814147,
      "grad_norm": 0.5139286518096924,
      "learning_rate": 1.757042402873205e-07,
      "loss": 0.44,
      "step": 12200
    },
    {
      "epoch": 0.6779472954230236,
      "grad_norm": 0.368712842464447,
      "learning_rate": 1.7466484582356212e-07,
      "loss": 0.4188,
      "step": 12220
    },
    {
      "epoch": 0.6790568654646325,
      "grad_norm": 0.3538356423377991,
      "learning_rate": 1.736268816567461e-07,
      "loss": 0.4575,
      "step": 12240
    },
    {
      "epoch": 0.6801664355062413,
      "grad_norm": 0.30539965629577637,
      "learning_rate": 1.725903674934357e-07,
      "loss": 0.4657,
      "step": 12260
    },
    {
      "epoch": 0.6812760055478502,
      "grad_norm": 0.2921498417854309,
      "learning_rate": 1.715553230126645e-07,
      "loss": 0.5406,
      "step": 12280
    },
    {
      "epoch": 0.6823855755894591,
      "grad_norm": 1.0049635171890259,
      "learning_rate": 1.705217678655633e-07,
      "loss": 0.5792,
      "step": 12300
    },
    {
      "epoch": 0.683495145631068,
      "grad_norm": 0.3126891851425171,
      "learning_rate": 1.6948972167498649e-07,
      "loss": 0.4519,
      "step": 12320
    },
    {
      "epoch": 0.6846047156726769,
      "grad_norm": 0.36534181237220764,
      "learning_rate": 1.684592040351398e-07,
      "loss": 0.4744,
      "step": 12340
    },
    {
      "epoch": 0.6857142857142857,
      "grad_norm": 0.32556411623954773,
      "learning_rate": 1.674302345112083e-07,
      "loss": 0.5786,
      "step": 12360
    },
    {
      "epoch": 0.6868238557558946,
      "grad_norm": 0.42282480001449585,
      "learning_rate": 1.664028326389847e-07,
      "loss": 0.6001,
      "step": 12380
    },
    {
      "epoch": 0.6879334257975035,
      "grad_norm": 0.2856026887893677,
      "learning_rate": 1.6537701792449882e-07,
      "loss": 0.4948,
      "step": 12400
    },
    {
      "epoch": 0.6890429958391123,
      "grad_norm": 0.30700036883354187,
      "learning_rate": 1.6435280984364692e-07,
      "loss": 0.515,
      "step": 12420
    },
    {
      "epoch": 0.6901525658807213,
      "grad_norm": 0.4423519968986511,
      "learning_rate": 1.633302278418221e-07,
      "loss": 0.4761,
      "step": 12440
    },
    {
      "epoch": 0.6912621359223301,
      "grad_norm": 0.3314474821090698,
      "learning_rate": 1.6230929133354506e-07,
      "loss": 0.5463,
      "step": 12460
    },
    {
      "epoch": 0.6923717059639389,
      "grad_norm": 0.5163138508796692,
      "learning_rate": 1.6129001970209552e-07,
      "loss": 0.4718,
      "step": 12480
    },
    {
      "epoch": 0.6934812760055479,
      "grad_norm": 0.3462437093257904,
      "learning_rate": 1.6027243229914414e-07,
      "loss": 0.4545,
      "step": 12500
    },
    {
      "epoch": 0.6945908460471567,
      "grad_norm": 0.32218673825263977,
      "learning_rate": 1.5925654844438536e-07,
      "loss": 0.5148,
      "step": 12520
    },
    {
      "epoch": 0.6957004160887656,
      "grad_norm": 0.44593995809555054,
      "learning_rate": 1.582423874251703e-07,
      "loss": 0.4836,
      "step": 12540
    },
    {
      "epoch": 0.6968099861303745,
      "grad_norm": 0.3792048990726471,
      "learning_rate": 1.5722996849614066e-07,
      "loss": 0.5882,
      "step": 12560
    },
    {
      "epoch": 0.6979195561719833,
      "grad_norm": 0.33511149883270264,
      "learning_rate": 1.5621931087886324e-07,
      "loss": 0.5293,
      "step": 12580
    },
    {
      "epoch": 0.6990291262135923,
      "grad_norm": 0.4749152660369873,
      "learning_rate": 1.5521043376146494e-07,
      "loss": 0.4484,
      "step": 12600
    },
    {
      "epoch": 0.7001386962552011,
      "grad_norm": 0.36789247393608093,
      "learning_rate": 1.5420335629826856e-07,
      "loss": 0.5205,
      "step": 12620
    },
    {
      "epoch": 0.70124826629681,
      "grad_norm": 0.3322623074054718,
      "learning_rate": 1.5319809760942896e-07,
      "loss": 0.4483,
      "step": 12640
    },
    {
      "epoch": 0.7023578363384189,
      "grad_norm": 0.3089485466480255,
      "learning_rate": 1.5219467678057017e-07,
      "loss": 0.4467,
      "step": 12660
    },
    {
      "epoch": 0.7034674063800277,
      "grad_norm": 0.3379858136177063,
      "learning_rate": 1.511931128624231e-07,
      "loss": 0.5278,
      "step": 12680
    },
    {
      "epoch": 0.7045769764216366,
      "grad_norm": 0.3589012622833252,
      "learning_rate": 1.5019342487046355e-07,
      "loss": 0.4508,
      "step": 12700
    },
    {
      "epoch": 0.7056865464632455,
      "grad_norm": 0.36919447779655457,
      "learning_rate": 1.4919563178455153e-07,
      "loss": 0.4822,
      "step": 12720
    },
    {
      "epoch": 0.7067961165048544,
      "grad_norm": 0.3971253037452698,
      "learning_rate": 1.4819975254857066e-07,
      "loss": 0.5558,
      "step": 12740
    },
    {
      "epoch": 0.7079056865464632,
      "grad_norm": 0.37145859003067017,
      "learning_rate": 1.472058060700689e-07,
      "loss": 0.5777,
      "step": 12760
    },
    {
      "epoch": 0.7090152565880721,
      "grad_norm": 0.33810412883758545,
      "learning_rate": 1.46213811219899e-07,
      "loss": 0.4974,
      "step": 12780
    },
    {
      "epoch": 0.710124826629681,
      "grad_norm": 0.28394588828086853,
      "learning_rate": 1.452237868318606e-07,
      "loss": 0.4865,
      "step": 12800
    },
    {
      "epoch": 0.7112343966712898,
      "grad_norm": 0.6261323690414429,
      "learning_rate": 1.4423575170234267e-07,
      "loss": 0.5135,
      "step": 12820
    },
    {
      "epoch": 0.7123439667128988,
      "grad_norm": 0.5439554452896118,
      "learning_rate": 1.4324972458996638e-07,
      "loss": 0.596,
      "step": 12840
    },
    {
      "epoch": 0.7134535367545076,
      "grad_norm": 0.37857553362846375,
      "learning_rate": 1.422657242152293e-07,
      "loss": 0.6137,
      "step": 12860
    },
    {
      "epoch": 0.7145631067961165,
      "grad_norm": 0.5687951445579529,
      "learning_rate": 1.4128376926014957e-07,
      "loss": 0.425,
      "step": 12880
    },
    {
      "epoch": 0.7156726768377254,
      "grad_norm": 0.39318007230758667,
      "learning_rate": 1.4030387836791164e-07,
      "loss": 0.4716,
      "step": 12900
    },
    {
      "epoch": 0.7167822468793342,
      "grad_norm": 0.4043221175670624,
      "learning_rate": 1.3932607014251218e-07,
      "loss": 0.5187,
      "step": 12920
    },
    {
      "epoch": 0.7178918169209432,
      "grad_norm": 0.2890317440032959,
      "learning_rate": 1.3835036314840643e-07,
      "loss": 0.5747,
      "step": 12940
    },
    {
      "epoch": 0.719001386962552,
      "grad_norm": 0.2909344434738159,
      "learning_rate": 1.3737677591015657e-07,
      "loss": 0.4737,
      "step": 12960
    },
    {
      "epoch": 0.7201109570041608,
      "grad_norm": 0.373444139957428,
      "learning_rate": 1.364053269120791e-07,
      "loss": 0.5127,
      "step": 12980
    },
    {
      "epoch": 0.7212205270457698,
      "grad_norm": 0.3974605202674866,
      "learning_rate": 1.3543603459789466e-07,
      "loss": 0.4898,
      "step": 13000
    },
    {
      "epoch": 0.7223300970873786,
      "grad_norm": 0.3636089563369751,
      "learning_rate": 1.3446891737037762e-07,
      "loss": 0.5415,
      "step": 13020
    },
    {
      "epoch": 0.7234396671289876,
      "grad_norm": 0.38493213057518005,
      "learning_rate": 1.3350399359100623e-07,
      "loss": 0.4693,
      "step": 13040
    },
    {
      "epoch": 0.7245492371705964,
      "grad_norm": 0.3705246150493622,
      "learning_rate": 1.3254128157961486e-07,
      "loss": 0.5556,
      "step": 13060
    },
    {
      "epoch": 0.7256588072122053,
      "grad_norm": 0.501947820186615,
      "learning_rate": 1.3158079961404534e-07,
      "loss": 0.5332,
      "step": 13080
    },
    {
      "epoch": 0.7267683772538142,
      "grad_norm": 0.2879910171031952,
      "learning_rate": 1.3062256592980064e-07,
      "loss": 0.5306,
      "step": 13100
    },
    {
      "epoch": 0.727877947295423,
      "grad_norm": 0.6388247013092041,
      "learning_rate": 1.296665987196983e-07,
      "loss": 0.4829,
      "step": 13120
    },
    {
      "epoch": 0.7289875173370319,
      "grad_norm": 0.34465721249580383,
      "learning_rate": 1.2871291613352477e-07,
      "loss": 0.4307,
      "step": 13140
    },
    {
      "epoch": 0.7300970873786408,
      "grad_norm": 0.34914347529411316,
      "learning_rate": 1.2776153627769159e-07,
      "loss": 0.5307,
      "step": 13160
    },
    {
      "epoch": 0.7312066574202497,
      "grad_norm": 0.3551192283630371,
      "learning_rate": 1.2681247721489074e-07,
      "loss": 0.5591,
      "step": 13180
    },
    {
      "epoch": 0.7323162274618585,
      "grad_norm": 0.43952858448028564,
      "learning_rate": 1.2586575696375238e-07,
      "loss": 0.5065,
      "step": 13200
    },
    {
      "epoch": 0.7334257975034674,
      "grad_norm": 0.4221738576889038,
      "learning_rate": 1.249213934985025e-07,
      "loss": 0.5212,
      "step": 13220
    },
    {
      "epoch": 0.7345353675450763,
      "grad_norm": 0.5308628678321838,
      "learning_rate": 1.2397940474862144e-07,
      "loss": 0.4936,
      "step": 13240
    },
    {
      "epoch": 0.7356449375866851,
      "grad_norm": 0.45746496319770813,
      "learning_rate": 1.2303980859850402e-07,
      "loss": 0.4479,
      "step": 13260
    },
    {
      "epoch": 0.7367545076282941,
      "grad_norm": 0.38960978388786316,
      "learning_rate": 1.2210262288711933e-07,
      "loss": 0.4848,
      "step": 13280
    },
    {
      "epoch": 0.7378640776699029,
      "grad_norm": 0.7810402512550354,
      "learning_rate": 1.2116786540767267e-07,
      "loss": 0.4522,
      "step": 13300
    },
    {
      "epoch": 0.7389736477115117,
      "grad_norm": 0.7264504432678223,
      "learning_rate": 1.2023555390726748e-07,
      "loss": 0.5517,
      "step": 13320
    },
    {
      "epoch": 0.7400832177531207,
      "grad_norm": 0.3533945381641388,
      "learning_rate": 1.1930570608656803e-07,
      "loss": 0.4049,
      "step": 13340
    },
    {
      "epoch": 0.7411927877947295,
      "grad_norm": 0.25890418887138367,
      "learning_rate": 1.183783395994641e-07,
      "loss": 0.5448,
      "step": 13360
    },
    {
      "epoch": 0.7423023578363385,
      "grad_norm": 0.279067724943161,
      "learning_rate": 1.1745347205273506e-07,
      "loss": 0.5113,
      "step": 13380
    },
    {
      "epoch": 0.7434119278779473,
      "grad_norm": 0.31982362270355225,
      "learning_rate": 1.1653112100571619e-07,
      "loss": 0.5634,
      "step": 13400
    },
    {
      "epoch": 0.7445214979195561,
      "grad_norm": 0.3901461064815521,
      "learning_rate": 1.1561130396996508e-07,
      "loss": 0.5766,
      "step": 13420
    },
    {
      "epoch": 0.7456310679611651,
      "grad_norm": 0.32104188203811646,
      "learning_rate": 1.146940384089288e-07,
      "loss": 0.4248,
      "step": 13440
    },
    {
      "epoch": 0.7467406380027739,
      "grad_norm": 0.27771735191345215,
      "learning_rate": 1.1377934173761311e-07,
      "loss": 0.4721,
      "step": 13460
    },
    {
      "epoch": 0.7478502080443828,
      "grad_norm": 0.4484061300754547,
      "learning_rate": 1.1286723132225095e-07,
      "loss": 0.4968,
      "step": 13480
    },
    {
      "epoch": 0.7489597780859917,
      "grad_norm": 0.2738656997680664,
      "learning_rate": 1.1195772447997348e-07,
      "loss": 0.5468,
      "step": 13500
    },
    {
      "epoch": 0.7500693481276005,
      "grad_norm": 0.3913407325744629,
      "learning_rate": 1.1105083847848101e-07,
      "loss": 0.5727,
      "step": 13520
    },
    {
      "epoch": 0.7511789181692095,
      "grad_norm": 0.46406638622283936,
      "learning_rate": 1.1014659053571476e-07,
      "loss": 0.4827,
      "step": 13540
    },
    {
      "epoch": 0.7522884882108183,
      "grad_norm": 0.30459386110305786,
      "learning_rate": 1.092449978195308e-07,
      "loss": 0.5731,
      "step": 13560
    },
    {
      "epoch": 0.7533980582524272,
      "grad_norm": 0.4219912588596344,
      "learning_rate": 1.0834607744737329e-07,
      "loss": 0.5629,
      "step": 13580
    },
    {
      "epoch": 0.7545076282940361,
      "grad_norm": 0.5843199491500854,
      "learning_rate": 1.0744984648595006e-07,
      "loss": 0.5359,
      "step": 13600
    },
    {
      "epoch": 0.7556171983356449,
      "grad_norm": 0.31454548239707947,
      "learning_rate": 1.0655632195090822e-07,
      "loss": 0.4659,
      "step": 13620
    },
    {
      "epoch": 0.7567267683772538,
      "grad_norm": 0.4814988374710083,
      "learning_rate": 1.0566552080651133e-07,
      "loss": 0.4961,
      "step": 13640
    },
    {
      "epoch": 0.7578363384188627,
      "grad_norm": 0.32601091265678406,
      "learning_rate": 1.0477745996531739e-07,
      "loss": 0.4892,
      "step": 13660
    },
    {
      "epoch": 0.7589459084604716,
      "grad_norm": 0.46707651019096375,
      "learning_rate": 1.0389215628785725e-07,
      "loss": 0.4755,
      "step": 13680
    },
    {
      "epoch": 0.7600554785020804,
      "grad_norm": 0.31303638219833374,
      "learning_rate": 1.0300962658231521e-07,
      "loss": 0.4734,
      "step": 13700
    },
    {
      "epoch": 0.7611650485436893,
      "grad_norm": 0.3001532554626465,
      "learning_rate": 1.0212988760420918e-07,
      "loss": 0.5897,
      "step": 13720
    },
    {
      "epoch": 0.7622746185852982,
      "grad_norm": 0.26135823130607605,
      "learning_rate": 1.0125295605607324e-07,
      "loss": 0.5347,
      "step": 13740
    },
    {
      "epoch": 0.763384188626907,
      "grad_norm": 0.5185014009475708,
      "learning_rate": 1.0037884858714012e-07,
      "loss": 0.4531,
      "step": 13760
    },
    {
      "epoch": 0.764493758668516,
      "grad_norm": 0.4258882999420166,
      "learning_rate": 9.950758179302504e-08,
      "loss": 0.5889,
      "step": 13780
    },
    {
      "epoch": 0.7656033287101248,
      "grad_norm": 0.2963704466819763,
      "learning_rate": 9.863917221541104e-08,
      "loss": 0.4763,
      "step": 13800
    },
    {
      "epoch": 0.7667128987517337,
      "grad_norm": 0.41408637166023254,
      "learning_rate": 9.777363634173436e-08,
      "loss": 0.4918,
      "step": 13820
    },
    {
      "epoch": 0.7678224687933426,
      "grad_norm": 0.563586950302124,
      "learning_rate": 9.691099060487196e-08,
      "loss": 0.5427,
      "step": 13840
    },
    {
      "epoch": 0.7689320388349514,
      "grad_norm": 0.41873815655708313,
      "learning_rate": 9.605125138282935e-08,
      "loss": 0.4846,
      "step": 13860
    },
    {
      "epoch": 0.7700416088765604,
      "grad_norm": 0.333218514919281,
      "learning_rate": 9.519443499842919e-08,
      "loss": 0.4646,
      "step": 13880
    },
    {
      "epoch": 0.7711511789181692,
      "grad_norm": 0.3226572573184967,
      "learning_rate": 9.434055771900227e-08,
      "loss": 0.5374,
      "step": 13900
    },
    {
      "epoch": 0.772260748959778,
      "grad_norm": 0.3277279734611511,
      "learning_rate": 9.348963575607771e-08,
      "loss": 0.5319,
      "step": 13920
    },
    {
      "epoch": 0.773370319001387,
      "grad_norm": 0.3328832983970642,
      "learning_rate": 9.264168526507593e-08,
      "loss": 0.593,
      "step": 13940
    },
    {
      "epoch": 0.7744798890429958,
      "grad_norm": 0.4502674341201782,
      "learning_rate": 9.179672234500166e-08,
      "loss": 0.4532,
      "step": 13960
    },
    {
      "epoch": 0.7755894590846047,
      "grad_norm": 0.3654020130634308,
      "learning_rate": 9.095476303813796e-08,
      "loss": 0.4858,
      "step": 13980
    },
    {
      "epoch": 0.7766990291262136,
      "grad_norm": 0.2795443832874298,
      "learning_rate": 9.011582332974227e-08,
      "loss": 0.4836,
      "step": 14000
    },
    {
      "epoch": 0.7778085991678225,
      "grad_norm": 0.4435333013534546,
      "learning_rate": 8.927991914774227e-08,
      "loss": 0.5314,
      "step": 14020
    },
    {
      "epoch": 0.7789181692094314,
      "grad_norm": 0.41879114508628845,
      "learning_rate": 8.844706636243404e-08,
      "loss": 0.4772,
      "step": 14040
    },
    {
      "epoch": 0.7800277392510402,
      "grad_norm": 0.2757185399532318,
      "learning_rate": 8.761728078618049e-08,
      "loss": 0.513,
      "step": 14060
    },
    {
      "epoch": 0.7811373092926491,
      "grad_norm": 0.4560401439666748,
      "learning_rate": 8.679057817311095e-08,
      "loss": 0.5303,
      "step": 14080
    },
    {
      "epoch": 0.782246879334258,
      "grad_norm": 0.3912280797958374,
      "learning_rate": 8.596697421882257e-08,
      "loss": 0.4567,
      "step": 14100
    },
    {
      "epoch": 0.7833564493758669,
      "grad_norm": 0.5057780146598816,
      "learning_rate": 8.514648456008173e-08,
      "loss": 0.4742,
      "step": 14120
    },
    {
      "epoch": 0.7844660194174757,
      "grad_norm": 0.33308303356170654,
      "learning_rate": 8.43291247745277e-08,
      "loss": 0.5547,
      "step": 14140
    },
    {
      "epoch": 0.7855755894590846,
      "grad_norm": 0.3485460877418518,
      "learning_rate": 8.351491038037662e-08,
      "loss": 0.4894,
      "step": 14160
    },
    {
      "epoch": 0.7866851595006935,
      "grad_norm": 0.46615713834762573,
      "learning_rate": 8.270385683612674e-08,
      "loss": 0.3763,
      "step": 14180
    },
    {
      "epoch": 0.7877947295423023,
      "grad_norm": 0.3317703902721405,
      "learning_rate": 8.189597954026539e-08,
      "loss": 0.4526,
      "step": 14200
    },
    {
      "epoch": 0.7889042995839113,
      "grad_norm": 0.4380096197128296,
      "learning_rate": 8.1091293830976e-08,
      "loss": 0.5891,
      "step": 14220
    },
    {
      "epoch": 0.7900138696255201,
      "grad_norm": 0.3943984806537628,
      "learning_rate": 8.028981498584745e-08,
      "loss": 0.563,
      "step": 14240
    },
    {
      "epoch": 0.791123439667129,
      "grad_norm": 0.3383914828300476,
      "learning_rate": 7.949155822158385e-08,
      "loss": 0.5196,
      "step": 14260
    },
    {
      "epoch": 0.7922330097087379,
      "grad_norm": 0.33651596307754517,
      "learning_rate": 7.869653869371528e-08,
      "loss": 0.5427,
      "step": 14280
    },
    {
      "epoch": 0.7933425797503467,
      "grad_norm": 0.42295658588409424,
      "learning_rate": 7.790477149631072e-08,
      "loss": 0.5018,
      "step": 14300
    },
    {
      "epoch": 0.7944521497919557,
      "grad_norm": 0.39132261276245117,
      "learning_rate": 7.711627166169073e-08,
      "loss": 0.4734,
      "step": 14320
    },
    {
      "epoch": 0.7955617198335645,
      "grad_norm": 0.36338910460472107,
      "learning_rate": 7.633105416014277e-08,
      "loss": 0.4265,
      "step": 14340
    },
    {
      "epoch": 0.7966712898751733,
      "grad_norm": 0.3046381175518036,
      "learning_rate": 7.554913389963646e-08,
      "loss": 0.4241,
      "step": 14360
    },
    {
      "epoch": 0.7977808599167823,
      "grad_norm": 0.3695002794265747,
      "learning_rate": 7.477052572554065e-08,
      "loss": 0.4685,
      "step": 14380
    },
    {
      "epoch": 0.7988904299583911,
      "grad_norm": 0.3680543601512909,
      "learning_rate": 7.399524442034188e-08,
      "loss": 0.5151,
      "step": 14400
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.4030877649784088,
      "learning_rate": 7.322330470336313e-08,
      "loss": 0.568,
      "step": 14420
    },
    {
      "epoch": 0.8011095700416089,
      "grad_norm": 0.5655102729797363,
      "learning_rate": 7.245472123048499e-08,
      "loss": 0.4919,
      "step": 14440
    },
    {
      "epoch": 0.8022191400832177,
      "grad_norm": 0.3112393319606781,
      "learning_rate": 7.168950859386714e-08,
      "loss": 0.5639,
      "step": 14460
    },
    {
      "epoch": 0.8033287101248267,
      "grad_norm": 2.190739393234253,
      "learning_rate": 7.092768132167098e-08,
      "loss": 0.5222,
      "step": 14480
    },
    {
      "epoch": 0.8044382801664355,
      "grad_norm": 0.33552542328834534,
      "learning_rate": 7.01692538777845e-08,
      "loss": 0.5515,
      "step": 14500
    },
    {
      "epoch": 0.8055478502080444,
      "grad_norm": 0.5092620253562927,
      "learning_rate": 6.941424066154697e-08,
      "loss": 0.6103,
      "step": 14520
    },
    {
      "epoch": 0.8066574202496533,
      "grad_norm": 0.4809440076351166,
      "learning_rate": 6.866265600747604e-08,
      "loss": 0.5302,
      "step": 14540
    },
    {
      "epoch": 0.8077669902912621,
      "grad_norm": 0.41583776473999023,
      "learning_rate": 6.79145141849955e-08,
      "loss": 0.5071,
      "step": 14560
    },
    {
      "epoch": 0.808876560332871,
      "grad_norm": 0.3800281882286072,
      "learning_rate": 6.716982939816398e-08,
      "loss": 0.42,
      "step": 14580
    },
    {
      "epoch": 0.8099861303744799,
      "grad_norm": 0.32890620827674866,
      "learning_rate": 6.642861578540595e-08,
      "loss": 0.422,
      "step": 14600
    },
    {
      "epoch": 0.8110957004160888,
      "grad_norm": 0.33213478326797485,
      "learning_rate": 6.569088741924261e-08,
      "loss": 0.4859,
      "step": 14620
    },
    {
      "epoch": 0.8122052704576976,
      "grad_norm": 0.2846478521823883,
      "learning_rate": 6.495665830602518e-08,
      "loss": 0.5174,
      "step": 14640
    },
    {
      "epoch": 0.8133148404993065,
      "grad_norm": 0.5666544437408447,
      "learning_rate": 6.42259423856689e-08,
      "loss": 0.4581,
      "step": 14660
    },
    {
      "epoch": 0.8144244105409154,
      "grad_norm": 0.39947426319122314,
      "learning_rate": 6.349875353138801e-08,
      "loss": 0.4929,
      "step": 14680
    },
    {
      "epoch": 0.8155339805825242,
      "grad_norm": 0.7812756299972534,
      "learning_rate": 6.277510554943294e-08,
      "loss": 0.5503,
      "step": 14700
    },
    {
      "epoch": 0.8166435506241332,
      "grad_norm": 0.2935800850391388,
      "learning_rate": 6.205501217882766e-08,
      "loss": 0.5464,
      "step": 14720
    },
    {
      "epoch": 0.817753120665742,
      "grad_norm": 0.5092408061027527,
      "learning_rate": 6.13384870911092e-08,
      "loss": 0.515,
      "step": 14740
    },
    {
      "epoch": 0.8188626907073508,
      "grad_norm": 0.3239974081516266,
      "learning_rate": 6.062554389006794e-08,
      "loss": 0.5617,
      "step": 14760
    },
    {
      "epoch": 0.8199722607489598,
      "grad_norm": 0.3289891481399536,
      "learning_rate": 5.991619611148918e-08,
      "loss": 0.4832,
      "step": 14780
    },
    {
      "epoch": 0.8210818307905686,
      "grad_norm": 0.37206193804740906,
      "learning_rate": 5.9210457222896524e-08,
      "loss": 0.4863,
      "step": 14800
    },
    {
      "epoch": 0.8221914008321776,
      "grad_norm": 0.3518655300140381,
      "learning_rate": 5.850834062329574e-08,
      "loss": 0.4942,
      "step": 14820
    },
    {
      "epoch": 0.8233009708737864,
      "grad_norm": 0.3711952567100525,
      "learning_rate": 5.780985964292079e-08,
      "loss": 0.5641,
      "step": 14840
    },
    {
      "epoch": 0.8244105409153952,
      "grad_norm": 0.41170036792755127,
      "learning_rate": 5.711502754298059e-08,
      "loss": 0.4882,
      "step": 14860
    },
    {
      "epoch": 0.8255201109570042,
      "grad_norm": 0.5306410193443298,
      "learning_rate": 5.6423857515406876e-08,
      "loss": 0.5864,
      "step": 14880
    },
    {
      "epoch": 0.826629680998613,
      "grad_norm": 0.34095829725265503,
      "learning_rate": 5.573636268260451e-08,
      "loss": 0.5834,
      "step": 14900
    },
    {
      "epoch": 0.827739251040222,
      "grad_norm": 0.245326429605484,
      "learning_rate": 5.5052556097201525e-08,
      "loss": 0.4505,
      "step": 14920
    },
    {
      "epoch": 0.8288488210818308,
      "grad_norm": 0.514102041721344,
      "learning_rate": 5.437245074180191e-08,
      "loss": 0.4891,
      "step": 14940
    },
    {
      "epoch": 0.8299583911234397,
      "grad_norm": 0.8509578704833984,
      "learning_rate": 5.369605952873887e-08,
      "loss": 0.6081,
      "step": 14960
    },
    {
      "epoch": 0.8310679611650486,
      "grad_norm": 0.35718920826911926,
      "learning_rate": 5.302339529982961e-08,
      "loss": 0.5393,
      "step": 14980
    },
    {
      "epoch": 0.8321775312066574,
      "grad_norm": 0.35100287199020386,
      "learning_rate": 5.2354470826131785e-08,
      "loss": 0.5476,
      "step": 15000
    },
    {
      "epoch": 0.8332871012482663,
      "grad_norm": 0.37764522433280945,
      "learning_rate": 5.168929880770062e-08,
      "loss": 0.475,
      "step": 15020
    },
    {
      "epoch": 0.8343966712898752,
      "grad_norm": 0.4380689859390259,
      "learning_rate": 5.102789187334827e-08,
      "loss": 0.4952,
      "step": 15040
    },
    {
      "epoch": 0.8355062413314841,
      "grad_norm": 0.40711092948913574,
      "learning_rate": 5.0370262580403775e-08,
      "loss": 0.4711,
      "step": 15060
    },
    {
      "epoch": 0.8366158113730929,
      "grad_norm": 0.3597396910190582,
      "learning_rate": 4.9716423414474515e-08,
      "loss": 0.4656,
      "step": 15080
    },
    {
      "epoch": 0.8377253814147018,
      "grad_norm": 0.30235543847084045,
      "learning_rate": 4.906638678920963e-08,
      "loss": 0.5144,
      "step": 15100
    },
    {
      "epoch": 0.8388349514563107,
      "grad_norm": 0.3047267198562622,
      "learning_rate": 4.842016504606375e-08,
      "loss": 0.4962,
      "step": 15120
    },
    {
      "epoch": 0.8399445214979195,
      "grad_norm": 0.296040415763855,
      "learning_rate": 4.777777045406314e-08,
      "loss": 0.4285,
      "step": 15140
    },
    {
      "epoch": 0.8410540915395285,
      "grad_norm": 0.2601630985736847,
      "learning_rate": 4.71392152095727e-08,
      "loss": 0.4683,
      "step": 15160
    },
    {
      "epoch": 0.8421636615811373,
      "grad_norm": 0.42486798763275146,
      "learning_rate": 4.6504511436064014e-08,
      "loss": 0.5188,
      "step": 15180
    },
    {
      "epoch": 0.8432732316227461,
      "grad_norm": 0.4439660310745239,
      "learning_rate": 4.587367118388577e-08,
      "loss": 0.4948,
      "step": 15200
    },
    {
      "epoch": 0.8443828016643551,
      "grad_norm": 0.36130619049072266,
      "learning_rate": 4.5246706430034445e-08,
      "loss": 0.52,
      "step": 15220
    },
    {
      "epoch": 0.8454923717059639,
      "grad_norm": 0.3554399609565735,
      "learning_rate": 4.4623629077927296e-08,
      "loss": 0.4171,
      "step": 15240
    },
    {
      "epoch": 0.8466019417475729,
      "grad_norm": 0.3487074673175812,
      "learning_rate": 4.40044509571762e-08,
      "loss": 0.4701,
      "step": 15260
    },
    {
      "epoch": 0.8477115117891817,
      "grad_norm": 0.7752673029899597,
      "learning_rate": 4.338918382336296e-08,
      "loss": 0.4984,
      "step": 15280
    },
    {
      "epoch": 0.8488210818307905,
      "grad_norm": 0.3020077347755432,
      "learning_rate": 4.277783935781637e-08,
      "loss": 0.4251,
      "step": 15300
    },
    {
      "epoch": 0.8499306518723995,
      "grad_norm": 3.607598304748535,
      "learning_rate": 4.217042916739011e-08,
      "loss": 0.4703,
      "step": 15320
    },
    {
      "epoch": 0.8510402219140083,
      "grad_norm": 0.4934079945087433,
      "learning_rate": 4.156696478424279e-08,
      "loss": 0.4898,
      "step": 15340
    },
    {
      "epoch": 0.8521497919556172,
      "grad_norm": 0.3440416753292084,
      "learning_rate": 4.096745766561857e-08,
      "loss": 0.4242,
      "step": 15360
    },
    {
      "epoch": 0.8532593619972261,
      "grad_norm": 0.43925386667251587,
      "learning_rate": 4.0371919193629975e-08,
      "loss": 0.5167,
      "step": 15380
    },
    {
      "epoch": 0.8543689320388349,
      "grad_norm": 0.40181154012680054,
      "learning_rate": 3.9780360675041675e-08,
      "loss": 0.4832,
      "step": 15400
    },
    {
      "epoch": 0.8554785020804438,
      "grad_norm": 0.49073562026023865,
      "learning_rate": 3.9192793341055655e-08,
      "loss": 0.4619,
      "step": 15420
    },
    {
      "epoch": 0.8565880721220527,
      "grad_norm": 0.3399178087711334,
      "learning_rate": 3.860922834709832e-08,
      "loss": 0.4904,
      "step": 15440
    },
    {
      "epoch": 0.8576976421636616,
      "grad_norm": 0.3309305012226105,
      "learning_rate": 3.8029676772608324e-08,
      "loss": 0.5175,
      "step": 15460
    },
    {
      "epoch": 0.8588072122052705,
      "grad_norm": 0.33893635869026184,
      "learning_rate": 3.745414962082655e-08,
      "loss": 0.5904,
      "step": 15480
    },
    {
      "epoch": 0.8599167822468793,
      "grad_norm": 0.4869129955768585,
      "learning_rate": 3.688265781858707e-08,
      "loss": 0.4194,
      "step": 15500
    },
    {
      "epoch": 0.8610263522884882,
      "grad_norm": 0.4826425015926361,
      "learning_rate": 3.631521221610953e-08,
      "loss": 0.4774,
      "step": 15520
    },
    {
      "epoch": 0.8621359223300971,
      "grad_norm": 0.4436647295951843,
      "learning_rate": 3.575182358679349e-08,
      "loss": 0.5091,
      "step": 15540
    },
    {
      "epoch": 0.863245492371706,
      "grad_norm": 0.3870086669921875,
      "learning_rate": 3.5192502627013535e-08,
      "loss": 0.4934,
      "step": 15560
    },
    {
      "epoch": 0.8643550624133148,
      "grad_norm": 0.3462676405906677,
      "learning_rate": 3.463725995591646e-08,
      "loss": 0.5185,
      "step": 15580
    },
    {
      "epoch": 0.8654646324549237,
      "grad_norm": 0.3750855028629303,
      "learning_rate": 3.408610611521959e-08,
      "loss": 0.4889,
      "step": 15600
    },
    {
      "epoch": 0.8665742024965326,
      "grad_norm": 0.39250943064689636,
      "learning_rate": 3.3539051569010376e-08,
      "loss": 0.5594,
      "step": 15620
    },
    {
      "epoch": 0.8676837725381414,
      "grad_norm": 0.6177974343299866,
      "learning_rate": 3.29961067035483e-08,
      "loss": 0.5567,
      "step": 15640
    },
    {
      "epoch": 0.8687933425797504,
      "grad_norm": 0.8788308501243591,
      "learning_rate": 3.245728182706695e-08,
      "loss": 0.5487,
      "step": 15660
    },
    {
      "epoch": 0.8699029126213592,
      "grad_norm": 0.3534790277481079,
      "learning_rate": 3.1922587169578965e-08,
      "loss": 0.5047,
      "step": 15680
    },
    {
      "epoch": 0.871012482662968,
      "grad_norm": 0.7439823746681213,
      "learning_rate": 3.1392032882681524e-08,
      "loss": 0.619,
      "step": 15700
    },
    {
      "epoch": 0.872122052704577,
      "grad_norm": 0.43660464882850647,
      "learning_rate": 3.086562903936343e-08,
      "loss": 0.5613,
      "step": 15720
    },
    {
      "epoch": 0.8732316227461858,
      "grad_norm": 0.2849920988082886,
      "learning_rate": 3.0343385633814336e-08,
      "loss": 0.5407,
      "step": 15740
    },
    {
      "epoch": 0.8743411927877948,
      "grad_norm": 0.46800124645233154,
      "learning_rate": 2.982531258123447e-08,
      "loss": 0.5268,
      "step": 15760
    },
    {
      "epoch": 0.8754507628294036,
      "grad_norm": 0.33923402428627014,
      "learning_rate": 2.931141971764675e-08,
      "loss": 0.5359,
      "step": 15780
    },
    {
      "epoch": 0.8765603328710125,
      "grad_norm": 0.5203589200973511,
      "learning_rate": 2.880171679971005e-08,
      "loss": 0.4298,
      "step": 15800
    },
    {
      "epoch": 0.8776699029126214,
      "grad_norm": 0.48814857006073,
      "learning_rate": 2.8296213504533596e-08,
      "loss": 0.4622,
      "step": 15820
    },
    {
      "epoch": 0.8787794729542302,
      "grad_norm": 0.32717978954315186,
      "learning_rate": 2.779491942949369e-08,
      "loss": 0.4351,
      "step": 15840
    },
    {
      "epoch": 0.8798890429958391,
      "grad_norm": 0.33301976323127747,
      "learning_rate": 2.7297844092051104e-08,
      "loss": 0.4853,
      "step": 15860
    },
    {
      "epoch": 0.880998613037448,
      "grad_norm": 0.42914196848869324,
      "learning_rate": 2.680499692957078e-08,
      "loss": 0.5133,
      "step": 15880
    },
    {
      "epoch": 0.8821081830790569,
      "grad_norm": 0.3375394344329834,
      "learning_rate": 2.6316387299142374e-08,
      "loss": 0.514,
      "step": 15900
    },
    {
      "epoch": 0.8832177531206657,
      "grad_norm": 0.33002445101737976,
      "learning_rate": 2.5832024477402543e-08,
      "loss": 0.4487,
      "step": 15920
    },
    {
      "epoch": 0.8843273231622746,
      "grad_norm": 0.5125333666801453,
      "learning_rate": 2.535191766035913e-08,
      "loss": 0.5942,
      "step": 15940
    },
    {
      "epoch": 0.8854368932038835,
      "grad_norm": 0.424376517534256,
      "learning_rate": 2.4876075963216226e-08,
      "loss": 0.5574,
      "step": 15960
    },
    {
      "epoch": 0.8865464632454924,
      "grad_norm": 1.1631702184677124,
      "learning_rate": 2.4404508420201446e-08,
      "loss": 0.5152,
      "step": 15980
    },
    {
      "epoch": 0.8876560332871013,
      "grad_norm": 0.42409613728523254,
      "learning_rate": 2.3937223984394212e-08,
      "loss": 0.5859,
      "step": 16000
    },
    {
      "epoch": 0.8887656033287101,
      "grad_norm": 0.32093319296836853,
      "learning_rate": 2.3474231527555595e-08,
      "loss": 0.5776,
      "step": 16020
    },
    {
      "epoch": 0.889875173370319,
      "grad_norm": 0.44207271933555603,
      "learning_rate": 2.301553983996041e-08,
      "loss": 0.5397,
      "step": 16040
    },
    {
      "epoch": 0.8909847434119279,
      "grad_norm": 0.29899469017982483,
      "learning_rate": 2.2561157630229673e-08,
      "loss": 0.5171,
      "step": 16060
    },
    {
      "epoch": 0.8920943134535367,
      "grad_norm": 0.49811238050460815,
      "learning_rate": 2.2111093525165826e-08,
      "loss": 0.5837,
      "step": 16080
    },
    {
      "epoch": 0.8932038834951457,
      "grad_norm": 0.36482110619544983,
      "learning_rate": 2.1665356069588607e-08,
      "loss": 0.5252,
      "step": 16100
    },
    {
      "epoch": 0.8943134535367545,
      "grad_norm": 0.40325450897216797,
      "learning_rate": 2.1223953726172917e-08,
      "loss": 0.5157,
      "step": 16120
    },
    {
      "epoch": 0.8954230235783633,
      "grad_norm": 0.2874036431312561,
      "learning_rate": 2.078689487528823e-08,
      "loss": 0.5223,
      "step": 16140
    },
    {
      "epoch": 0.8965325936199723,
      "grad_norm": 0.460287481546402,
      "learning_rate": 2.0354187814839248e-08,
      "loss": 0.6041,
      "step": 16160
    },
    {
      "epoch": 0.8976421636615811,
      "grad_norm": 0.3832845687866211,
      "learning_rate": 1.992584076010867e-08,
      "loss": 0.5905,
      "step": 16180
    },
    {
      "epoch": 0.8987517337031901,
      "grad_norm": 0.47805336117744446,
      "learning_rate": 1.9501861843601114e-08,
      "loss": 0.4894,
      "step": 16200
    },
    {
      "epoch": 0.8998613037447989,
      "grad_norm": 0.34525078535079956,
      "learning_rate": 1.9082259114888477e-08,
      "loss": 0.555,
      "step": 16220
    },
    {
      "epoch": 0.9009708737864077,
      "grad_norm": 0.37707841396331787,
      "learning_rate": 1.8667040540457423e-08,
      "loss": 0.4523,
      "step": 16240
    },
    {
      "epoch": 0.9020804438280167,
      "grad_norm": 0.4305242896080017,
      "learning_rate": 1.8256214003558035e-08,
      "loss": 0.5538,
      "step": 16260
    },
    {
      "epoch": 0.9031900138696255,
      "grad_norm": 0.4085891842842102,
      "learning_rate": 1.7849787304054093e-08,
      "loss": 0.5101,
      "step": 16280
    },
    {
      "epoch": 0.9042995839112344,
      "grad_norm": 0.44601982831954956,
      "learning_rate": 1.7447768158274923e-08,
      "loss": 0.4732,
      "step": 16300
    },
    {
      "epoch": 0.9054091539528433,
      "grad_norm": 0.3422205150127411,
      "learning_rate": 1.7050164198869148e-08,
      "loss": 0.4478,
      "step": 16320
    },
    {
      "epoch": 0.9065187239944521,
      "grad_norm": 0.26549020409584045,
      "learning_rate": 1.6656982974659563e-08,
      "loss": 0.5429,
      "step": 16340
    },
    {
      "epoch": 0.907628294036061,
      "grad_norm": 0.47383949160575867,
      "learning_rate": 1.6268231950499727e-08,
      "loss": 0.5087,
      "step": 16360
    },
    {
      "epoch": 0.9087378640776699,
      "grad_norm": 1.3218978643417358,
      "learning_rate": 1.5883918507132637e-08,
      "loss": 0.5044,
      "step": 16380
    },
    {
      "epoch": 0.9098474341192788,
      "grad_norm": 0.3553486168384552,
      "learning_rate": 1.550404994105009e-08,
      "loss": 0.5442,
      "step": 16400
    },
    {
      "epoch": 0.9109570041608877,
      "grad_norm": 0.2937772274017334,
      "learning_rate": 1.5128633464354584e-08,
      "loss": 0.4458,
      "step": 16420
    },
    {
      "epoch": 0.9120665742024965,
      "grad_norm": 0.49511197209358215,
      "learning_rate": 1.475767620462215e-08,
      "loss": 0.4199,
      "step": 16440
    },
    {
      "epoch": 0.9131761442441054,
      "grad_norm": 0.7402114272117615,
      "learning_rate": 1.439118520476701e-08,
      "loss": 0.5255,
      "step": 16460
    },
    {
      "epoch": 0.9142857142857143,
      "grad_norm": 0.44214996695518494,
      "learning_rate": 1.4029167422908105e-08,
      "loss": 0.4961,
      "step": 16480
    },
    {
      "epoch": 0.9153952843273232,
      "grad_norm": 0.42241570353507996,
      "learning_rate": 1.3671629732236679e-08,
      "loss": 0.5096,
      "step": 16500
    },
    {
      "epoch": 0.916504854368932,
      "grad_norm": 0.5015203952789307,
      "learning_rate": 1.3318578920886003e-08,
      "loss": 0.549,
      "step": 16520
    },
    {
      "epoch": 0.9176144244105409,
      "grad_norm": 0.4513172209262848,
      "learning_rate": 1.2970021691802475e-08,
      "loss": 0.5027,
      "step": 16540
    },
    {
      "epoch": 0.9187239944521498,
      "grad_norm": 0.3667598366737366,
      "learning_rate": 1.2625964662618172e-08,
      "loss": 0.4524,
      "step": 16560
    },
    {
      "epoch": 0.9198335644937586,
      "grad_norm": 0.3975818157196045,
      "learning_rate": 1.2286414365525494e-08,
      "loss": 0.4872,
      "step": 16580
    },
    {
      "epoch": 0.9209431345353676,
      "grad_norm": 0.4363032281398773,
      "learning_rate": 1.1951377247152867e-08,
      "loss": 0.6175,
      "step": 16600
    },
    {
      "epoch": 0.9220527045769764,
      "grad_norm": 0.2995266318321228,
      "learning_rate": 1.162085966844259e-08,
      "loss": 0.5223,
      "step": 16620
    },
    {
      "epoch": 0.9231622746185852,
      "grad_norm": 0.29473140835762024,
      "learning_rate": 1.1294867904529992e-08,
      "loss": 0.5011,
      "step": 16640
    },
    {
      "epoch": 0.9242718446601942,
      "grad_norm": 0.3131171464920044,
      "learning_rate": 1.097340814462408e-08,
      "loss": 0.5525,
      "step": 16660
    },
    {
      "epoch": 0.925381414701803,
      "grad_norm": 0.29646238684654236,
      "learning_rate": 1.065648649189041e-08,
      "loss": 0.4261,
      "step": 16680
    },
    {
      "epoch": 0.926490984743412,
      "grad_norm": 0.3700522482395172,
      "learning_rate": 1.0344108963334847e-08,
      "loss": 0.4667,
      "step": 16700
    },
    {
      "epoch": 0.9276005547850208,
      "grad_norm": 0.5135470628738403,
      "learning_rate": 1.003628148968963e-08,
      "loss": 0.5734,
      "step": 16720
    },
    {
      "epoch": 0.9287101248266297,
      "grad_norm": 0.3333655595779419,
      "learning_rate": 9.733009915300628e-09,
      "loss": 0.5045,
      "step": 16740
    },
    {
      "epoch": 0.9298196948682386,
      "grad_norm": 0.5014081001281738,
      "learning_rate": 9.434299998016287e-09,
      "loss": 0.5693,
      "step": 16760
    },
    {
      "epoch": 0.9309292649098474,
      "grad_norm": 0.33439749479293823,
      "learning_rate": 9.140157409078559e-09,
      "loss": 0.5434,
      "step": 16780
    },
    {
      "epoch": 0.9320388349514563,
      "grad_norm": 0.28119370341300964,
      "learning_rate": 8.850587733014947e-09,
      "loss": 0.5789,
      "step": 16800
    },
    {
      "epoch": 0.9331484049930652,
      "grad_norm": 0.3373952805995941,
      "learning_rate": 8.565596467532715e-09,
      "loss": 0.4614,
      "step": 16820
    },
    {
      "epoch": 0.9342579750346741,
      "grad_norm": 0.4669179916381836,
      "learning_rate": 8.28518902341438e-09,
      "loss": 0.5021,
      "step": 16840
    },
    {
      "epoch": 0.935367545076283,
      "grad_norm": 0.45809802412986755,
      "learning_rate": 8.009370724415015e-09,
      "loss": 0.5104,
      "step": 16860
    },
    {
      "epoch": 0.9364771151178918,
      "grad_norm": 0.40159252285957336,
      "learning_rate": 7.738146807161255e-09,
      "loss": 0.5569,
      "step": 16880
    },
    {
      "epoch": 0.9375866851595007,
      "grad_norm": 0.34096261858940125,
      "learning_rate": 7.471522421051618e-09,
      "loss": 0.5477,
      "step": 16900
    },
    {
      "epoch": 0.9386962552011096,
      "grad_norm": 0.4823736250400543,
      "learning_rate": 7.209502628159142e-09,
      "loss": 0.4552,
      "step": 16920
    },
    {
      "epoch": 0.9398058252427185,
      "grad_norm": 0.37772753834724426,
      "learning_rate": 6.952092403134851e-09,
      "loss": 0.4999,
      "step": 16940
    },
    {
      "epoch": 0.9409153952843273,
      "grad_norm": 0.44477227330207825,
      "learning_rate": 6.69929663311361e-09,
      "loss": 0.596,
      "step": 16960
    },
    {
      "epoch": 0.9420249653259362,
      "grad_norm": 0.27438193559646606,
      "learning_rate": 6.451120117621306e-09,
      "loss": 0.483,
      "step": 16980
    },
    {
      "epoch": 0.9431345353675451,
      "grad_norm": 0.3919523060321808,
      "learning_rate": 6.2075675684835075e-09,
      "loss": 0.4991,
      "step": 17000
    },
    {
      "epoch": 0.9442441054091539,
      "grad_norm": 0.36142680048942566,
      "learning_rate": 5.968643609736257e-09,
      "loss": 0.4884,
      "step": 17020
    },
    {
      "epoch": 0.9453536754507629,
      "grad_norm": 0.37626388669013977,
      "learning_rate": 5.734352777538143e-09,
      "loss": 0.473,
      "step": 17040
    },
    {
      "epoch": 0.9464632454923717,
      "grad_norm": 0.8457902073860168,
      "learning_rate": 5.504699520084227e-09,
      "loss": 0.4457,
      "step": 17060
    },
    {
      "epoch": 0.9475728155339805,
      "grad_norm": 0.26701247692108154,
      "learning_rate": 5.279688197521643e-09,
      "loss": 0.506,
      "step": 17080
    },
    {
      "epoch": 0.9486823855755895,
      "grad_norm": 0.33689752221107483,
      "learning_rate": 5.059323081866601e-09,
      "loss": 0.4893,
      "step": 17100
    },
    {
      "epoch": 0.9497919556171983,
      "grad_norm": 0.30724218487739563,
      "learning_rate": 4.8436083569236004e-09,
      "loss": 0.4775,
      "step": 17120
    },
    {
      "epoch": 0.9509015256588073,
      "grad_norm": 0.3921775817871094,
      "learning_rate": 4.632548118205681e-09,
      "loss": 0.6024,
      "step": 17140
    },
    {
      "epoch": 0.9520110957004161,
      "grad_norm": 0.465593159198761,
      "learning_rate": 4.4261463728569315e-09,
      "loss": 0.5698,
      "step": 17160
    },
    {
      "epoch": 0.9531206657420249,
      "grad_norm": 0.3770931661128998,
      "learning_rate": 4.224407039576244e-09,
      "loss": 0.4477,
      "step": 17180
    },
    {
      "epoch": 0.9542302357836339,
      "grad_norm": 0.7365812063217163,
      "learning_rate": 4.027333948542932e-09,
      "loss": 0.4356,
      "step": 17200
    },
    {
      "epoch": 0.9553398058252427,
      "grad_norm": 0.4825473427772522,
      "learning_rate": 3.834930841344119e-09,
      "loss": 0.5003,
      "step": 17220
    },
    {
      "epoch": 0.9564493758668516,
      "grad_norm": 0.4335998296737671,
      "learning_rate": 3.6472013709035464e-09,
      "loss": 0.5292,
      "step": 17240
    },
    {
      "epoch": 0.9575589459084605,
      "grad_norm": 0.3976069688796997,
      "learning_rate": 3.4641491014123224e-09,
      "loss": 0.6148,
      "step": 17260
    },
    {
      "epoch": 0.9586685159500693,
      "grad_norm": 0.45377933979034424,
      "learning_rate": 3.2857775082613115e-09,
      "loss": 0.5478,
      "step": 17280
    },
    {
      "epoch": 0.9597780859916782,
      "grad_norm": 0.35374292731285095,
      "learning_rate": 3.1120899779749354e-09,
      "loss": 0.4997,
      "step": 17300
    },
    {
      "epoch": 0.9608876560332871,
      "grad_norm": 0.9163030385971069,
      "learning_rate": 2.9430898081471144e-09,
      "loss": 0.5127,
      "step": 17320
    },
    {
      "epoch": 0.961997226074896,
      "grad_norm": 0.4174667000770569,
      "learning_rate": 2.7787802073784563e-09,
      "loss": 0.4448,
      "step": 17340
    },
    {
      "epoch": 0.9631067961165048,
      "grad_norm": 0.41283664107322693,
      "learning_rate": 2.619164295215581e-09,
      "loss": 0.5467,
      "step": 17360
    },
    {
      "epoch": 0.9642163661581137,
      "grad_norm": 0.39738011360168457,
      "learning_rate": 2.4642451020916165e-09,
      "loss": 0.5459,
      "step": 17380
    },
    {
      "epoch": 0.9653259361997226,
      "grad_norm": 0.4814308285713196,
      "learning_rate": 2.314025569268879e-09,
      "loss": 0.4956,
      "step": 17400
    },
    {
      "epoch": 0.9664355062413315,
      "grad_norm": 0.3121536076068878,
      "learning_rate": 2.1685085487829493e-09,
      "loss": 0.5044,
      "step": 17420
    },
    {
      "epoch": 0.9675450762829404,
      "grad_norm": 0.5314778089523315,
      "learning_rate": 2.0276968033884347e-09,
      "loss": 0.5479,
      "step": 17440
    },
    {
      "epoch": 0.9686546463245492,
      "grad_norm": 0.49265632033348083,
      "learning_rate": 1.8915930065067365e-09,
      "loss": 0.4362,
      "step": 17460
    },
    {
      "epoch": 0.9697642163661581,
      "grad_norm": 0.3280515670776367,
      "learning_rate": 1.760199742175089e-09,
      "loss": 0.4533,
      "step": 17480
    },
    {
      "epoch": 0.970873786407767,
      "grad_norm": 0.3209471106529236,
      "learning_rate": 1.6335195049975992e-09,
      "loss": 0.523,
      "step": 17500
    },
    {
      "epoch": 0.9719833564493758,
      "grad_norm": 0.4209744334220886,
      "learning_rate": 1.5115547000978113e-09,
      "loss": 0.4551,
      "step": 17520
    },
    {
      "epoch": 0.9730929264909848,
      "grad_norm": 0.4232068359851837,
      "learning_rate": 1.3943076430731614e-09,
      "loss": 0.4994,
      "step": 17540
    },
    {
      "epoch": 0.9742024965325936,
      "grad_norm": 0.513118326663971,
      "learning_rate": 1.2817805599509014e-09,
      "loss": 0.5737,
      "step": 17560
    },
    {
      "epoch": 0.9753120665742026,
      "grad_norm": 0.4443225860595703,
      "learning_rate": 1.173975587145909e-09,
      "loss": 0.5862,
      "step": 17580
    },
    {
      "epoch": 0.9764216366158114,
      "grad_norm": 0.40851354598999023,
      "learning_rate": 1.0708947714200557e-09,
      "loss": 0.5229,
      "step": 17600
    },
    {
      "epoch": 0.9775312066574202,
      "grad_norm": 0.2527603209018707,
      "learning_rate": 9.725400698434583e-10,
      "loss": 0.5171,
      "step": 17620
    },
    {
      "epoch": 0.9786407766990292,
      "grad_norm": 0.43800926208496094,
      "learning_rate": 8.789133497571488e-10,
      "loss": 0.5315,
      "step": 17640
    },
    {
      "epoch": 0.979750346740638,
      "grad_norm": 0.33943334221839905,
      "learning_rate": 7.900163887377964e-10,
      "loss": 0.5071,
      "step": 17660
    },
    {
      "epoch": 0.9808599167822469,
      "grad_norm": 0.5179576277732849,
      "learning_rate": 7.058508745639014e-10,
      "loss": 0.5261,
      "step": 17680
    },
    {
      "epoch": 0.9819694868238558,
      "grad_norm": 0.6271900534629822,
      "learning_rate": 6.264184051837096e-10,
      "loss": 0.4876,
      "step": 17700
    },
    {
      "epoch": 0.9830790568654646,
      "grad_norm": 0.5471246838569641,
      "learning_rate": 5.517204886848758e-10,
      "loss": 0.4974,
      "step": 17720
    },
    {
      "epoch": 0.9841886269070735,
      "grad_norm": 0.30157485604286194,
      "learning_rate": 4.817585432659032e-10,
      "loss": 0.4899,
      "step": 17740
    },
    {
      "epoch": 0.9852981969486824,
      "grad_norm": 0.7236303091049194,
      "learning_rate": 4.1653389720916474e-10,
      "loss": 0.5567,
      "step": 17760
    },
    {
      "epoch": 0.9864077669902913,
      "grad_norm": 0.3008541762828827,
      "learning_rate": 3.5604778885564567e-10,
      "loss": 0.5912,
      "step": 17780
    },
    {
      "epoch": 0.9875173370319001,
      "grad_norm": 0.4916051924228668,
      "learning_rate": 3.0030136658157343e-10,
      "loss": 0.5003,
      "step": 17800
    },
    {
      "epoch": 0.988626907073509,
      "grad_norm": 0.34007692337036133,
      "learning_rate": 2.492956887764075e-10,
      "loss": 0.5146,
      "step": 17820
    },
    {
      "epoch": 0.9897364771151179,
      "grad_norm": 0.35913458466529846,
      "learning_rate": 2.0303172382293843e-10,
      "loss": 0.5003,
      "step": 17840
    },
    {
      "epoch": 0.9908460471567268,
      "grad_norm": 0.38319680094718933,
      "learning_rate": 1.6151035007883062e-10,
      "loss": 0.458,
      "step": 17860
    },
    {
      "epoch": 0.9919556171983357,
      "grad_norm": 0.4455585181713104,
      "learning_rate": 1.2473235585983012e-10,
      "loss": 0.4971,
      "step": 17880
    },
    {
      "epoch": 0.9930651872399445,
      "grad_norm": 0.521392822265625,
      "learning_rate": 9.269843942505407e-11,
      "loss": 0.5466,
      "step": 17900
    },
    {
      "epoch": 0.9941747572815534,
      "grad_norm": 0.2978236675262451,
      "learning_rate": 6.54092089634739e-11,
      "loss": 0.5089,
      "step": 17920
    },
    {
      "epoch": 0.9952843273231623,
      "grad_norm": 0.4199017584323883,
      "learning_rate": 4.286518258250771e-11,
      "loss": 0.4961,
      "step": 17940
    },
    {
      "epoch": 0.9963938973647711,
      "grad_norm": 0.39249876141548157,
      "learning_rate": 2.506678829819475e-11,
      "loss": 0.4764,
      "step": 17960
    },
    {
      "epoch": 0.9975034674063801,
      "grad_norm": 0.41262540221214294,
      "learning_rate": 1.2014364026979862e-11,
      "loss": 0.5336,
      "step": 17980
    },
    {
      "epoch": 0.9986130374479889,
      "grad_norm": 0.33444827795028687,
      "learning_rate": 3.708157579357385e-12,
      "loss": 0.5269,
      "step": 18000
    },
    {
      "epoch": 0.9997226074895977,
      "grad_norm": 0.423446387052536,
      "learning_rate": 1.4832665518049737e-13,
      "loss": 0.4982,
      "step": 18020
    },
    {
      "epoch": 1.0,
      "step": 18025,
      "total_flos": 4.409726465817354e+17,
      "train_loss": 0.606607598028302,
      "train_runtime": 18158.8522,
      "train_samples_per_second": 0.993,
      "train_steps_per_second": 0.993
    }
  ],
  "logging_steps": 20,
  "max_steps": 18025,
  "num_input_tokens_seen": 0,
  "num_train_epochs": 1,
  "save_steps": 100,
  "stateful_callbacks": {
    "TrainerControl": {
      "args": {
        "should_epoch_stop": false,
        "should_evaluate": false,
        "should_log": false,
        "should_save": true,
        "should_training_stop": true
      },
      "attributes": {}
    }
  },
  "total_flos": 4.409726465817354e+17,
  "train_batch_size": 1,
  "trial_name": null,
  "trial_params": null
}