{
  "best_metric": 0.22382062673568726,
  "best_model_checkpoint": "./ryan03282024/checkpoint-1600",
  "epoch": 4.0,
  "eval_steps": 100,
  "global_step": 9160,
  "is_hyper_param_search": false,
  "is_local_process_zero": true,
  "is_world_process_zero": true,
  "log_history": [
    {
      "epoch": 0.01,
      "grad_norm": 0.5509258508682251,
      "learning_rate": 9.972707423580787e-05,
      "loss": 0.4095,
      "step": 25
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.37446415424346924,
      "learning_rate": 9.945414847161573e-05,
      "loss": 0.3743,
      "step": 50
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.36988818645477295,
      "learning_rate": 9.918122270742359e-05,
      "loss": 0.3766,
      "step": 75
    },
    {
      "epoch": 0.04,
      "grad_norm": 2.9999992847442627,
      "learning_rate": 9.890829694323145e-05,
      "loss": 0.3421,
      "step": 100
    },
    {
      "epoch": 0.04,
      "eval_loss": 0.33314022421836853,
      "eval_na_accuracy": 0.6911196708679199,
      "eval_ordinal_accuracy": 0.3816815912723541,
      "eval_ordinal_mae": 0.8749264478683472,
      "eval_runtime": 311.6641,
      "eval_samples_per_second": 12.767,
      "eval_steps_per_second": 1.598,
      "step": 100
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.6784111261367798,
      "learning_rate": 9.863537117903931e-05,
      "loss": 0.3422,
      "step": 125
    },
    {
      "epoch": 0.07,
      "grad_norm": 1.7690809965133667,
      "learning_rate": 9.836244541484717e-05,
      "loss": 0.3059,
      "step": 150
    },
    {
      "epoch": 0.08,
      "grad_norm": 0.5732172131538391,
      "learning_rate": 9.808951965065503e-05,
      "loss": 0.3654,
      "step": 175
    },
    {
      "epoch": 0.09,
      "grad_norm": 0.3226703405380249,
      "learning_rate": 9.781659388646288e-05,
      "loss": 0.2813,
      "step": 200
    },
    {
      "epoch": 0.09,
      "eval_loss": 0.29995909333229065,
      "eval_na_accuracy": 0.7953668236732483,
      "eval_ordinal_accuracy": 0.5117018222808838,
      "eval_ordinal_mae": 0.7492409348487854,
      "eval_runtime": 157.8026,
      "eval_samples_per_second": 25.215,
      "eval_steps_per_second": 3.156,
      "step": 200
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.44079920649528503,
      "learning_rate": 9.754366812227075e-05,
      "loss": 0.3359,
      "step": 225
    },
    {
      "epoch": 0.11,
      "grad_norm": 1.5866132974624634,
      "learning_rate": 9.727074235807861e-05,
      "loss": 0.3131,
      "step": 250
    },
    {
      "epoch": 0.12,
      "grad_norm": 1.5666215419769287,
      "learning_rate": 9.699781659388647e-05,
      "loss": 0.2723,
      "step": 275
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.2892788350582123,
      "learning_rate": 9.672489082969433e-05,
      "loss": 0.2619,
      "step": 300
    },
    {
      "epoch": 0.13,
      "eval_loss": 0.3019290566444397,
      "eval_na_accuracy": 0.7046331763267517,
      "eval_ordinal_accuracy": 0.5273042321205139,
      "eval_ordinal_mae": 0.6840940117835999,
      "eval_runtime": 158.8663,
      "eval_samples_per_second": 25.046,
      "eval_steps_per_second": 3.135,
      "step": 300
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.24918608367443085,
      "learning_rate": 9.645196506550219e-05,
      "loss": 0.303,
      "step": 325
    },
    {
      "epoch": 0.15,
      "grad_norm": 1.626022219657898,
      "learning_rate": 9.617903930131005e-05,
      "loss": 0.2867,
      "step": 350
    },
    {
      "epoch": 0.16,
      "grad_norm": 2.1694397926330566,
      "learning_rate": 9.59061135371179e-05,
      "loss": 0.249,
      "step": 375
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.3259480893611908,
      "learning_rate": 9.563318777292577e-05,
      "loss": 0.2863,
      "step": 400
    },
    {
      "epoch": 0.17,
      "eval_loss": 0.2960417866706848,
      "eval_na_accuracy": 0.7335907220840454,
      "eval_ordinal_accuracy": 0.5096792578697205,
      "eval_ordinal_mae": 0.6538078188896179,
      "eval_runtime": 159.697,
      "eval_samples_per_second": 24.916,
      "eval_steps_per_second": 3.118,
      "step": 400
    },
    {
      "epoch": 0.19,
      "grad_norm": 1.1021641492843628,
      "learning_rate": 9.536026200873362e-05,
      "loss": 0.2729,
      "step": 425
    },
    {
      "epoch": 0.2,
      "grad_norm": 1.3987232446670532,
      "learning_rate": 9.50873362445415e-05,
      "loss": 0.2544,
      "step": 450
    },
    {
      "epoch": 0.21,
      "grad_norm": 1.7090039253234863,
      "learning_rate": 9.481441048034934e-05,
      "loss": 0.2776,
      "step": 475
    },
    {
      "epoch": 0.22,
      "grad_norm": 1.737847924232483,
      "learning_rate": 9.454148471615721e-05,
      "loss": 0.2159,
      "step": 500
    },
    {
      "epoch": 0.22,
      "eval_loss": 0.26019465923309326,
      "eval_na_accuracy": 0.8243243098258972,
      "eval_ordinal_accuracy": 0.5660213828086853,
      "eval_ordinal_mae": 0.540388822555542,
      "eval_runtime": 157.5356,
      "eval_samples_per_second": 25.258,
      "eval_steps_per_second": 3.161,
      "step": 500
    },
    {
      "epoch": 0.23,
      "grad_norm": 3.2989888191223145,
      "learning_rate": 9.426855895196508e-05,
      "loss": 0.2554,
      "step": 525
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.7149298191070557,
      "learning_rate": 9.399563318777294e-05,
      "loss": 0.272,
      "step": 550
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.90256667137146,
      "learning_rate": 9.37227074235808e-05,
      "loss": 0.2595,
      "step": 575
    },
    {
      "epoch": 0.26,
      "grad_norm": 2.2932217121124268,
      "learning_rate": 9.344978165938864e-05,
      "loss": 0.2235,
      "step": 600
    },
    {
      "epoch": 0.26,
      "eval_loss": 0.2556995749473572,
      "eval_na_accuracy": 0.7779922485351562,
      "eval_ordinal_accuracy": 0.5874024629592896,
      "eval_ordinal_mae": 0.5014671683311462,
      "eval_runtime": 159.648,
      "eval_samples_per_second": 24.924,
      "eval_steps_per_second": 3.119,
      "step": 600
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.8714308142662048,
      "learning_rate": 9.317685589519652e-05,
      "loss": 0.2871,
      "step": 625
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.699299156665802,
      "learning_rate": 9.290393013100436e-05,
      "loss": 0.2489,
      "step": 650
    },
    {
      "epoch": 0.29,
      "grad_norm": 3.3761134147644043,
      "learning_rate": 9.263100436681224e-05,
      "loss": 0.2696,
      "step": 675
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.5747934579849243,
      "learning_rate": 9.235807860262009e-05,
      "loss": 0.285,
      "step": 700
    },
    {
      "epoch": 0.31,
      "eval_loss": 0.2563941180706024,
      "eval_na_accuracy": 0.6853281855583191,
      "eval_ordinal_accuracy": 0.6180294752120972,
      "eval_ordinal_mae": 0.49999386072158813,
      "eval_runtime": 160.5887,
      "eval_samples_per_second": 24.778,
      "eval_steps_per_second": 3.101,
      "step": 700
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.6936819553375244,
      "learning_rate": 9.208515283842796e-05,
      "loss": 0.2379,
      "step": 725
    },
    {
      "epoch": 0.33,
      "grad_norm": 1.3918662071228027,
      "learning_rate": 9.18122270742358e-05,
      "loss": 0.2509,
      "step": 750
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.46456876397132874,
      "learning_rate": 9.153930131004367e-05,
      "loss": 0.2531,
      "step": 775
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.516076385974884,
      "learning_rate": 9.126637554585154e-05,
      "loss": 0.2028,
      "step": 800
    },
    {
      "epoch": 0.35,
      "eval_loss": 0.2861529588699341,
      "eval_na_accuracy": 0.7220077514648438,
      "eval_ordinal_accuracy": 0.5067899227142334,
      "eval_ordinal_mae": 0.633837103843689,
      "eval_runtime": 154.4747,
      "eval_samples_per_second": 25.758,
      "eval_steps_per_second": 3.224,
      "step": 800
    },
    {
      "epoch": 0.36,
      "grad_norm": 1.5322011709213257,
      "learning_rate": 9.100436681222709e-05,
      "loss": 0.2936,
      "step": 825
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.49020248651504517,
      "learning_rate": 9.073144104803494e-05,
      "loss": 0.2206,
      "step": 850
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.8753771781921387,
      "learning_rate": 9.045851528384281e-05,
      "loss": 0.2031,
      "step": 875
    },
    {
      "epoch": 0.39,
      "grad_norm": 1.1926873922348022,
      "learning_rate": 9.018558951965066e-05,
      "loss": 0.2006,
      "step": 900
    },
    {
      "epoch": 0.39,
      "eval_loss": 0.24949249625205994,
      "eval_na_accuracy": 0.7586872577667236,
      "eval_ordinal_accuracy": 0.6298757791519165,
      "eval_ordinal_mae": 0.4830287992954254,
      "eval_runtime": 156.0615,
      "eval_samples_per_second": 25.496,
      "eval_steps_per_second": 3.191,
      "step": 900
    },
    {
      "epoch": 0.4,
      "grad_norm": 1.0290440320968628,
      "learning_rate": 8.991266375545852e-05,
      "loss": 0.2648,
      "step": 925
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.615397036075592,
      "learning_rate": 8.963973799126638e-05,
      "loss": 0.1866,
      "step": 950
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.4451560974121094,
      "learning_rate": 8.936681222707424e-05,
      "loss": 0.2641,
      "step": 975
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.9014913439750671,
      "learning_rate": 8.90938864628821e-05,
      "loss": 0.2663,
      "step": 1000
    },
    {
      "epoch": 0.44,
      "eval_loss": 0.26604241132736206,
      "eval_na_accuracy": 0.8610038757324219,
      "eval_ordinal_accuracy": 0.602138102054596,
      "eval_ordinal_mae": 0.4893138110637665,
      "eval_runtime": 155.5971,
      "eval_samples_per_second": 25.572,
      "eval_steps_per_second": 3.201,
      "step": 1000
    },
    {
      "epoch": 0.45,
      "grad_norm": 2.4818997383117676,
      "learning_rate": 8.882096069868996e-05,
      "loss": 0.2454,
      "step": 1025
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.7425001859664917,
      "learning_rate": 8.854803493449782e-05,
      "loss": 0.215,
      "step": 1050
    },
    {
      "epoch": 0.47,
      "grad_norm": 2.0386316776275635,
      "learning_rate": 8.827510917030568e-05,
      "loss": 0.2368,
      "step": 1075
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.6596102714538574,
      "learning_rate": 8.800218340611354e-05,
      "loss": 0.2062,
      "step": 1100
    },
    {
      "epoch": 0.48,
      "eval_loss": 0.24812151491641998,
      "eval_na_accuracy": 0.8436293601989746,
      "eval_ordinal_accuracy": 0.6266974806785583,
      "eval_ordinal_mae": 0.47125470638275146,
      "eval_runtime": 157.0377,
      "eval_samples_per_second": 25.338,
      "eval_steps_per_second": 3.171,
      "step": 1100
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.4987865686416626,
      "learning_rate": 8.77292576419214e-05,
      "loss": 0.2151,
      "step": 1125
    },
    {
      "epoch": 0.5,
      "grad_norm": 1.0523704290390015,
      "learning_rate": 8.745633187772926e-05,
      "loss": 0.2092,
      "step": 1150
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.6211034059524536,
      "learning_rate": 8.718340611353712e-05,
      "loss": 0.2452,
      "step": 1175
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.49944034218788147,
      "learning_rate": 8.691048034934498e-05,
      "loss": 0.1749,
      "step": 1200
    },
    {
      "epoch": 0.52,
      "eval_loss": 0.25862136483192444,
      "eval_na_accuracy": 0.6737451553344727,
      "eval_ordinal_accuracy": 0.6422998905181885,
      "eval_ordinal_mae": 0.4958673417568207,
      "eval_runtime": 161.9513,
      "eval_samples_per_second": 24.569,
      "eval_steps_per_second": 3.075,
      "step": 1200
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.30346667766571045,
      "learning_rate": 8.663755458515284e-05,
      "loss": 0.202,
      "step": 1225
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.8414286971092224,
      "learning_rate": 8.63646288209607e-05,
      "loss": 0.2019,
      "step": 1250
    },
    {
      "epoch": 0.56,
      "grad_norm": 1.1115086078643799,
      "learning_rate": 8.609170305676856e-05,
      "loss": 0.2037,
      "step": 1275
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.7678231000900269,
      "learning_rate": 8.581877729257642e-05,
      "loss": 0.2197,
      "step": 1300
    },
    {
      "epoch": 0.57,
      "eval_loss": 0.23488646745681763,
      "eval_na_accuracy": 0.8030887842178345,
      "eval_ordinal_accuracy": 0.5980930328369141,
      "eval_ordinal_mae": 0.48411670327186584,
      "eval_runtime": 161.4724,
      "eval_samples_per_second": 24.642,
      "eval_steps_per_second": 3.084,
      "step": 1300
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.33742988109588623,
      "learning_rate": 8.554585152838429e-05,
      "loss": 0.226,
      "step": 1325
    },
    {
      "epoch": 0.59,
      "grad_norm": 1.6625324487686157,
      "learning_rate": 8.527292576419215e-05,
      "loss": 0.2612,
      "step": 1350
    },
    {
      "epoch": 0.6,
      "grad_norm": 1.5984447002410889,
      "learning_rate": 8.5e-05,
      "loss": 0.1812,
      "step": 1375
    },
    {
      "epoch": 0.61,
      "grad_norm": 1.39547860622406,
      "learning_rate": 8.472707423580787e-05,
      "loss": 0.2073,
      "step": 1400
    },
    {
      "epoch": 0.61,
      "eval_loss": 0.25865110754966736,
      "eval_na_accuracy": 0.6949806809425354,
      "eval_ordinal_accuracy": 0.6012713313102722,
      "eval_ordinal_mae": 0.4877532422542572,
      "eval_runtime": 161.5877,
      "eval_samples_per_second": 24.624,
      "eval_steps_per_second": 3.082,
      "step": 1400
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.6241037845611572,
      "learning_rate": 8.445414847161573e-05,
      "loss": 0.2783,
      "step": 1425
    },
    {
      "epoch": 0.63,
      "grad_norm": 1.9269381761550903,
      "learning_rate": 8.418122270742357e-05,
      "loss": 0.1813,
      "step": 1450
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.5204592943191528,
      "learning_rate": 8.390829694323145e-05,
      "loss": 0.1753,
      "step": 1475
    },
    {
      "epoch": 0.66,
      "grad_norm": 1.156730055809021,
      "learning_rate": 8.36353711790393e-05,
      "loss": 0.1915,
      "step": 1500
    },
    {
      "epoch": 0.66,
      "eval_loss": 0.23929761350154877,
      "eval_na_accuracy": 0.7683397531509399,
      "eval_ordinal_accuracy": 0.6321872472763062,
      "eval_ordinal_mae": 0.4770694673061371,
      "eval_runtime": 160.096,
      "eval_samples_per_second": 24.854,
      "eval_steps_per_second": 3.111,
      "step": 1500
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.31671473383903503,
      "learning_rate": 8.336244541484717e-05,
      "loss": 0.2555,
      "step": 1525
    },
    {
      "epoch": 0.68,
      "grad_norm": 1.452971339225769,
      "learning_rate": 8.308951965065503e-05,
      "loss": 0.2321,
      "step": 1550
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.7802607417106628,
      "learning_rate": 8.281659388646289e-05,
      "loss": 0.2013,
      "step": 1575
    },
    {
      "epoch": 0.7,
      "grad_norm": 1.238275408744812,
      "learning_rate": 8.254366812227075e-05,
      "loss": 0.2374,
      "step": 1600
    },
    {
      "epoch": 0.7,
      "eval_loss": 0.22382062673568726,
      "eval_na_accuracy": 0.799227774143219,
      "eval_ordinal_accuracy": 0.6446113586425781,
      "eval_ordinal_mae": 0.4441048502922058,
      "eval_runtime": 161.7443,
      "eval_samples_per_second": 24.601,
      "eval_steps_per_second": 3.079,
      "step": 1600
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.8560432195663452,
      "learning_rate": 8.227074235807861e-05,
      "loss": 0.2055,
      "step": 1625
    },
    {
      "epoch": 0.72,
      "grad_norm": 2.4070003032684326,
      "learning_rate": 8.199781659388647e-05,
      "loss": 0.2357,
      "step": 1650
    },
    {
      "epoch": 0.73,
      "grad_norm": 2.119150161743164,
      "learning_rate": 8.172489082969432e-05,
      "loss": 0.2145,
      "step": 1675
    },
    {
      "epoch": 0.74,
      "grad_norm": 1.8999676704406738,
      "learning_rate": 8.145196506550219e-05,
      "loss": 0.2278,
      "step": 1700
    },
    {
      "epoch": 0.74,
      "eval_loss": 0.24530412256717682,
      "eval_na_accuracy": 0.7277992367744446,
      "eval_ordinal_accuracy": 0.6538572907447815,
      "eval_ordinal_mae": 0.4410313367843628,
      "eval_runtime": 160.7764,
      "eval_samples_per_second": 24.749,
      "eval_steps_per_second": 3.097,
      "step": 1700
    },
    {
      "epoch": 0.75,
      "grad_norm": 1.699065923690796,
      "learning_rate": 8.117903930131004e-05,
      "loss": 0.2416,
      "step": 1725
    },
    {
      "epoch": 0.76,
      "grad_norm": 1.4952126741409302,
      "learning_rate": 8.090611353711791e-05,
      "loss": 0.188,
      "step": 1750
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.4823841452598572,
      "learning_rate": 8.063318777292576e-05,
      "loss": 0.2148,
      "step": 1775
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.9280353784561157,
      "learning_rate": 8.036026200873363e-05,
      "loss": 0.2033,
      "step": 1800
    },
    {
      "epoch": 0.79,
      "eval_loss": 0.22507312893867493,
      "eval_na_accuracy": 0.8185328245162964,
      "eval_ordinal_accuracy": 0.6298757791519165,
      "eval_ordinal_mae": 0.4584101140499115,
      "eval_runtime": 160.7672,
      "eval_samples_per_second": 24.75,
      "eval_steps_per_second": 3.098,
      "step": 1800
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.9829936027526855,
      "learning_rate": 8.00873362445415e-05,
      "loss": 0.2015,
      "step": 1825
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.5692235231399536,
      "learning_rate": 7.981441048034934e-05,
      "loss": 0.2213,
      "step": 1850
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.4303966164588928,
      "learning_rate": 7.954148471615722e-05,
      "loss": 0.2156,
      "step": 1875
    },
    {
      "epoch": 0.83,
      "grad_norm": 1.4745689630508423,
      "learning_rate": 7.926855895196506e-05,
      "loss": 0.1843,
      "step": 1900
    },
    {
      "epoch": 0.83,
      "eval_loss": 0.2280111461877823,
      "eval_na_accuracy": 0.8127413392066956,
      "eval_ordinal_accuracy": 0.6512568593025208,
      "eval_ordinal_mae": 0.4446066617965698,
      "eval_runtime": 160.0671,
      "eval_samples_per_second": 24.858,
      "eval_steps_per_second": 3.111,
      "step": 1900
    },
    {
      "epoch": 0.84,
      "grad_norm": 2.2468037605285645,
      "learning_rate": 7.899563318777294e-05,
      "loss": 0.2182,
      "step": 1925
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.6818110346794128,
      "learning_rate": 7.872270742358078e-05,
      "loss": 0.1822,
      "step": 1950
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.802448034286499,
      "learning_rate": 7.844978165938866e-05,
      "loss": 0.2289,
      "step": 1975
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.6907253861427307,
      "learning_rate": 7.81768558951965e-05,
      "loss": 0.1878,
      "step": 2000
    },
    {
      "epoch": 0.87,
      "eval_loss": 0.22766314446926117,
      "eval_na_accuracy": 0.8127413392066956,
      "eval_ordinal_accuracy": 0.6492343544960022,
      "eval_ordinal_mae": 0.4454284906387329,
      "eval_runtime": 159.2821,
      "eval_samples_per_second": 24.981,
      "eval_steps_per_second": 3.127,
      "step": 2000
    },
    {
      "epoch": 0.88,
      "grad_norm": 1.1525399684906006,
      "learning_rate": 7.790393013100437e-05,
      "loss": 0.2146,
      "step": 2025
    },
    {
      "epoch": 0.9,
      "grad_norm": 1.3308144807815552,
      "learning_rate": 7.763100436681223e-05,
      "loss": 0.2367,
      "step": 2050
    },
    {
      "epoch": 0.91,
      "grad_norm": 1.1532173156738281,
      "learning_rate": 7.735807860262009e-05,
      "loss": 0.1948,
      "step": 2075
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.9184058904647827,
      "learning_rate": 7.708515283842796e-05,
      "loss": 0.2608,
      "step": 2100
    },
    {
      "epoch": 0.92,
      "eval_loss": 0.23085492849349976,
      "eval_na_accuracy": 0.8494208455085754,
      "eval_ordinal_accuracy": 0.619185209274292,
      "eval_ordinal_mae": 0.4517284035682678,
      "eval_runtime": 158.9549,
      "eval_samples_per_second": 25.032,
      "eval_steps_per_second": 3.133,
      "step": 2100
    },
    {
      "epoch": 0.93,
      "grad_norm": 1.0736620426177979,
      "learning_rate": 7.681222707423581e-05,
      "loss": 0.2409,
      "step": 2125
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.5520908832550049,
      "learning_rate": 7.653930131004368e-05,
      "loss": 0.1722,
      "step": 2150
    },
    {
      "epoch": 0.95,
      "grad_norm": 1.255903720855713,
      "learning_rate": 7.626637554585153e-05,
      "loss": 0.1996,
      "step": 2175
    },
    {
      "epoch": 0.96,
      "grad_norm": 1.3203591108322144,
      "learning_rate": 7.599344978165939e-05,
      "loss": 0.201,
      "step": 2200
    },
    {
      "epoch": 0.96,
      "eval_loss": 0.24588599801063538,
      "eval_na_accuracy": 0.7277992367744446,
      "eval_ordinal_accuracy": 0.6405662894248962,
      "eval_ordinal_mae": 0.46535709500312805,
      "eval_runtime": 163.7913,
      "eval_samples_per_second": 24.293,
      "eval_steps_per_second": 3.04,
      "step": 2200
    },
    {
      "epoch": 0.97,
      "grad_norm": 1.3525117635726929,
      "learning_rate": 7.572052401746725e-05,
      "loss": 0.2068,
      "step": 2225
    },
    {
      "epoch": 0.98,
      "grad_norm": 2.916431188583374,
      "learning_rate": 7.544759825327511e-05,
      "loss": 0.243,
      "step": 2250
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.8391708135604858,
      "learning_rate": 7.517467248908297e-05,
      "loss": 0.1842,
      "step": 2275
    },
    {
      "epoch": 1.0,
      "grad_norm": 0.5292081236839294,
      "learning_rate": 7.490174672489083e-05,
      "loss": 0.1736,
      "step": 2300
    },
    {
      "epoch": 1.0,
      "eval_loss": 0.24380208551883698,
      "eval_na_accuracy": 0.7200772166252136,
      "eval_ordinal_accuracy": 0.6475006937980652,
      "eval_ordinal_mae": 0.44738492369651794,
      "eval_runtime": 161.0768,
      "eval_samples_per_second": 24.703,
      "eval_steps_per_second": 3.092,
      "step": 2300
    },
    {
      "epoch": 1.02,
      "grad_norm": 0.6971523761749268,
      "learning_rate": 7.462882096069869e-05,
      "loss": 0.1683,
      "step": 2325
    },
    {
      "epoch": 1.03,
      "grad_norm": 0.8325093388557434,
      "learning_rate": 7.435589519650655e-05,
      "loss": 0.1177,
      "step": 2350
    },
    {
      "epoch": 1.04,
      "grad_norm": 0.8595998883247375,
      "learning_rate": 7.408296943231441e-05,
      "loss": 0.1626,
      "step": 2375
    },
    {
      "epoch": 1.05,
      "grad_norm": 0.38421839475631714,
      "learning_rate": 7.381004366812227e-05,
      "loss": 0.1374,
      "step": 2400
    },
    {
      "epoch": 1.05,
      "eval_loss": 0.23675759136676788,
      "eval_na_accuracy": 0.7799227833747864,
      "eval_ordinal_accuracy": 0.6622363328933716,
      "eval_ordinal_mae": 0.41446253657341003,
      "eval_runtime": 160.4437,
      "eval_samples_per_second": 24.8,
      "eval_steps_per_second": 3.104,
      "step": 2400
    },
    {
      "epoch": 1.06,
      "grad_norm": 0.7025877237319946,
      "learning_rate": 7.353711790393013e-05,
      "loss": 0.1484,
      "step": 2425
    },
    {
      "epoch": 1.07,
      "grad_norm": 1.4461692571640015,
      "learning_rate": 7.3264192139738e-05,
      "loss": 0.1564,
      "step": 2450
    },
    {
      "epoch": 1.08,
      "grad_norm": 1.1262603998184204,
      "learning_rate": 7.299126637554585e-05,
      "loss": 0.1252,
      "step": 2475
    },
    {
      "epoch": 1.09,
      "grad_norm": 1.1054977178573608,
      "learning_rate": 7.271834061135371e-05,
      "loss": 0.1334,
      "step": 2500
    },
    {
      "epoch": 1.09,
      "eval_loss": 0.2424485832452774,
      "eval_na_accuracy": 0.7509652376174927,
      "eval_ordinal_accuracy": 0.6732158064842224,
      "eval_ordinal_mae": 0.4105488359928131,
      "eval_runtime": 161.6129,
      "eval_samples_per_second": 24.621,
      "eval_steps_per_second": 3.081,
      "step": 2500
    },
    {
      "epoch": 1.1,
      "grad_norm": 0.6302788257598877,
      "learning_rate": 7.244541484716158e-05,
      "loss": 0.1252,
      "step": 2525
    },
    {
      "epoch": 1.11,
      "grad_norm": 3.8645241260528564,
      "learning_rate": 7.217248908296944e-05,
      "loss": 0.1444,
      "step": 2550
    },
    {
      "epoch": 1.12,
      "grad_norm": 0.7108765244483948,
      "learning_rate": 7.18995633187773e-05,
      "loss": 0.1273,
      "step": 2575
    },
    {
      "epoch": 1.14,
      "grad_norm": 2.3784756660461426,
      "learning_rate": 7.162663755458516e-05,
      "loss": 0.1319,
      "step": 2600
    },
    {
      "epoch": 1.14,
      "eval_loss": 0.23355962336063385,
      "eval_na_accuracy": 0.7741312980651855,
      "eval_ordinal_accuracy": 0.6711933016777039,
      "eval_ordinal_mae": 0.41552796959877014,
      "eval_runtime": 161.0033,
      "eval_samples_per_second": 24.714,
      "eval_steps_per_second": 3.093,
      "step": 2600
    },
    {
      "epoch": 1.15,
      "grad_norm": 0.6569415330886841,
      "learning_rate": 7.135371179039302e-05,
      "loss": 0.1418,
      "step": 2625
    },
    {
      "epoch": 1.16,
      "grad_norm": 2.9414520263671875,
      "learning_rate": 7.108078602620088e-05,
      "loss": 0.123,
      "step": 2650
    },
    {
      "epoch": 1.17,
      "grad_norm": 0.5435983538627625,
      "learning_rate": 7.080786026200874e-05,
      "loss": 0.1458,
      "step": 2675
    },
    {
      "epoch": 1.18,
      "grad_norm": 0.6094673275947571,
      "learning_rate": 7.05349344978166e-05,
      "loss": 0.1549,
      "step": 2700
    },
    {
      "epoch": 1.18,
      "eval_loss": 0.25251126289367676,
      "eval_na_accuracy": 0.7586872577667236,
      "eval_ordinal_accuracy": 0.6625252962112427,
      "eval_ordinal_mae": 0.40396779775619507,
      "eval_runtime": 162.435,
      "eval_samples_per_second": 24.496,
      "eval_steps_per_second": 3.066,
      "step": 2700
    },
    {
      "epoch": 1.19,
      "grad_norm": 0.4580962359905243,
      "learning_rate": 7.026200873362446e-05,
      "loss": 0.1382,
      "step": 2725
    },
    {
      "epoch": 1.2,
      "grad_norm": 0.7736852765083313,
      "learning_rate": 6.998908296943232e-05,
      "loss": 0.1595,
      "step": 2750
    },
    {
      "epoch": 1.21,
      "grad_norm": 1.1125404834747314,
      "learning_rate": 6.971615720524018e-05,
      "loss": 0.1071,
      "step": 2775
    },
    {
      "epoch": 1.22,
      "grad_norm": 0.529449999332428,
      "learning_rate": 6.944323144104804e-05,
      "loss": 0.116,
      "step": 2800
    },
    {
      "epoch": 1.22,
      "eval_loss": 0.25007495284080505,
      "eval_na_accuracy": 0.7664092779159546,
      "eval_ordinal_accuracy": 0.6370990872383118,
      "eval_ordinal_mae": 0.44249165058135986,
      "eval_runtime": 161.6127,
      "eval_samples_per_second": 24.621,
      "eval_steps_per_second": 3.081,
      "step": 2800
    },
    {
      "epoch": 1.23,
      "grad_norm": 3.3976492881774902,
      "learning_rate": 6.91703056768559e-05,
      "loss": 0.1238,
      "step": 2825
    },
    {
      "epoch": 1.24,
      "grad_norm": 0.9712594747543335,
      "learning_rate": 6.889737991266376e-05,
      "loss": 0.1313,
      "step": 2850
    },
    {
      "epoch": 1.26,
      "grad_norm": 0.35930949449539185,
      "learning_rate": 6.862445414847162e-05,
      "loss": 0.1228,
      "step": 2875
    },
    {
      "epoch": 1.27,
      "grad_norm": 1.873953938484192,
      "learning_rate": 6.835152838427948e-05,
      "loss": 0.1358,
      "step": 2900
    },
    {
      "epoch": 1.27,
      "eval_loss": 0.23235850036144257,
      "eval_na_accuracy": 0.8185328245162964,
      "eval_ordinal_accuracy": 0.6498122215270996,
      "eval_ordinal_mae": 0.4136166572570801,
      "eval_runtime": 162.283,
      "eval_samples_per_second": 24.519,
      "eval_steps_per_second": 3.069,
      "step": 2900
    },
    {
      "epoch": 1.28,
      "grad_norm": 1.8601669073104858,
      "learning_rate": 6.807860262008734e-05,
      "loss": 0.101,
      "step": 2925
    },
    {
      "epoch": 1.29,
      "grad_norm": 0.9282914996147156,
      "learning_rate": 6.780567685589519e-05,
      "loss": 0.1435,
      "step": 2950
    },
    {
      "epoch": 1.3,
      "grad_norm": 1.7728241682052612,
      "learning_rate": 6.753275109170306e-05,
      "loss": 0.1125,
      "step": 2975
    },
    {
      "epoch": 1.31,
      "grad_norm": 0.5749986171722412,
      "learning_rate": 6.725982532751091e-05,
      "loss": 0.1614,
      "step": 3000
    },
    {
      "epoch": 1.31,
      "eval_loss": 0.26365962624549866,
      "eval_na_accuracy": 0.7915058135986328,
      "eval_ordinal_accuracy": 0.6316093802452087,
      "eval_ordinal_mae": 0.43529626727104187,
      "eval_runtime": 161.4837,
      "eval_samples_per_second": 24.64,
      "eval_steps_per_second": 3.084,
      "step": 3000
    },
    {
      "epoch": 1.32,
      "grad_norm": 0.44815096259117126,
      "learning_rate": 6.698689956331879e-05,
      "loss": 0.1436,
      "step": 3025
    },
    {
      "epoch": 1.33,
      "grad_norm": 0.4672500491142273,
      "learning_rate": 6.671397379912665e-05,
      "loss": 0.0943,
      "step": 3050
    },
    {
      "epoch": 1.34,
      "grad_norm": 0.8902660608291626,
      "learning_rate": 6.64410480349345e-05,
      "loss": 0.1258,
      "step": 3075
    },
    {
      "epoch": 1.35,
      "grad_norm": 0.7342121005058289,
      "learning_rate": 6.616812227074237e-05,
      "loss": 0.1395,
      "step": 3100
    },
    {
      "epoch": 1.35,
      "eval_loss": 0.2445780336856842,
      "eval_na_accuracy": 0.8011583089828491,
      "eval_ordinal_accuracy": 0.672637939453125,
      "eval_ordinal_mae": 0.4019619822502136,
      "eval_runtime": 161.5727,
      "eval_samples_per_second": 24.627,
      "eval_steps_per_second": 3.082,
      "step": 3100
    },
    {
      "epoch": 1.36,
      "grad_norm": 0.7066202163696289,
      "learning_rate": 6.589519650655021e-05,
      "loss": 0.1627,
      "step": 3125
    },
    {
      "epoch": 1.38,
      "grad_norm": 0.8218971490859985,
      "learning_rate": 6.562227074235809e-05,
      "loss": 0.1116,
      "step": 3150
    },
    {
      "epoch": 1.39,
      "grad_norm": 2.74863600730896,
      "learning_rate": 6.534934497816593e-05,
      "loss": 0.1151,
      "step": 3175
    },
    {
      "epoch": 1.4,
      "grad_norm": 2.341121196746826,
      "learning_rate": 6.507641921397381e-05,
      "loss": 0.1208,
      "step": 3200
    },
    {
      "epoch": 1.4,
      "eval_loss": 0.24651078879833221,
      "eval_na_accuracy": 0.8243243098258972,
      "eval_ordinal_accuracy": 0.6763941049575806,
      "eval_ordinal_mae": 0.394586980342865,
      "eval_runtime": 160.9965,
      "eval_samples_per_second": 24.715,
      "eval_steps_per_second": 3.093,
      "step": 3200
    },
    {
      "epoch": 1.41,
      "grad_norm": 0.610958993434906,
      "learning_rate": 6.480349344978166e-05,
      "loss": 0.1145,
      "step": 3225
    },
    {
      "epoch": 1.42,
      "grad_norm": 0.43066203594207764,
      "learning_rate": 6.453056768558953e-05,
      "loss": 0.1322,
      "step": 3250
    },
    {
      "epoch": 1.43,
      "grad_norm": 0.21925854682922363,
      "learning_rate": 6.425764192139738e-05,
      "loss": 0.1602,
      "step": 3275
    },
    {
      "epoch": 1.44,
      "grad_norm": 0.34638360142707825,
      "learning_rate": 6.398471615720524e-05,
      "loss": 0.1432,
      "step": 3300
    },
    {
      "epoch": 1.44,
      "eval_loss": 0.25519701838493347,
      "eval_na_accuracy": 0.8899613618850708,
      "eval_ordinal_accuracy": 0.6576133966445923,
      "eval_ordinal_mae": 0.3918676972389221,
      "eval_runtime": 160.3951,
      "eval_samples_per_second": 24.807,
      "eval_steps_per_second": 3.105,
      "step": 3300
    },
    {
      "epoch": 1.45,
      "grad_norm": 0.5949413776397705,
      "learning_rate": 6.371179039301311e-05,
      "loss": 0.1249,
      "step": 3325
    },
    {
      "epoch": 1.46,
      "grad_norm": 0.8993425965309143,
      "learning_rate": 6.343886462882096e-05,
      "loss": 0.1139,
      "step": 3350
    },
    {
      "epoch": 1.47,
      "grad_norm": 0.7099699974060059,
      "learning_rate": 6.316593886462883e-05,
      "loss": 0.1019,
      "step": 3375
    },
    {
      "epoch": 1.48,
      "grad_norm": 2.9975674152374268,
      "learning_rate": 6.289301310043668e-05,
      "loss": 0.1358,
      "step": 3400
    },
    {
      "epoch": 1.48,
      "eval_loss": 0.2561098039150238,
      "eval_na_accuracy": 0.7895752787590027,
      "eval_ordinal_accuracy": 0.6795724034309387,
      "eval_ordinal_mae": 0.39841172099113464,
      "eval_runtime": 162.3278,
      "eval_samples_per_second": 24.512,
      "eval_steps_per_second": 3.068,
      "step": 3400
    },
    {
      "epoch": 1.5,
      "grad_norm": 0.259729266166687,
      "learning_rate": 6.262008733624455e-05,
      "loss": 0.1486,
      "step": 3425
    },
    {
      "epoch": 1.51,
      "grad_norm": 4.243904113769531,
      "learning_rate": 6.23471615720524e-05,
      "loss": 0.1652,
      "step": 3450
    },
    {
      "epoch": 1.52,
      "grad_norm": 2.9280548095703125,
      "learning_rate": 6.207423580786027e-05,
      "loss": 0.1699,
      "step": 3475
    },
    {
      "epoch": 1.53,
      "grad_norm": 0.5303541421890259,
      "learning_rate": 6.180131004366812e-05,
      "loss": 0.0877,
      "step": 3500
    },
    {
      "epoch": 1.53,
      "eval_loss": 0.23811650276184082,
      "eval_na_accuracy": 0.7876448035240173,
      "eval_ordinal_accuracy": 0.6821727752685547,
      "eval_ordinal_mae": 0.3901168704032898,
      "eval_runtime": 162.0532,
      "eval_samples_per_second": 24.554,
      "eval_steps_per_second": 3.073,
      "step": 3500
    },
    {
      "epoch": 1.54,
      "grad_norm": 0.42129185795783997,
      "learning_rate": 6.152838427947598e-05,
      "loss": 0.1149,
      "step": 3525
    },
    {
      "epoch": 1.55,
      "grad_norm": 1.27903151512146,
      "learning_rate": 6.125545851528384e-05,
      "loss": 0.1208,
      "step": 3550
    },
    {
      "epoch": 1.56,
      "grad_norm": 3.1208670139312744,
      "learning_rate": 6.09825327510917e-05,
      "loss": 0.1106,
      "step": 3575
    },
    {
      "epoch": 1.57,
      "grad_norm": 3.3159916400909424,
      "learning_rate": 6.070960698689957e-05,
      "loss": 0.1212,
      "step": 3600
    },
    {
      "epoch": 1.57,
      "eval_loss": 0.2600151598453522,
      "eval_na_accuracy": 0.7258687019348145,
      "eval_ordinal_accuracy": 0.6948858499526978,
      "eval_ordinal_mae": 0.400073766708374,
      "eval_runtime": 160.7378,
      "eval_samples_per_second": 24.755,
      "eval_steps_per_second": 3.098,
      "step": 3600
    },
    {
      "epoch": 1.58,
      "grad_norm": 3.3880019187927246,
      "learning_rate": 6.043668122270742e-05,
      "loss": 0.1593,
      "step": 3625
    },
    {
      "epoch": 1.59,
      "grad_norm": 2.641679286956787,
      "learning_rate": 6.016375545851529e-05,
      "loss": 0.1489,
      "step": 3650
    },
    {
      "epoch": 1.6,
      "grad_norm": 1.1284505128860474,
      "learning_rate": 5.9890829694323144e-05,
      "loss": 0.1097,
      "step": 3675
    },
    {
      "epoch": 1.62,
      "grad_norm": 1.6684277057647705,
      "learning_rate": 5.9617903930131005e-05,
      "loss": 0.1917,
      "step": 3700
    },
    {
      "epoch": 1.62,
      "eval_loss": 0.24585247039794922,
      "eval_na_accuracy": 0.7818532586097717,
      "eval_ordinal_accuracy": 0.6893961429595947,
      "eval_ordinal_mae": 0.3889385461807251,
      "eval_runtime": 161.3514,
      "eval_samples_per_second": 24.66,
      "eval_steps_per_second": 3.086,
      "step": 3700
    },
    {
      "epoch": 1.63,
      "grad_norm": 0.4128411114215851,
      "learning_rate": 5.934497816593887e-05,
      "loss": 0.1423,
      "step": 3725
    },
    {
      "epoch": 1.64,
      "grad_norm": 1.0505822896957397,
      "learning_rate": 5.9072052401746726e-05,
      "loss": 0.1257,
      "step": 3750
    },
    {
      "epoch": 1.65,
      "grad_norm": 0.8468612432479858,
      "learning_rate": 5.879912663755459e-05,
      "loss": 0.1296,
      "step": 3775
    },
    {
      "epoch": 1.66,
      "grad_norm": 0.31584060192108154,
      "learning_rate": 5.852620087336245e-05,
      "loss": 0.1175,
      "step": 3800
    },
    {
      "epoch": 1.66,
      "eval_loss": 0.2443784475326538,
      "eval_na_accuracy": 0.7741312980651855,
      "eval_ordinal_accuracy": 0.6818838715553284,
      "eval_ordinal_mae": 0.3937167227268219,
      "eval_runtime": 161.5524,
      "eval_samples_per_second": 24.63,
      "eval_steps_per_second": 3.083,
      "step": 3800
    },
    {
      "epoch": 1.67,
      "grad_norm": 5.336325645446777,
      "learning_rate": 5.8253275109170314e-05,
      "loss": 0.1638,
      "step": 3825
    },
    {
      "epoch": 1.68,
      "grad_norm": 1.6800111532211304,
      "learning_rate": 5.798034934497817e-05,
      "loss": 0.1314,
      "step": 3850
    },
    {
      "epoch": 1.69,
      "grad_norm": 1.491882085800171,
      "learning_rate": 5.770742358078602e-05,
      "loss": 0.1415,
      "step": 3875
    },
    {
      "epoch": 1.7,
      "grad_norm": 2.401737928390503,
      "learning_rate": 5.743449781659389e-05,
      "loss": 0.1522,
      "step": 3900
    },
    {
      "epoch": 1.7,
      "eval_loss": 0.2472807914018631,
      "eval_na_accuracy": 0.8050193190574646,
      "eval_ordinal_accuracy": 0.6607916951179504,
      "eval_ordinal_mae": 0.40097710490226746,
      "eval_runtime": 160.8219,
      "eval_samples_per_second": 24.742,
      "eval_steps_per_second": 3.097,
      "step": 3900
    },
    {
      "epoch": 1.71,
      "grad_norm": 0.6720189452171326,
      "learning_rate": 5.716157205240175e-05,
      "loss": 0.1239,
      "step": 3925
    },
    {
      "epoch": 1.72,
      "grad_norm": 2.349804639816284,
      "learning_rate": 5.688864628820961e-05,
      "loss": 0.1318,
      "step": 3950
    },
    {
      "epoch": 1.74,
      "grad_norm": 0.732524573802948,
      "learning_rate": 5.661572052401747e-05,
      "loss": 0.1423,
      "step": 3975
    },
    {
      "epoch": 1.75,
      "grad_norm": 0.4833851456642151,
      "learning_rate": 5.634279475982534e-05,
      "loss": 0.1027,
      "step": 4000
    },
    {
      "epoch": 1.75,
      "eval_loss": 0.23541530966758728,
      "eval_na_accuracy": 0.7837837934494019,
      "eval_ordinal_accuracy": 0.6477896571159363,
      "eval_ordinal_mae": 0.420841783285141,
      "eval_runtime": 155.1162,
      "eval_samples_per_second": 25.652,
      "eval_steps_per_second": 3.21,
      "step": 4000
    },
    {
      "epoch": 1.76,
      "grad_norm": 0.5297548174858093,
      "learning_rate": 5.606986899563319e-05,
      "loss": 0.1596,
      "step": 4025
    },
    {
      "epoch": 1.77,
      "grad_norm": 0.5475759506225586,
      "learning_rate": 5.5796943231441045e-05,
      "loss": 0.1272,
      "step": 4050
    },
    {
      "epoch": 1.78,
      "grad_norm": 2.1666433811187744,
      "learning_rate": 5.552401746724891e-05,
      "loss": 0.1382,
      "step": 4075
    },
    {
      "epoch": 1.79,
      "grad_norm": 3.707628011703491,
      "learning_rate": 5.5251091703056766e-05,
      "loss": 0.1343,
      "step": 4100
    },
    {
      "epoch": 1.79,
      "eval_loss": 0.228408545255661,
      "eval_na_accuracy": 0.799227774143219,
      "eval_ordinal_accuracy": 0.6743715405464172,
      "eval_ordinal_mae": 0.3976960778236389,
      "eval_runtime": 158.2657,
      "eval_samples_per_second": 25.141,
      "eval_steps_per_second": 3.147,
      "step": 4100
    },
    {
      "epoch": 1.8,
      "grad_norm": 0.5081536769866943,
      "learning_rate": 5.497816593886463e-05,
      "loss": 0.1381,
      "step": 4125
    },
    {
      "epoch": 1.81,
      "grad_norm": 0.7783900499343872,
      "learning_rate": 5.470524017467249e-05,
      "loss": 0.1391,
      "step": 4150
    },
    {
      "epoch": 1.82,
      "grad_norm": 0.6352062821388245,
      "learning_rate": 5.4432314410480354e-05,
      "loss": 0.1157,
      "step": 4175
    },
    {
      "epoch": 1.83,
      "grad_norm": 0.6280900835990906,
      "learning_rate": 5.4159388646288215e-05,
      "loss": 0.1552,
      "step": 4200
    },
    {
      "epoch": 1.83,
      "eval_loss": 0.2606957256793976,
      "eval_na_accuracy": 0.7779922485351562,
      "eval_ordinal_accuracy": 0.6714822053909302,
      "eval_ordinal_mae": 0.4044625461101532,
      "eval_runtime": 154.4211,
      "eval_samples_per_second": 25.767,
      "eval_steps_per_second": 3.225,
      "step": 4200
    },
    {
      "epoch": 1.84,
      "grad_norm": 1.6340936422348022,
      "learning_rate": 5.388646288209607e-05,
      "loss": 0.1448,
      "step": 4225
    },
    {
      "epoch": 1.86,
      "grad_norm": 3.3546087741851807,
      "learning_rate": 5.3613537117903936e-05,
      "loss": 0.1485,
      "step": 4250
    },
    {
      "epoch": 1.87,
      "grad_norm": 0.5650043487548828,
      "learning_rate": 5.334061135371179e-05,
      "loss": 0.127,
      "step": 4275
    },
    {
      "epoch": 1.88,
      "grad_norm": 0.8098490834236145,
      "learning_rate": 5.306768558951966e-05,
      "loss": 0.1172,
      "step": 4300
    },
    {
      "epoch": 1.88,
      "eval_loss": 0.24209196865558624,
      "eval_na_accuracy": 0.8281853199005127,
      "eval_ordinal_accuracy": 0.6665703654289246,
      "eval_ordinal_mae": 0.3971348702907562,
      "eval_runtime": 153.4086,
      "eval_samples_per_second": 25.937,
      "eval_steps_per_second": 3.246,
      "step": 4300
    },
    {
      "epoch": 1.89,
      "grad_norm": 1.127596378326416,
      "learning_rate": 5.2805676855895205e-05,
      "loss": 0.1808,
      "step": 4325
    },
    {
      "epoch": 1.9,
      "grad_norm": 0.5557155609130859,
      "learning_rate": 5.253275109170306e-05,
      "loss": 0.09,
      "step": 4350
    },
    {
      "epoch": 1.91,
      "grad_norm": 0.31405743956565857,
      "learning_rate": 5.2259825327510926e-05,
      "loss": 0.1061,
      "step": 4375
    },
    {
      "epoch": 1.92,
      "grad_norm": 6.475078105926514,
      "learning_rate": 5.198689956331878e-05,
      "loss": 0.1381,
      "step": 4400
    },
    {
      "epoch": 1.92,
      "eval_loss": 0.2253342866897583,
      "eval_na_accuracy": 0.7857142686843872,
      "eval_ordinal_accuracy": 0.6792834401130676,
      "eval_ordinal_mae": 0.3813394010066986,
      "eval_runtime": 156.7221,
      "eval_samples_per_second": 25.389,
      "eval_steps_per_second": 3.178,
      "step": 4400
    },
    {
      "epoch": 1.93,
      "grad_norm": 0.5825577974319458,
      "learning_rate": 5.171397379912663e-05,
      "loss": 0.1524,
      "step": 4425
    },
    {
      "epoch": 1.94,
      "grad_norm": 1.4979143142700195,
      "learning_rate": 5.14410480349345e-05,
      "loss": 0.142,
      "step": 4450
    },
    {
      "epoch": 1.95,
      "grad_norm": 0.5638359785079956,
      "learning_rate": 5.116812227074236e-05,
      "loss": 0.1225,
      "step": 4475
    },
    {
      "epoch": 1.97,
      "grad_norm": 0.44809991121292114,
      "learning_rate": 5.089519650655022e-05,
      "loss": 0.1282,
      "step": 4500
    },
    {
      "epoch": 1.97,
      "eval_loss": 0.23353050649166107,
      "eval_na_accuracy": 0.8436293601989746,
      "eval_ordinal_accuracy": 0.6509679555892944,
      "eval_ordinal_mae": 0.41455620527267456,
      "eval_runtime": 154.2508,
      "eval_samples_per_second": 25.796,
      "eval_steps_per_second": 3.229,
      "step": 4500
    },
    {
      "epoch": 1.98,
      "grad_norm": 2.9640109539031982,
      "learning_rate": 5.062227074235808e-05,
      "loss": 0.1057,
      "step": 4525
    },
    {
      "epoch": 1.99,
      "grad_norm": 0.44117558002471924,
      "learning_rate": 5.034934497816595e-05,
      "loss": 0.13,
      "step": 4550
    },
    {
      "epoch": 2.0,
      "grad_norm": 0.6167708039283752,
      "learning_rate": 5.00764192139738e-05,
      "loss": 0.1223,
      "step": 4575
    },
    {
      "epoch": 2.01,
      "grad_norm": 2.0781619548797607,
      "learning_rate": 4.9803493449781664e-05,
      "loss": 0.0734,
      "step": 4600
    },
    {
      "epoch": 2.01,
      "eval_loss": 0.23820282518863678,
      "eval_na_accuracy": 0.7895752787590027,
      "eval_ordinal_accuracy": 0.689685046672821,
      "eval_ordinal_mae": 0.38021621108055115,
      "eval_runtime": 154.7894,
      "eval_samples_per_second": 25.706,
      "eval_steps_per_second": 3.217,
      "step": 4600
    },
    {
      "epoch": 2.02,
      "grad_norm": 3.165067195892334,
      "learning_rate": 4.9530567685589524e-05,
      "loss": 0.08,
      "step": 4625
    },
    {
      "epoch": 2.03,
      "grad_norm": 0.43800944089889526,
      "learning_rate": 4.9257641921397385e-05,
      "loss": 0.0571,
      "step": 4650
    },
    {
      "epoch": 2.04,
      "grad_norm": 0.45996198058128357,
      "learning_rate": 4.898471615720524e-05,
      "loss": 0.0579,
      "step": 4675
    },
    {
      "epoch": 2.05,
      "grad_norm": 1.9151467084884644,
      "learning_rate": 4.87117903930131e-05,
      "loss": 0.1046,
      "step": 4700
    },
    {
      "epoch": 2.05,
      "eval_loss": 0.2358408272266388,
      "eval_na_accuracy": 0.8011583089828491,
      "eval_ordinal_accuracy": 0.6873735785484314,
      "eval_ordinal_mae": 0.36946654319763184,
      "eval_runtime": 150.9459,
      "eval_samples_per_second": 26.36,
      "eval_steps_per_second": 3.299,
      "step": 4700
    },
    {
      "epoch": 2.06,
      "grad_norm": 0.5566712021827698,
      "learning_rate": 4.843886462882096e-05,
      "loss": 0.068,
      "step": 4725
    },
    {
      "epoch": 2.07,
      "grad_norm": 0.5846825838088989,
      "learning_rate": 4.8165938864628827e-05,
      "loss": 0.0487,
      "step": 4750
    },
    {
      "epoch": 2.09,
      "grad_norm": 0.3993060290813446,
      "learning_rate": 4.789301310043669e-05,
      "loss": 0.0546,
      "step": 4775
    },
    {
      "epoch": 2.1,
      "grad_norm": 0.6624791026115417,
      "learning_rate": 4.762008733624455e-05,
      "loss": 0.0529,
      "step": 4800
    },
    {
      "epoch": 2.1,
      "eval_loss": 0.246298685669899,
      "eval_na_accuracy": 0.7934362888336182,
      "eval_ordinal_accuracy": 0.7096214890480042,
      "eval_ordinal_mae": 0.3595849573612213,
      "eval_runtime": 153.0771,
      "eval_samples_per_second": 25.993,
      "eval_steps_per_second": 3.253,
      "step": 4800
    },
    {
      "epoch": 2.11,
      "grad_norm": 0.33810412883758545,
      "learning_rate": 4.734716157205241e-05,
      "loss": 0.0625,
      "step": 4825
    },
    {
      "epoch": 2.12,
      "grad_norm": 3.0376217365264893,
      "learning_rate": 4.707423580786026e-05,
      "loss": 0.0749,
      "step": 4850
    },
    {
      "epoch": 2.13,
      "grad_norm": 0.4552454352378845,
      "learning_rate": 4.680131004366812e-05,
      "loss": 0.0536,
      "step": 4875
    },
    {
      "epoch": 2.14,
      "grad_norm": 0.6775699257850647,
      "learning_rate": 4.652838427947598e-05,
      "loss": 0.0687,
      "step": 4900
    },
    {
      "epoch": 2.14,
      "eval_loss": 0.26146525144577026,
      "eval_na_accuracy": 0.7857142686843872,
      "eval_ordinal_accuracy": 0.6737936735153198,
      "eval_ordinal_mae": 0.39211294054985046,
      "eval_runtime": 156.1814,
      "eval_samples_per_second": 25.477,
      "eval_steps_per_second": 3.189,
      "step": 4900
    },
    {
      "epoch": 2.15,
      "grad_norm": 0.623523473739624,
      "learning_rate": 4.625545851528384e-05,
      "loss": 0.068,
      "step": 4925
    },
    {
      "epoch": 2.16,
      "grad_norm": 0.5668771862983704,
      "learning_rate": 4.5982532751091704e-05,
      "loss": 0.0684,
      "step": 4950
    },
    {
      "epoch": 2.17,
      "grad_norm": 0.16505593061447144,
      "learning_rate": 4.5709606986899564e-05,
      "loss": 0.0546,
      "step": 4975
    },
    {
      "epoch": 2.18,
      "grad_norm": 0.5560925602912903,
      "learning_rate": 4.5436681222707425e-05,
      "loss": 0.0613,
      "step": 5000
    },
    {
      "epoch": 2.18,
      "eval_loss": 0.2542937397956848,
      "eval_na_accuracy": 0.8108108043670654,
      "eval_ordinal_accuracy": 0.6876625418663025,
      "eval_ordinal_mae": 0.36514100432395935,
      "eval_runtime": 156.8699,
      "eval_samples_per_second": 25.365,
      "eval_steps_per_second": 3.175,
      "step": 5000
    },
    {
      "epoch": 2.19,
      "grad_norm": 0.440731942653656,
      "learning_rate": 4.5163755458515285e-05,
      "loss": 0.0967,
      "step": 5025
    },
    {
      "epoch": 2.21,
      "grad_norm": 0.26702314615249634,
      "learning_rate": 4.4890829694323146e-05,
      "loss": 0.0499,
      "step": 5050
    },
    {
      "epoch": 2.22,
      "grad_norm": 0.9617776274681091,
      "learning_rate": 4.4617903930131006e-05,
      "loss": 0.0792,
      "step": 5075
    },
    {
      "epoch": 2.23,
      "grad_norm": 3.431128740310669,
      "learning_rate": 4.434497816593887e-05,
      "loss": 0.0591,
      "step": 5100
    },
    {
      "epoch": 2.23,
      "eval_loss": 0.25389814376831055,
      "eval_na_accuracy": 0.7915058135986328,
      "eval_ordinal_accuracy": 0.6885293126106262,
      "eval_ordinal_mae": 0.3693314790725708,
      "eval_runtime": 155.0651,
      "eval_samples_per_second": 25.66,
      "eval_steps_per_second": 3.212,
      "step": 5100
    },
    {
      "epoch": 2.24,
      "grad_norm": 5.355819225311279,
      "learning_rate": 4.407205240174673e-05,
      "loss": 0.0611,
      "step": 5125
    },
    {
      "epoch": 2.25,
      "grad_norm": 0.4490479528903961,
      "learning_rate": 4.379912663755459e-05,
      "loss": 0.0676,
      "step": 5150
    },
    {
      "epoch": 2.26,
      "grad_norm": 0.594838559627533,
      "learning_rate": 4.352620087336245e-05,
      "loss": 0.0567,
      "step": 5175
    },
    {
      "epoch": 2.27,
      "grad_norm": 0.2912037670612335,
      "learning_rate": 4.325327510917031e-05,
      "loss": 0.0474,
      "step": 5200
    },
    {
      "epoch": 2.27,
      "eval_loss": 0.26495158672332764,
      "eval_na_accuracy": 0.799227774143219,
      "eval_ordinal_accuracy": 0.6836174726486206,
      "eval_ordinal_mae": 0.3721810579299927,
      "eval_runtime": 156.3533,
      "eval_samples_per_second": 25.449,
      "eval_steps_per_second": 3.185,
      "step": 5200
    },
    {
      "epoch": 2.28,
      "grad_norm": 1.547253966331482,
      "learning_rate": 4.298034934497817e-05,
      "loss": 0.0583,
      "step": 5225
    },
    {
      "epoch": 2.29,
      "grad_norm": 0.6619101762771606,
      "learning_rate": 4.270742358078603e-05,
      "loss": 0.0713,
      "step": 5250
    },
    {
      "epoch": 2.3,
      "grad_norm": 0.6830999255180359,
      "learning_rate": 4.243449781659389e-05,
      "loss": 0.068,
      "step": 5275
    },
    {
      "epoch": 2.31,
      "grad_norm": 0.6977065801620483,
      "learning_rate": 4.216157205240175e-05,
      "loss": 0.0511,
      "step": 5300
    },
    {
      "epoch": 2.31,
      "eval_loss": 0.2630845010280609,
      "eval_na_accuracy": 0.8127413392066956,
      "eval_ordinal_accuracy": 0.686795711517334,
      "eval_ordinal_mae": 0.36813271045684814,
      "eval_runtime": 154.8738,
      "eval_samples_per_second": 25.692,
      "eval_steps_per_second": 3.216,
      "step": 5300
    },
    {
      "epoch": 2.33,
      "grad_norm": 0.8015338778495789,
      "learning_rate": 4.188864628820961e-05,
      "loss": 0.0511,
      "step": 5325
    },
    {
      "epoch": 2.34,
      "grad_norm": 2.4707908630371094,
      "learning_rate": 4.161572052401747e-05,
      "loss": 0.0704,
      "step": 5350
    },
    {
      "epoch": 2.35,
      "grad_norm": 0.526327908039093,
      "learning_rate": 4.134279475982533e-05,
      "loss": 0.0824,
      "step": 5375
    },
    {
      "epoch": 2.36,
      "grad_norm": 1.6800874471664429,
      "learning_rate": 4.1069868995633186e-05,
      "loss": 0.0683,
      "step": 5400
    },
    {
      "epoch": 2.36,
      "eval_loss": 0.271382212638855,
      "eval_na_accuracy": 0.7837837934494019,
      "eval_ordinal_accuracy": 0.6954637169837952,
      "eval_ordinal_mae": 0.36302649974823,
      "eval_runtime": 155.8769,
      "eval_samples_per_second": 25.527,
      "eval_steps_per_second": 3.195,
      "step": 5400
    },
    {
      "epoch": 2.37,
      "grad_norm": 0.5335835814476013,
      "learning_rate": 4.0796943231441046e-05,
      "loss": 0.056,
      "step": 5425
    },
    {
      "epoch": 2.38,
      "grad_norm": 2.2423832416534424,
      "learning_rate": 4.052401746724891e-05,
      "loss": 0.0787,
      "step": 5450
    },
    {
      "epoch": 2.39,
      "grad_norm": 0.558754026889801,
      "learning_rate": 4.025109170305677e-05,
      "loss": 0.0481,
      "step": 5475
    },
    {
      "epoch": 2.4,
      "grad_norm": 0.6908044815063477,
      "learning_rate": 3.9978165938864635e-05,
      "loss": 0.0654,
      "step": 5500
    },
    {
      "epoch": 2.4,
      "eval_loss": 0.27688542008399963,
      "eval_na_accuracy": 0.799227774143219,
      "eval_ordinal_accuracy": 0.6787055730819702,
      "eval_ordinal_mae": 0.3673117458820343,
      "eval_runtime": 154.1587,
      "eval_samples_per_second": 25.811,
      "eval_steps_per_second": 3.23,
      "step": 5500
    },
    {
      "epoch": 2.41,
      "grad_norm": 0.43695610761642456,
      "learning_rate": 3.9705240174672495e-05,
      "loss": 0.0539,
      "step": 5525
    },
    {
      "epoch": 2.42,
      "grad_norm": 3.290745496749878,
      "learning_rate": 3.9432314410480356e-05,
      "loss": 0.0608,
      "step": 5550
    },
    {
      "epoch": 2.43,
      "grad_norm": 0.25710350275039673,
      "learning_rate": 3.9159388646288216e-05,
      "loss": 0.0657,
      "step": 5575
    },
    {
      "epoch": 2.45,
      "grad_norm": 0.6269412040710449,
      "learning_rate": 3.888646288209607e-05,
      "loss": 0.0581,
      "step": 5600
    },
    {
      "epoch": 2.45,
      "eval_loss": 0.27770209312438965,
      "eval_na_accuracy": 0.799227774143219,
      "eval_ordinal_accuracy": 0.6951748132705688,
      "eval_ordinal_mae": 0.3627748191356659,
      "eval_runtime": 157.9718,
      "eval_samples_per_second": 25.188,
      "eval_steps_per_second": 3.152,
      "step": 5600
    },
    {
      "epoch": 2.46,
      "grad_norm": 1.5579248666763306,
      "learning_rate": 3.861353711790393e-05,
      "loss": 0.0556,
      "step": 5625
    },
    {
      "epoch": 2.47,
      "grad_norm": 0.30162498354911804,
      "learning_rate": 3.834061135371179e-05,
      "loss": 0.0845,
      "step": 5650
    },
    {
      "epoch": 2.48,
      "grad_norm": 0.43656259775161743,
      "learning_rate": 3.806768558951965e-05,
      "loss": 0.0616,
      "step": 5675
    },
    {
      "epoch": 2.49,
      "grad_norm": 0.4505567252635956,
      "learning_rate": 3.779475982532751e-05,
      "loss": 0.072,
      "step": 5700
    },
    {
      "epoch": 2.49,
      "eval_loss": 0.29192212224006653,
      "eval_na_accuracy": 0.7683397531509399,
      "eval_ordinal_accuracy": 0.6888182759284973,
      "eval_ordinal_mae": 0.36100971698760986,
      "eval_runtime": 155.17,
      "eval_samples_per_second": 25.643,
      "eval_steps_per_second": 3.209,
      "step": 5700
    },
    {
      "epoch": 2.5,
      "grad_norm": 0.8305689692497253,
      "learning_rate": 3.752183406113537e-05,
      "loss": 0.0489,
      "step": 5725
    },
    {
      "epoch": 2.51,
      "grad_norm": 0.5452491641044617,
      "learning_rate": 3.724890829694323e-05,
      "loss": 0.0584,
      "step": 5750
    },
    {
      "epoch": 2.52,
      "grad_norm": 0.7193215489387512,
      "learning_rate": 3.697598253275109e-05,
      "loss": 0.0643,
      "step": 5775
    },
    {
      "epoch": 2.53,
      "grad_norm": 0.4529215693473816,
      "learning_rate": 3.6703056768558954e-05,
      "loss": 0.0737,
      "step": 5800
    },
    {
      "epoch": 2.53,
      "eval_loss": 0.2807420790195465,
      "eval_na_accuracy": 0.7837837934494019,
      "eval_ordinal_accuracy": 0.6983530521392822,
      "eval_ordinal_mae": 0.3611612617969513,
      "eval_runtime": 154.656,
      "eval_samples_per_second": 25.728,
      "eval_steps_per_second": 3.22,
      "step": 5800
    },
    {
      "epoch": 2.54,
      "grad_norm": 0.4110221564769745,
      "learning_rate": 3.6430131004366814e-05,
      "loss": 0.0548,
      "step": 5825
    },
    {
      "epoch": 2.55,
      "grad_norm": 0.7328541278839111,
      "learning_rate": 3.6157205240174675e-05,
      "loss": 0.0728,
      "step": 5850
    },
    {
      "epoch": 2.57,
      "grad_norm": 0.2497873455286026,
      "learning_rate": 3.5884279475982535e-05,
      "loss": 0.0608,
      "step": 5875
    },
    {
      "epoch": 2.58,
      "grad_norm": 6.07706880569458,
      "learning_rate": 3.5611353711790396e-05,
      "loss": 0.0667,
      "step": 5900
    },
    {
      "epoch": 2.58,
      "eval_loss": 0.292630672454834,
      "eval_na_accuracy": 0.7509652376174927,
      "eval_ordinal_accuracy": 0.7000866532325745,
      "eval_ordinal_mae": 0.36070069670677185,
      "eval_runtime": 156.6494,
      "eval_samples_per_second": 25.401,
      "eval_steps_per_second": 3.179,
      "step": 5900
    },
    {
      "epoch": 2.59,
      "grad_norm": 2.6040148735046387,
      "learning_rate": 3.5338427947598256e-05,
      "loss": 0.0745,
      "step": 5925
    },
    {
      "epoch": 2.6,
      "grad_norm": 0.428023099899292,
      "learning_rate": 3.506550218340611e-05,
      "loss": 0.0455,
      "step": 5950
    },
    {
      "epoch": 2.61,
      "grad_norm": 0.24350808560848236,
      "learning_rate": 3.479257641921397e-05,
      "loss": 0.067,
      "step": 5975
    },
    {
      "epoch": 2.62,
      "grad_norm": 0.6879482865333557,
      "learning_rate": 3.451965065502184e-05,
      "loss": 0.0669,
      "step": 6000
    },
    {
      "epoch": 2.62,
      "eval_loss": 0.2874707579612732,
      "eval_na_accuracy": 0.799227774143219,
      "eval_ordinal_accuracy": 0.6891071796417236,
      "eval_ordinal_mae": 0.36164331436157227,
      "eval_runtime": 155.7575,
      "eval_samples_per_second": 25.546,
      "eval_steps_per_second": 3.197,
      "step": 6000
    },
    {
      "epoch": 2.63,
      "grad_norm": 0.786054790019989,
      "learning_rate": 3.42467248908297e-05,
      "loss": 0.0736,
      "step": 6025
    },
    {
      "epoch": 2.64,
      "grad_norm": 0.7047861218452454,
      "learning_rate": 3.397379912663756e-05,
      "loss": 0.0551,
      "step": 6050
    },
    {
      "epoch": 2.65,
      "grad_norm": 0.6863640546798706,
      "learning_rate": 3.370087336244542e-05,
      "loss": 0.0643,
      "step": 6075
    },
    {
      "epoch": 2.66,
      "grad_norm": 0.6037794947624207,
      "learning_rate": 3.342794759825328e-05,
      "loss": 0.0535,
      "step": 6100
    },
    {
      "epoch": 2.66,
      "eval_loss": 0.2853965759277344,
      "eval_na_accuracy": 0.7683397531509399,
      "eval_ordinal_accuracy": 0.6960415840148926,
      "eval_ordinal_mae": 0.35648074746131897,
      "eval_runtime": 156.7027,
      "eval_samples_per_second": 25.392,
      "eval_steps_per_second": 3.178,
      "step": 6100
    },
    {
      "epoch": 2.67,
      "grad_norm": 0.30565348267555237,
      "learning_rate": 3.315502183406114e-05,
      "loss": 0.0412,
      "step": 6125
    },
    {
      "epoch": 2.69,
      "grad_norm": 0.3631564974784851,
      "learning_rate": 3.2882096069868994e-05,
      "loss": 0.0584,
      "step": 6150
    },
    {
      "epoch": 2.7,
      "grad_norm": 0.6103675365447998,
      "learning_rate": 3.2609170305676854e-05,
      "loss": 0.0481,
      "step": 6175
    },
    {
      "epoch": 2.71,
      "grad_norm": 10.306925773620605,
      "learning_rate": 3.2336244541484715e-05,
      "loss": 0.06,
      "step": 6200
    },
    {
      "epoch": 2.71,
      "eval_loss": 0.28473392128944397,
      "eval_na_accuracy": 0.7741312980651855,
      "eval_ordinal_accuracy": 0.7015313506126404,
      "eval_ordinal_mae": 0.3500910997390747,
      "eval_runtime": 154.8993,
      "eval_samples_per_second": 25.688,
      "eval_steps_per_second": 3.215,
      "step": 6200
    },
    {
      "epoch": 2.72,
      "grad_norm": 0.5145313143730164,
      "learning_rate": 3.2063318777292575e-05,
      "loss": 0.0586,
      "step": 6225
    },
    {
      "epoch": 2.73,
      "grad_norm": 0.9967983365058899,
      "learning_rate": 3.1790393013100436e-05,
      "loss": 0.0736,
      "step": 6250
    },
    {
      "epoch": 2.74,
      "grad_norm": 0.652100682258606,
      "learning_rate": 3.15174672489083e-05,
      "loss": 0.057,
      "step": 6275
    },
    {
      "epoch": 2.75,
      "grad_norm": 0.64205402135849,
      "learning_rate": 3.1244541484716164e-05,
      "loss": 0.0534,
      "step": 6300
    },
    {
      "epoch": 2.75,
      "eval_loss": 0.28205522894859314,
      "eval_na_accuracy": 0.7625482678413391,
      "eval_ordinal_accuracy": 0.7006645202636719,
      "eval_ordinal_mae": 0.34947505593299866,
      "eval_runtime": 155.841,
      "eval_samples_per_second": 25.532,
      "eval_steps_per_second": 3.196,
      "step": 6300
    },
    {
      "epoch": 2.76,
      "grad_norm": 1.2284494638442993,
      "learning_rate": 3.097161572052402e-05,
      "loss": 0.0517,
      "step": 6325
    },
    {
      "epoch": 2.77,
      "grad_norm": 0.4024920165538788,
      "learning_rate": 3.069868995633188e-05,
      "loss": 0.0857,
      "step": 6350
    },
    {
      "epoch": 2.78,
      "grad_norm": 0.5288224816322327,
      "learning_rate": 3.0425764192139738e-05,
      "loss": 0.051,
      "step": 6375
    },
    {
      "epoch": 2.79,
      "grad_norm": 0.6519297957420349,
      "learning_rate": 3.01528384279476e-05,
      "loss": 0.0526,
      "step": 6400
    },
    {
      "epoch": 2.79,
      "eval_loss": 0.28344523906707764,
      "eval_na_accuracy": 0.7625482678413391,
      "eval_ordinal_accuracy": 0.670037567615509,
      "eval_ordinal_mae": 0.385305792093277,
      "eval_runtime": 154.7606,
      "eval_samples_per_second": 25.711,
      "eval_steps_per_second": 3.218,
      "step": 6400
    },
    {
      "epoch": 2.81,
      "grad_norm": 0.7009943723678589,
      "learning_rate": 2.987991266375546e-05,
      "loss": 0.0624,
      "step": 6425
    },
    {
      "epoch": 2.82,
      "grad_norm": 0.6731958389282227,
      "learning_rate": 2.9606986899563323e-05,
      "loss": 0.068,
      "step": 6450
    },
    {
      "epoch": 2.83,
      "grad_norm": 0.3764704763889313,
      "learning_rate": 2.9334061135371184e-05,
      "loss": 0.0677,
      "step": 6475
    },
    {
      "epoch": 2.84,
      "grad_norm": 0.5906457901000977,
      "learning_rate": 2.9061135371179037e-05,
      "loss": 0.0841,
      "step": 6500
    },
    {
      "epoch": 2.84,
      "eval_loss": 0.28385940194129944,
      "eval_na_accuracy": 0.7490347623825073,
      "eval_ordinal_accuracy": 0.7044206857681274,
      "eval_ordinal_mae": 0.35039493441581726,
      "eval_runtime": 156.3169,
      "eval_samples_per_second": 25.455,
      "eval_steps_per_second": 3.186,
      "step": 6500
    },
    {
      "epoch": 2.85,
      "grad_norm": 0.3305709958076477,
      "learning_rate": 2.8788209606986898e-05,
      "loss": 0.0652,
      "step": 6525
    },
    {
      "epoch": 2.86,
      "grad_norm": 0.4257774353027344,
      "learning_rate": 2.8526200873362446e-05,
      "loss": 0.0608,
      "step": 6550
    },
    {
      "epoch": 2.87,
      "grad_norm": 0.9447629451751709,
      "learning_rate": 2.8253275109170307e-05,
      "loss": 0.0527,
      "step": 6575
    },
    {
      "epoch": 2.88,
      "grad_norm": 0.8275176882743835,
      "learning_rate": 2.7980349344978167e-05,
      "loss": 0.0529,
      "step": 6600
    },
    {
      "epoch": 2.88,
      "eval_loss": 0.2857762575149536,
      "eval_na_accuracy": 0.7818532586097717,
      "eval_ordinal_accuracy": 0.689685046672821,
      "eval_ordinal_mae": 0.35951367020606995,
      "eval_runtime": 156.4554,
      "eval_samples_per_second": 25.432,
      "eval_steps_per_second": 3.183,
      "step": 6600
    },
    {
      "epoch": 2.89,
      "grad_norm": 0.5894487500190735,
      "learning_rate": 2.770742358078603e-05,
      "loss": 0.0643,
      "step": 6625
    },
    {
      "epoch": 2.9,
      "grad_norm": 0.8872036933898926,
      "learning_rate": 2.743449781659389e-05,
      "loss": 0.0578,
      "step": 6650
    },
    {
      "epoch": 2.91,
      "grad_norm": 0.511169970035553,
      "learning_rate": 2.7161572052401745e-05,
      "loss": 0.0393,
      "step": 6675
    },
    {
      "epoch": 2.93,
      "grad_norm": 0.5537564754486084,
      "learning_rate": 2.6888646288209606e-05,
      "loss": 0.0811,
      "step": 6700
    },
    {
      "epoch": 2.93,
      "eval_loss": 0.28430891036987305,
      "eval_na_accuracy": 0.7799227833747864,
      "eval_ordinal_accuracy": 0.7047096490859985,
      "eval_ordinal_mae": 0.34795093536376953,
      "eval_runtime": 156.7138,
      "eval_samples_per_second": 25.39,
      "eval_steps_per_second": 3.178,
      "step": 6700
    },
    {
      "epoch": 2.94,
      "grad_norm": 1.6618189811706543,
      "learning_rate": 2.661572052401747e-05,
      "loss": 0.0705,
      "step": 6725
    },
    {
      "epoch": 2.95,
      "grad_norm": 0.4913439154624939,
      "learning_rate": 2.634279475982533e-05,
      "loss": 0.0875,
      "step": 6750
    },
    {
      "epoch": 2.96,
      "grad_norm": 0.5068143010139465,
      "learning_rate": 2.606986899563319e-05,
      "loss": 0.0557,
      "step": 6775
    },
    {
      "epoch": 2.97,
      "grad_norm": 0.23857353627681732,
      "learning_rate": 2.579694323144105e-05,
      "loss": 0.0502,
      "step": 6800
    },
    {
      "epoch": 2.97,
      "eval_loss": 0.28915783762931824,
      "eval_na_accuracy": 0.7818532586097717,
      "eval_ordinal_accuracy": 0.700953483581543,
      "eval_ordinal_mae": 0.34830236434936523,
      "eval_runtime": 155.6303,
      "eval_samples_per_second": 25.567,
      "eval_steps_per_second": 3.2,
      "step": 6800
    },
    {
      "epoch": 2.98,
      "grad_norm": 0.9183345437049866,
      "learning_rate": 2.552401746724891e-05,
      "loss": 0.0514,
      "step": 6825
    },
    {
      "epoch": 2.99,
      "grad_norm": 0.6839514374732971,
      "learning_rate": 2.525109170305677e-05,
      "loss": 0.0682,
      "step": 6850
    },
    {
      "epoch": 3.0,
      "grad_norm": 2.4068310260772705,
      "learning_rate": 2.4978165938864632e-05,
      "loss": 0.0709,
      "step": 6875
    },
    {
      "epoch": 3.01,
      "grad_norm": 0.04599028080701828,
      "learning_rate": 2.470524017467249e-05,
      "loss": 0.0273,
      "step": 6900
    },
    {
      "epoch": 3.01,
      "eval_loss": 0.2801385819911957,
      "eval_na_accuracy": 0.8108108043670654,
      "eval_ordinal_accuracy": 0.6957526803016663,
      "eval_ordinal_mae": 0.34536227583885193,
      "eval_runtime": 156.2048,
      "eval_samples_per_second": 25.473,
      "eval_steps_per_second": 3.188,
      "step": 6900
    },
    {
      "epoch": 3.02,
      "grad_norm": 0.5234570503234863,
      "learning_rate": 2.443231441048035e-05,
      "loss": 0.0273,
      "step": 6925
    },
    {
      "epoch": 3.03,
      "grad_norm": 0.4756307601928711,
      "learning_rate": 2.415938864628821e-05,
      "loss": 0.0338,
      "step": 6950
    },
    {
      "epoch": 3.05,
      "grad_norm": 0.40528589487075806,
      "learning_rate": 2.388646288209607e-05,
      "loss": 0.0391,
      "step": 6975
    },
    {
      "epoch": 3.06,
      "grad_norm": 0.4284062385559082,
      "learning_rate": 2.361353711790393e-05,
      "loss": 0.0306,
      "step": 7000
    },
    {
      "epoch": 3.06,
      "eval_loss": 0.2782219350337982,
      "eval_na_accuracy": 0.8030887842178345,
      "eval_ordinal_accuracy": 0.7023981213569641,
      "eval_ordinal_mae": 0.3443802297115326,
      "eval_runtime": 155.0659,
      "eval_samples_per_second": 25.66,
      "eval_steps_per_second": 3.212,
      "step": 7000
    },
    {
      "epoch": 3.07,
      "grad_norm": 0.420832097530365,
      "learning_rate": 2.3340611353711792e-05,
      "loss": 0.0308,
      "step": 7025
    },
    {
      "epoch": 3.08,
      "grad_norm": 0.12627224624156952,
      "learning_rate": 2.3067685589519653e-05,
      "loss": 0.0219,
      "step": 7050
    },
    {
      "epoch": 3.09,
      "grad_norm": 0.6852111220359802,
      "learning_rate": 2.279475982532751e-05,
      "loss": 0.0289,
      "step": 7075
    },
    {
      "epoch": 3.1,
      "grad_norm": 0.4591895043849945,
      "learning_rate": 2.252183406113537e-05,
      "loss": 0.0257,
      "step": 7100
    },
    {
      "epoch": 3.1,
      "eval_loss": 0.2796567380428314,
      "eval_na_accuracy": 0.7934362888336182,
      "eval_ordinal_accuracy": 0.7084657549858093,
      "eval_ordinal_mae": 0.33523455262184143,
      "eval_runtime": 155.6618,
      "eval_samples_per_second": 25.562,
      "eval_steps_per_second": 3.199,
      "step": 7100
    },
    {
      "epoch": 3.11,
      "grad_norm": 7.793858051300049,
      "learning_rate": 2.2248908296943234e-05,
      "loss": 0.0323,
      "step": 7125
    },
    {
      "epoch": 3.12,
      "grad_norm": 0.4730852246284485,
      "learning_rate": 2.1975982532751095e-05,
      "loss": 0.0432,
      "step": 7150
    },
    {
      "epoch": 3.13,
      "grad_norm": 0.9067665338516235,
      "learning_rate": 2.170305676855895e-05,
      "loss": 0.0324,
      "step": 7175
    },
    {
      "epoch": 3.14,
      "grad_norm": 0.23400144279003143,
      "learning_rate": 2.1430131004366812e-05,
      "loss": 0.0241,
      "step": 7200
    },
    {
      "epoch": 3.14,
      "eval_loss": 0.2827575206756592,
      "eval_na_accuracy": 0.7953668236732483,
      "eval_ordinal_accuracy": 0.7058653831481934,
      "eval_ordinal_mae": 0.33425432443618774,
      "eval_runtime": 157.3972,
      "eval_samples_per_second": 25.28,
      "eval_steps_per_second": 3.164,
      "step": 7200
    },
    {
      "epoch": 3.16,
      "grad_norm": 0.6540936231613159,
      "learning_rate": 2.1157205240174673e-05,
      "loss": 0.0266,
      "step": 7225
    },
    {
      "epoch": 3.17,
      "grad_norm": 0.3661313056945801,
      "learning_rate": 2.0884279475982536e-05,
      "loss": 0.03,
      "step": 7250
    },
    {
      "epoch": 3.18,
      "grad_norm": 0.48538270592689514,
      "learning_rate": 2.0611353711790394e-05,
      "loss": 0.0268,
      "step": 7275
    },
    {
      "epoch": 3.19,
      "grad_norm": 0.12903234362602234,
      "learning_rate": 2.0338427947598254e-05,
      "loss": 0.0255,
      "step": 7300
    },
    {
      "epoch": 3.19,
      "eval_loss": 0.28903913497924805,
      "eval_na_accuracy": 0.8050193190574646,
      "eval_ordinal_accuracy": 0.6980641484260559,
      "eval_ordinal_mae": 0.3364236354827881,
      "eval_runtime": 155.705,
      "eval_samples_per_second": 25.555,
      "eval_steps_per_second": 3.198,
      "step": 7300
    },
    {
      "epoch": 3.2,
      "grad_norm": 0.6427123546600342,
      "learning_rate": 2.0065502183406115e-05,
      "loss": 0.0256,
      "step": 7325
    },
    {
      "epoch": 3.21,
      "grad_norm": 0.9630228281021118,
      "learning_rate": 1.9792576419213975e-05,
      "loss": 0.0261,
      "step": 7350
    },
    {
      "epoch": 3.22,
      "grad_norm": 0.4561873972415924,
      "learning_rate": 1.9519650655021836e-05,
      "loss": 0.0337,
      "step": 7375
    },
    {
      "epoch": 3.23,
      "grad_norm": 0.40141957998275757,
      "learning_rate": 1.9246724890829696e-05,
      "loss": 0.0245,
      "step": 7400
    },
    {
      "epoch": 3.23,
      "eval_loss": 0.29058343172073364,
      "eval_na_accuracy": 0.799227774143219,
      "eval_ordinal_accuracy": 0.7044206857681274,
      "eval_ordinal_mae": 0.3391839265823364,
      "eval_runtime": 156.6469,
      "eval_samples_per_second": 25.401,
      "eval_steps_per_second": 3.179,
      "step": 7400
    },
    {
      "epoch": 3.24,
      "grad_norm": 0.04360814392566681,
      "learning_rate": 1.8973799126637557e-05,
      "loss": 0.0271,
      "step": 7425
    },
    {
      "epoch": 3.25,
      "grad_norm": 0.2782489061355591,
      "learning_rate": 1.8700873362445414e-05,
      "loss": 0.0345,
      "step": 7450
    },
    {
      "epoch": 3.26,
      "grad_norm": 0.4086083769798279,
      "learning_rate": 1.8427947598253274e-05,
      "loss": 0.0519,
      "step": 7475
    },
    {
      "epoch": 3.28,
      "grad_norm": 0.37691470980644226,
      "learning_rate": 1.8155021834061138e-05,
      "loss": 0.0232,
      "step": 7500
    },
    {
      "epoch": 3.28,
      "eval_loss": 0.28911489248275757,
      "eval_na_accuracy": 0.7857142686843872,
      "eval_ordinal_accuracy": 0.7035539150238037,
      "eval_ordinal_mae": 0.3337612450122833,
      "eval_runtime": 155.5823,
      "eval_samples_per_second": 25.575,
      "eval_steps_per_second": 3.201,
      "step": 7500
    },
    {
      "epoch": 3.29,
      "grad_norm": 0.220824733376503,
      "learning_rate": 1.7882096069869e-05,
      "loss": 0.0274,
      "step": 7525
    },
    {
      "epoch": 3.3,
      "grad_norm": 0.38171643018722534,
      "learning_rate": 1.7609170305676856e-05,
      "loss": 0.0257,
      "step": 7550
    },
    {
      "epoch": 3.31,
      "grad_norm": 0.6748324632644653,
      "learning_rate": 1.7336244541484716e-05,
      "loss": 0.0212,
      "step": 7575
    },
    {
      "epoch": 3.32,
      "grad_norm": 0.42487770318984985,
      "learning_rate": 1.7063318777292577e-05,
      "loss": 0.0352,
      "step": 7600
    },
    {
      "epoch": 3.32,
      "eval_loss": 0.2908113896846771,
      "eval_na_accuracy": 0.7895752787590027,
      "eval_ordinal_accuracy": 0.6925743818283081,
      "eval_ordinal_mae": 0.34433993697166443,
      "eval_runtime": 154.4772,
      "eval_samples_per_second": 25.758,
      "eval_steps_per_second": 3.224,
      "step": 7600
    },
    {
      "epoch": 3.33,
      "grad_norm": 0.6813647747039795,
      "learning_rate": 1.6790393013100437e-05,
      "loss": 0.0405,
      "step": 7625
    },
    {
      "epoch": 3.34,
      "grad_norm": 0.46545034646987915,
      "learning_rate": 1.6517467248908298e-05,
      "loss": 0.0252,
      "step": 7650
    },
    {
      "epoch": 3.35,
      "grad_norm": 0.32561665773391724,
      "learning_rate": 1.6244541484716158e-05,
      "loss": 0.0287,
      "step": 7675
    },
    {
      "epoch": 3.36,
      "grad_norm": 0.38751569390296936,
      "learning_rate": 1.597161572052402e-05,
      "loss": 0.0376,
      "step": 7700
    },
    {
      "epoch": 3.36,
      "eval_loss": 0.2876608371734619,
      "eval_na_accuracy": 0.7915058135986328,
      "eval_ordinal_accuracy": 0.7049985527992249,
      "eval_ordinal_mae": 0.331503301858902,
      "eval_runtime": 155.51,
      "eval_samples_per_second": 25.587,
      "eval_steps_per_second": 3.202,
      "step": 7700
    },
    {
      "epoch": 3.37,
      "grad_norm": 0.317843496799469,
      "learning_rate": 1.5698689956331876e-05,
      "loss": 0.0346,
      "step": 7725
    },
    {
      "epoch": 3.38,
      "grad_norm": 0.09346043318510056,
      "learning_rate": 1.542576419213974e-05,
      "loss": 0.0187,
      "step": 7750
    },
    {
      "epoch": 3.4,
      "grad_norm": 0.508102536201477,
      "learning_rate": 1.51528384279476e-05,
      "loss": 0.0208,
      "step": 7775
    },
    {
      "epoch": 3.41,
      "grad_norm": 0.3914332687854767,
      "learning_rate": 1.487991266375546e-05,
      "loss": 0.025,
      "step": 7800
    },
    {
      "epoch": 3.41,
      "eval_loss": 0.2889249920845032,
      "eval_na_accuracy": 0.7895752787590027,
      "eval_ordinal_accuracy": 0.7075989842414856,
      "eval_ordinal_mae": 0.33163872361183167,
      "eval_runtime": 156.7804,
      "eval_samples_per_second": 25.379,
      "eval_steps_per_second": 3.176,
      "step": 7800
    },
    {
      "epoch": 3.42,
      "grad_norm": 0.6848337054252625,
      "learning_rate": 1.460698689956332e-05,
      "loss": 0.0259,
      "step": 7825
    },
    {
      "epoch": 3.43,
      "grad_norm": 0.5732501149177551,
      "learning_rate": 1.433406113537118e-05,
      "loss": 0.0222,
      "step": 7850
    },
    {
      "epoch": 3.44,
      "grad_norm": 0.638590931892395,
      "learning_rate": 1.406113537117904e-05,
      "loss": 0.0237,
      "step": 7875
    },
    {
      "epoch": 3.45,
      "grad_norm": 0.7879953980445862,
      "learning_rate": 1.37882096069869e-05,
      "loss": 0.0225,
      "step": 7900
    },
    {
      "epoch": 3.45,
      "eval_loss": 0.2901510000228882,
      "eval_na_accuracy": 0.7818532586097717,
      "eval_ordinal_accuracy": 0.7070211172103882,
      "eval_ordinal_mae": 0.32855790853500366,
      "eval_runtime": 157.7082,
      "eval_samples_per_second": 25.23,
      "eval_steps_per_second": 3.158,
      "step": 7900
    },
    {
      "epoch": 3.46,
      "grad_norm": 0.45826223492622375,
      "learning_rate": 1.351528384279476e-05,
      "loss": 0.03,
      "step": 7925
    },
    {
      "epoch": 3.47,
      "grad_norm": 0.5093744993209839,
      "learning_rate": 1.324235807860262e-05,
      "loss": 0.023,
      "step": 7950
    },
    {
      "epoch": 3.48,
      "grad_norm": 0.39197513461112976,
      "learning_rate": 1.2969432314410482e-05,
      "loss": 0.0263,
      "step": 7975
    },
    {
      "epoch": 3.49,
      "grad_norm": 0.4618348777294159,
      "learning_rate": 1.269650655021834e-05,
      "loss": 0.024,
      "step": 8000
    },
    {
      "epoch": 3.49,
      "eval_loss": 0.29018646478652954,
      "eval_na_accuracy": 0.7953668236732483,
      "eval_ordinal_accuracy": 0.7101993560791016,
      "eval_ordinal_mae": 0.3269650340080261,
      "eval_runtime": 155.3244,
      "eval_samples_per_second": 25.617,
      "eval_steps_per_second": 3.206,
      "step": 8000
    },
    {
      "epoch": 3.5,
      "grad_norm": 0.5933849811553955,
      "learning_rate": 1.2423580786026202e-05,
      "loss": 0.0225,
      "step": 8025
    },
    {
      "epoch": 3.52,
      "grad_norm": 0.04294763505458832,
      "learning_rate": 1.2150655021834062e-05,
      "loss": 0.0228,
      "step": 8050
    },
    {
      "epoch": 3.53,
      "grad_norm": 0.5286312699317932,
      "learning_rate": 1.1877729257641921e-05,
      "loss": 0.03,
      "step": 8075
    },
    {
      "epoch": 3.54,
      "grad_norm": 0.09194879978895187,
      "learning_rate": 1.1604803493449783e-05,
      "loss": 0.0404,
      "step": 8100
    },
    {
      "epoch": 3.54,
      "eval_loss": 0.29498282074928284,
      "eval_na_accuracy": 0.7895752787590027,
      "eval_ordinal_accuracy": 0.705287516117096,
      "eval_ordinal_mae": 0.32936587929725647,
      "eval_runtime": 156.3299,
      "eval_samples_per_second": 25.453,
      "eval_steps_per_second": 3.186,
      "step": 8100
    },
    {
      "epoch": 3.55,
      "grad_norm": 0.3700167238712311,
      "learning_rate": 1.1331877729257642e-05,
      "loss": 0.0293,
      "step": 8125
    },
    {
      "epoch": 3.56,
      "grad_norm": 0.10644713789224625,
      "learning_rate": 1.1058951965065504e-05,
      "loss": 0.0265,
      "step": 8150
    },
    {
      "epoch": 3.57,
      "grad_norm": 0.4384317994117737,
      "learning_rate": 1.0786026200873363e-05,
      "loss": 0.0274,
      "step": 8175
    },
    {
      "epoch": 3.58,
      "grad_norm": 0.4964589774608612,
      "learning_rate": 1.0513100436681223e-05,
      "loss": 0.0221,
      "step": 8200
    },
    {
      "epoch": 3.58,
      "eval_loss": 0.2923668920993805,
      "eval_na_accuracy": 0.7934362888336182,
      "eval_ordinal_accuracy": 0.7093325853347778,
      "eval_ordinal_mae": 0.3270767033100128,
      "eval_runtime": 156.0459,
      "eval_samples_per_second": 25.499,
      "eval_steps_per_second": 3.191,
      "step": 8200
    },
    {
      "epoch": 3.59,
      "grad_norm": 0.20232300460338593,
      "learning_rate": 1.0240174672489084e-05,
      "loss": 0.0253,
      "step": 8225
    },
    {
      "epoch": 3.6,
      "grad_norm": 0.19642572104930878,
      "learning_rate": 9.967248908296943e-06,
      "loss": 0.0309,
      "step": 8250
    },
    {
      "epoch": 3.61,
      "grad_norm": 0.12435351312160492,
      "learning_rate": 9.694323144104805e-06,
      "loss": 0.0275,
      "step": 8275
    },
    {
      "epoch": 3.62,
      "grad_norm": 0.3429054915904999,
      "learning_rate": 9.421397379912664e-06,
      "loss": 0.0182,
      "step": 8300
    },
    {
      "epoch": 3.62,
      "eval_loss": 0.29207319021224976,
      "eval_na_accuracy": 0.7934362888336182,
      "eval_ordinal_accuracy": 0.7104883193969727,
      "eval_ordinal_mae": 0.32371771335601807,
      "eval_runtime": 157.1157,
      "eval_samples_per_second": 25.325,
      "eval_steps_per_second": 3.17,
      "step": 8300
    },
    {
      "epoch": 3.64,
      "grad_norm": 0.11734936386346817,
      "learning_rate": 9.148471615720524e-06,
      "loss": 0.0341,
      "step": 8325
    },
    {
      "epoch": 3.65,
      "grad_norm": 0.18671230971813202,
      "learning_rate": 8.875545851528385e-06,
      "loss": 0.0257,
      "step": 8350
    },
    {
      "epoch": 3.66,
      "grad_norm": 0.1754232794046402,
      "learning_rate": 8.602620087336245e-06,
      "loss": 0.0306,
      "step": 8375
    },
    {
      "epoch": 3.67,
      "grad_norm": 0.2011016607284546,
      "learning_rate": 8.329694323144106e-06,
      "loss": 0.0304,
      "step": 8400
    },
    {
      "epoch": 3.67,
      "eval_loss": 0.29112711548805237,
      "eval_na_accuracy": 0.7857142686843872,
      "eval_ordinal_accuracy": 0.7133776545524597,
      "eval_ordinal_mae": 0.3231416344642639,
      "eval_runtime": 155.6543,
      "eval_samples_per_second": 25.563,
      "eval_steps_per_second": 3.199,
      "step": 8400
    },
    {
      "epoch": 3.68,
      "grad_norm": 1.1286529302597046,
      "learning_rate": 8.056768558951966e-06,
      "loss": 0.0285,
      "step": 8425
    },
    {
      "epoch": 3.69,
      "grad_norm": 0.4878416359424591,
      "learning_rate": 7.783842794759825e-06,
      "loss": 0.0163,
      "step": 8450
    },
    {
      "epoch": 3.7,
      "grad_norm": 0.5879720449447632,
      "learning_rate": 7.510917030567686e-06,
      "loss": 0.0433,
      "step": 8475
    },
    {
      "epoch": 3.71,
      "grad_norm": 0.8792235255241394,
      "learning_rate": 7.237991266375546e-06,
      "loss": 0.0193,
      "step": 8500
    },
    {
      "epoch": 3.71,
      "eval_loss": 0.2914559841156006,
      "eval_na_accuracy": 0.7837837934494019,
      "eval_ordinal_accuracy": 0.7165558934211731,
      "eval_ordinal_mae": 0.32214629650115967,
      "eval_runtime": 155.1881,
      "eval_samples_per_second": 25.64,
      "eval_steps_per_second": 3.209,
      "step": 8500
    },
    {
      "epoch": 3.72,
      "grad_norm": 0.2179727405309677,
      "learning_rate": 6.9650655021834055e-06,
      "loss": 0.0237,
      "step": 8525
    },
    {
      "epoch": 3.73,
      "grad_norm": 0.7128644585609436,
      "learning_rate": 6.692139737991267e-06,
      "loss": 0.0306,
      "step": 8550
    },
    {
      "epoch": 3.74,
      "grad_norm": 0.2546403110027313,
      "learning_rate": 6.4192139737991265e-06,
      "loss": 0.0215,
      "step": 8575
    },
    {
      "epoch": 3.76,
      "grad_norm": 0.39975956082344055,
      "learning_rate": 6.146288209606987e-06,
      "loss": 0.0223,
      "step": 8600
    },
    {
      "epoch": 3.76,
      "eval_loss": 0.29310527443885803,
      "eval_na_accuracy": 0.7895752787590027,
      "eval_ordinal_accuracy": 0.7122219204902649,
      "eval_ordinal_mae": 0.32349658012390137,
      "eval_runtime": 154.5645,
      "eval_samples_per_second": 25.743,
      "eval_steps_per_second": 3.222,
      "step": 8600
    },
    {
      "epoch": 3.77,
      "grad_norm": 0.34678587317466736,
      "learning_rate": 5.884279475982533e-06,
      "loss": 0.0203,
      "step": 8625
    },
    {
      "epoch": 3.78,
      "grad_norm": 0.46029233932495117,
      "learning_rate": 5.611353711790393e-06,
      "loss": 0.0252,
      "step": 8650
    },
    {
      "epoch": 3.79,
      "grad_norm": 0.3514462113380432,
      "learning_rate": 5.338427947598254e-06,
      "loss": 0.0182,
      "step": 8675
    },
    {
      "epoch": 3.8,
      "grad_norm": 0.9231175184249878,
      "learning_rate": 5.065502183406113e-06,
      "loss": 0.0254,
      "step": 8700
    },
    {
      "epoch": 3.8,
      "eval_loss": 0.2946593761444092,
      "eval_na_accuracy": 0.7876448035240173,
      "eval_ordinal_accuracy": 0.7174227237701416,
      "eval_ordinal_mae": 0.32142174243927,
      "eval_runtime": 155.717,
      "eval_samples_per_second": 25.553,
      "eval_steps_per_second": 3.198,
      "step": 8700
    },
    {
      "epoch": 3.81,
      "grad_norm": 0.659959614276886,
      "learning_rate": 4.792576419213974e-06,
      "loss": 0.0277,
      "step": 8725
    },
    {
      "epoch": 3.82,
      "grad_norm": 1.0919593572616577,
      "learning_rate": 4.519650655021834e-06,
      "loss": 0.0306,
      "step": 8750
    },
    {
      "epoch": 3.83,
      "grad_norm": 0.9325423836708069,
      "learning_rate": 4.246724890829695e-06,
      "loss": 0.0331,
      "step": 8775
    },
    {
      "epoch": 3.84,
      "grad_norm": 0.21544720232486725,
      "learning_rate": 3.9737991266375545e-06,
      "loss": 0.0215,
      "step": 8800
    },
    {
      "epoch": 3.84,
      "eval_loss": 0.293580025434494,
      "eval_na_accuracy": 0.7857142686843872,
      "eval_ordinal_accuracy": 0.7127997875213623,
      "eval_ordinal_mae": 0.3201707899570465,
      "eval_runtime": 157.8828,
      "eval_samples_per_second": 25.202,
      "eval_steps_per_second": 3.154,
      "step": 8800
    },
    {
      "epoch": 3.85,
      "grad_norm": 0.43472805619239807,
      "learning_rate": 3.7008733624454154e-06,
      "loss": 0.0345,
      "step": 8825
    },
    {
      "epoch": 3.86,
      "grad_norm": 0.18415190279483795,
      "learning_rate": 3.4279475982532755e-06,
      "loss": 0.0268,
      "step": 8850
    },
    {
      "epoch": 3.88,
      "grad_norm": 0.8721256852149963,
      "learning_rate": 3.155021834061136e-06,
      "loss": 0.0258,
      "step": 8875
    },
    {
      "epoch": 3.89,
      "grad_norm": 0.1487792730331421,
      "learning_rate": 2.8820960698689956e-06,
      "loss": 0.0312,
      "step": 8900
    },
    {
      "epoch": 3.89,
      "eval_loss": 0.295600026845932,
      "eval_na_accuracy": 0.7857142686843872,
      "eval_ordinal_accuracy": 0.7133776545524597,
      "eval_ordinal_mae": 0.32104432582855225,
      "eval_runtime": 158.1112,
      "eval_samples_per_second": 25.166,
      "eval_steps_per_second": 3.15,
      "step": 8900
    },
    {
      "epoch": 3.9,
      "grad_norm": 0.6491901278495789,
      "learning_rate": 2.609170305676856e-06,
      "loss": 0.0526,
      "step": 8925
    },
    {
      "epoch": 3.91,
      "grad_norm": 0.40534159541130066,
      "learning_rate": 2.336244541484716e-06,
      "loss": 0.0227,
      "step": 8950
    },
    {
      "epoch": 3.92,
      "grad_norm": 0.06649890542030334,
      "learning_rate": 2.0633187772925767e-06,
      "loss": 0.0157,
      "step": 8975
    },
    {
      "epoch": 3.93,
      "grad_norm": 0.2696411609649658,
      "learning_rate": 1.7903930131004367e-06,
      "loss": 0.0189,
      "step": 9000
    },
    {
      "epoch": 3.93,
      "eval_loss": 0.2945779263973236,
      "eval_na_accuracy": 0.7876448035240173,
      "eval_ordinal_accuracy": 0.7125108242034912,
      "eval_ordinal_mae": 0.3210395574569702,
      "eval_runtime": 157.9016,
      "eval_samples_per_second": 25.199,
      "eval_steps_per_second": 3.154,
      "step": 9000
    },
    {
      "epoch": 3.94,
      "grad_norm": 0.16493447124958038,
      "learning_rate": 1.517467248908297e-06,
      "loss": 0.0204,
      "step": 9025
    },
    {
      "epoch": 3.95,
      "grad_norm": 0.5132766366004944,
      "learning_rate": 1.2445414847161573e-06,
      "loss": 0.0182,
      "step": 9050
    },
    {
      "epoch": 3.96,
      "grad_norm": 0.06899993121623993,
      "learning_rate": 9.716157205240176e-07,
      "loss": 0.0194,
      "step": 9075
    },
    {
      "epoch": 3.97,
      "grad_norm": 0.09663155674934387,
      "learning_rate": 6.986899563318777e-07,
      "loss": 0.021,
      "step": 9100
    },
    {
      "epoch": 3.97,
      "eval_loss": 0.2948993146419525,
      "eval_na_accuracy": 0.7876448035240173,
      "eval_ordinal_accuracy": 0.7145333886146545,
      "eval_ordinal_mae": 0.31944769620895386,
      "eval_runtime": 157.4833,
      "eval_samples_per_second": 25.266,
      "eval_steps_per_second": 3.162,
      "step": 9100
    },
    {
      "epoch": 3.98,
      "grad_norm": 0.4333021342754364,
      "learning_rate": 4.2576419213973797e-07,
      "loss": 0.0262,
      "step": 9125
    },
    {
      "epoch": 4.0,
      "grad_norm": 0.39392733573913574,
      "learning_rate": 1.5283842794759825e-07,
      "loss": 0.0247,
      "step": 9150
    },
    {
      "epoch": 4.0,
      "step": 9160,
      "total_flos": 1.1353293455817277e+19,
      "train_loss": 0.11642740544208273,
      "train_runtime": 27830.4973,
      "train_samples_per_second": 5.264,
      "train_steps_per_second": 0.329
    }
  ],
  "logging_steps": 25,
  "max_steps": 9160,
  "num_input_tokens_seen": 0,
  "num_train_epochs": 4,
  "save_steps": 100,
  "total_flos": 1.1353293455817277e+19,
  "train_batch_size": 16,
  "trial_name": null,
  "trial_params": null
}