{
  "best_metric": 0.1913878321647644,
  "best_model_checkpoint": "./ryan03312024_lr_2e-5_wd_001_v2/checkpoint-2800",
  "epoch": 2.0,
  "eval_steps": 100,
  "global_step": 5546,
  "is_hyper_param_search": false,
  "is_local_process_zero": true,
  "is_world_process_zero": true,
  "log_history": [
    {
      "epoch": 0.01,
      "grad_norm": 0.6066028475761414,
      "learning_rate": 1.9909844933285253e-05,
      "loss": 0.571,
      "step": 25
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.938056468963623,
      "learning_rate": 1.9819689866570503e-05,
      "loss": 0.4613,
      "step": 50
    },
    {
      "epoch": 0.03,
      "grad_norm": 1.0178314447402954,
      "learning_rate": 1.9729534799855754e-05,
      "loss": 0.4095,
      "step": 75
    },
    {
      "epoch": 0.04,
      "grad_norm": 1.4264367818832397,
      "learning_rate": 1.9639379733141005e-05,
      "loss": 0.4426,
      "step": 100
    },
    {
      "epoch": 0.04,
      "eval_loss": 0.37066397070884705,
      "eval_na_accuracy": 0.80756014585495,
      "eval_ordinal_accuracy": 0.3408682346343994,
      "eval_ordinal_mae": 0.8707404732704163,
      "eval_runtime": 343.5137,
      "eval_samples_per_second": 13.027,
      "eval_steps_per_second": 1.63,
      "step": 100
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.495741069316864,
      "learning_rate": 1.9549224666426253e-05,
      "loss": 0.3942,
      "step": 125
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.6811705827713013,
      "learning_rate": 1.9459069599711507e-05,
      "loss": 0.3667,
      "step": 150
    },
    {
      "epoch": 0.06,
      "grad_norm": 1.048833966255188,
      "learning_rate": 1.9368914532996758e-05,
      "loss": 0.3792,
      "step": 175
    },
    {
      "epoch": 0.07,
      "grad_norm": 0.517120897769928,
      "learning_rate": 1.9278759466282005e-05,
      "loss": 0.3133,
      "step": 200
    },
    {
      "epoch": 0.07,
      "eval_loss": 0.3202913999557495,
      "eval_na_accuracy": 0.7749140858650208,
      "eval_ordinal_accuracy": 0.4300025701522827,
      "eval_ordinal_mae": 0.8544500470161438,
      "eval_runtime": 201.5044,
      "eval_samples_per_second": 22.208,
      "eval_steps_per_second": 2.779,
      "step": 200
    },
    {
      "epoch": 0.08,
      "grad_norm": 1.3575235605239868,
      "learning_rate": 1.9188604399567256e-05,
      "loss": 0.3491,
      "step": 225
    },
    {
      "epoch": 0.09,
      "grad_norm": 1.0482808351516724,
      "learning_rate": 1.909844933285251e-05,
      "loss": 0.3655,
      "step": 250
    },
    {
      "epoch": 0.1,
      "grad_norm": 2.861924409866333,
      "learning_rate": 1.9008294266137758e-05,
      "loss": 0.3141,
      "step": 275
    },
    {
      "epoch": 0.11,
      "grad_norm": 1.2723792791366577,
      "learning_rate": 1.891813919942301e-05,
      "loss": 0.3349,
      "step": 300
    },
    {
      "epoch": 0.11,
      "eval_loss": 0.2997085452079773,
      "eval_na_accuracy": 0.8419243693351746,
      "eval_ordinal_accuracy": 0.4592858850955963,
      "eval_ordinal_mae": 0.8338660001754761,
      "eval_runtime": 195.4794,
      "eval_samples_per_second": 22.892,
      "eval_steps_per_second": 2.865,
      "step": 300
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.6588447690010071,
      "learning_rate": 1.882798413270826e-05,
      "loss": 0.2944,
      "step": 325
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.5759406089782715,
      "learning_rate": 1.873782906599351e-05,
      "loss": 0.2992,
      "step": 350
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.21856403350830078,
      "learning_rate": 1.864767399927876e-05,
      "loss": 0.305,
      "step": 375
    },
    {
      "epoch": 0.14,
      "grad_norm": 1.3187426328659058,
      "learning_rate": 1.8557518932564012e-05,
      "loss": 0.3173,
      "step": 400
    },
    {
      "epoch": 0.14,
      "eval_loss": 0.28696730732917786,
      "eval_na_accuracy": 0.8659793734550476,
      "eval_ordinal_accuracy": 0.48189055919647217,
      "eval_ordinal_mae": 0.7992547750473022,
      "eval_runtime": 196.5851,
      "eval_samples_per_second": 22.764,
      "eval_steps_per_second": 2.849,
      "step": 400
    },
    {
      "epoch": 0.15,
      "grad_norm": 1.22658109664917,
      "learning_rate": 1.846736386584926e-05,
      "loss": 0.2952,
      "step": 425
    },
    {
      "epoch": 0.16,
      "grad_norm": 2.2546815872192383,
      "learning_rate": 1.8377208799134514e-05,
      "loss": 0.3037,
      "step": 450
    },
    {
      "epoch": 0.17,
      "grad_norm": 2.8508079051971436,
      "learning_rate": 1.8287053732419765e-05,
      "loss": 0.3114,
      "step": 475
    },
    {
      "epoch": 0.18,
      "grad_norm": 1.2727103233337402,
      "learning_rate": 1.8196898665705013e-05,
      "loss": 0.2946,
      "step": 500
    },
    {
      "epoch": 0.18,
      "eval_loss": 0.2856302559375763,
      "eval_na_accuracy": 0.8934707641601562,
      "eval_ordinal_accuracy": 0.5111739039421082,
      "eval_ordinal_mae": 0.7690269351005554,
      "eval_runtime": 195.0911,
      "eval_samples_per_second": 22.938,
      "eval_steps_per_second": 2.87,
      "step": 500
    },
    {
      "epoch": 0.19,
      "grad_norm": 1.6682345867156982,
      "learning_rate": 1.8106743598990263e-05,
      "loss": 0.301,
      "step": 525
    },
    {
      "epoch": 0.2,
      "grad_norm": 1.0263928174972534,
      "learning_rate": 1.8016588532275518e-05,
      "loss": 0.2746,
      "step": 550
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.3150116503238678,
      "learning_rate": 1.7926433465560765e-05,
      "loss": 0.2676,
      "step": 575
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.7826778888702393,
      "learning_rate": 1.7836278398846016e-05,
      "loss": 0.3002,
      "step": 600
    },
    {
      "epoch": 0.22,
      "eval_loss": 0.2724498510360718,
      "eval_na_accuracy": 0.9209622144699097,
      "eval_ordinal_accuracy": 0.5345491766929626,
      "eval_ordinal_mae": 0.7232748866081238,
      "eval_runtime": 200.1722,
      "eval_samples_per_second": 22.356,
      "eval_steps_per_second": 2.798,
      "step": 600
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.7353246808052063,
      "learning_rate": 1.7746123332131267e-05,
      "loss": 0.303,
      "step": 625
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.5406638383865356,
      "learning_rate": 1.7655968265416518e-05,
      "loss": 0.2909,
      "step": 650
    },
    {
      "epoch": 0.24,
      "grad_norm": 1.443584680557251,
      "learning_rate": 1.756581319870177e-05,
      "loss": 0.2512,
      "step": 675
    },
    {
      "epoch": 0.25,
      "grad_norm": 2.028228521347046,
      "learning_rate": 1.747565813198702e-05,
      "loss": 0.2817,
      "step": 700
    },
    {
      "epoch": 0.25,
      "eval_loss": 0.26569074392318726,
      "eval_na_accuracy": 0.8625429272651672,
      "eval_ordinal_accuracy": 0.5566401481628418,
      "eval_ordinal_mae": 0.6927691102027893,
      "eval_runtime": 201.4281,
      "eval_samples_per_second": 22.216,
      "eval_steps_per_second": 2.78,
      "step": 700
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.40620824694633484,
      "learning_rate": 1.738550306527227e-05,
      "loss": 0.2732,
      "step": 725
    },
    {
      "epoch": 0.27,
      "grad_norm": 1.9016238451004028,
      "learning_rate": 1.729534799855752e-05,
      "loss": 0.2552,
      "step": 750
    },
    {
      "epoch": 0.28,
      "grad_norm": 2.003732919692993,
      "learning_rate": 1.7205192931842772e-05,
      "loss": 0.2538,
      "step": 775
    },
    {
      "epoch": 0.29,
      "grad_norm": 2.751253604888916,
      "learning_rate": 1.711503786512802e-05,
      "loss": 0.2939,
      "step": 800
    },
    {
      "epoch": 0.29,
      "eval_loss": 0.25960293412208557,
      "eval_na_accuracy": 0.7920961976051331,
      "eval_ordinal_accuracy": 0.5861803293228149,
      "eval_ordinal_mae": 0.6425440907478333,
      "eval_runtime": 200.1318,
      "eval_samples_per_second": 22.36,
      "eval_steps_per_second": 2.798,
      "step": 800
    },
    {
      "epoch": 0.3,
      "grad_norm": 2.241727113723755,
      "learning_rate": 1.702488279841327e-05,
      "loss": 0.2659,
      "step": 825
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.8341929912567139,
      "learning_rate": 1.6938333934367113e-05,
      "loss": 0.2654,
      "step": 850
    },
    {
      "epoch": 0.32,
      "grad_norm": 1.8499948978424072,
      "learning_rate": 1.6848178867652364e-05,
      "loss": 0.2774,
      "step": 875
    },
    {
      "epoch": 0.32,
      "grad_norm": 1.8184664249420166,
      "learning_rate": 1.6758023800937615e-05,
      "loss": 0.2525,
      "step": 900
    },
    {
      "epoch": 0.32,
      "eval_loss": 0.24585017561912537,
      "eval_na_accuracy": 0.8264604806900024,
      "eval_ordinal_accuracy": 0.6046750545501709,
      "eval_ordinal_mae": 0.6053154468536377,
      "eval_runtime": 203.517,
      "eval_samples_per_second": 21.988,
      "eval_steps_per_second": 2.752,
      "step": 900
    },
    {
      "epoch": 0.33,
      "grad_norm": 1.131605625152588,
      "learning_rate": 1.6667868734222866e-05,
      "loss": 0.2701,
      "step": 925
    },
    {
      "epoch": 0.34,
      "grad_norm": 2.05479097366333,
      "learning_rate": 1.6577713667508117e-05,
      "loss": 0.2956,
      "step": 950
    },
    {
      "epoch": 0.35,
      "grad_norm": 1.3499815464019775,
      "learning_rate": 1.6487558600793364e-05,
      "loss": 0.2446,
      "step": 975
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.7456974983215332,
      "learning_rate": 1.639740353407862e-05,
      "loss": 0.2163,
      "step": 1000
    },
    {
      "epoch": 0.36,
      "eval_loss": 0.2399808019399643,
      "eval_na_accuracy": 0.8109965920448303,
      "eval_ordinal_accuracy": 0.624454140663147,
      "eval_ordinal_mae": 0.5777344703674316,
      "eval_runtime": 217.2069,
      "eval_samples_per_second": 20.602,
      "eval_steps_per_second": 2.578,
      "step": 1000
    },
    {
      "epoch": 0.37,
      "grad_norm": 1.0125823020935059,
      "learning_rate": 1.6307248467363866e-05,
      "loss": 0.2168,
      "step": 1025
    },
    {
      "epoch": 0.38,
      "grad_norm": 1.9579060077667236,
      "learning_rate": 1.6217093400649117e-05,
      "loss": 0.2364,
      "step": 1050
    },
    {
      "epoch": 0.39,
      "grad_norm": 2.975767135620117,
      "learning_rate": 1.6126938333934368e-05,
      "loss": 0.2716,
      "step": 1075
    },
    {
      "epoch": 0.4,
      "grad_norm": 2.9766530990600586,
      "learning_rate": 1.603678326721962e-05,
      "loss": 0.2181,
      "step": 1100
    },
    {
      "epoch": 0.4,
      "eval_loss": 0.23391272127628326,
      "eval_na_accuracy": 0.876288652420044,
      "eval_ordinal_accuracy": 0.6023632287979126,
      "eval_ordinal_mae": 0.5429691672325134,
      "eval_runtime": 209.4151,
      "eval_samples_per_second": 21.369,
      "eval_steps_per_second": 2.674,
      "step": 1100
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.6088442802429199,
      "learning_rate": 1.594662820050487e-05,
      "loss": 0.2128,
      "step": 1125
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.3461017906665802,
      "learning_rate": 1.585647313379012e-05,
      "loss": 0.2145,
      "step": 1150
    },
    {
      "epoch": 0.42,
      "grad_norm": 1.341361165046692,
      "learning_rate": 1.576631806707537e-05,
      "loss": 0.2554,
      "step": 1175
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.5675370693206787,
      "learning_rate": 1.5676163000360622e-05,
      "loss": 0.1949,
      "step": 1200
    },
    {
      "epoch": 0.43,
      "eval_loss": 0.23308563232421875,
      "eval_na_accuracy": 0.7955326437950134,
      "eval_ordinal_accuracy": 0.6285640597343445,
      "eval_ordinal_mae": 0.5328579545021057,
      "eval_runtime": 209.3844,
      "eval_samples_per_second": 21.372,
      "eval_steps_per_second": 2.675,
      "step": 1200
    },
    {
      "epoch": 0.44,
      "grad_norm": 2.334373712539673,
      "learning_rate": 1.5586007933645873e-05,
      "loss": 0.2739,
      "step": 1225
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.5421320199966431,
      "learning_rate": 1.5495852866931124e-05,
      "loss": 0.2294,
      "step": 1250
    },
    {
      "epoch": 0.46,
      "grad_norm": 2.599592924118042,
      "learning_rate": 1.540569780021637e-05,
      "loss": 0.255,
      "step": 1275
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.6286499500274658,
      "learning_rate": 1.5315542733501626e-05,
      "loss": 0.214,
      "step": 1300
    },
    {
      "epoch": 0.47,
      "eval_loss": 0.24238631129264832,
      "eval_na_accuracy": 0.7628865838050842,
      "eval_ordinal_accuracy": 0.6182892322540283,
      "eval_ordinal_mae": 0.5243560671806335,
      "eval_runtime": 212.87,
      "eval_samples_per_second": 21.022,
      "eval_steps_per_second": 2.631,
      "step": 1300
    },
    {
      "epoch": 0.48,
      "grad_norm": 1.9406815767288208,
      "learning_rate": 1.5225387666786875e-05,
      "loss": 0.2225,
      "step": 1325
    },
    {
      "epoch": 0.49,
      "grad_norm": 3.733520269393921,
      "learning_rate": 1.5135232600072126e-05,
      "loss": 0.1918,
      "step": 1350
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.8792308568954468,
      "learning_rate": 1.5045077533357375e-05,
      "loss": 0.2179,
      "step": 1375
    },
    {
      "epoch": 0.5,
      "grad_norm": 1.1207715272903442,
      "learning_rate": 1.4954922466642628e-05,
      "loss": 0.27,
      "step": 1400
    },
    {
      "epoch": 0.5,
      "eval_loss": 0.22981737554073334,
      "eval_na_accuracy": 0.7869415879249573,
      "eval_ordinal_accuracy": 0.6367839574813843,
      "eval_ordinal_mae": 0.4994910955429077,
      "eval_runtime": 212.2299,
      "eval_samples_per_second": 21.086,
      "eval_steps_per_second": 2.639,
      "step": 1400
    },
    {
      "epoch": 0.51,
      "grad_norm": 2.47230863571167,
      "learning_rate": 1.4864767399927877e-05,
      "loss": 0.2348,
      "step": 1425
    },
    {
      "epoch": 0.52,
      "grad_norm": 1.015068769454956,
      "learning_rate": 1.4774612333213128e-05,
      "loss": 0.2593,
      "step": 1450
    },
    {
      "epoch": 0.53,
      "grad_norm": 1.7957515716552734,
      "learning_rate": 1.4684457266498377e-05,
      "loss": 0.2049,
      "step": 1475
    },
    {
      "epoch": 0.54,
      "grad_norm": 1.9824727773666382,
      "learning_rate": 1.459430219978363e-05,
      "loss": 0.2117,
      "step": 1500
    },
    {
      "epoch": 0.54,
      "eval_loss": 0.2300846129655838,
      "eval_na_accuracy": 0.7783505320549011,
      "eval_ordinal_accuracy": 0.6473156809806824,
      "eval_ordinal_mae": 0.4949534237384796,
      "eval_runtime": 210.619,
      "eval_samples_per_second": 21.247,
      "eval_steps_per_second": 2.659,
      "step": 1500
    },
    {
      "epoch": 0.55,
      "grad_norm": 1.1805070638656616,
      "learning_rate": 1.450414713306888e-05,
      "loss": 0.189,
      "step": 1525
    },
    {
      "epoch": 0.56,
      "grad_norm": 1.2057225704193115,
      "learning_rate": 1.441399206635413e-05,
      "loss": 0.2254,
      "step": 1550
    },
    {
      "epoch": 0.57,
      "grad_norm": 3.1214776039123535,
      "learning_rate": 1.432383699963938e-05,
      "loss": 0.2286,
      "step": 1575
    },
    {
      "epoch": 0.58,
      "grad_norm": 2.340480089187622,
      "learning_rate": 1.4233681932924633e-05,
      "loss": 0.2038,
      "step": 1600
    },
    {
      "epoch": 0.58,
      "eval_loss": 0.21561825275421143,
      "eval_na_accuracy": 0.8367697596549988,
      "eval_ordinal_accuracy": 0.6550218462944031,
      "eval_ordinal_mae": 0.48989665508270264,
      "eval_runtime": 207.7331,
      "eval_samples_per_second": 21.542,
      "eval_steps_per_second": 2.696,
      "step": 1600
    },
    {
      "epoch": 0.59,
      "grad_norm": 2.1810505390167236,
      "learning_rate": 1.4143526866209882e-05,
      "loss": 0.2223,
      "step": 1625
    },
    {
      "epoch": 0.6,
      "grad_norm": 2.7431559562683105,
      "learning_rate": 1.4053371799495133e-05,
      "loss": 0.2644,
      "step": 1650
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.8831206560134888,
      "learning_rate": 1.3963216732780382e-05,
      "loss": 0.2589,
      "step": 1675
    },
    {
      "epoch": 0.61,
      "grad_norm": 2.4183125495910645,
      "learning_rate": 1.3873061666065635e-05,
      "loss": 0.1974,
      "step": 1700
    },
    {
      "epoch": 0.61,
      "eval_loss": 0.22116971015930176,
      "eval_na_accuracy": 0.8281787037849426,
      "eval_ordinal_accuracy": 0.6347290277481079,
      "eval_ordinal_mae": 0.4638911485671997,
      "eval_runtime": 207.3318,
      "eval_samples_per_second": 21.584,
      "eval_steps_per_second": 2.701,
      "step": 1700
    },
    {
      "epoch": 0.62,
      "grad_norm": 1.318037986755371,
      "learning_rate": 1.3782906599350886e-05,
      "loss": 0.2456,
      "step": 1725
    },
    {
      "epoch": 0.63,
      "grad_norm": 3.9031999111175537,
      "learning_rate": 1.3692751532636135e-05,
      "loss": 0.2208,
      "step": 1750
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.9724792838096619,
      "learning_rate": 1.3602596465921386e-05,
      "loss": 0.1694,
      "step": 1775
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.8028815984725952,
      "learning_rate": 1.3512441399206637e-05,
      "loss": 0.1916,
      "step": 1800
    },
    {
      "epoch": 0.65,
      "eval_loss": 0.2150656282901764,
      "eval_na_accuracy": 0.8797250986099243,
      "eval_ordinal_accuracy": 0.6439763903617859,
      "eval_ordinal_mae": 0.4789562523365021,
      "eval_runtime": 211.6606,
      "eval_samples_per_second": 21.142,
      "eval_steps_per_second": 2.646,
      "step": 1800
    },
    {
      "epoch": 0.66,
      "grad_norm": 5.7413434982299805,
      "learning_rate": 1.3422286332491887e-05,
      "loss": 0.2097,
      "step": 1825
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.586044192314148,
      "learning_rate": 1.3332131265777137e-05,
      "loss": 0.1926,
      "step": 1850
    },
    {
      "epoch": 0.68,
      "grad_norm": 3.0166969299316406,
      "learning_rate": 1.3241976199062388e-05,
      "loss": 0.1901,
      "step": 1875
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.5749481916427612,
      "learning_rate": 1.315182113234764e-05,
      "loss": 0.1921,
      "step": 1900
    },
    {
      "epoch": 0.69,
      "eval_loss": 0.20500315725803375,
      "eval_na_accuracy": 0.8728522062301636,
      "eval_ordinal_accuracy": 0.6609298586845398,
      "eval_ordinal_mae": 0.4614226222038269,
      "eval_runtime": 179.4835,
      "eval_samples_per_second": 24.933,
      "eval_steps_per_second": 3.12,
      "step": 1900
    },
    {
      "epoch": 0.69,
      "grad_norm": 1.0241420269012451,
      "learning_rate": 1.306166606563289e-05,
      "loss": 0.2312,
      "step": 1925
    },
    {
      "epoch": 0.7,
      "grad_norm": 2.6365127563476562,
      "learning_rate": 1.2975117201586732e-05,
      "loss": 0.2358,
      "step": 1950
    },
    {
      "epoch": 0.71,
      "grad_norm": 2.4451234340667725,
      "learning_rate": 1.2884962134871981e-05,
      "loss": 0.2349,
      "step": 1975
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.9995049238204956,
      "learning_rate": 1.2794807068157232e-05,
      "loss": 0.1936,
      "step": 2000
    },
    {
      "epoch": 0.72,
      "eval_loss": 0.20607997477054596,
      "eval_na_accuracy": 0.8573883175849915,
      "eval_ordinal_accuracy": 0.6496275663375854,
      "eval_ordinal_mae": 0.4565940797328949,
      "eval_runtime": 175.1471,
      "eval_samples_per_second": 25.55,
      "eval_steps_per_second": 3.197,
      "step": 2000
    },
    {
      "epoch": 0.73,
      "grad_norm": 5.394123554229736,
      "learning_rate": 1.2704652001442481e-05,
      "loss": 0.2187,
      "step": 2025
    },
    {
      "epoch": 0.74,
      "grad_norm": 2.7290520668029785,
      "learning_rate": 1.2614496934727734e-05,
      "loss": 0.2144,
      "step": 2050
    },
    {
      "epoch": 0.75,
      "grad_norm": 2.8023579120635986,
      "learning_rate": 1.2524341868012983e-05,
      "loss": 0.1749,
      "step": 2075
    },
    {
      "epoch": 0.76,
      "grad_norm": 2.040255308151245,
      "learning_rate": 1.2434186801298234e-05,
      "loss": 0.1939,
      "step": 2100
    },
    {
      "epoch": 0.76,
      "eval_loss": 0.22944368422031403,
      "eval_na_accuracy": 0.9089347124099731,
      "eval_ordinal_accuracy": 0.6362702250480652,
      "eval_ordinal_mae": 0.4656565487384796,
      "eval_runtime": 176.4408,
      "eval_samples_per_second": 25.363,
      "eval_steps_per_second": 3.174,
      "step": 2100
    },
    {
      "epoch": 0.77,
      "grad_norm": 1.4762495756149292,
      "learning_rate": 1.2344031734583483e-05,
      "loss": 0.2211,
      "step": 2125
    },
    {
      "epoch": 0.78,
      "grad_norm": 3.6306161880493164,
      "learning_rate": 1.2253876667868735e-05,
      "loss": 0.2548,
      "step": 2150
    },
    {
      "epoch": 0.78,
      "grad_norm": 1.6580469608306885,
      "learning_rate": 1.2163721601153986e-05,
      "loss": 0.211,
      "step": 2175
    },
    {
      "epoch": 0.79,
      "grad_norm": 2.7241005897521973,
      "learning_rate": 1.2073566534439236e-05,
      "loss": 0.257,
      "step": 2200
    },
    {
      "epoch": 0.79,
      "eval_loss": 0.20536109805107117,
      "eval_na_accuracy": 0.8608247637748718,
      "eval_ordinal_accuracy": 0.6527100205421448,
      "eval_ordinal_mae": 0.45671576261520386,
      "eval_runtime": 176.5474,
      "eval_samples_per_second": 25.347,
      "eval_steps_per_second": 3.172,
      "step": 2200
    },
    {
      "epoch": 0.8,
      "grad_norm": 2.5323662757873535,
      "learning_rate": 1.1983411467724486e-05,
      "loss": 0.176,
      "step": 2225
    },
    {
      "epoch": 0.81,
      "grad_norm": 5.30660343170166,
      "learning_rate": 1.1893256401009739e-05,
      "loss": 0.195,
      "step": 2250
    },
    {
      "epoch": 0.82,
      "grad_norm": 3.7983808517456055,
      "learning_rate": 1.1803101334294988e-05,
      "loss": 0.1746,
      "step": 2275
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.6054279804229736,
      "learning_rate": 1.1712946267580239e-05,
      "loss": 0.2236,
      "step": 2300
    },
    {
      "epoch": 0.83,
      "eval_loss": 0.2043941468000412,
      "eval_na_accuracy": 0.876288652420044,
      "eval_ordinal_accuracy": 0.6640123128890991,
      "eval_ordinal_mae": 0.45420947670936584,
      "eval_runtime": 176.816,
      "eval_samples_per_second": 25.309,
      "eval_steps_per_second": 3.167,
      "step": 2300
    },
    {
      "epoch": 0.84,
      "grad_norm": 1.1567206382751465,
      "learning_rate": 1.1622791200865488e-05,
      "loss": 0.2479,
      "step": 2325
    },
    {
      "epoch": 0.85,
      "grad_norm": 5.52665376663208,
      "learning_rate": 1.1532636134150741e-05,
      "loss": 0.2271,
      "step": 2350
    },
    {
      "epoch": 0.86,
      "grad_norm": 2.9272255897521973,
      "learning_rate": 1.1442481067435992e-05,
      "loss": 0.2303,
      "step": 2375
    },
    {
      "epoch": 0.87,
      "grad_norm": 3.266409397125244,
      "learning_rate": 1.1352326000721241e-05,
      "loss": 0.1925,
      "step": 2400
    },
    {
      "epoch": 0.87,
      "eval_loss": 0.20849426090717316,
      "eval_na_accuracy": 0.80756014585495,
      "eval_ordinal_accuracy": 0.6886719465255737,
      "eval_ordinal_mae": 0.44630834460258484,
      "eval_runtime": 175.9296,
      "eval_samples_per_second": 25.436,
      "eval_steps_per_second": 3.183,
      "step": 2400
    },
    {
      "epoch": 0.87,
      "grad_norm": 2.3057689666748047,
      "learning_rate": 1.1262170934006492e-05,
      "loss": 0.2049,
      "step": 2425
    },
    {
      "epoch": 0.88,
      "grad_norm": 3.8032755851745605,
      "learning_rate": 1.1172015867291743e-05,
      "loss": 0.1909,
      "step": 2450
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.4314826428890228,
      "learning_rate": 1.1081860800576994e-05,
      "loss": 0.1825,
      "step": 2475
    },
    {
      "epoch": 0.9,
      "grad_norm": 3.1127185821533203,
      "learning_rate": 1.0991705733862243e-05,
      "loss": 0.1657,
      "step": 2500
    },
    {
      "epoch": 0.9,
      "eval_loss": 0.2033829689025879,
      "eval_na_accuracy": 0.8522336483001709,
      "eval_ordinal_accuracy": 0.6768559217453003,
      "eval_ordinal_mae": 0.4391731023788452,
      "eval_runtime": 174.624,
      "eval_samples_per_second": 25.626,
      "eval_steps_per_second": 3.207,
      "step": 2500
    },
    {
      "epoch": 0.91,
      "grad_norm": 4.6838459968566895,
      "learning_rate": 1.0901550667147494e-05,
      "loss": 0.2397,
      "step": 2525
    },
    {
      "epoch": 0.92,
      "grad_norm": 2.440084218978882,
      "learning_rate": 1.0811395600432746e-05,
      "loss": 0.1921,
      "step": 2550
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.5833593606948853,
      "learning_rate": 1.0721240533717995e-05,
      "loss": 0.188,
      "step": 2575
    },
    {
      "epoch": 0.94,
      "grad_norm": 2.1624417304992676,
      "learning_rate": 1.0631085467003246e-05,
      "loss": 0.1723,
      "step": 2600
    },
    {
      "epoch": 0.94,
      "eval_loss": 0.1956964135169983,
      "eval_na_accuracy": 0.838487982749939,
      "eval_ordinal_accuracy": 0.6755715608596802,
      "eval_ordinal_mae": 0.42565402388572693,
      "eval_runtime": 175.3137,
      "eval_samples_per_second": 25.526,
      "eval_steps_per_second": 3.194,
      "step": 2600
    },
    {
      "epoch": 0.95,
      "grad_norm": 3.9376819133758545,
      "learning_rate": 1.0540930400288496e-05,
      "loss": 0.1889,
      "step": 2625
    },
    {
      "epoch": 0.96,
      "grad_norm": 2.2914035320281982,
      "learning_rate": 1.0450775333573748e-05,
      "loss": 0.2078,
      "step": 2650
    },
    {
      "epoch": 0.96,
      "grad_norm": 1.2271968126296997,
      "learning_rate": 1.0360620266858999e-05,
      "loss": 0.2247,
      "step": 2675
    },
    {
      "epoch": 0.97,
      "grad_norm": 2.135350227355957,
      "learning_rate": 1.0270465200144248e-05,
      "loss": 0.2279,
      "step": 2700
    },
    {
      "epoch": 0.97,
      "eval_loss": 0.19462928175926208,
      "eval_na_accuracy": 0.8642611503601074,
      "eval_ordinal_accuracy": 0.6740303039550781,
      "eval_ordinal_mae": 0.4287233054637909,
      "eval_runtime": 175.4222,
      "eval_samples_per_second": 25.51,
      "eval_steps_per_second": 3.192,
      "step": 2700
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.897443413734436,
      "learning_rate": 1.0180310133429499e-05,
      "loss": 0.1672,
      "step": 2725
    },
    {
      "epoch": 0.99,
      "grad_norm": 1.9568222761154175,
      "learning_rate": 1.009015506671475e-05,
      "loss": 0.1597,
      "step": 2750
    },
    {
      "epoch": 1.0,
      "grad_norm": 1.222680926322937,
      "learning_rate": 1e-05,
      "loss": 0.2093,
      "step": 2775
    },
    {
      "epoch": 1.01,
      "grad_norm": 1.0666455030441284,
      "learning_rate": 9.909844933285252e-06,
      "loss": 0.1421,
      "step": 2800
    },
    {
      "epoch": 1.01,
      "eval_loss": 0.1913878321647644,
      "eval_na_accuracy": 0.8505154848098755,
      "eval_ordinal_accuracy": 0.6843051910400391,
      "eval_ordinal_mae": 0.41984713077545166,
      "eval_runtime": 177.3377,
      "eval_samples_per_second": 25.234,
      "eval_steps_per_second": 3.158,
      "step": 2800
    },
    {
      "epoch": 1.02,
      "grad_norm": 0.9463332295417786,
      "learning_rate": 9.819689866570503e-06,
      "loss": 0.1563,
      "step": 2825
    },
    {
      "epoch": 1.03,
      "grad_norm": 0.9075057506561279,
      "learning_rate": 9.729534799855753e-06,
      "loss": 0.1411,
      "step": 2850
    },
    {
      "epoch": 1.04,
      "grad_norm": 3.121269702911377,
      "learning_rate": 9.639379733141003e-06,
      "loss": 0.1368,
      "step": 2875
    },
    {
      "epoch": 1.05,
      "grad_norm": 0.5694425702095032,
      "learning_rate": 9.549224666426255e-06,
      "loss": 0.1116,
      "step": 2900
    },
    {
      "epoch": 1.05,
      "eval_loss": 0.20190554857254028,
      "eval_na_accuracy": 0.8230240345001221,
      "eval_ordinal_accuracy": 0.6704341173171997,
      "eval_ordinal_mae": 0.4214249551296234,
      "eval_runtime": 175.5908,
      "eval_samples_per_second": 25.485,
      "eval_steps_per_second": 3.189,
      "step": 2900
    },
    {
      "epoch": 1.05,
      "grad_norm": 4.727107048034668,
      "learning_rate": 9.459069599711504e-06,
      "loss": 0.1485,
      "step": 2925
    },
    {
      "epoch": 1.06,
      "grad_norm": 0.6293134689331055,
      "learning_rate": 9.368914532996755e-06,
      "loss": 0.116,
      "step": 2950
    },
    {
      "epoch": 1.07,
      "grad_norm": 0.7508574724197388,
      "learning_rate": 9.278759466282006e-06,
      "loss": 0.1656,
      "step": 2975
    },
    {
      "epoch": 1.08,
      "grad_norm": 0.7475106120109558,
      "learning_rate": 9.188604399567257e-06,
      "loss": 0.1194,
      "step": 3000
    },
    {
      "epoch": 1.08,
      "eval_loss": 0.19536536931991577,
      "eval_na_accuracy": 0.8367697596549988,
      "eval_ordinal_accuracy": 0.6807089447975159,
      "eval_ordinal_mae": 0.41778308153152466,
      "eval_runtime": 175.6051,
      "eval_samples_per_second": 25.483,
      "eval_steps_per_second": 3.189,
      "step": 3000
    },
    {
      "epoch": 1.09,
      "grad_norm": 0.6796592473983765,
      "learning_rate": 9.098449332852506e-06,
      "loss": 0.1483,
      "step": 3025
    },
    {
      "epoch": 1.1,
      "grad_norm": 1.7480669021606445,
      "learning_rate": 9.008294266137759e-06,
      "loss": 0.1083,
      "step": 3050
    },
    {
      "epoch": 1.11,
      "grad_norm": 0.7591541409492493,
      "learning_rate": 8.918139199423008e-06,
      "loss": 0.1463,
      "step": 3075
    },
    {
      "epoch": 1.12,
      "grad_norm": 0.8966888189315796,
      "learning_rate": 8.827984132708259e-06,
      "loss": 0.1312,
      "step": 3100
    },
    {
      "epoch": 1.12,
      "eval_loss": 0.19300740957260132,
      "eval_na_accuracy": 0.8591065406799316,
      "eval_ordinal_accuracy": 0.6873876452445984,
      "eval_ordinal_mae": 0.4165719449520111,
      "eval_runtime": 179.7716,
      "eval_samples_per_second": 24.893,
      "eval_steps_per_second": 3.115,
      "step": 3100
    },
    {
      "epoch": 1.13,
      "grad_norm": 1.1078789234161377,
      "learning_rate": 8.73782906599351e-06,
      "loss": 0.154,
      "step": 3125
    },
    {
      "epoch": 1.14,
      "grad_norm": 4.303710460662842,
      "learning_rate": 8.64767399927876e-06,
      "loss": 0.1328,
      "step": 3150
    },
    {
      "epoch": 1.14,
      "grad_norm": 0.9769502282142639,
      "learning_rate": 8.55751893256401e-06,
      "loss": 0.1253,
      "step": 3175
    },
    {
      "epoch": 1.15,
      "grad_norm": 2.734457492828369,
      "learning_rate": 8.467363865849262e-06,
      "loss": 0.1836,
      "step": 3200
    },
    {
      "epoch": 1.15,
      "eval_loss": 0.19890539348125458,
      "eval_na_accuracy": 0.8642611503601074,
      "eval_ordinal_accuracy": 0.6794245839118958,
      "eval_ordinal_mae": 0.4106903374195099,
      "eval_runtime": 173.0922,
      "eval_samples_per_second": 25.853,
      "eval_steps_per_second": 3.235,
      "step": 3200
    },
    {
      "epoch": 1.16,
      "grad_norm": 1.8836966753005981,
      "learning_rate": 8.377208799134512e-06,
      "loss": 0.1168,
      "step": 3225
    },
    {
      "epoch": 1.17,
      "grad_norm": 1.8701483011245728,
      "learning_rate": 8.287053732419763e-06,
      "loss": 0.1184,
      "step": 3250
    },
    {
      "epoch": 1.18,
      "grad_norm": 1.9784027338027954,
      "learning_rate": 8.196898665705013e-06,
      "loss": 0.1085,
      "step": 3275
    },
    {
      "epoch": 1.19,
      "grad_norm": 3.046462059020996,
      "learning_rate": 8.106743598990264e-06,
      "loss": 0.1282,
      "step": 3300
    },
    {
      "epoch": 1.19,
      "eval_loss": 0.195108100771904,
      "eval_na_accuracy": 0.8539518713951111,
      "eval_ordinal_accuracy": 0.6971487402915955,
      "eval_ordinal_mae": 0.412724107503891,
      "eval_runtime": 175.5444,
      "eval_samples_per_second": 25.492,
      "eval_steps_per_second": 3.19,
      "step": 3300
    },
    {
      "epoch": 1.2,
      "grad_norm": 1.4622838497161865,
      "learning_rate": 8.016588532275515e-06,
      "loss": 0.1428,
      "step": 3325
    },
    {
      "epoch": 1.21,
      "grad_norm": 1.1039246320724487,
      "learning_rate": 7.926433465560766e-06,
      "loss": 0.1027,
      "step": 3350
    },
    {
      "epoch": 1.22,
      "grad_norm": 1.006919503211975,
      "learning_rate": 7.836278398846015e-06,
      "loss": 0.1545,
      "step": 3375
    },
    {
      "epoch": 1.23,
      "grad_norm": 0.9952992796897888,
      "learning_rate": 7.746123332131266e-06,
      "loss": 0.1406,
      "step": 3400
    },
    {
      "epoch": 1.23,
      "eval_loss": 0.19589702785015106,
      "eval_na_accuracy": 0.8505154848098755,
      "eval_ordinal_accuracy": 0.6974055767059326,
      "eval_ordinal_mae": 0.4036160707473755,
      "eval_runtime": 176.1463,
      "eval_samples_per_second": 25.405,
      "eval_steps_per_second": 3.179,
      "step": 3400
    },
    {
      "epoch": 1.24,
      "grad_norm": 2.206535577774048,
      "learning_rate": 7.655968265416517e-06,
      "loss": 0.1523,
      "step": 3425
    },
    {
      "epoch": 1.24,
      "grad_norm": 2.8195924758911133,
      "learning_rate": 7.565813198701768e-06,
      "loss": 0.1357,
      "step": 3450
    },
    {
      "epoch": 1.25,
      "grad_norm": 1.383612871170044,
      "learning_rate": 7.475658131987018e-06,
      "loss": 0.1262,
      "step": 3475
    },
    {
      "epoch": 1.26,
      "grad_norm": 1.5167522430419922,
      "learning_rate": 7.385503065272269e-06,
      "loss": 0.0929,
      "step": 3500
    },
    {
      "epoch": 1.26,
      "eval_loss": 0.19685061275959015,
      "eval_na_accuracy": 0.8453608155250549,
      "eval_ordinal_accuracy": 0.6976624727249146,
      "eval_ordinal_mae": 0.4020155370235443,
      "eval_runtime": 170.2776,
      "eval_samples_per_second": 26.281,
      "eval_steps_per_second": 3.289,
      "step": 3500
    },
    {
      "epoch": 1.27,
      "grad_norm": 0.9208484888076782,
      "learning_rate": 7.295347998557519e-06,
      "loss": 0.1193,
      "step": 3525
    },
    {
      "epoch": 1.28,
      "grad_norm": 1.8087962865829468,
      "learning_rate": 7.205192931842771e-06,
      "loss": 0.1109,
      "step": 3550
    },
    {
      "epoch": 1.29,
      "grad_norm": 1.2812789678573608,
      "learning_rate": 7.115037865128021e-06,
      "loss": 0.1778,
      "step": 3575
    },
    {
      "epoch": 1.3,
      "grad_norm": 0.8966251015663147,
      "learning_rate": 7.0248827984132715e-06,
      "loss": 0.1135,
      "step": 3600
    },
    {
      "epoch": 1.3,
      "eval_loss": 0.19572028517723083,
      "eval_na_accuracy": 0.831615149974823,
      "eval_ordinal_accuracy": 0.6981762051582336,
      "eval_ordinal_mae": 0.4026087522506714,
      "eval_runtime": 171.6945,
      "eval_samples_per_second": 26.064,
      "eval_steps_per_second": 3.262,
      "step": 3600
    },
    {
      "epoch": 1.31,
      "grad_norm": 0.8011260628700256,
      "learning_rate": 6.9347277316985216e-06,
      "loss": 0.1078,
      "step": 3625
    },
    {
      "epoch": 1.32,
      "grad_norm": 0.5675652623176575,
      "learning_rate": 6.844572664983773e-06,
      "loss": 0.1114,
      "step": 3650
    },
    {
      "epoch": 1.33,
      "grad_norm": 1.0519170761108398,
      "learning_rate": 6.754417598269023e-06,
      "loss": 0.1435,
      "step": 3675
    },
    {
      "epoch": 1.33,
      "grad_norm": 1.6201542615890503,
      "learning_rate": 6.664262531554274e-06,
      "loss": 0.1345,
      "step": 3700
    },
    {
      "epoch": 1.33,
      "eval_loss": 0.1987382471561432,
      "eval_na_accuracy": 0.8814433217048645,
      "eval_ordinal_accuracy": 0.6832776665687561,
      "eval_ordinal_mae": 0.4107266366481781,
      "eval_runtime": 173.6501,
      "eval_samples_per_second": 25.77,
      "eval_steps_per_second": 3.225,
      "step": 3700
    },
    {
      "epoch": 1.34,
      "grad_norm": 8.633369445800781,
      "learning_rate": 6.574107464839524e-06,
      "loss": 0.174,
      "step": 3725
    },
    {
      "epoch": 1.35,
      "grad_norm": 1.3483631610870361,
      "learning_rate": 6.483952398124775e-06,
      "loss": 0.1104,
      "step": 3750
    },
    {
      "epoch": 1.36,
      "grad_norm": 2.3553450107574463,
      "learning_rate": 6.393797331410025e-06,
      "loss": 0.1401,
      "step": 3775
    },
    {
      "epoch": 1.37,
      "grad_norm": 6.273257732391357,
      "learning_rate": 6.303642264695277e-06,
      "loss": 0.1198,
      "step": 3800
    },
    {
      "epoch": 1.37,
      "eval_loss": 0.19685617089271545,
      "eval_na_accuracy": 0.8522336483001709,
      "eval_ordinal_accuracy": 0.6992036700248718,
      "eval_ordinal_mae": 0.3987964391708374,
      "eval_runtime": 176.5744,
      "eval_samples_per_second": 25.343,
      "eval_steps_per_second": 3.171,
      "step": 3800
    },
    {
      "epoch": 1.38,
      "grad_norm": 0.8371938467025757,
      "learning_rate": 6.213487197980527e-06,
      "loss": 0.1131,
      "step": 3825
    },
    {
      "epoch": 1.39,
      "grad_norm": 4.081541061401367,
      "learning_rate": 6.123332131265778e-06,
      "loss": 0.1345,
      "step": 3850
    },
    {
      "epoch": 1.4,
      "grad_norm": 2.648669958114624,
      "learning_rate": 6.033177064551028e-06,
      "loss": 0.1376,
      "step": 3875
    },
    {
      "epoch": 1.41,
      "grad_norm": 0.8278887867927551,
      "learning_rate": 5.943021997836279e-06,
      "loss": 0.1281,
      "step": 3900
    },
    {
      "epoch": 1.41,
      "eval_loss": 0.19765153527259827,
      "eval_na_accuracy": 0.8402062058448792,
      "eval_ordinal_accuracy": 0.6966350078582764,
      "eval_ordinal_mae": 0.4065835475921631,
      "eval_runtime": 174.9801,
      "eval_samples_per_second": 25.574,
      "eval_steps_per_second": 3.2,
      "step": 3900
    },
    {
      "epoch": 1.42,
      "grad_norm": 0.9070908427238464,
      "learning_rate": 5.852866931121529e-06,
      "loss": 0.1278,
      "step": 3925
    },
    {
      "epoch": 1.42,
      "grad_norm": 0.7849236130714417,
      "learning_rate": 5.7627118644067805e-06,
      "loss": 0.1575,
      "step": 3950
    },
    {
      "epoch": 1.43,
      "grad_norm": 2.2364354133605957,
      "learning_rate": 5.6761630003606205e-06,
      "loss": 0.116,
      "step": 3975
    },
    {
      "epoch": 1.44,
      "grad_norm": 0.8048083782196045,
      "learning_rate": 5.586007933645871e-06,
      "loss": 0.1153,
      "step": 4000
    },
    {
      "epoch": 1.44,
      "eval_loss": 0.20141063630580902,
      "eval_na_accuracy": 0.8436425924301147,
      "eval_ordinal_accuracy": 0.693552553653717,
      "eval_ordinal_mae": 0.4091172218322754,
      "eval_runtime": 175.0141,
      "eval_samples_per_second": 25.569,
      "eval_steps_per_second": 3.2,
      "step": 4000
    },
    {
      "epoch": 1.45,
      "grad_norm": 1.209535002708435,
      "learning_rate": 5.495852866931121e-06,
      "loss": 0.106,
      "step": 4025
    },
    {
      "epoch": 1.46,
      "grad_norm": 3.71276593208313,
      "learning_rate": 5.405697800216373e-06,
      "loss": 0.1316,
      "step": 4050
    },
    {
      "epoch": 1.47,
      "grad_norm": 18.610342025756836,
      "learning_rate": 5.315542733501623e-06,
      "loss": 0.132,
      "step": 4075
    },
    {
      "epoch": 1.48,
      "grad_norm": 2.119901180267334,
      "learning_rate": 5.225387666786874e-06,
      "loss": 0.1485,
      "step": 4100
    },
    {
      "epoch": 1.48,
      "eval_loss": 0.19648495316505432,
      "eval_na_accuracy": 0.838487982749939,
      "eval_ordinal_accuracy": 0.7038273811340332,
      "eval_ordinal_mae": 0.39893215894699097,
      "eval_runtime": 174.9887,
      "eval_samples_per_second": 25.573,
      "eval_steps_per_second": 3.2,
      "step": 4100
    },
    {
      "epoch": 1.49,
      "grad_norm": 1.489436388015747,
      "learning_rate": 5.135232600072124e-06,
      "loss": 0.1158,
      "step": 4125
    },
    {
      "epoch": 1.5,
      "grad_norm": 3.2724928855895996,
      "learning_rate": 5.045077533357375e-06,
      "loss": 0.1098,
      "step": 4150
    },
    {
      "epoch": 1.51,
      "grad_norm": 1.0074132680892944,
      "learning_rate": 4.954922466642626e-06,
      "loss": 0.1094,
      "step": 4175
    },
    {
      "epoch": 1.51,
      "grad_norm": 0.7578648924827576,
      "learning_rate": 4.864767399927877e-06,
      "loss": 0.1292,
      "step": 4200
    },
    {
      "epoch": 1.51,
      "eval_loss": 0.19693247973918915,
      "eval_na_accuracy": 0.8470790386199951,
      "eval_ordinal_accuracy": 0.7030567526817322,
      "eval_ordinal_mae": 0.39781150221824646,
      "eval_runtime": 173.0859,
      "eval_samples_per_second": 25.854,
      "eval_steps_per_second": 3.235,
      "step": 4200
    },
    {
      "epoch": 1.52,
      "grad_norm": 3.7381861209869385,
      "learning_rate": 4.774612333213128e-06,
      "loss": 0.1309,
      "step": 4225
    },
    {
      "epoch": 1.53,
      "grad_norm": 0.8151502013206482,
      "learning_rate": 4.684457266498378e-06,
      "loss": 0.1428,
      "step": 4250
    },
    {
      "epoch": 1.54,
      "grad_norm": 0.9021445512771606,
      "learning_rate": 4.5943021997836285e-06,
      "loss": 0.1104,
      "step": 4275
    },
    {
      "epoch": 1.55,
      "grad_norm": 3.8359127044677734,
      "learning_rate": 4.5041471330688794e-06,
      "loss": 0.1233,
      "step": 4300
    },
    {
      "epoch": 1.55,
      "eval_loss": 0.1989138424396515,
      "eval_na_accuracy": 0.8659793734550476,
      "eval_ordinal_accuracy": 0.6950937509536743,
      "eval_ordinal_mae": 0.39926186203956604,
      "eval_runtime": 173.9,
      "eval_samples_per_second": 25.733,
      "eval_steps_per_second": 3.22,
      "step": 4300
    },
    {
      "epoch": 1.56,
      "grad_norm": 2.0737252235412598,
      "learning_rate": 4.4139920663541295e-06,
      "loss": 0.1343,
      "step": 4325
    },
    {
      "epoch": 1.57,
      "grad_norm": 0.8500555753707886,
      "learning_rate": 4.32383699963938e-06,
      "loss": 0.1274,
      "step": 4350
    },
    {
      "epoch": 1.58,
      "grad_norm": 4.984882354736328,
      "learning_rate": 4.233681932924631e-06,
      "loss": 0.1131,
      "step": 4375
    },
    {
      "epoch": 1.59,
      "grad_norm": 0.8260248899459839,
      "learning_rate": 4.143526866209881e-06,
      "loss": 0.1128,
      "step": 4400
    },
    {
      "epoch": 1.59,
      "eval_loss": 0.1998443454504013,
      "eval_na_accuracy": 0.8522336483001709,
      "eval_ordinal_accuracy": 0.6971487402915955,
      "eval_ordinal_mae": 0.391990065574646,
      "eval_runtime": 175.0388,
      "eval_samples_per_second": 25.566,
      "eval_steps_per_second": 3.199,
      "step": 4400
    },
    {
      "epoch": 1.6,
      "grad_norm": 0.7347471117973328,
      "learning_rate": 4.053371799495132e-06,
      "loss": 0.1095,
      "step": 4425
    },
    {
      "epoch": 1.6,
      "grad_norm": 0.6527581810951233,
      "learning_rate": 3.963216732780383e-06,
      "loss": 0.1627,
      "step": 4450
    },
    {
      "epoch": 1.61,
      "grad_norm": 1.0017273426055908,
      "learning_rate": 3.873061666065633e-06,
      "loss": 0.1047,
      "step": 4475
    },
    {
      "epoch": 1.62,
      "grad_norm": 1.7798997163772583,
      "learning_rate": 3.782906599350884e-06,
      "loss": 0.0964,
      "step": 4500
    },
    {
      "epoch": 1.62,
      "eval_loss": 0.20046481490135193,
      "eval_na_accuracy": 0.8625429272651672,
      "eval_ordinal_accuracy": 0.6981762051582336,
      "eval_ordinal_mae": 0.39257487654685974,
      "eval_runtime": 174.4806,
      "eval_samples_per_second": 25.648,
      "eval_steps_per_second": 3.21,
      "step": 4500
    },
    {
      "epoch": 1.63,
      "grad_norm": 0.8608025312423706,
      "learning_rate": 3.6927515326361344e-06,
      "loss": 0.0974,
      "step": 4525
    },
    {
      "epoch": 1.64,
      "grad_norm": 0.5464763641357422,
      "learning_rate": 3.6025964659213853e-06,
      "loss": 0.1224,
      "step": 4550
    },
    {
      "epoch": 1.65,
      "grad_norm": 1.151129961013794,
      "learning_rate": 3.5124413992066358e-06,
      "loss": 0.1514,
      "step": 4575
    },
    {
      "epoch": 1.66,
      "grad_norm": 0.5751986503601074,
      "learning_rate": 3.4222863324918867e-06,
      "loss": 0.1184,
      "step": 4600
    },
    {
      "epoch": 1.66,
      "eval_loss": 0.20076720416545868,
      "eval_na_accuracy": 0.8711340427398682,
      "eval_ordinal_accuracy": 0.6968918442726135,
      "eval_ordinal_mae": 0.3860225975513458,
      "eval_runtime": 175.6222,
      "eval_samples_per_second": 25.481,
      "eval_steps_per_second": 3.189,
      "step": 4600
    },
    {
      "epoch": 1.67,
      "grad_norm": 0.8153491020202637,
      "learning_rate": 3.332131265777137e-06,
      "loss": 0.1344,
      "step": 4625
    },
    {
      "epoch": 1.68,
      "grad_norm": 1.3030270338058472,
      "learning_rate": 3.2419761990623876e-06,
      "loss": 0.136,
      "step": 4650
    },
    {
      "epoch": 1.69,
      "grad_norm": 6.30877685546875,
      "learning_rate": 3.1518211323476385e-06,
      "loss": 0.1224,
      "step": 4675
    },
    {
      "epoch": 1.69,
      "grad_norm": 4.972422122955322,
      "learning_rate": 3.061666065632889e-06,
      "loss": 0.108,
      "step": 4700
    },
    {
      "epoch": 1.69,
      "eval_loss": 0.19936376810073853,
      "eval_na_accuracy": 0.8573883175849915,
      "eval_ordinal_accuracy": 0.702029287815094,
      "eval_ordinal_mae": 0.39070162177085876,
      "eval_runtime": 174.219,
      "eval_samples_per_second": 25.686,
      "eval_steps_per_second": 3.214,
      "step": 4700
    },
    {
      "epoch": 1.7,
      "grad_norm": 1.4122976064682007,
      "learning_rate": 2.9715109989181394e-06,
      "loss": 0.1496,
      "step": 4725
    },
    {
      "epoch": 1.71,
      "grad_norm": 2.4173316955566406,
      "learning_rate": 2.8813559322033903e-06,
      "loss": 0.1276,
      "step": 4750
    },
    {
      "epoch": 1.72,
      "grad_norm": 1.0301971435546875,
      "learning_rate": 2.7912008654886407e-06,
      "loss": 0.1315,
      "step": 4775
    },
    {
      "epoch": 1.73,
      "grad_norm": 0.8369082808494568,
      "learning_rate": 2.701045798773891e-06,
      "loss": 0.129,
      "step": 4800
    },
    {
      "epoch": 1.73,
      "eval_loss": 0.19854028522968292,
      "eval_na_accuracy": 0.8591065406799316,
      "eval_ordinal_accuracy": 0.7033136487007141,
      "eval_ordinal_mae": 0.3896414041519165,
      "eval_runtime": 176.1539,
      "eval_samples_per_second": 25.404,
      "eval_steps_per_second": 3.179,
      "step": 4800
    },
    {
      "epoch": 1.74,
      "grad_norm": 3.947204113006592,
      "learning_rate": 2.610890732059142e-06,
      "loss": 0.1106,
      "step": 4825
    },
    {
      "epoch": 1.75,
      "grad_norm": 2.09318470954895,
      "learning_rate": 2.5207356653443925e-06,
      "loss": 0.164,
      "step": 4850
    },
    {
      "epoch": 1.76,
      "grad_norm": 0.7661357522010803,
      "learning_rate": 2.4305805986296434e-06,
      "loss": 0.1214,
      "step": 4875
    },
    {
      "epoch": 1.77,
      "grad_norm": 0.4803534746170044,
      "learning_rate": 2.340425531914894e-06,
      "loss": 0.1396,
      "step": 4900
    },
    {
      "epoch": 1.77,
      "eval_loss": 0.19975444674491882,
      "eval_na_accuracy": 0.8573883175849915,
      "eval_ordinal_accuracy": 0.6984331011772156,
      "eval_ordinal_mae": 0.3833589553833008,
      "eval_runtime": 176.3204,
      "eval_samples_per_second": 25.38,
      "eval_steps_per_second": 3.176,
      "step": 4900
    },
    {
      "epoch": 1.78,
      "grad_norm": 0.7548712491989136,
      "learning_rate": 2.2502704652001443e-06,
      "loss": 0.0977,
      "step": 4925
    },
    {
      "epoch": 1.79,
      "grad_norm": 1.906633734703064,
      "learning_rate": 2.1601153984853952e-06,
      "loss": 0.1385,
      "step": 4950
    },
    {
      "epoch": 1.79,
      "grad_norm": 0.6955749988555908,
      "learning_rate": 2.0699603317706457e-06,
      "loss": 0.1017,
      "step": 4975
    },
    {
      "epoch": 1.8,
      "grad_norm": 6.211479663848877,
      "learning_rate": 1.979805265055896e-06,
      "loss": 0.1323,
      "step": 5000
    },
    {
      "epoch": 1.8,
      "eval_loss": 0.19856412708759308,
      "eval_na_accuracy": 0.8453608155250549,
      "eval_ordinal_accuracy": 0.7051117420196533,
      "eval_ordinal_mae": 0.3843817710876465,
      "eval_runtime": 173.9748,
      "eval_samples_per_second": 25.722,
      "eval_steps_per_second": 3.219,
      "step": 5000
    },
    {
      "epoch": 1.81,
      "grad_norm": 0.5422704219818115,
      "learning_rate": 1.889650198341147e-06,
      "loss": 0.1367,
      "step": 5025
    },
    {
      "epoch": 1.82,
      "grad_norm": 2.782160997390747,
      "learning_rate": 1.7994951316263975e-06,
      "loss": 0.1002,
      "step": 5050
    },
    {
      "epoch": 1.83,
      "grad_norm": 4.336767673492432,
      "learning_rate": 1.7093400649116482e-06,
      "loss": 0.1407,
      "step": 5075
    },
    {
      "epoch": 1.84,
      "grad_norm": 3.1643247604370117,
      "learning_rate": 1.6191849981968988e-06,
      "loss": 0.1079,
      "step": 5100
    },
    {
      "epoch": 1.84,
      "eval_loss": 0.19739079475402832,
      "eval_na_accuracy": 0.8402062058448792,
      "eval_ordinal_accuracy": 0.7053686380386353,
      "eval_ordinal_mae": 0.3832775950431824,
      "eval_runtime": 174.8648,
      "eval_samples_per_second": 25.591,
      "eval_steps_per_second": 3.202,
      "step": 5100
    },
    {
      "epoch": 1.85,
      "grad_norm": 5.217372894287109,
      "learning_rate": 1.5290299314821493e-06,
      "loss": 0.122,
      "step": 5125
    },
    {
      "epoch": 1.86,
      "grad_norm": 4.001213073730469,
      "learning_rate": 1.4388748647674e-06,
      "loss": 0.1102,
      "step": 5150
    },
    {
      "epoch": 1.87,
      "grad_norm": 2.418731451034546,
      "learning_rate": 1.3487197980526506e-06,
      "loss": 0.1117,
      "step": 5175
    },
    {
      "epoch": 1.88,
      "grad_norm": 1.181628704071045,
      "learning_rate": 1.2585647313379013e-06,
      "loss": 0.0802,
      "step": 5200
    },
    {
      "epoch": 1.88,
      "eval_loss": 0.19647686183452606,
      "eval_na_accuracy": 0.8487972617149353,
      "eval_ordinal_accuracy": 0.7074235677719116,
      "eval_ordinal_mae": 0.38217073678970337,
      "eval_runtime": 175.4662,
      "eval_samples_per_second": 25.503,
      "eval_steps_per_second": 3.191,
      "step": 5200
    },
    {
      "epoch": 1.88,
      "grad_norm": 0.654973030090332,
      "learning_rate": 1.1684096646231518e-06,
      "loss": 0.1139,
      "step": 5225
    },
    {
      "epoch": 1.89,
      "grad_norm": 0.66736900806427,
      "learning_rate": 1.0782545979084025e-06,
      "loss": 0.1099,
      "step": 5250
    },
    {
      "epoch": 1.9,
      "grad_norm": 0.9291324615478516,
      "learning_rate": 9.880995311936533e-07,
      "loss": 0.1233,
      "step": 5275
    },
    {
      "epoch": 1.91,
      "grad_norm": 0.864493191242218,
      "learning_rate": 8.979444644789038e-07,
      "loss": 0.1391,
      "step": 5300
    },
    {
      "epoch": 1.91,
      "eval_loss": 0.19748477637767792,
      "eval_na_accuracy": 0.8453608155250549,
      "eval_ordinal_accuracy": 0.7051117420196533,
      "eval_ordinal_mae": 0.3809069097042084,
      "eval_runtime": 174.5451,
      "eval_samples_per_second": 25.638,
      "eval_steps_per_second": 3.208,
      "step": 5300
    },
    {
      "epoch": 1.92,
      "grad_norm": 0.844887912273407,
      "learning_rate": 8.077893977641545e-07,
      "loss": 0.1364,
      "step": 5325
    },
    {
      "epoch": 1.93,
      "grad_norm": 0.7487301230430603,
      "learning_rate": 7.17634331049405e-07,
      "loss": 0.1137,
      "step": 5350
    },
    {
      "epoch": 1.94,
      "grad_norm": 3.6033549308776855,
      "learning_rate": 6.274792643346557e-07,
      "loss": 0.1508,
      "step": 5375
    },
    {
      "epoch": 1.95,
      "grad_norm": 0.7312902212142944,
      "learning_rate": 5.373241976199063e-07,
      "loss": 0.1183,
      "step": 5400
    },
    {
      "epoch": 1.95,
      "eval_loss": 0.19734890758991241,
      "eval_na_accuracy": 0.8350515365600586,
      "eval_ordinal_accuracy": 0.7087079286575317,
      "eval_ordinal_mae": 0.38272929191589355,
      "eval_runtime": 176.207,
      "eval_samples_per_second": 25.396,
      "eval_steps_per_second": 3.178,
      "step": 5400
    },
    {
      "epoch": 1.96,
      "grad_norm": 0.859944760799408,
      "learning_rate": 4.4716913090515695e-07,
      "loss": 0.1125,
      "step": 5425
    },
    {
      "epoch": 1.97,
      "grad_norm": 0.7725034952163696,
      "learning_rate": 3.5701406419040757e-07,
      "loss": 0.1009,
      "step": 5450
    },
    {
      "epoch": 1.97,
      "grad_norm": 8.944540977478027,
      "learning_rate": 2.6685899747565814e-07,
      "loss": 0.1115,
      "step": 5475
    },
    {
      "epoch": 1.98,
      "grad_norm": 0.6212058663368225,
      "learning_rate": 1.7670393076090878e-07,
      "loss": 0.1368,
      "step": 5500
    },
    {
      "epoch": 1.98,
      "eval_loss": 0.19747541844844818,
      "eval_na_accuracy": 0.8333333134651184,
      "eval_ordinal_accuracy": 0.7081941962242126,
      "eval_ordinal_mae": 0.38125094771385193,
      "eval_runtime": 170.5638,
      "eval_samples_per_second": 26.237,
      "eval_steps_per_second": 3.283,
      "step": 5500
    },
    {
      "epoch": 1.99,
      "grad_norm": 0.4366958737373352,
      "learning_rate": 8.65488640461594e-08,
      "loss": 0.108,
      "step": 5525
    },
    {
      "epoch": 2.0,
      "step": 5546,
      "total_flos": 6.876575711894569e+18,
      "train_loss": 0.18857605492557122,
      "train_runtime": 17933.6413,
      "train_samples_per_second": 4.948,
      "train_steps_per_second": 0.309
    }
  ],
  "logging_steps": 25,
  "max_steps": 5546,
  "num_input_tokens_seen": 0,
  "num_train_epochs": 2,
  "save_steps": 100,
  "total_flos": 6.876575711894569e+18,
  "train_batch_size": 16,
  "trial_name": null,
  "trial_params": null
}