CrackDetectionLowRes / trainer_state.json
Taki3d's picture
End of training
4c2ab7e
raw
history blame
59.5 kB
{
"best_metric": 0.018311450257897377,
"best_model_checkpoint": "./CrackDetectionLowRes/checkpoint-4760",
"epoch": 5.0,
"global_step": 4760,
"is_hyper_param_search": false,
"is_local_process_zero": true,
"is_world_process_zero": true,
"log_history": [
{
"epoch": 0.01,
"learning_rate": 1.995967741935484e-05,
"loss": 0.6268,
"step": 10
},
{
"epoch": 0.02,
"learning_rate": 1.9919354838709678e-05,
"loss": 0.4477,
"step": 20
},
{
"epoch": 0.03,
"learning_rate": 1.9879032258064516e-05,
"loss": 0.2904,
"step": 30
},
{
"epoch": 0.04,
"learning_rate": 1.9838709677419358e-05,
"loss": 0.3415,
"step": 40
},
{
"epoch": 0.05,
"learning_rate": 1.9798387096774193e-05,
"loss": 0.2829,
"step": 50
},
{
"epoch": 0.06,
"learning_rate": 1.9758064516129035e-05,
"loss": 0.2098,
"step": 60
},
{
"epoch": 0.07,
"learning_rate": 1.9717741935483873e-05,
"loss": 0.2949,
"step": 70
},
{
"epoch": 0.08,
"learning_rate": 1.967741935483871e-05,
"loss": 0.215,
"step": 80
},
{
"epoch": 0.09,
"learning_rate": 1.963709677419355e-05,
"loss": 0.2315,
"step": 90
},
{
"epoch": 0.1,
"learning_rate": 1.9596774193548388e-05,
"loss": 0.1707,
"step": 100
},
{
"epoch": 0.11,
"learning_rate": 1.9556451612903226e-05,
"loss": 0.1596,
"step": 110
},
{
"epoch": 0.12,
"learning_rate": 1.9516129032258068e-05,
"loss": 0.1451,
"step": 120
},
{
"epoch": 0.13,
"learning_rate": 1.9475806451612903e-05,
"loss": 0.1504,
"step": 130
},
{
"epoch": 0.14,
"learning_rate": 1.9435483870967744e-05,
"loss": 0.1132,
"step": 140
},
{
"epoch": 0.15,
"learning_rate": 1.9395161290322583e-05,
"loss": 0.1199,
"step": 150
},
{
"epoch": 0.16,
"learning_rate": 1.935483870967742e-05,
"loss": 0.0533,
"step": 160
},
{
"epoch": 0.17,
"learning_rate": 1.931451612903226e-05,
"loss": 0.2408,
"step": 170
},
{
"epoch": 0.18,
"learning_rate": 1.9274193548387097e-05,
"loss": 0.1581,
"step": 180
},
{
"epoch": 0.19,
"learning_rate": 1.9233870967741936e-05,
"loss": 0.0945,
"step": 190
},
{
"epoch": 0.2,
"learning_rate": 1.9193548387096777e-05,
"loss": 0.1388,
"step": 200
},
{
"epoch": 0.21,
"learning_rate": 1.9153225806451616e-05,
"loss": 0.1212,
"step": 210
},
{
"epoch": 0.22,
"learning_rate": 1.9112903225806454e-05,
"loss": 0.0485,
"step": 220
},
{
"epoch": 0.23,
"learning_rate": 1.9072580645161292e-05,
"loss": 0.091,
"step": 230
},
{
"epoch": 0.24,
"learning_rate": 1.903225806451613e-05,
"loss": 0.2035,
"step": 240
},
{
"epoch": 0.25,
"learning_rate": 1.899193548387097e-05,
"loss": 0.201,
"step": 250
},
{
"epoch": 0.26,
"learning_rate": 1.895161290322581e-05,
"loss": 0.1092,
"step": 260
},
{
"epoch": 0.27,
"learning_rate": 1.8911290322580645e-05,
"loss": 0.1697,
"step": 270
},
{
"epoch": 0.28,
"learning_rate": 1.8870967741935487e-05,
"loss": 0.0249,
"step": 280
},
{
"epoch": 0.29,
"learning_rate": 1.8830645161290325e-05,
"loss": 0.1491,
"step": 290
},
{
"epoch": 0.3,
"learning_rate": 1.8790322580645163e-05,
"loss": 0.0702,
"step": 300
},
{
"epoch": 0.31,
"learning_rate": 1.8750000000000002e-05,
"loss": 0.1092,
"step": 310
},
{
"epoch": 0.32,
"learning_rate": 1.870967741935484e-05,
"loss": 0.1212,
"step": 320
},
{
"epoch": 0.33,
"learning_rate": 1.8669354838709678e-05,
"loss": 0.0811,
"step": 330
},
{
"epoch": 0.34,
"learning_rate": 1.8629032258064517e-05,
"loss": 0.0601,
"step": 340
},
{
"epoch": 0.35,
"learning_rate": 1.8588709677419355e-05,
"loss": 0.0169,
"step": 350
},
{
"epoch": 0.36,
"learning_rate": 1.8548387096774193e-05,
"loss": 0.0909,
"step": 360
},
{
"epoch": 0.37,
"learning_rate": 1.8508064516129035e-05,
"loss": 0.1427,
"step": 370
},
{
"epoch": 0.38,
"learning_rate": 1.8467741935483873e-05,
"loss": 0.0313,
"step": 380
},
{
"epoch": 0.39,
"learning_rate": 1.842741935483871e-05,
"loss": 0.1578,
"step": 390
},
{
"epoch": 0.4,
"learning_rate": 1.838709677419355e-05,
"loss": 0.0671,
"step": 400
},
{
"epoch": 0.41,
"learning_rate": 1.8346774193548388e-05,
"loss": 0.0415,
"step": 410
},
{
"epoch": 0.42,
"learning_rate": 1.8306451612903226e-05,
"loss": 0.0236,
"step": 420
},
{
"epoch": 0.43,
"learning_rate": 1.8266129032258068e-05,
"loss": 0.1072,
"step": 430
},
{
"epoch": 0.44,
"learning_rate": 1.8225806451612903e-05,
"loss": 0.1273,
"step": 440
},
{
"epoch": 0.45,
"learning_rate": 1.8185483870967744e-05,
"loss": 0.0715,
"step": 450
},
{
"epoch": 0.46,
"learning_rate": 1.8145161290322583e-05,
"loss": 0.0228,
"step": 460
},
{
"epoch": 0.47,
"learning_rate": 1.810483870967742e-05,
"loss": 0.0112,
"step": 470
},
{
"epoch": 0.48,
"learning_rate": 1.806451612903226e-05,
"loss": 0.1342,
"step": 480
},
{
"epoch": 0.49,
"learning_rate": 1.8024193548387097e-05,
"loss": 0.0673,
"step": 490
},
{
"epoch": 0.5,
"learning_rate": 1.7983870967741936e-05,
"loss": 0.037,
"step": 500
},
{
"epoch": 0.51,
"learning_rate": 1.7943548387096777e-05,
"loss": 0.0879,
"step": 510
},
{
"epoch": 0.52,
"learning_rate": 1.7903225806451612e-05,
"loss": 0.0344,
"step": 520
},
{
"epoch": 0.53,
"learning_rate": 1.7862903225806454e-05,
"loss": 0.064,
"step": 530
},
{
"epoch": 0.54,
"learning_rate": 1.7822580645161292e-05,
"loss": 0.012,
"step": 540
},
{
"epoch": 0.55,
"learning_rate": 1.778225806451613e-05,
"loss": 0.1869,
"step": 550
},
{
"epoch": 0.56,
"learning_rate": 1.774193548387097e-05,
"loss": 0.1631,
"step": 560
},
{
"epoch": 0.57,
"learning_rate": 1.770161290322581e-05,
"loss": 0.1175,
"step": 570
},
{
"epoch": 0.58,
"learning_rate": 1.7661290322580645e-05,
"loss": 0.0674,
"step": 580
},
{
"epoch": 0.59,
"learning_rate": 1.7620967741935487e-05,
"loss": 0.0543,
"step": 590
},
{
"epoch": 0.6,
"learning_rate": 1.7580645161290325e-05,
"loss": 0.077,
"step": 600
},
{
"epoch": 0.61,
"learning_rate": 1.7540322580645164e-05,
"loss": 0.1112,
"step": 610
},
{
"epoch": 0.62,
"learning_rate": 1.7500000000000002e-05,
"loss": 0.075,
"step": 620
},
{
"epoch": 0.64,
"learning_rate": 1.745967741935484e-05,
"loss": 0.0393,
"step": 630
},
{
"epoch": 0.65,
"learning_rate": 1.741935483870968e-05,
"loss": 0.0329,
"step": 640
},
{
"epoch": 0.66,
"learning_rate": 1.7379032258064517e-05,
"loss": 0.0841,
"step": 650
},
{
"epoch": 0.67,
"learning_rate": 1.7338709677419355e-05,
"loss": 0.0531,
"step": 660
},
{
"epoch": 0.68,
"learning_rate": 1.7298387096774193e-05,
"loss": 0.0647,
"step": 670
},
{
"epoch": 0.69,
"learning_rate": 1.7258064516129035e-05,
"loss": 0.0142,
"step": 680
},
{
"epoch": 0.7,
"learning_rate": 1.7217741935483873e-05,
"loss": 0.1275,
"step": 690
},
{
"epoch": 0.71,
"learning_rate": 1.717741935483871e-05,
"loss": 0.0745,
"step": 700
},
{
"epoch": 0.72,
"learning_rate": 1.713709677419355e-05,
"loss": 0.0159,
"step": 710
},
{
"epoch": 0.73,
"learning_rate": 1.7096774193548388e-05,
"loss": 0.1443,
"step": 720
},
{
"epoch": 0.74,
"learning_rate": 1.7056451612903226e-05,
"loss": 0.2165,
"step": 730
},
{
"epoch": 0.75,
"learning_rate": 1.7016129032258068e-05,
"loss": 0.0651,
"step": 740
},
{
"epoch": 0.76,
"learning_rate": 1.6975806451612903e-05,
"loss": 0.0842,
"step": 750
},
{
"epoch": 0.77,
"learning_rate": 1.6935483870967744e-05,
"loss": 0.0606,
"step": 760
},
{
"epoch": 0.78,
"learning_rate": 1.6895161290322583e-05,
"loss": 0.0346,
"step": 770
},
{
"epoch": 0.79,
"learning_rate": 1.685483870967742e-05,
"loss": 0.0426,
"step": 780
},
{
"epoch": 0.8,
"learning_rate": 1.681451612903226e-05,
"loss": 0.052,
"step": 790
},
{
"epoch": 0.81,
"learning_rate": 1.6774193548387098e-05,
"loss": 0.0294,
"step": 800
},
{
"epoch": 0.82,
"learning_rate": 1.6733870967741936e-05,
"loss": 0.1751,
"step": 810
},
{
"epoch": 0.83,
"learning_rate": 1.6693548387096778e-05,
"loss": 0.0452,
"step": 820
},
{
"epoch": 0.84,
"learning_rate": 1.6653225806451612e-05,
"loss": 0.1524,
"step": 830
},
{
"epoch": 0.85,
"learning_rate": 1.6612903225806454e-05,
"loss": 0.0378,
"step": 840
},
{
"epoch": 0.86,
"learning_rate": 1.6572580645161292e-05,
"loss": 0.1454,
"step": 850
},
{
"epoch": 0.87,
"learning_rate": 1.653225806451613e-05,
"loss": 0.0107,
"step": 860
},
{
"epoch": 0.88,
"learning_rate": 1.649193548387097e-05,
"loss": 0.0224,
"step": 870
},
{
"epoch": 0.89,
"learning_rate": 1.6451612903225807e-05,
"loss": 0.093,
"step": 880
},
{
"epoch": 0.9,
"learning_rate": 1.6411290322580645e-05,
"loss": 0.0118,
"step": 890
},
{
"epoch": 0.91,
"learning_rate": 1.6370967741935487e-05,
"loss": 0.0426,
"step": 900
},
{
"epoch": 0.92,
"learning_rate": 1.6330645161290325e-05,
"loss": 0.0524,
"step": 910
},
{
"epoch": 0.93,
"learning_rate": 1.6290322580645164e-05,
"loss": 0.0648,
"step": 920
},
{
"epoch": 0.94,
"learning_rate": 1.6250000000000002e-05,
"loss": 0.0221,
"step": 930
},
{
"epoch": 0.95,
"learning_rate": 1.620967741935484e-05,
"loss": 0.0318,
"step": 940
},
{
"epoch": 0.96,
"learning_rate": 1.616935483870968e-05,
"loss": 0.0485,
"step": 950
},
{
"epoch": 0.97,
"learning_rate": 1.6129032258064517e-05,
"loss": 0.2064,
"step": 960
},
{
"epoch": 0.98,
"learning_rate": 1.6088709677419355e-05,
"loss": 0.2422,
"step": 970
},
{
"epoch": 0.99,
"learning_rate": 1.6048387096774193e-05,
"loss": 0.0453,
"step": 980
},
{
"epoch": 1.0,
"learning_rate": 1.6008064516129035e-05,
"loss": 0.0126,
"step": 990
},
{
"epoch": 1.0,
"eval_accuracy": 0.9878658101356174,
"eval_loss": 0.03435095399618149,
"eval_runtime": 693.6855,
"eval_samples_per_second": 2.02,
"eval_steps_per_second": 0.254,
"step": 992
},
{
"epoch": 1.05,
"learning_rate": 1.5798319327731094e-05,
"loss": 0.0081,
"step": 1000
},
{
"epoch": 1.06,
"learning_rate": 1.5756302521008403e-05,
"loss": 0.1512,
"step": 1010
},
{
"epoch": 1.07,
"learning_rate": 1.5714285714285715e-05,
"loss": 0.1389,
"step": 1020
},
{
"epoch": 1.08,
"learning_rate": 1.5672268907563028e-05,
"loss": 0.0544,
"step": 1030
},
{
"epoch": 1.09,
"learning_rate": 1.5630252100840337e-05,
"loss": 0.0738,
"step": 1040
},
{
"epoch": 1.1,
"learning_rate": 1.558823529411765e-05,
"loss": 0.0664,
"step": 1050
},
{
"epoch": 1.11,
"learning_rate": 1.5546218487394958e-05,
"loss": 0.075,
"step": 1060
},
{
"epoch": 1.12,
"learning_rate": 1.550420168067227e-05,
"loss": 0.0077,
"step": 1070
},
{
"epoch": 1.13,
"learning_rate": 1.5462184873949583e-05,
"loss": 0.1675,
"step": 1080
},
{
"epoch": 1.14,
"learning_rate": 1.5420168067226892e-05,
"loss": 0.1071,
"step": 1090
},
{
"epoch": 1.16,
"learning_rate": 1.5378151260504204e-05,
"loss": 0.071,
"step": 1100
},
{
"epoch": 1.17,
"learning_rate": 1.5336134453781513e-05,
"loss": 0.0273,
"step": 1110
},
{
"epoch": 1.18,
"learning_rate": 1.5294117647058822e-05,
"loss": 0.0076,
"step": 1120
},
{
"epoch": 1.19,
"learning_rate": 1.5252100840336135e-05,
"loss": 0.0329,
"step": 1130
},
{
"epoch": 1.2,
"learning_rate": 1.5210084033613445e-05,
"loss": 0.0737,
"step": 1140
},
{
"epoch": 1.21,
"learning_rate": 1.5168067226890758e-05,
"loss": 0.0247,
"step": 1150
},
{
"epoch": 1.22,
"learning_rate": 1.5126050420168068e-05,
"loss": 0.0819,
"step": 1160
},
{
"epoch": 1.23,
"learning_rate": 1.5084033613445379e-05,
"loss": 0.0652,
"step": 1170
},
{
"epoch": 1.24,
"learning_rate": 1.504201680672269e-05,
"loss": 0.0286,
"step": 1180
},
{
"epoch": 1.25,
"learning_rate": 1.5000000000000002e-05,
"loss": 0.1009,
"step": 1190
},
{
"epoch": 1.26,
"learning_rate": 1.4957983193277313e-05,
"loss": 0.0141,
"step": 1200
},
{
"epoch": 1.27,
"learning_rate": 1.4915966386554623e-05,
"loss": 0.0301,
"step": 1210
},
{
"epoch": 1.28,
"learning_rate": 1.4873949579831934e-05,
"loss": 0.0066,
"step": 1220
},
{
"epoch": 1.29,
"learning_rate": 1.4831932773109245e-05,
"loss": 0.0669,
"step": 1230
},
{
"epoch": 1.3,
"learning_rate": 1.4789915966386557e-05,
"loss": 0.021,
"step": 1240
},
{
"epoch": 1.31,
"learning_rate": 1.4747899159663868e-05,
"loss": 0.1255,
"step": 1250
},
{
"epoch": 1.32,
"learning_rate": 1.4705882352941179e-05,
"loss": 0.1764,
"step": 1260
},
{
"epoch": 1.33,
"learning_rate": 1.4663865546218488e-05,
"loss": 0.137,
"step": 1270
},
{
"epoch": 1.34,
"learning_rate": 1.4621848739495798e-05,
"loss": 0.0158,
"step": 1280
},
{
"epoch": 1.36,
"learning_rate": 1.4579831932773109e-05,
"loss": 0.0091,
"step": 1290
},
{
"epoch": 1.37,
"learning_rate": 1.4537815126050421e-05,
"loss": 0.1019,
"step": 1300
},
{
"epoch": 1.38,
"learning_rate": 1.4495798319327732e-05,
"loss": 0.1512,
"step": 1310
},
{
"epoch": 1.39,
"learning_rate": 1.4453781512605043e-05,
"loss": 0.1021,
"step": 1320
},
{
"epoch": 1.4,
"learning_rate": 1.4411764705882353e-05,
"loss": 0.0807,
"step": 1330
},
{
"epoch": 1.41,
"learning_rate": 1.4369747899159666e-05,
"loss": 0.0722,
"step": 1340
},
{
"epoch": 1.42,
"learning_rate": 1.4327731092436976e-05,
"loss": 0.0254,
"step": 1350
},
{
"epoch": 1.43,
"learning_rate": 1.4285714285714287e-05,
"loss": 0.0453,
"step": 1360
},
{
"epoch": 1.44,
"learning_rate": 1.4243697478991598e-05,
"loss": 0.035,
"step": 1370
},
{
"epoch": 1.45,
"learning_rate": 1.4201680672268908e-05,
"loss": 0.0787,
"step": 1380
},
{
"epoch": 1.46,
"learning_rate": 1.4159663865546221e-05,
"loss": 0.0783,
"step": 1390
},
{
"epoch": 1.47,
"learning_rate": 1.4117647058823532e-05,
"loss": 0.0498,
"step": 1400
},
{
"epoch": 1.48,
"learning_rate": 1.4075630252100842e-05,
"loss": 0.1108,
"step": 1410
},
{
"epoch": 1.49,
"learning_rate": 1.4033613445378151e-05,
"loss": 0.0292,
"step": 1420
},
{
"epoch": 1.5,
"learning_rate": 1.3991596638655462e-05,
"loss": 0.0845,
"step": 1430
},
{
"epoch": 1.51,
"learning_rate": 1.3949579831932774e-05,
"loss": 0.0348,
"step": 1440
},
{
"epoch": 1.52,
"learning_rate": 1.3907563025210085e-05,
"loss": 0.0147,
"step": 1450
},
{
"epoch": 1.53,
"learning_rate": 1.3865546218487396e-05,
"loss": 0.0705,
"step": 1460
},
{
"epoch": 1.54,
"learning_rate": 1.3823529411764706e-05,
"loss": 0.0076,
"step": 1470
},
{
"epoch": 1.55,
"learning_rate": 1.3781512605042017e-05,
"loss": 0.1588,
"step": 1480
},
{
"epoch": 1.57,
"learning_rate": 1.373949579831933e-05,
"loss": 0.0286,
"step": 1490
},
{
"epoch": 1.58,
"learning_rate": 1.369747899159664e-05,
"loss": 0.0058,
"step": 1500
},
{
"epoch": 1.59,
"learning_rate": 1.365546218487395e-05,
"loss": 0.0636,
"step": 1510
},
{
"epoch": 1.6,
"learning_rate": 1.3613445378151261e-05,
"loss": 0.03,
"step": 1520
},
{
"epoch": 1.61,
"learning_rate": 1.3571428571428574e-05,
"loss": 0.1051,
"step": 1530
},
{
"epoch": 1.62,
"learning_rate": 1.3529411764705885e-05,
"loss": 0.0054,
"step": 1540
},
{
"epoch": 1.63,
"learning_rate": 1.3487394957983195e-05,
"loss": 0.1695,
"step": 1550
},
{
"epoch": 1.64,
"learning_rate": 1.3445378151260506e-05,
"loss": 0.0063,
"step": 1560
},
{
"epoch": 1.65,
"learning_rate": 1.3403361344537815e-05,
"loss": 0.0724,
"step": 1570
},
{
"epoch": 1.66,
"learning_rate": 1.3361344537815126e-05,
"loss": 0.0053,
"step": 1580
},
{
"epoch": 1.67,
"learning_rate": 1.3319327731092438e-05,
"loss": 0.0074,
"step": 1590
},
{
"epoch": 1.68,
"learning_rate": 1.3277310924369749e-05,
"loss": 0.1233,
"step": 1600
},
{
"epoch": 1.69,
"learning_rate": 1.323529411764706e-05,
"loss": 0.1316,
"step": 1610
},
{
"epoch": 1.7,
"learning_rate": 1.319327731092437e-05,
"loss": 0.1144,
"step": 1620
},
{
"epoch": 1.71,
"learning_rate": 1.3151260504201682e-05,
"loss": 0.1086,
"step": 1630
},
{
"epoch": 1.72,
"learning_rate": 1.3109243697478993e-05,
"loss": 0.0699,
"step": 1640
},
{
"epoch": 1.73,
"learning_rate": 1.3067226890756304e-05,
"loss": 0.0796,
"step": 1650
},
{
"epoch": 1.74,
"learning_rate": 1.3025210084033614e-05,
"loss": 0.0893,
"step": 1660
},
{
"epoch": 1.75,
"learning_rate": 1.2983193277310925e-05,
"loss": 0.0555,
"step": 1670
},
{
"epoch": 1.76,
"learning_rate": 1.2941176470588238e-05,
"loss": 0.0469,
"step": 1680
},
{
"epoch": 1.78,
"learning_rate": 1.2899159663865548e-05,
"loss": 0.159,
"step": 1690
},
{
"epoch": 1.79,
"learning_rate": 1.2857142857142859e-05,
"loss": 0.0689,
"step": 1700
},
{
"epoch": 1.8,
"learning_rate": 1.281512605042017e-05,
"loss": 0.0428,
"step": 1710
},
{
"epoch": 1.81,
"learning_rate": 1.2773109243697479e-05,
"loss": 0.0741,
"step": 1720
},
{
"epoch": 1.82,
"learning_rate": 1.2731092436974791e-05,
"loss": 0.0675,
"step": 1730
},
{
"epoch": 1.83,
"learning_rate": 1.2689075630252102e-05,
"loss": 0.0087,
"step": 1740
},
{
"epoch": 1.84,
"learning_rate": 1.2647058823529412e-05,
"loss": 0.1111,
"step": 1750
},
{
"epoch": 1.85,
"learning_rate": 1.2605042016806723e-05,
"loss": 0.0188,
"step": 1760
},
{
"epoch": 1.86,
"learning_rate": 1.2563025210084034e-05,
"loss": 0.0179,
"step": 1770
},
{
"epoch": 1.87,
"learning_rate": 1.2521008403361346e-05,
"loss": 0.0448,
"step": 1780
},
{
"epoch": 1.88,
"learning_rate": 1.2478991596638657e-05,
"loss": 0.1377,
"step": 1790
},
{
"epoch": 1.89,
"learning_rate": 1.2436974789915967e-05,
"loss": 0.0297,
"step": 1800
},
{
"epoch": 1.9,
"learning_rate": 1.2394957983193278e-05,
"loss": 0.0721,
"step": 1810
},
{
"epoch": 1.91,
"learning_rate": 1.235294117647059e-05,
"loss": 0.0193,
"step": 1820
},
{
"epoch": 1.92,
"learning_rate": 1.2310924369747901e-05,
"loss": 0.0372,
"step": 1830
},
{
"epoch": 1.93,
"learning_rate": 1.2268907563025212e-05,
"loss": 0.0045,
"step": 1840
},
{
"epoch": 1.94,
"learning_rate": 1.2226890756302523e-05,
"loss": 0.0087,
"step": 1850
},
{
"epoch": 1.95,
"learning_rate": 1.2184873949579832e-05,
"loss": 0.0064,
"step": 1860
},
{
"epoch": 1.96,
"learning_rate": 1.2142857142857142e-05,
"loss": 0.0046,
"step": 1870
},
{
"epoch": 1.97,
"learning_rate": 1.2100840336134455e-05,
"loss": 0.0043,
"step": 1880
},
{
"epoch": 1.99,
"learning_rate": 1.2058823529411765e-05,
"loss": 0.1367,
"step": 1890
},
{
"epoch": 2.0,
"learning_rate": 1.2016806722689076e-05,
"loss": 0.0788,
"step": 1900
},
{
"epoch": 2.0,
"eval_accuracy": 0.9933035714285714,
"eval_loss": 0.021986495703458786,
"eval_runtime": 493.5244,
"eval_samples_per_second": 2.723,
"eval_steps_per_second": 0.34,
"step": 1904
},
{
"epoch": 2.01,
"learning_rate": 1.1974789915966387e-05,
"loss": 0.0948,
"step": 1910
},
{
"epoch": 2.02,
"learning_rate": 1.1932773109243699e-05,
"loss": 0.0061,
"step": 1920
},
{
"epoch": 2.03,
"learning_rate": 1.189075630252101e-05,
"loss": 0.116,
"step": 1930
},
{
"epoch": 2.04,
"learning_rate": 1.184873949579832e-05,
"loss": 0.0164,
"step": 1940
},
{
"epoch": 2.05,
"learning_rate": 1.1806722689075631e-05,
"loss": 0.0638,
"step": 1950
},
{
"epoch": 2.06,
"learning_rate": 1.1764705882352942e-05,
"loss": 0.0044,
"step": 1960
},
{
"epoch": 2.07,
"learning_rate": 1.1722689075630254e-05,
"loss": 0.0639,
"step": 1970
},
{
"epoch": 2.08,
"learning_rate": 1.1680672268907565e-05,
"loss": 0.1186,
"step": 1980
},
{
"epoch": 2.09,
"learning_rate": 1.1638655462184876e-05,
"loss": 0.0501,
"step": 1990
},
{
"epoch": 2.1,
"learning_rate": 1.1596638655462186e-05,
"loss": 0.0043,
"step": 2000
},
{
"epoch": 2.11,
"learning_rate": 1.1554621848739495e-05,
"loss": 0.0789,
"step": 2010
},
{
"epoch": 2.12,
"learning_rate": 1.1512605042016806e-05,
"loss": 0.0905,
"step": 2020
},
{
"epoch": 2.13,
"learning_rate": 1.1470588235294118e-05,
"loss": 0.0738,
"step": 2030
},
{
"epoch": 2.14,
"learning_rate": 1.1428571428571429e-05,
"loss": 0.0096,
"step": 2040
},
{
"epoch": 2.15,
"learning_rate": 1.138655462184874e-05,
"loss": 0.1941,
"step": 2050
},
{
"epoch": 2.16,
"learning_rate": 1.134453781512605e-05,
"loss": 0.0887,
"step": 2060
},
{
"epoch": 2.17,
"learning_rate": 1.1302521008403363e-05,
"loss": 0.1173,
"step": 2070
},
{
"epoch": 2.18,
"learning_rate": 1.1260504201680673e-05,
"loss": 0.0231,
"step": 2080
},
{
"epoch": 2.2,
"learning_rate": 1.1218487394957984e-05,
"loss": 0.0695,
"step": 2090
},
{
"epoch": 2.21,
"learning_rate": 1.1176470588235295e-05,
"loss": 0.0108,
"step": 2100
},
{
"epoch": 2.22,
"learning_rate": 1.1134453781512606e-05,
"loss": 0.0848,
"step": 2110
},
{
"epoch": 2.23,
"learning_rate": 1.1092436974789918e-05,
"loss": 0.0689,
"step": 2120
},
{
"epoch": 2.24,
"learning_rate": 1.1050420168067229e-05,
"loss": 0.0587,
"step": 2130
},
{
"epoch": 2.25,
"learning_rate": 1.100840336134454e-05,
"loss": 0.0542,
"step": 2140
},
{
"epoch": 2.26,
"learning_rate": 1.096638655462185e-05,
"loss": 0.0228,
"step": 2150
},
{
"epoch": 2.27,
"learning_rate": 1.0924369747899159e-05,
"loss": 0.0636,
"step": 2160
},
{
"epoch": 2.28,
"learning_rate": 1.0882352941176471e-05,
"loss": 0.0465,
"step": 2170
},
{
"epoch": 2.29,
"learning_rate": 1.0840336134453782e-05,
"loss": 0.0174,
"step": 2180
},
{
"epoch": 2.3,
"learning_rate": 1.0798319327731093e-05,
"loss": 0.064,
"step": 2190
},
{
"epoch": 2.31,
"learning_rate": 1.0756302521008403e-05,
"loss": 0.0166,
"step": 2200
},
{
"epoch": 2.32,
"learning_rate": 1.0714285714285714e-05,
"loss": 0.098,
"step": 2210
},
{
"epoch": 2.33,
"learning_rate": 1.0672268907563026e-05,
"loss": 0.007,
"step": 2220
},
{
"epoch": 2.34,
"learning_rate": 1.0630252100840337e-05,
"loss": 0.0089,
"step": 2230
},
{
"epoch": 2.35,
"learning_rate": 1.0588235294117648e-05,
"loss": 0.0159,
"step": 2240
},
{
"epoch": 2.36,
"learning_rate": 1.0546218487394959e-05,
"loss": 0.0467,
"step": 2250
},
{
"epoch": 2.37,
"learning_rate": 1.0504201680672271e-05,
"loss": 0.0114,
"step": 2260
},
{
"epoch": 2.38,
"learning_rate": 1.0462184873949582e-05,
"loss": 0.1498,
"step": 2270
},
{
"epoch": 2.39,
"learning_rate": 1.0420168067226892e-05,
"loss": 0.0412,
"step": 2280
},
{
"epoch": 2.41,
"learning_rate": 1.0378151260504203e-05,
"loss": 0.0264,
"step": 2290
},
{
"epoch": 2.42,
"learning_rate": 1.0336134453781514e-05,
"loss": 0.1393,
"step": 2300
},
{
"epoch": 2.43,
"learning_rate": 1.0294117647058823e-05,
"loss": 0.026,
"step": 2310
},
{
"epoch": 2.44,
"learning_rate": 1.0252100840336135e-05,
"loss": 0.1609,
"step": 2320
},
{
"epoch": 2.45,
"learning_rate": 1.0210084033613446e-05,
"loss": 0.0484,
"step": 2330
},
{
"epoch": 2.46,
"learning_rate": 1.0168067226890756e-05,
"loss": 0.0391,
"step": 2340
},
{
"epoch": 2.47,
"learning_rate": 1.0126050420168067e-05,
"loss": 0.0363,
"step": 2350
},
{
"epoch": 2.48,
"learning_rate": 1.008403361344538e-05,
"loss": 0.0495,
"step": 2360
},
{
"epoch": 2.49,
"learning_rate": 1.004201680672269e-05,
"loss": 0.0376,
"step": 2370
},
{
"epoch": 2.5,
"learning_rate": 1e-05,
"loss": 0.0347,
"step": 2380
},
{
"epoch": 2.51,
"learning_rate": 9.957983193277312e-06,
"loss": 0.0811,
"step": 2390
},
{
"epoch": 2.52,
"learning_rate": 9.915966386554622e-06,
"loss": 0.0226,
"step": 2400
},
{
"epoch": 2.53,
"learning_rate": 9.873949579831935e-06,
"loss": 0.0189,
"step": 2410
},
{
"epoch": 2.54,
"learning_rate": 9.831932773109244e-06,
"loss": 0.0102,
"step": 2420
},
{
"epoch": 2.55,
"learning_rate": 9.789915966386554e-06,
"loss": 0.1182,
"step": 2430
},
{
"epoch": 2.56,
"learning_rate": 9.747899159663867e-06,
"loss": 0.0222,
"step": 2440
},
{
"epoch": 2.57,
"learning_rate": 9.705882352941177e-06,
"loss": 0.0034,
"step": 2450
},
{
"epoch": 2.58,
"learning_rate": 9.663865546218488e-06,
"loss": 0.0887,
"step": 2460
},
{
"epoch": 2.59,
"learning_rate": 9.621848739495799e-06,
"loss": 0.141,
"step": 2470
},
{
"epoch": 2.61,
"learning_rate": 9.579831932773111e-06,
"loss": 0.0734,
"step": 2480
},
{
"epoch": 2.62,
"learning_rate": 9.537815126050422e-06,
"loss": 0.0044,
"step": 2490
},
{
"epoch": 2.63,
"learning_rate": 9.49579831932773e-06,
"loss": 0.0266,
"step": 2500
},
{
"epoch": 2.64,
"learning_rate": 9.453781512605043e-06,
"loss": 0.1665,
"step": 2510
},
{
"epoch": 2.65,
"learning_rate": 9.411764705882354e-06,
"loss": 0.0165,
"step": 2520
},
{
"epoch": 2.66,
"learning_rate": 9.369747899159664e-06,
"loss": 0.007,
"step": 2530
},
{
"epoch": 2.67,
"learning_rate": 9.327731092436975e-06,
"loss": 0.0042,
"step": 2540
},
{
"epoch": 2.68,
"learning_rate": 9.285714285714288e-06,
"loss": 0.1296,
"step": 2550
},
{
"epoch": 2.69,
"learning_rate": 9.243697478991598e-06,
"loss": 0.0035,
"step": 2560
},
{
"epoch": 2.7,
"learning_rate": 9.201680672268907e-06,
"loss": 0.0164,
"step": 2570
},
{
"epoch": 2.71,
"learning_rate": 9.15966386554622e-06,
"loss": 0.0077,
"step": 2580
},
{
"epoch": 2.72,
"learning_rate": 9.11764705882353e-06,
"loss": 0.0301,
"step": 2590
},
{
"epoch": 2.73,
"learning_rate": 9.075630252100841e-06,
"loss": 0.0841,
"step": 2600
},
{
"epoch": 2.74,
"learning_rate": 9.033613445378152e-06,
"loss": 0.1664,
"step": 2610
},
{
"epoch": 2.75,
"learning_rate": 8.991596638655462e-06,
"loss": 0.0046,
"step": 2620
},
{
"epoch": 2.76,
"learning_rate": 8.949579831932775e-06,
"loss": 0.0469,
"step": 2630
},
{
"epoch": 2.77,
"learning_rate": 8.907563025210085e-06,
"loss": 0.0463,
"step": 2640
},
{
"epoch": 2.78,
"learning_rate": 8.865546218487396e-06,
"loss": 0.0459,
"step": 2650
},
{
"epoch": 2.79,
"learning_rate": 8.823529411764707e-06,
"loss": 0.0565,
"step": 2660
},
{
"epoch": 2.8,
"learning_rate": 8.781512605042017e-06,
"loss": 0.0247,
"step": 2670
},
{
"epoch": 2.82,
"learning_rate": 8.739495798319328e-06,
"loss": 0.0074,
"step": 2680
},
{
"epoch": 2.83,
"learning_rate": 8.697478991596639e-06,
"loss": 0.0677,
"step": 2690
},
{
"epoch": 2.84,
"learning_rate": 8.655462184873951e-06,
"loss": 0.0345,
"step": 2700
},
{
"epoch": 2.85,
"learning_rate": 8.613445378151262e-06,
"loss": 0.0406,
"step": 2710
},
{
"epoch": 2.86,
"learning_rate": 8.571428571428571e-06,
"loss": 0.16,
"step": 2720
},
{
"epoch": 2.87,
"learning_rate": 8.529411764705883e-06,
"loss": 0.084,
"step": 2730
},
{
"epoch": 2.88,
"learning_rate": 8.487394957983194e-06,
"loss": 0.0075,
"step": 2740
},
{
"epoch": 2.89,
"learning_rate": 8.445378151260505e-06,
"loss": 0.0047,
"step": 2750
},
{
"epoch": 2.9,
"learning_rate": 8.403361344537815e-06,
"loss": 0.0893,
"step": 2760
},
{
"epoch": 2.91,
"learning_rate": 8.361344537815128e-06,
"loss": 0.0034,
"step": 2770
},
{
"epoch": 2.92,
"learning_rate": 8.319327731092438e-06,
"loss": 0.0137,
"step": 2780
},
{
"epoch": 2.93,
"learning_rate": 8.277310924369747e-06,
"loss": 0.1151,
"step": 2790
},
{
"epoch": 2.94,
"learning_rate": 8.23529411764706e-06,
"loss": 0.0848,
"step": 2800
},
{
"epoch": 2.95,
"learning_rate": 8.19327731092437e-06,
"loss": 0.1494,
"step": 2810
},
{
"epoch": 2.96,
"learning_rate": 8.151260504201681e-06,
"loss": 0.027,
"step": 2820
},
{
"epoch": 2.97,
"learning_rate": 8.109243697478992e-06,
"loss": 0.0517,
"step": 2830
},
{
"epoch": 2.98,
"learning_rate": 8.067226890756303e-06,
"loss": 0.0029,
"step": 2840
},
{
"epoch": 2.99,
"learning_rate": 8.025210084033615e-06,
"loss": 0.1336,
"step": 2850
},
{
"epoch": 3.0,
"eval_accuracy": 0.9933035714285714,
"eval_loss": 0.02223922871053219,
"eval_runtime": 494.1467,
"eval_samples_per_second": 2.72,
"eval_steps_per_second": 0.34,
"step": 2856
},
{
"epoch": 3.0,
"learning_rate": 7.983193277310926e-06,
"loss": 0.0377,
"step": 2860
},
{
"epoch": 3.01,
"learning_rate": 7.941176470588236e-06,
"loss": 0.0625,
"step": 2870
},
{
"epoch": 3.03,
"learning_rate": 7.899159663865547e-06,
"loss": 0.0395,
"step": 2880
},
{
"epoch": 3.04,
"learning_rate": 7.857142857142858e-06,
"loss": 0.0509,
"step": 2890
},
{
"epoch": 3.05,
"learning_rate": 7.815126050420168e-06,
"loss": 0.0188,
"step": 2900
},
{
"epoch": 3.06,
"learning_rate": 7.773109243697479e-06,
"loss": 0.0059,
"step": 2910
},
{
"epoch": 3.07,
"learning_rate": 7.731092436974791e-06,
"loss": 0.0099,
"step": 2920
},
{
"epoch": 3.08,
"learning_rate": 7.689075630252102e-06,
"loss": 0.0569,
"step": 2930
},
{
"epoch": 3.09,
"learning_rate": 7.647058823529411e-06,
"loss": 0.0386,
"step": 2940
},
{
"epoch": 3.1,
"learning_rate": 7.605042016806723e-06,
"loss": 0.0117,
"step": 2950
},
{
"epoch": 3.11,
"learning_rate": 7.563025210084034e-06,
"loss": 0.0802,
"step": 2960
},
{
"epoch": 3.12,
"learning_rate": 7.521008403361345e-06,
"loss": 0.038,
"step": 2970
},
{
"epoch": 3.13,
"learning_rate": 7.478991596638656e-06,
"loss": 0.0344,
"step": 2980
},
{
"epoch": 3.14,
"learning_rate": 7.436974789915967e-06,
"loss": 0.0122,
"step": 2990
},
{
"epoch": 3.15,
"learning_rate": 7.394957983193279e-06,
"loss": 0.0303,
"step": 3000
},
{
"epoch": 3.16,
"learning_rate": 7.352941176470589e-06,
"loss": 0.0057,
"step": 3010
},
{
"epoch": 3.17,
"learning_rate": 7.310924369747899e-06,
"loss": 0.0749,
"step": 3020
},
{
"epoch": 3.18,
"learning_rate": 7.268907563025211e-06,
"loss": 0.0517,
"step": 3030
},
{
"epoch": 3.19,
"learning_rate": 7.226890756302521e-06,
"loss": 0.0469,
"step": 3040
},
{
"epoch": 3.2,
"learning_rate": 7.184873949579833e-06,
"loss": 0.0844,
"step": 3050
},
{
"epoch": 3.21,
"learning_rate": 7.1428571428571436e-06,
"loss": 0.0038,
"step": 3060
},
{
"epoch": 3.22,
"learning_rate": 7.100840336134454e-06,
"loss": 0.0552,
"step": 3070
},
{
"epoch": 3.24,
"learning_rate": 7.058823529411766e-06,
"loss": 0.0029,
"step": 3080
},
{
"epoch": 3.25,
"learning_rate": 7.016806722689076e-06,
"loss": 0.0034,
"step": 3090
},
{
"epoch": 3.26,
"learning_rate": 6.974789915966387e-06,
"loss": 0.0106,
"step": 3100
},
{
"epoch": 3.27,
"learning_rate": 6.932773109243698e-06,
"loss": 0.0039,
"step": 3110
},
{
"epoch": 3.28,
"learning_rate": 6.8907563025210085e-06,
"loss": 0.1056,
"step": 3120
},
{
"epoch": 3.29,
"learning_rate": 6.84873949579832e-06,
"loss": 0.0024,
"step": 3130
},
{
"epoch": 3.3,
"learning_rate": 6.806722689075631e-06,
"loss": 0.072,
"step": 3140
},
{
"epoch": 3.31,
"learning_rate": 6.764705882352942e-06,
"loss": 0.0797,
"step": 3150
},
{
"epoch": 3.32,
"learning_rate": 6.722689075630253e-06,
"loss": 0.0058,
"step": 3160
},
{
"epoch": 3.33,
"learning_rate": 6.680672268907563e-06,
"loss": 0.0496,
"step": 3170
},
{
"epoch": 3.34,
"learning_rate": 6.638655462184874e-06,
"loss": 0.0135,
"step": 3180
},
{
"epoch": 3.35,
"learning_rate": 6.596638655462185e-06,
"loss": 0.0598,
"step": 3190
},
{
"epoch": 3.36,
"learning_rate": 6.5546218487394966e-06,
"loss": 0.0109,
"step": 3200
},
{
"epoch": 3.37,
"learning_rate": 6.512605042016807e-06,
"loss": 0.0427,
"step": 3210
},
{
"epoch": 3.38,
"learning_rate": 6.470588235294119e-06,
"loss": 0.0447,
"step": 3220
},
{
"epoch": 3.39,
"learning_rate": 6.4285714285714295e-06,
"loss": 0.0406,
"step": 3230
},
{
"epoch": 3.4,
"learning_rate": 6.386554621848739e-06,
"loss": 0.0454,
"step": 3240
},
{
"epoch": 3.41,
"learning_rate": 6.344537815126051e-06,
"loss": 0.0045,
"step": 3250
},
{
"epoch": 3.42,
"learning_rate": 6.3025210084033615e-06,
"loss": 0.0024,
"step": 3260
},
{
"epoch": 3.43,
"learning_rate": 6.260504201680673e-06,
"loss": 0.0551,
"step": 3270
},
{
"epoch": 3.45,
"learning_rate": 6.218487394957984e-06,
"loss": 0.0848,
"step": 3280
},
{
"epoch": 3.46,
"learning_rate": 6.176470588235295e-06,
"loss": 0.0044,
"step": 3290
},
{
"epoch": 3.47,
"learning_rate": 6.134453781512606e-06,
"loss": 0.0307,
"step": 3300
},
{
"epoch": 3.48,
"learning_rate": 6.092436974789916e-06,
"loss": 0.0616,
"step": 3310
},
{
"epoch": 3.49,
"learning_rate": 6.050420168067227e-06,
"loss": 0.0287,
"step": 3320
},
{
"epoch": 3.5,
"learning_rate": 6.008403361344538e-06,
"loss": 0.0344,
"step": 3330
},
{
"epoch": 3.51,
"learning_rate": 5.9663865546218495e-06,
"loss": 0.0749,
"step": 3340
},
{
"epoch": 3.52,
"learning_rate": 5.92436974789916e-06,
"loss": 0.0031,
"step": 3350
},
{
"epoch": 3.53,
"learning_rate": 5.882352941176471e-06,
"loss": 0.0819,
"step": 3360
},
{
"epoch": 3.54,
"learning_rate": 5.8403361344537825e-06,
"loss": 0.0616,
"step": 3370
},
{
"epoch": 3.55,
"learning_rate": 5.798319327731093e-06,
"loss": 0.1642,
"step": 3380
},
{
"epoch": 3.56,
"learning_rate": 5.756302521008403e-06,
"loss": 0.0435,
"step": 3390
},
{
"epoch": 3.57,
"learning_rate": 5.7142857142857145e-06,
"loss": 0.0449,
"step": 3400
},
{
"epoch": 3.58,
"learning_rate": 5.672268907563025e-06,
"loss": 0.07,
"step": 3410
},
{
"epoch": 3.59,
"learning_rate": 5.630252100840337e-06,
"loss": 0.0704,
"step": 3420
},
{
"epoch": 3.6,
"learning_rate": 5.588235294117647e-06,
"loss": 0.0206,
"step": 3430
},
{
"epoch": 3.61,
"learning_rate": 5.546218487394959e-06,
"loss": 0.0741,
"step": 3440
},
{
"epoch": 3.62,
"learning_rate": 5.50420168067227e-06,
"loss": 0.1549,
"step": 3450
},
{
"epoch": 3.63,
"learning_rate": 5.4621848739495795e-06,
"loss": 0.0267,
"step": 3460
},
{
"epoch": 3.64,
"learning_rate": 5.420168067226891e-06,
"loss": 0.0077,
"step": 3470
},
{
"epoch": 3.66,
"learning_rate": 5.378151260504202e-06,
"loss": 0.008,
"step": 3480
},
{
"epoch": 3.67,
"learning_rate": 5.336134453781513e-06,
"loss": 0.027,
"step": 3490
},
{
"epoch": 3.68,
"learning_rate": 5.294117647058824e-06,
"loss": 0.0068,
"step": 3500
},
{
"epoch": 3.69,
"learning_rate": 5.2521008403361354e-06,
"loss": 0.0555,
"step": 3510
},
{
"epoch": 3.7,
"learning_rate": 5.210084033613446e-06,
"loss": 0.0118,
"step": 3520
},
{
"epoch": 3.71,
"learning_rate": 5.168067226890757e-06,
"loss": 0.0109,
"step": 3530
},
{
"epoch": 3.72,
"learning_rate": 5.1260504201680675e-06,
"loss": 0.0258,
"step": 3540
},
{
"epoch": 3.73,
"learning_rate": 5.084033613445378e-06,
"loss": 0.0085,
"step": 3550
},
{
"epoch": 3.74,
"learning_rate": 5.04201680672269e-06,
"loss": 0.0694,
"step": 3560
},
{
"epoch": 3.75,
"learning_rate": 5e-06,
"loss": 0.0618,
"step": 3570
},
{
"epoch": 3.76,
"learning_rate": 4.957983193277311e-06,
"loss": 0.0876,
"step": 3580
},
{
"epoch": 3.77,
"learning_rate": 4.915966386554622e-06,
"loss": 0.0744,
"step": 3590
},
{
"epoch": 3.78,
"learning_rate": 4.873949579831933e-06,
"loss": 0.0114,
"step": 3600
},
{
"epoch": 3.79,
"learning_rate": 4.831932773109244e-06,
"loss": 0.0578,
"step": 3610
},
{
"epoch": 3.8,
"learning_rate": 4.7899159663865555e-06,
"loss": 0.0823,
"step": 3620
},
{
"epoch": 3.81,
"learning_rate": 4.747899159663865e-06,
"loss": 0.0895,
"step": 3630
},
{
"epoch": 3.82,
"learning_rate": 4.705882352941177e-06,
"loss": 0.0066,
"step": 3640
},
{
"epoch": 3.83,
"learning_rate": 4.663865546218488e-06,
"loss": 0.0034,
"step": 3650
},
{
"epoch": 3.84,
"learning_rate": 4.621848739495799e-06,
"loss": 0.0334,
"step": 3660
},
{
"epoch": 3.86,
"learning_rate": 4.57983193277311e-06,
"loss": 0.0022,
"step": 3670
},
{
"epoch": 3.87,
"learning_rate": 4.5378151260504205e-06,
"loss": 0.1133,
"step": 3680
},
{
"epoch": 3.88,
"learning_rate": 4.495798319327731e-06,
"loss": 0.0362,
"step": 3690
},
{
"epoch": 3.89,
"learning_rate": 4.453781512605043e-06,
"loss": 0.061,
"step": 3700
},
{
"epoch": 3.9,
"learning_rate": 4.411764705882353e-06,
"loss": 0.0813,
"step": 3710
},
{
"epoch": 3.91,
"learning_rate": 4.369747899159664e-06,
"loss": 0.0977,
"step": 3720
},
{
"epoch": 3.92,
"learning_rate": 4.327731092436976e-06,
"loss": 0.0219,
"step": 3730
},
{
"epoch": 3.93,
"learning_rate": 4.2857142857142855e-06,
"loss": 0.0894,
"step": 3740
},
{
"epoch": 3.94,
"learning_rate": 4.243697478991597e-06,
"loss": 0.0717,
"step": 3750
},
{
"epoch": 3.95,
"learning_rate": 4.201680672268908e-06,
"loss": 0.024,
"step": 3760
},
{
"epoch": 3.96,
"learning_rate": 4.159663865546219e-06,
"loss": 0.0415,
"step": 3770
},
{
"epoch": 3.97,
"learning_rate": 4.11764705882353e-06,
"loss": 0.0315,
"step": 3780
},
{
"epoch": 3.98,
"learning_rate": 4.075630252100841e-06,
"loss": 0.057,
"step": 3790
},
{
"epoch": 3.99,
"learning_rate": 4.033613445378151e-06,
"loss": 0.0066,
"step": 3800
},
{
"epoch": 4.0,
"eval_accuracy": 0.9933035714285714,
"eval_loss": 0.01902906596660614,
"eval_runtime": 608.5715,
"eval_samples_per_second": 2.208,
"eval_steps_per_second": 0.276,
"step": 3808
},
{
"epoch": 4.0,
"learning_rate": 3.991596638655463e-06,
"loss": 0.0044,
"step": 3810
},
{
"epoch": 4.01,
"learning_rate": 3.9495798319327735e-06,
"loss": 0.0688,
"step": 3820
},
{
"epoch": 4.02,
"learning_rate": 3.907563025210084e-06,
"loss": 0.0083,
"step": 3830
},
{
"epoch": 4.03,
"learning_rate": 3.865546218487396e-06,
"loss": 0.0352,
"step": 3840
},
{
"epoch": 4.04,
"learning_rate": 3.8235294117647055e-06,
"loss": 0.0203,
"step": 3850
},
{
"epoch": 4.05,
"learning_rate": 3.781512605042017e-06,
"loss": 0.0046,
"step": 3860
},
{
"epoch": 4.07,
"learning_rate": 3.739495798319328e-06,
"loss": 0.0327,
"step": 3870
},
{
"epoch": 4.08,
"learning_rate": 3.6974789915966393e-06,
"loss": 0.1025,
"step": 3880
},
{
"epoch": 4.09,
"learning_rate": 3.6554621848739496e-06,
"loss": 0.0306,
"step": 3890
},
{
"epoch": 4.1,
"learning_rate": 3.6134453781512607e-06,
"loss": 0.0027,
"step": 3900
},
{
"epoch": 4.11,
"learning_rate": 3.5714285714285718e-06,
"loss": 0.0315,
"step": 3910
},
{
"epoch": 4.12,
"learning_rate": 3.529411764705883e-06,
"loss": 0.0025,
"step": 3920
},
{
"epoch": 4.13,
"learning_rate": 3.4873949579831936e-06,
"loss": 0.0279,
"step": 3930
},
{
"epoch": 4.14,
"learning_rate": 3.4453781512605043e-06,
"loss": 0.0222,
"step": 3940
},
{
"epoch": 4.15,
"learning_rate": 3.4033613445378154e-06,
"loss": 0.0056,
"step": 3950
},
{
"epoch": 4.16,
"learning_rate": 3.3613445378151265e-06,
"loss": 0.0712,
"step": 3960
},
{
"epoch": 4.17,
"learning_rate": 3.319327731092437e-06,
"loss": 0.0021,
"step": 3970
},
{
"epoch": 4.18,
"learning_rate": 3.2773109243697483e-06,
"loss": 0.0086,
"step": 3980
},
{
"epoch": 4.19,
"learning_rate": 3.2352941176470594e-06,
"loss": 0.0034,
"step": 3990
},
{
"epoch": 4.2,
"learning_rate": 3.1932773109243696e-06,
"loss": 0.0221,
"step": 4000
},
{
"epoch": 4.21,
"learning_rate": 3.1512605042016808e-06,
"loss": 0.0685,
"step": 4010
},
{
"epoch": 4.22,
"learning_rate": 3.109243697478992e-06,
"loss": 0.0028,
"step": 4020
},
{
"epoch": 4.23,
"learning_rate": 3.067226890756303e-06,
"loss": 0.0066,
"step": 4030
},
{
"epoch": 4.24,
"learning_rate": 3.0252100840336137e-06,
"loss": 0.0623,
"step": 4040
},
{
"epoch": 4.25,
"learning_rate": 2.9831932773109248e-06,
"loss": 0.027,
"step": 4050
},
{
"epoch": 4.26,
"learning_rate": 2.9411764705882355e-06,
"loss": 0.0162,
"step": 4060
},
{
"epoch": 4.28,
"learning_rate": 2.8991596638655466e-06,
"loss": 0.0022,
"step": 4070
},
{
"epoch": 4.29,
"learning_rate": 2.8571428571428573e-06,
"loss": 0.0021,
"step": 4080
},
{
"epoch": 4.3,
"learning_rate": 2.8151260504201684e-06,
"loss": 0.0027,
"step": 4090
},
{
"epoch": 4.31,
"learning_rate": 2.7731092436974795e-06,
"loss": 0.0871,
"step": 4100
},
{
"epoch": 4.32,
"learning_rate": 2.7310924369747897e-06,
"loss": 0.0418,
"step": 4110
},
{
"epoch": 4.33,
"learning_rate": 2.689075630252101e-06,
"loss": 0.0026,
"step": 4120
},
{
"epoch": 4.34,
"learning_rate": 2.647058823529412e-06,
"loss": 0.0512,
"step": 4130
},
{
"epoch": 4.35,
"learning_rate": 2.605042016806723e-06,
"loss": 0.0251,
"step": 4140
},
{
"epoch": 4.36,
"learning_rate": 2.5630252100840338e-06,
"loss": 0.0894,
"step": 4150
},
{
"epoch": 4.37,
"learning_rate": 2.521008403361345e-06,
"loss": 0.0692,
"step": 4160
},
{
"epoch": 4.38,
"learning_rate": 2.4789915966386555e-06,
"loss": 0.0885,
"step": 4170
},
{
"epoch": 4.39,
"learning_rate": 2.4369747899159667e-06,
"loss": 0.003,
"step": 4180
},
{
"epoch": 4.4,
"learning_rate": 2.3949579831932778e-06,
"loss": 0.0235,
"step": 4190
},
{
"epoch": 4.41,
"learning_rate": 2.3529411764705885e-06,
"loss": 0.0286,
"step": 4200
},
{
"epoch": 4.42,
"learning_rate": 2.3109243697478996e-06,
"loss": 0.0523,
"step": 4210
},
{
"epoch": 4.43,
"learning_rate": 2.2689075630252102e-06,
"loss": 0.0258,
"step": 4220
},
{
"epoch": 4.44,
"learning_rate": 2.2268907563025214e-06,
"loss": 0.0149,
"step": 4230
},
{
"epoch": 4.45,
"learning_rate": 2.184873949579832e-06,
"loss": 0.0177,
"step": 4240
},
{
"epoch": 4.46,
"learning_rate": 2.1428571428571427e-06,
"loss": 0.1035,
"step": 4250
},
{
"epoch": 4.47,
"learning_rate": 2.100840336134454e-06,
"loss": 0.0618,
"step": 4260
},
{
"epoch": 4.49,
"learning_rate": 2.058823529411765e-06,
"loss": 0.0492,
"step": 4270
},
{
"epoch": 4.5,
"learning_rate": 2.0168067226890756e-06,
"loss": 0.0194,
"step": 4280
},
{
"epoch": 4.51,
"learning_rate": 1.9747899159663867e-06,
"loss": 0.0216,
"step": 4290
},
{
"epoch": 4.52,
"learning_rate": 1.932773109243698e-06,
"loss": 0.0582,
"step": 4300
},
{
"epoch": 4.53,
"learning_rate": 1.8907563025210085e-06,
"loss": 0.0085,
"step": 4310
},
{
"epoch": 4.54,
"learning_rate": 1.8487394957983196e-06,
"loss": 0.0805,
"step": 4320
},
{
"epoch": 4.55,
"learning_rate": 1.8067226890756303e-06,
"loss": 0.0094,
"step": 4330
},
{
"epoch": 4.56,
"learning_rate": 1.7647058823529414e-06,
"loss": 0.076,
"step": 4340
},
{
"epoch": 4.57,
"learning_rate": 1.7226890756302521e-06,
"loss": 0.0233,
"step": 4350
},
{
"epoch": 4.58,
"learning_rate": 1.6806722689075632e-06,
"loss": 0.062,
"step": 4360
},
{
"epoch": 4.59,
"learning_rate": 1.6386554621848741e-06,
"loss": 0.0194,
"step": 4370
},
{
"epoch": 4.6,
"learning_rate": 1.5966386554621848e-06,
"loss": 0.0811,
"step": 4380
},
{
"epoch": 4.61,
"learning_rate": 1.554621848739496e-06,
"loss": 0.0289,
"step": 4390
},
{
"epoch": 4.62,
"learning_rate": 1.5126050420168068e-06,
"loss": 0.0045,
"step": 4400
},
{
"epoch": 4.63,
"learning_rate": 1.4705882352941177e-06,
"loss": 0.0027,
"step": 4410
},
{
"epoch": 4.64,
"learning_rate": 1.4285714285714286e-06,
"loss": 0.0233,
"step": 4420
},
{
"epoch": 4.65,
"learning_rate": 1.3865546218487397e-06,
"loss": 0.0236,
"step": 4430
},
{
"epoch": 4.66,
"learning_rate": 1.3445378151260504e-06,
"loss": 0.0026,
"step": 4440
},
{
"epoch": 4.67,
"learning_rate": 1.3025210084033615e-06,
"loss": 0.002,
"step": 4450
},
{
"epoch": 4.68,
"learning_rate": 1.2605042016806724e-06,
"loss": 0.0032,
"step": 4460
},
{
"epoch": 4.7,
"learning_rate": 1.2184873949579833e-06,
"loss": 0.0047,
"step": 4470
},
{
"epoch": 4.71,
"learning_rate": 1.1764705882352942e-06,
"loss": 0.0075,
"step": 4480
},
{
"epoch": 4.72,
"learning_rate": 1.1344537815126051e-06,
"loss": 0.0444,
"step": 4490
},
{
"epoch": 4.73,
"learning_rate": 1.092436974789916e-06,
"loss": 0.002,
"step": 4500
},
{
"epoch": 4.74,
"learning_rate": 1.050420168067227e-06,
"loss": 0.0787,
"step": 4510
},
{
"epoch": 4.75,
"learning_rate": 1.0084033613445378e-06,
"loss": 0.0027,
"step": 4520
},
{
"epoch": 4.76,
"learning_rate": 9.66386554621849e-07,
"loss": 0.0037,
"step": 4530
},
{
"epoch": 4.77,
"learning_rate": 9.243697478991598e-07,
"loss": 0.0096,
"step": 4540
},
{
"epoch": 4.78,
"learning_rate": 8.823529411764707e-07,
"loss": 0.0138,
"step": 4550
},
{
"epoch": 4.79,
"learning_rate": 8.403361344537816e-07,
"loss": 0.0051,
"step": 4560
},
{
"epoch": 4.8,
"learning_rate": 7.983193277310924e-07,
"loss": 0.0464,
"step": 4570
},
{
"epoch": 4.81,
"learning_rate": 7.563025210084034e-07,
"loss": 0.0153,
"step": 4580
},
{
"epoch": 4.82,
"learning_rate": 7.142857142857143e-07,
"loss": 0.0393,
"step": 4590
},
{
"epoch": 4.83,
"learning_rate": 6.722689075630252e-07,
"loss": 0.0512,
"step": 4600
},
{
"epoch": 4.84,
"learning_rate": 6.302521008403362e-07,
"loss": 0.003,
"step": 4610
},
{
"epoch": 4.85,
"learning_rate": 5.882352941176471e-07,
"loss": 0.0025,
"step": 4620
},
{
"epoch": 4.86,
"learning_rate": 5.46218487394958e-07,
"loss": 0.0058,
"step": 4630
},
{
"epoch": 4.87,
"learning_rate": 5.042016806722689e-07,
"loss": 0.1052,
"step": 4640
},
{
"epoch": 4.88,
"learning_rate": 4.621848739495799e-07,
"loss": 0.0746,
"step": 4650
},
{
"epoch": 4.89,
"learning_rate": 4.201680672268908e-07,
"loss": 0.0711,
"step": 4660
},
{
"epoch": 4.91,
"learning_rate": 3.781512605042017e-07,
"loss": 0.0026,
"step": 4670
},
{
"epoch": 4.92,
"learning_rate": 3.361344537815126e-07,
"loss": 0.002,
"step": 4680
},
{
"epoch": 4.93,
"learning_rate": 2.9411764705882356e-07,
"loss": 0.0097,
"step": 4690
},
{
"epoch": 4.94,
"learning_rate": 2.5210084033613445e-07,
"loss": 0.0272,
"step": 4700
},
{
"epoch": 4.95,
"learning_rate": 2.100840336134454e-07,
"loss": 0.0071,
"step": 4710
},
{
"epoch": 4.96,
"learning_rate": 1.680672268907563e-07,
"loss": 0.1572,
"step": 4720
},
{
"epoch": 4.97,
"learning_rate": 1.2605042016806723e-07,
"loss": 0.1464,
"step": 4730
},
{
"epoch": 4.98,
"learning_rate": 8.403361344537815e-08,
"loss": 0.0028,
"step": 4740
},
{
"epoch": 4.99,
"learning_rate": 4.2016806722689076e-08,
"loss": 0.0018,
"step": 4750
},
{
"epoch": 5.0,
"learning_rate": 0.0,
"loss": 0.0528,
"step": 4760
},
{
"epoch": 5.0,
"eval_accuracy": 0.9940476190476191,
"eval_loss": 0.018311450257897377,
"eval_runtime": 575.8026,
"eval_samples_per_second": 2.334,
"eval_steps_per_second": 0.292,
"step": 4760
},
{
"epoch": 5.0,
"step": 4760,
"total_flos": 2.949422616720986e+18,
"train_loss": 0.0,
"train_runtime": 0.7379,
"train_samples_per_second": 51580.17,
"train_steps_per_second": 6450.909
}
],
"max_steps": 4760,
"num_train_epochs": 5,
"total_flos": 2.949422616720986e+18,
"trial_name": null,
"trial_params": null
}