{
  "best_metric": 0.03933868557214737,
  "best_model_checkpoint": "./rorshark_outputs/checkpoint-1840",
  "epoch": 5.0,
  "eval_steps": 500,
  "global_step": 1840,
  "is_hyper_param_search": false,
  "is_local_process_zero": true,
  "is_world_process_zero": true,
  "log_history": [
    {
      "epoch": 0.03,
      "learning_rate": 1.989130434782609e-05,
      "loss": 0.5675,
      "step": 10
    },
    {
      "epoch": 0.05,
      "learning_rate": 1.9782608695652176e-05,
      "loss": 0.4112,
      "step": 20
    },
    {
      "epoch": 0.08,
      "learning_rate": 1.9673913043478263e-05,
      "loss": 0.3002,
      "step": 30
    },
    {
      "epoch": 0.11,
      "learning_rate": 1.956521739130435e-05,
      "loss": 0.3774,
      "step": 40
    },
    {
      "epoch": 0.14,
      "learning_rate": 1.9456521739130436e-05,
      "loss": 0.3295,
      "step": 50
    },
    {
      "epoch": 0.16,
      "learning_rate": 1.9347826086956523e-05,
      "loss": 0.3067,
      "step": 60
    },
    {
      "epoch": 0.19,
      "learning_rate": 1.923913043478261e-05,
      "loss": 0.2457,
      "step": 70
    },
    {
      "epoch": 0.22,
      "learning_rate": 1.9130434782608697e-05,
      "loss": 0.3318,
      "step": 80
    },
    {
      "epoch": 0.24,
      "learning_rate": 1.9021739130434784e-05,
      "loss": 0.1932,
      "step": 90
    },
    {
      "epoch": 0.27,
      "learning_rate": 1.891304347826087e-05,
      "loss": 0.1948,
      "step": 100
    },
    {
      "epoch": 0.3,
      "learning_rate": 1.8804347826086958e-05,
      "loss": 0.2475,
      "step": 110
    },
    {
      "epoch": 0.33,
      "learning_rate": 1.8695652173913045e-05,
      "loss": 0.1432,
      "step": 120
    },
    {
      "epoch": 0.35,
      "learning_rate": 1.8586956521739132e-05,
      "loss": 0.2069,
      "step": 130
    },
    {
      "epoch": 0.38,
      "learning_rate": 1.847826086956522e-05,
      "loss": 0.1986,
      "step": 140
    },
    {
      "epoch": 0.41,
      "learning_rate": 1.8369565217391306e-05,
      "loss": 0.2156,
      "step": 150
    },
    {
      "epoch": 0.43,
      "learning_rate": 1.8260869565217393e-05,
      "loss": 0.1187,
      "step": 160
    },
    {
      "epoch": 0.46,
      "learning_rate": 1.815217391304348e-05,
      "loss": 0.1192,
      "step": 170
    },
    {
      "epoch": 0.49,
      "learning_rate": 1.8043478260869567e-05,
      "loss": 0.1748,
      "step": 180
    },
    {
      "epoch": 0.52,
      "learning_rate": 1.7934782608695654e-05,
      "loss": 0.0779,
      "step": 190
    },
    {
      "epoch": 0.54,
      "learning_rate": 1.782608695652174e-05,
      "loss": 0.1075,
      "step": 200
    },
    {
      "epoch": 0.57,
      "learning_rate": 1.7717391304347828e-05,
      "loss": 0.1298,
      "step": 210
    },
    {
      "epoch": 0.6,
      "learning_rate": 1.7608695652173915e-05,
      "loss": 0.0728,
      "step": 220
    },
    {
      "epoch": 0.62,
      "learning_rate": 1.7500000000000002e-05,
      "loss": 0.1189,
      "step": 230
    },
    {
      "epoch": 0.65,
      "learning_rate": 1.739130434782609e-05,
      "loss": 0.1102,
      "step": 240
    },
    {
      "epoch": 0.68,
      "learning_rate": 1.7282608695652176e-05,
      "loss": 0.1183,
      "step": 250
    },
    {
      "epoch": 0.71,
      "learning_rate": 1.7173913043478263e-05,
      "loss": 0.3006,
      "step": 260
    },
    {
      "epoch": 0.73,
      "learning_rate": 1.706521739130435e-05,
      "loss": 0.1408,
      "step": 270
    },
    {
      "epoch": 0.76,
      "learning_rate": 1.6956521739130437e-05,
      "loss": 0.141,
      "step": 280
    },
    {
      "epoch": 0.79,
      "learning_rate": 1.6847826086956524e-05,
      "loss": 0.1208,
      "step": 290
    },
    {
      "epoch": 0.82,
      "learning_rate": 1.673913043478261e-05,
      "loss": 0.1004,
      "step": 300
    },
    {
      "epoch": 0.84,
      "learning_rate": 1.6630434782608698e-05,
      "loss": 0.206,
      "step": 310
    },
    {
      "epoch": 0.87,
      "learning_rate": 1.6521739130434785e-05,
      "loss": 0.12,
      "step": 320
    },
    {
      "epoch": 0.9,
      "learning_rate": 1.641304347826087e-05,
      "loss": 0.0705,
      "step": 330
    },
    {
      "epoch": 0.92,
      "learning_rate": 1.630434782608696e-05,
      "loss": 0.1018,
      "step": 340
    },
    {
      "epoch": 0.95,
      "learning_rate": 1.6195652173913045e-05,
      "loss": 0.1501,
      "step": 350
    },
    {
      "epoch": 0.98,
      "learning_rate": 1.6086956521739132e-05,
      "loss": 0.0597,
      "step": 360
    },
    {
      "epoch": 1.0,
      "eval_accuracy": 0.9865125240847784,
      "eval_loss": 0.05456383526325226,
      "eval_runtime": 6.5116,
      "eval_samples_per_second": 79.704,
      "eval_steps_per_second": 9.982,
      "step": 368
    },
    {
      "epoch": 1.01,
      "learning_rate": 1.597826086956522e-05,
      "loss": 0.0878,
      "step": 370
    },
    {
      "epoch": 1.03,
      "learning_rate": 1.5869565217391306e-05,
      "loss": 0.1651,
      "step": 380
    },
    {
      "epoch": 1.06,
      "learning_rate": 1.576086956521739e-05,
      "loss": 0.0645,
      "step": 390
    },
    {
      "epoch": 1.09,
      "learning_rate": 1.565217391304348e-05,
      "loss": 0.1085,
      "step": 400
    },
    {
      "epoch": 1.11,
      "learning_rate": 1.5543478260869567e-05,
      "loss": 0.0967,
      "step": 410
    },
    {
      "epoch": 1.14,
      "learning_rate": 1.5434782608695654e-05,
      "loss": 0.1178,
      "step": 420
    },
    {
      "epoch": 1.17,
      "learning_rate": 1.532608695652174e-05,
      "loss": 0.0605,
      "step": 430
    },
    {
      "epoch": 1.2,
      "learning_rate": 1.5217391304347828e-05,
      "loss": 0.1394,
      "step": 440
    },
    {
      "epoch": 1.22,
      "learning_rate": 1.5108695652173915e-05,
      "loss": 0.1113,
      "step": 450
    },
    {
      "epoch": 1.25,
      "learning_rate": 1.5000000000000002e-05,
      "loss": 0.0225,
      "step": 460
    },
    {
      "epoch": 1.28,
      "learning_rate": 1.4891304347826087e-05,
      "loss": 0.1861,
      "step": 470
    },
    {
      "epoch": 1.3,
      "learning_rate": 1.4782608695652174e-05,
      "loss": 0.0879,
      "step": 480
    },
    {
      "epoch": 1.33,
      "learning_rate": 1.4673913043478263e-05,
      "loss": 0.094,
      "step": 490
    },
    {
      "epoch": 1.36,
      "learning_rate": 1.456521739130435e-05,
      "loss": 0.1837,
      "step": 500
    },
    {
      "epoch": 1.39,
      "learning_rate": 1.4456521739130435e-05,
      "loss": 0.057,
      "step": 510
    },
    {
      "epoch": 1.41,
      "learning_rate": 1.4347826086956522e-05,
      "loss": 0.0504,
      "step": 520
    },
    {
      "epoch": 1.44,
      "learning_rate": 1.423913043478261e-05,
      "loss": 0.03,
      "step": 530
    },
    {
      "epoch": 1.47,
      "learning_rate": 1.4130434782608698e-05,
      "loss": 0.0637,
      "step": 540
    },
    {
      "epoch": 1.49,
      "learning_rate": 1.4021739130434783e-05,
      "loss": 0.1572,
      "step": 550
    },
    {
      "epoch": 1.52,
      "learning_rate": 1.391304347826087e-05,
      "loss": 0.2074,
      "step": 560
    },
    {
      "epoch": 1.55,
      "learning_rate": 1.3804347826086957e-05,
      "loss": 0.1031,
      "step": 570
    },
    {
      "epoch": 1.58,
      "learning_rate": 1.3695652173913046e-05,
      "loss": 0.075,
      "step": 580
    },
    {
      "epoch": 1.6,
      "learning_rate": 1.3586956521739133e-05,
      "loss": 0.0854,
      "step": 590
    },
    {
      "epoch": 1.63,
      "learning_rate": 1.3478260869565218e-05,
      "loss": 0.0897,
      "step": 600
    },
    {
      "epoch": 1.66,
      "learning_rate": 1.3369565217391305e-05,
      "loss": 0.1017,
      "step": 610
    },
    {
      "epoch": 1.68,
      "learning_rate": 1.3260869565217392e-05,
      "loss": 0.132,
      "step": 620
    },
    {
      "epoch": 1.71,
      "learning_rate": 1.315217391304348e-05,
      "loss": 0.0471,
      "step": 630
    },
    {
      "epoch": 1.74,
      "learning_rate": 1.3043478260869566e-05,
      "loss": 0.0707,
      "step": 640
    },
    {
      "epoch": 1.77,
      "learning_rate": 1.2934782608695653e-05,
      "loss": 0.0506,
      "step": 650
    },
    {
      "epoch": 1.79,
      "learning_rate": 1.282608695652174e-05,
      "loss": 0.1308,
      "step": 660
    },
    {
      "epoch": 1.82,
      "learning_rate": 1.2717391304347828e-05,
      "loss": 0.1188,
      "step": 670
    },
    {
      "epoch": 1.85,
      "learning_rate": 1.2608695652173915e-05,
      "loss": 0.1021,
      "step": 680
    },
    {
      "epoch": 1.88,
      "learning_rate": 1.25e-05,
      "loss": 0.1199,
      "step": 690
    },
    {
      "epoch": 1.9,
      "learning_rate": 1.2391304347826088e-05,
      "loss": 0.1068,
      "step": 700
    },
    {
      "epoch": 1.93,
      "learning_rate": 1.2282608695652175e-05,
      "loss": 0.0535,
      "step": 710
    },
    {
      "epoch": 1.96,
      "learning_rate": 1.2173913043478263e-05,
      "loss": 0.0723,
      "step": 720
    },
    {
      "epoch": 1.98,
      "learning_rate": 1.2065217391304348e-05,
      "loss": 0.2009,
      "step": 730
    },
    {
      "epoch": 2.0,
      "eval_accuracy": 0.9865125240847784,
      "eval_loss": 0.05307452380657196,
      "eval_runtime": 6.4841,
      "eval_samples_per_second": 80.043,
      "eval_steps_per_second": 10.025,
      "step": 736
    },
    {
      "epoch": 2.01,
      "learning_rate": 1.1956521739130435e-05,
      "loss": 0.0156,
      "step": 740
    },
    {
      "epoch": 2.04,
      "learning_rate": 1.1847826086956522e-05,
      "loss": 0.169,
      "step": 750
    },
    {
      "epoch": 2.07,
      "learning_rate": 1.1739130434782611e-05,
      "loss": 0.0866,
      "step": 760
    },
    {
      "epoch": 2.09,
      "learning_rate": 1.1630434782608698e-05,
      "loss": 0.0973,
      "step": 770
    },
    {
      "epoch": 2.12,
      "learning_rate": 1.1521739130434783e-05,
      "loss": 0.0427,
      "step": 780
    },
    {
      "epoch": 2.15,
      "learning_rate": 1.141304347826087e-05,
      "loss": 0.1296,
      "step": 790
    },
    {
      "epoch": 2.17,
      "learning_rate": 1.1304347826086957e-05,
      "loss": 0.0265,
      "step": 800
    },
    {
      "epoch": 2.2,
      "learning_rate": 1.1195652173913046e-05,
      "loss": 0.1574,
      "step": 810
    },
    {
      "epoch": 2.23,
      "learning_rate": 1.1086956521739131e-05,
      "loss": 0.0655,
      "step": 820
    },
    {
      "epoch": 2.26,
      "learning_rate": 1.0978260869565218e-05,
      "loss": 0.0785,
      "step": 830
    },
    {
      "epoch": 2.28,
      "learning_rate": 1.0869565217391305e-05,
      "loss": 0.1273,
      "step": 840
    },
    {
      "epoch": 2.31,
      "learning_rate": 1.076086956521739e-05,
      "loss": 0.0374,
      "step": 850
    },
    {
      "epoch": 2.34,
      "learning_rate": 1.0652173913043479e-05,
      "loss": 0.2576,
      "step": 860
    },
    {
      "epoch": 2.36,
      "learning_rate": 1.0543478260869566e-05,
      "loss": 0.0417,
      "step": 870
    },
    {
      "epoch": 2.39,
      "learning_rate": 1.0434782608695653e-05,
      "loss": 0.115,
      "step": 880
    },
    {
      "epoch": 2.42,
      "learning_rate": 1.032608695652174e-05,
      "loss": 0.105,
      "step": 890
    },
    {
      "epoch": 2.45,
      "learning_rate": 1.0217391304347829e-05,
      "loss": 0.1704,
      "step": 900
    },
    {
      "epoch": 2.47,
      "learning_rate": 1.0108695652173914e-05,
      "loss": 0.0442,
      "step": 910
    },
    {
      "epoch": 2.5,
      "learning_rate": 1e-05,
      "loss": 0.079,
      "step": 920
    },
    {
      "epoch": 2.53,
      "learning_rate": 9.891304347826088e-06,
      "loss": 0.0214,
      "step": 930
    },
    {
      "epoch": 2.55,
      "learning_rate": 9.782608695652175e-06,
      "loss": 0.112,
      "step": 940
    },
    {
      "epoch": 2.58,
      "learning_rate": 9.673913043478262e-06,
      "loss": 0.0467,
      "step": 950
    },
    {
      "epoch": 2.61,
      "learning_rate": 9.565217391304349e-06,
      "loss": 0.0944,
      "step": 960
    },
    {
      "epoch": 2.64,
      "learning_rate": 9.456521739130436e-06,
      "loss": 0.0195,
      "step": 970
    },
    {
      "epoch": 2.66,
      "learning_rate": 9.347826086956523e-06,
      "loss": 0.1084,
      "step": 980
    },
    {
      "epoch": 2.69,
      "learning_rate": 9.23913043478261e-06,
      "loss": 0.0598,
      "step": 990
    },
    {
      "epoch": 2.72,
      "learning_rate": 9.130434782608697e-06,
      "loss": 0.0563,
      "step": 1000
    },
    {
      "epoch": 2.74,
      "learning_rate": 9.021739130434784e-06,
      "loss": 0.1212,
      "step": 1010
    },
    {
      "epoch": 2.77,
      "learning_rate": 8.91304347826087e-06,
      "loss": 0.103,
      "step": 1020
    },
    {
      "epoch": 2.8,
      "learning_rate": 8.804347826086957e-06,
      "loss": 0.0708,
      "step": 1030
    },
    {
      "epoch": 2.83,
      "learning_rate": 8.695652173913044e-06,
      "loss": 0.0639,
      "step": 1040
    },
    {
      "epoch": 2.85,
      "learning_rate": 8.586956521739131e-06,
      "loss": 0.0153,
      "step": 1050
    },
    {
      "epoch": 2.88,
      "learning_rate": 8.478260869565218e-06,
      "loss": 0.028,
      "step": 1060
    },
    {
      "epoch": 2.91,
      "learning_rate": 8.369565217391305e-06,
      "loss": 0.029,
      "step": 1070
    },
    {
      "epoch": 2.93,
      "learning_rate": 8.260869565217392e-06,
      "loss": 0.0915,
      "step": 1080
    },
    {
      "epoch": 2.96,
      "learning_rate": 8.15217391304348e-06,
      "loss": 0.0186,
      "step": 1090
    },
    {
      "epoch": 2.99,
      "learning_rate": 8.043478260869566e-06,
      "loss": 0.0114,
      "step": 1100
    },
    {
      "epoch": 3.0,
      "eval_accuracy": 0.9903660886319846,
      "eval_loss": 0.04182479530572891,
      "eval_runtime": 6.3668,
      "eval_samples_per_second": 81.517,
      "eval_steps_per_second": 10.209,
      "step": 1104
    },
    {
      "epoch": 3.02,
      "learning_rate": 7.934782608695653e-06,
      "loss": 0.2106,
      "step": 1110
    },
    {
      "epoch": 3.04,
      "learning_rate": 7.82608695652174e-06,
      "loss": 0.0515,
      "step": 1120
    },
    {
      "epoch": 3.07,
      "learning_rate": 7.717391304347827e-06,
      "loss": 0.0406,
      "step": 1130
    },
    {
      "epoch": 3.1,
      "learning_rate": 7.608695652173914e-06,
      "loss": 0.0355,
      "step": 1140
    },
    {
      "epoch": 3.12,
      "learning_rate": 7.500000000000001e-06,
      "loss": 0.1842,
      "step": 1150
    },
    {
      "epoch": 3.15,
      "learning_rate": 7.391304347826087e-06,
      "loss": 0.0545,
      "step": 1160
    },
    {
      "epoch": 3.18,
      "learning_rate": 7.282608695652175e-06,
      "loss": 0.1349,
      "step": 1170
    },
    {
      "epoch": 3.21,
      "learning_rate": 7.173913043478261e-06,
      "loss": 0.0104,
      "step": 1180
    },
    {
      "epoch": 3.23,
      "learning_rate": 7.065217391304349e-06,
      "loss": 0.1324,
      "step": 1190
    },
    {
      "epoch": 3.26,
      "learning_rate": 6.956521739130435e-06,
      "loss": 0.0934,
      "step": 1200
    },
    {
      "epoch": 3.29,
      "learning_rate": 6.847826086956523e-06,
      "loss": 0.0966,
      "step": 1210
    },
    {
      "epoch": 3.32,
      "learning_rate": 6.739130434782609e-06,
      "loss": 0.0588,
      "step": 1220
    },
    {
      "epoch": 3.34,
      "learning_rate": 6.630434782608696e-06,
      "loss": 0.0802,
      "step": 1230
    },
    {
      "epoch": 3.37,
      "learning_rate": 6.521739130434783e-06,
      "loss": 0.0576,
      "step": 1240
    },
    {
      "epoch": 3.4,
      "learning_rate": 6.41304347826087e-06,
      "loss": 0.0419,
      "step": 1250
    },
    {
      "epoch": 3.42,
      "learning_rate": 6.304347826086958e-06,
      "loss": 0.0481,
      "step": 1260
    },
    {
      "epoch": 3.45,
      "learning_rate": 6.195652173913044e-06,
      "loss": 0.0861,
      "step": 1270
    },
    {
      "epoch": 3.48,
      "learning_rate": 6.086956521739132e-06,
      "loss": 0.1023,
      "step": 1280
    },
    {
      "epoch": 3.51,
      "learning_rate": 5.978260869565218e-06,
      "loss": 0.0584,
      "step": 1290
    },
    {
      "epoch": 3.53,
      "learning_rate": 5.8695652173913055e-06,
      "loss": 0.1282,
      "step": 1300
    },
    {
      "epoch": 3.56,
      "learning_rate": 5.760869565217392e-06,
      "loss": 0.0277,
      "step": 1310
    },
    {
      "epoch": 3.59,
      "learning_rate": 5.652173913043479e-06,
      "loss": 0.1837,
      "step": 1320
    },
    {
      "epoch": 3.61,
      "learning_rate": 5.543478260869566e-06,
      "loss": 0.0264,
      "step": 1330
    },
    {
      "epoch": 3.64,
      "learning_rate": 5.4347826086956525e-06,
      "loss": 0.1224,
      "step": 1340
    },
    {
      "epoch": 3.67,
      "learning_rate": 5.3260869565217395e-06,
      "loss": 0.0434,
      "step": 1350
    },
    {
      "epoch": 3.7,
      "learning_rate": 5.2173913043478265e-06,
      "loss": 0.1337,
      "step": 1360
    },
    {
      "epoch": 3.72,
      "learning_rate": 5.108695652173914e-06,
      "loss": 0.0071,
      "step": 1370
    },
    {
      "epoch": 3.75,
      "learning_rate": 5e-06,
      "loss": 0.0568,
      "step": 1380
    },
    {
      "epoch": 3.78,
      "learning_rate": 4.891304347826087e-06,
      "loss": 0.043,
      "step": 1390
    },
    {
      "epoch": 3.8,
      "learning_rate": 4.782608695652174e-06,
      "loss": 0.0719,
      "step": 1400
    },
    {
      "epoch": 3.83,
      "learning_rate": 4.673913043478261e-06,
      "loss": 0.1128,
      "step": 1410
    },
    {
      "epoch": 3.86,
      "learning_rate": 4.565217391304348e-06,
      "loss": 0.0477,
      "step": 1420
    },
    {
      "epoch": 3.89,
      "learning_rate": 4.456521739130435e-06,
      "loss": 0.0791,
      "step": 1430
    },
    {
      "epoch": 3.91,
      "learning_rate": 4.347826086956522e-06,
      "loss": 0.037,
      "step": 1440
    },
    {
      "epoch": 3.94,
      "learning_rate": 4.239130434782609e-06,
      "loss": 0.1466,
      "step": 1450
    },
    {
      "epoch": 3.97,
      "learning_rate": 4.130434782608696e-06,
      "loss": 0.0467,
      "step": 1460
    },
    {
      "epoch": 3.99,
      "learning_rate": 4.021739130434783e-06,
      "loss": 0.0998,
      "step": 1470
    },
    {
      "epoch": 4.0,
      "eval_accuracy": 0.9903660886319846,
      "eval_loss": 0.04251210391521454,
      "eval_runtime": 6.5268,
      "eval_samples_per_second": 79.518,
      "eval_steps_per_second": 9.959,
      "step": 1472
    },
    {
      "epoch": 4.02,
      "learning_rate": 3.91304347826087e-06,
      "loss": 0.1285,
      "step": 1480
    },
    {
      "epoch": 4.05,
      "learning_rate": 3.804347826086957e-06,
      "loss": 0.1634,
      "step": 1490
    },
    {
      "epoch": 4.08,
      "learning_rate": 3.6956521739130436e-06,
      "loss": 0.0462,
      "step": 1500
    },
    {
      "epoch": 4.1,
      "learning_rate": 3.5869565217391305e-06,
      "loss": 0.0846,
      "step": 1510
    },
    {
      "epoch": 4.13,
      "learning_rate": 3.4782608695652175e-06,
      "loss": 0.1239,
      "step": 1520
    },
    {
      "epoch": 4.16,
      "learning_rate": 3.3695652173913045e-06,
      "loss": 0.1818,
      "step": 1530
    },
    {
      "epoch": 4.18,
      "learning_rate": 3.2608695652173914e-06,
      "loss": 0.021,
      "step": 1540
    },
    {
      "epoch": 4.21,
      "learning_rate": 3.152173913043479e-06,
      "loss": 0.0741,
      "step": 1550
    },
    {
      "epoch": 4.24,
      "learning_rate": 3.043478260869566e-06,
      "loss": 0.182,
      "step": 1560
    },
    {
      "epoch": 4.27,
      "learning_rate": 2.9347826086956528e-06,
      "loss": 0.0433,
      "step": 1570
    },
    {
      "epoch": 4.29,
      "learning_rate": 2.8260869565217393e-06,
      "loss": 0.0437,
      "step": 1580
    },
    {
      "epoch": 4.32,
      "learning_rate": 2.7173913043478263e-06,
      "loss": 0.0382,
      "step": 1590
    },
    {
      "epoch": 4.35,
      "learning_rate": 2.6086956521739132e-06,
      "loss": 0.046,
      "step": 1600
    },
    {
      "epoch": 4.38,
      "learning_rate": 2.5e-06,
      "loss": 0.0213,
      "step": 1610
    },
    {
      "epoch": 4.4,
      "learning_rate": 2.391304347826087e-06,
      "loss": 0.0186,
      "step": 1620
    },
    {
      "epoch": 4.43,
      "learning_rate": 2.282608695652174e-06,
      "loss": 0.0671,
      "step": 1630
    },
    {
      "epoch": 4.46,
      "learning_rate": 2.173913043478261e-06,
      "loss": 0.0908,
      "step": 1640
    },
    {
      "epoch": 4.48,
      "learning_rate": 2.065217391304348e-06,
      "loss": 0.0697,
      "step": 1650
    },
    {
      "epoch": 4.51,
      "learning_rate": 1.956521739130435e-06,
      "loss": 0.0637,
      "step": 1660
    },
    {
      "epoch": 4.54,
      "learning_rate": 1.8478260869565218e-06,
      "loss": 0.0819,
      "step": 1670
    },
    {
      "epoch": 4.57,
      "learning_rate": 1.7391304347826088e-06,
      "loss": 0.0623,
      "step": 1680
    },
    {
      "epoch": 4.59,
      "learning_rate": 1.6304347826086957e-06,
      "loss": 0.0114,
      "step": 1690
    },
    {
      "epoch": 4.62,
      "learning_rate": 1.521739130434783e-06,
      "loss": 0.0342,
      "step": 1700
    },
    {
      "epoch": 4.65,
      "learning_rate": 1.4130434782608697e-06,
      "loss": 0.0859,
      "step": 1710
    },
    {
      "epoch": 4.67,
      "learning_rate": 1.3043478260869566e-06,
      "loss": 0.0462,
      "step": 1720
    },
    {
      "epoch": 4.7,
      "learning_rate": 1.1956521739130436e-06,
      "loss": 0.1022,
      "step": 1730
    },
    {
      "epoch": 4.73,
      "learning_rate": 1.0869565217391306e-06,
      "loss": 0.0571,
      "step": 1740
    },
    {
      "epoch": 4.76,
      "learning_rate": 9.782608695652175e-07,
      "loss": 0.0108,
      "step": 1750
    },
    {
      "epoch": 4.78,
      "learning_rate": 8.695652173913044e-07,
      "loss": 0.0893,
      "step": 1760
    },
    {
      "epoch": 4.81,
      "learning_rate": 7.608695652173914e-07,
      "loss": 0.0214,
      "step": 1770
    },
    {
      "epoch": 4.84,
      "learning_rate": 6.521739130434783e-07,
      "loss": 0.0416,
      "step": 1780
    },
    {
      "epoch": 4.86,
      "learning_rate": 5.434782608695653e-07,
      "loss": 0.1022,
      "step": 1790
    },
    {
      "epoch": 4.89,
      "learning_rate": 4.347826086956522e-07,
      "loss": 0.0628,
      "step": 1800
    },
    {
      "epoch": 4.92,
      "learning_rate": 3.2608695652173915e-07,
      "loss": 0.0691,
      "step": 1810
    },
    {
      "epoch": 4.95,
      "learning_rate": 2.173913043478261e-07,
      "loss": 0.0371,
      "step": 1820
    },
    {
      "epoch": 4.97,
      "learning_rate": 1.0869565217391305e-07,
      "loss": 0.0714,
      "step": 1830
    },
    {
      "epoch": 5.0,
      "learning_rate": 0.0,
      "loss": 0.1244,
      "step": 1840
    },
    {
      "epoch": 5.0,
      "eval_accuracy": 0.9922928709055877,
      "eval_loss": 0.03933868557214737,
      "eval_runtime": 6.3674,
      "eval_samples_per_second": 81.509,
      "eval_steps_per_second": 10.208,
      "step": 1840
    },
    {
      "epoch": 5.0,
      "step": 1840,
      "total_flos": 1.1387447873864294e+18,
      "train_loss": 0.10440107471431079,
      "train_runtime": 430.0921,
      "train_samples_per_second": 34.167,
      "train_steps_per_second": 4.278
    }
  ],
  "logging_steps": 10,
  "max_steps": 1840,
  "num_input_tokens_seen": 0,
  "num_train_epochs": 5,
  "save_steps": 500,
  "total_flos": 1.1387447873864294e+18,
  "trial_name": null,
  "trial_params": null
}