SWIN-AI-Image-Detector / trainer_state.json
mmanikanta's picture
End of training
545a75f
{
"best_metric": 0.9833333333333333,
"best_model_checkpoint": "SWIN-AI-Image-Detector/checkpoint-3829",
"epoch": 10.0,
"global_step": 5470,
"is_hyper_param_search": false,
"is_local_process_zero": true,
"is_world_process_zero": true,
"log_history": [
{
"epoch": 0.02,
"learning_rate": 9.140767824497258e-07,
"loss": 0.7202,
"step": 10
},
{
"epoch": 0.04,
"learning_rate": 1.8281535648994516e-06,
"loss": 0.7217,
"step": 20
},
{
"epoch": 0.05,
"learning_rate": 2.7422303473491773e-06,
"loss": 0.7037,
"step": 30
},
{
"epoch": 0.07,
"learning_rate": 3.6563071297989032e-06,
"loss": 0.6712,
"step": 40
},
{
"epoch": 0.09,
"learning_rate": 4.570383912248629e-06,
"loss": 0.6445,
"step": 50
},
{
"epoch": 0.11,
"learning_rate": 5.484460694698355e-06,
"loss": 0.61,
"step": 60
},
{
"epoch": 0.13,
"learning_rate": 6.398537477148081e-06,
"loss": 0.5873,
"step": 70
},
{
"epoch": 0.15,
"learning_rate": 7.3126142595978065e-06,
"loss": 0.5315,
"step": 80
},
{
"epoch": 0.16,
"learning_rate": 8.226691042047533e-06,
"loss": 0.5032,
"step": 90
},
{
"epoch": 0.18,
"learning_rate": 9.140767824497258e-06,
"loss": 0.4722,
"step": 100
},
{
"epoch": 0.2,
"learning_rate": 1.0054844606946984e-05,
"loss": 0.4301,
"step": 110
},
{
"epoch": 0.22,
"learning_rate": 1.096892138939671e-05,
"loss": 0.4045,
"step": 120
},
{
"epoch": 0.24,
"learning_rate": 1.1882998171846435e-05,
"loss": 0.3834,
"step": 130
},
{
"epoch": 0.26,
"learning_rate": 1.2797074954296162e-05,
"loss": 0.3638,
"step": 140
},
{
"epoch": 0.27,
"learning_rate": 1.3711151736745886e-05,
"loss": 0.3989,
"step": 150
},
{
"epoch": 0.29,
"learning_rate": 1.4625228519195613e-05,
"loss": 0.3732,
"step": 160
},
{
"epoch": 0.31,
"learning_rate": 1.553930530164534e-05,
"loss": 0.3499,
"step": 170
},
{
"epoch": 0.33,
"learning_rate": 1.6453382084095066e-05,
"loss": 0.3414,
"step": 180
},
{
"epoch": 0.35,
"learning_rate": 1.7367458866544793e-05,
"loss": 0.3469,
"step": 190
},
{
"epoch": 0.37,
"learning_rate": 1.8281535648994517e-05,
"loss": 0.3556,
"step": 200
},
{
"epoch": 0.38,
"learning_rate": 1.9195612431444244e-05,
"loss": 0.3356,
"step": 210
},
{
"epoch": 0.4,
"learning_rate": 2.0109689213893968e-05,
"loss": 0.3244,
"step": 220
},
{
"epoch": 0.42,
"learning_rate": 2.1023765996343695e-05,
"loss": 0.3274,
"step": 230
},
{
"epoch": 0.44,
"learning_rate": 2.193784277879342e-05,
"loss": 0.301,
"step": 240
},
{
"epoch": 0.46,
"learning_rate": 2.2851919561243146e-05,
"loss": 0.3249,
"step": 250
},
{
"epoch": 0.48,
"learning_rate": 2.376599634369287e-05,
"loss": 0.2922,
"step": 260
},
{
"epoch": 0.49,
"learning_rate": 2.4680073126142597e-05,
"loss": 0.3002,
"step": 270
},
{
"epoch": 0.51,
"learning_rate": 2.5594149908592324e-05,
"loss": 0.2998,
"step": 280
},
{
"epoch": 0.53,
"learning_rate": 2.6508226691042048e-05,
"loss": 0.3251,
"step": 290
},
{
"epoch": 0.55,
"learning_rate": 2.742230347349177e-05,
"loss": 0.3045,
"step": 300
},
{
"epoch": 0.57,
"learning_rate": 2.8336380255941502e-05,
"loss": 0.2822,
"step": 310
},
{
"epoch": 0.59,
"learning_rate": 2.9250457038391226e-05,
"loss": 0.2582,
"step": 320
},
{
"epoch": 0.6,
"learning_rate": 3.016453382084095e-05,
"loss": 0.3063,
"step": 330
},
{
"epoch": 0.62,
"learning_rate": 3.107861060329068e-05,
"loss": 0.2475,
"step": 340
},
{
"epoch": 0.64,
"learning_rate": 3.1992687385740404e-05,
"loss": 0.3006,
"step": 350
},
{
"epoch": 0.66,
"learning_rate": 3.290676416819013e-05,
"loss": 0.3132,
"step": 360
},
{
"epoch": 0.68,
"learning_rate": 3.382084095063985e-05,
"loss": 0.2733,
"step": 370
},
{
"epoch": 0.69,
"learning_rate": 3.4734917733089586e-05,
"loss": 0.2682,
"step": 380
},
{
"epoch": 0.71,
"learning_rate": 3.5648994515539306e-05,
"loss": 0.3016,
"step": 390
},
{
"epoch": 0.73,
"learning_rate": 3.656307129798903e-05,
"loss": 0.2722,
"step": 400
},
{
"epoch": 0.75,
"learning_rate": 3.7477148080438754e-05,
"loss": 0.2517,
"step": 410
},
{
"epoch": 0.77,
"learning_rate": 3.839122486288849e-05,
"loss": 0.2823,
"step": 420
},
{
"epoch": 0.79,
"learning_rate": 3.930530164533821e-05,
"loss": 0.2712,
"step": 430
},
{
"epoch": 0.8,
"learning_rate": 4.0219378427787935e-05,
"loss": 0.297,
"step": 440
},
{
"epoch": 0.82,
"learning_rate": 4.113345521023766e-05,
"loss": 0.2538,
"step": 450
},
{
"epoch": 0.84,
"learning_rate": 4.204753199268739e-05,
"loss": 0.2803,
"step": 460
},
{
"epoch": 0.86,
"learning_rate": 4.296160877513711e-05,
"loss": 0.2442,
"step": 470
},
{
"epoch": 0.88,
"learning_rate": 4.387568555758684e-05,
"loss": 0.2415,
"step": 480
},
{
"epoch": 0.9,
"learning_rate": 4.4789762340036564e-05,
"loss": 0.2304,
"step": 490
},
{
"epoch": 0.91,
"learning_rate": 4.570383912248629e-05,
"loss": 0.2743,
"step": 500
},
{
"epoch": 0.93,
"learning_rate": 4.661791590493602e-05,
"loss": 0.2507,
"step": 510
},
{
"epoch": 0.95,
"learning_rate": 4.753199268738574e-05,
"loss": 0.2488,
"step": 520
},
{
"epoch": 0.97,
"learning_rate": 4.844606946983547e-05,
"loss": 0.2456,
"step": 530
},
{
"epoch": 0.99,
"learning_rate": 4.936014625228519e-05,
"loss": 0.2411,
"step": 540
},
{
"epoch": 1.0,
"eval_accuracy": 0.9013333333333333,
"eval_loss": 0.24297115206718445,
"eval_runtime": 107.1641,
"eval_samples_per_second": 279.944,
"eval_steps_per_second": 8.753,
"step": 547
},
{
"epoch": 1.01,
"learning_rate": 4.996953077391834e-05,
"loss": 0.2475,
"step": 550
},
{
"epoch": 1.02,
"learning_rate": 4.986796668697948e-05,
"loss": 0.2483,
"step": 560
},
{
"epoch": 1.04,
"learning_rate": 4.976640260004063e-05,
"loss": 0.2289,
"step": 570
},
{
"epoch": 1.06,
"learning_rate": 4.966483851310177e-05,
"loss": 0.2345,
"step": 580
},
{
"epoch": 1.08,
"learning_rate": 4.956327442616291e-05,
"loss": 0.2676,
"step": 590
},
{
"epoch": 1.1,
"learning_rate": 4.946171033922405e-05,
"loss": 0.2422,
"step": 600
},
{
"epoch": 1.12,
"learning_rate": 4.936014625228519e-05,
"loss": 0.2348,
"step": 610
},
{
"epoch": 1.13,
"learning_rate": 4.9258582165346334e-05,
"loss": 0.2316,
"step": 620
},
{
"epoch": 1.15,
"learning_rate": 4.915701807840748e-05,
"loss": 0.232,
"step": 630
},
{
"epoch": 1.17,
"learning_rate": 4.905545399146862e-05,
"loss": 0.2271,
"step": 640
},
{
"epoch": 1.19,
"learning_rate": 4.8953889904529756e-05,
"loss": 0.2086,
"step": 650
},
{
"epoch": 1.21,
"learning_rate": 4.8852325817590903e-05,
"loss": 0.2284,
"step": 660
},
{
"epoch": 1.22,
"learning_rate": 4.8750761730652044e-05,
"loss": 0.2265,
"step": 670
},
{
"epoch": 1.24,
"learning_rate": 4.8649197643713185e-05,
"loss": 0.227,
"step": 680
},
{
"epoch": 1.26,
"learning_rate": 4.8547633556774326e-05,
"loss": 0.226,
"step": 690
},
{
"epoch": 1.28,
"learning_rate": 4.844606946983547e-05,
"loss": 0.204,
"step": 700
},
{
"epoch": 1.3,
"learning_rate": 4.834450538289661e-05,
"loss": 0.2187,
"step": 710
},
{
"epoch": 1.32,
"learning_rate": 4.824294129595775e-05,
"loss": 0.2079,
"step": 720
},
{
"epoch": 1.33,
"learning_rate": 4.8141377209018895e-05,
"loss": 0.2215,
"step": 730
},
{
"epoch": 1.35,
"learning_rate": 4.8039813122080036e-05,
"loss": 0.206,
"step": 740
},
{
"epoch": 1.37,
"learning_rate": 4.7938249035141176e-05,
"loss": 0.2405,
"step": 750
},
{
"epoch": 1.39,
"learning_rate": 4.7836684948202324e-05,
"loss": 0.2155,
"step": 760
},
{
"epoch": 1.41,
"learning_rate": 4.773512086126346e-05,
"loss": 0.1981,
"step": 770
},
{
"epoch": 1.43,
"learning_rate": 4.76335567743246e-05,
"loss": 0.2124,
"step": 780
},
{
"epoch": 1.44,
"learning_rate": 4.753199268738574e-05,
"loss": 0.2483,
"step": 790
},
{
"epoch": 1.46,
"learning_rate": 4.7430428600446886e-05,
"loss": 0.2463,
"step": 800
},
{
"epoch": 1.48,
"learning_rate": 4.732886451350803e-05,
"loss": 0.2262,
"step": 810
},
{
"epoch": 1.5,
"learning_rate": 4.722730042656917e-05,
"loss": 0.1905,
"step": 820
},
{
"epoch": 1.52,
"learning_rate": 4.712573633963031e-05,
"loss": 0.213,
"step": 830
},
{
"epoch": 1.54,
"learning_rate": 4.702417225269145e-05,
"loss": 0.1902,
"step": 840
},
{
"epoch": 1.55,
"learning_rate": 4.692260816575259e-05,
"loss": 0.18,
"step": 850
},
{
"epoch": 1.57,
"learning_rate": 4.682104407881374e-05,
"loss": 0.2204,
"step": 860
},
{
"epoch": 1.59,
"learning_rate": 4.671947999187488e-05,
"loss": 0.2002,
"step": 870
},
{
"epoch": 1.61,
"learning_rate": 4.661791590493602e-05,
"loss": 0.2229,
"step": 880
},
{
"epoch": 1.63,
"learning_rate": 4.651635181799715e-05,
"loss": 0.1787,
"step": 890
},
{
"epoch": 1.65,
"learning_rate": 4.64147877310583e-05,
"loss": 0.2312,
"step": 900
},
{
"epoch": 1.66,
"learning_rate": 4.631322364411944e-05,
"loss": 0.2002,
"step": 910
},
{
"epoch": 1.68,
"learning_rate": 4.621165955718058e-05,
"loss": 0.2167,
"step": 920
},
{
"epoch": 1.7,
"learning_rate": 4.611009547024173e-05,
"loss": 0.1949,
"step": 930
},
{
"epoch": 1.72,
"learning_rate": 4.600853138330287e-05,
"loss": 0.2002,
"step": 940
},
{
"epoch": 1.74,
"learning_rate": 4.5906967296364e-05,
"loss": 0.1973,
"step": 950
},
{
"epoch": 1.76,
"learning_rate": 4.580540320942515e-05,
"loss": 0.2049,
"step": 960
},
{
"epoch": 1.77,
"learning_rate": 4.570383912248629e-05,
"loss": 0.1919,
"step": 970
},
{
"epoch": 1.79,
"learning_rate": 4.560227503554743e-05,
"loss": 0.2147,
"step": 980
},
{
"epoch": 1.81,
"learning_rate": 4.550071094860857e-05,
"loss": 0.1964,
"step": 990
},
{
"epoch": 1.83,
"learning_rate": 4.539914686166972e-05,
"loss": 0.1764,
"step": 1000
},
{
"epoch": 1.85,
"learning_rate": 4.5297582774730854e-05,
"loss": 0.1741,
"step": 1010
},
{
"epoch": 1.86,
"learning_rate": 4.5196018687791995e-05,
"loss": 0.1963,
"step": 1020
},
{
"epoch": 1.88,
"learning_rate": 4.509445460085314e-05,
"loss": 0.2276,
"step": 1030
},
{
"epoch": 1.9,
"learning_rate": 4.499289051391428e-05,
"loss": 0.1977,
"step": 1040
},
{
"epoch": 1.92,
"learning_rate": 4.4891326426975424e-05,
"loss": 0.1676,
"step": 1050
},
{
"epoch": 1.94,
"learning_rate": 4.4789762340036564e-05,
"loss": 0.1924,
"step": 1060
},
{
"epoch": 1.96,
"learning_rate": 4.4688198253097705e-05,
"loss": 0.2131,
"step": 1070
},
{
"epoch": 1.97,
"learning_rate": 4.4586634166158846e-05,
"loss": 0.2047,
"step": 1080
},
{
"epoch": 1.99,
"learning_rate": 4.448507007921999e-05,
"loss": 0.2011,
"step": 1090
},
{
"epoch": 2.0,
"eval_accuracy": 0.9592666666666667,
"eval_loss": 0.10533071309328079,
"eval_runtime": 107.4335,
"eval_samples_per_second": 279.243,
"eval_steps_per_second": 8.731,
"step": 1094
},
{
"epoch": 2.01,
"learning_rate": 4.4383505992281134e-05,
"loss": 0.1827,
"step": 1100
},
{
"epoch": 2.03,
"learning_rate": 4.4281941905342274e-05,
"loss": 0.1878,
"step": 1110
},
{
"epoch": 2.05,
"learning_rate": 4.4180377818403415e-05,
"loss": 0.1796,
"step": 1120
},
{
"epoch": 2.07,
"learning_rate": 4.4078813731464556e-05,
"loss": 0.1562,
"step": 1130
},
{
"epoch": 2.08,
"learning_rate": 4.3977249644525696e-05,
"loss": 0.1898,
"step": 1140
},
{
"epoch": 2.1,
"learning_rate": 4.387568555758684e-05,
"loss": 0.1798,
"step": 1150
},
{
"epoch": 2.12,
"learning_rate": 4.3774121470647984e-05,
"loss": 0.1691,
"step": 1160
},
{
"epoch": 2.14,
"learning_rate": 4.3672557383709125e-05,
"loss": 0.2052,
"step": 1170
},
{
"epoch": 2.16,
"learning_rate": 4.3570993296770266e-05,
"loss": 0.1673,
"step": 1180
},
{
"epoch": 2.18,
"learning_rate": 4.3469429209831407e-05,
"loss": 0.1788,
"step": 1190
},
{
"epoch": 2.19,
"learning_rate": 4.336786512289255e-05,
"loss": 0.1986,
"step": 1200
},
{
"epoch": 2.21,
"learning_rate": 4.326630103595369e-05,
"loss": 0.1988,
"step": 1210
},
{
"epoch": 2.23,
"learning_rate": 4.316473694901483e-05,
"loss": 0.1873,
"step": 1220
},
{
"epoch": 2.25,
"learning_rate": 4.3063172862075976e-05,
"loss": 0.1555,
"step": 1230
},
{
"epoch": 2.27,
"learning_rate": 4.296160877513711e-05,
"loss": 0.1616,
"step": 1240
},
{
"epoch": 2.29,
"learning_rate": 4.286004468819825e-05,
"loss": 0.1578,
"step": 1250
},
{
"epoch": 2.3,
"learning_rate": 4.27584806012594e-05,
"loss": 0.1689,
"step": 1260
},
{
"epoch": 2.32,
"learning_rate": 4.265691651432054e-05,
"loss": 0.1566,
"step": 1270
},
{
"epoch": 2.34,
"learning_rate": 4.255535242738168e-05,
"loss": 0.1789,
"step": 1280
},
{
"epoch": 2.36,
"learning_rate": 4.245378834044283e-05,
"loss": 0.1798,
"step": 1290
},
{
"epoch": 2.38,
"learning_rate": 4.235222425350396e-05,
"loss": 0.1831,
"step": 1300
},
{
"epoch": 2.39,
"learning_rate": 4.22506601665651e-05,
"loss": 0.1817,
"step": 1310
},
{
"epoch": 2.41,
"learning_rate": 4.214909607962624e-05,
"loss": 0.1636,
"step": 1320
},
{
"epoch": 2.43,
"learning_rate": 4.204753199268739e-05,
"loss": 0.2407,
"step": 1330
},
{
"epoch": 2.45,
"learning_rate": 4.194596790574853e-05,
"loss": 0.1733,
"step": 1340
},
{
"epoch": 2.47,
"learning_rate": 4.184440381880967e-05,
"loss": 0.1833,
"step": 1350
},
{
"epoch": 2.49,
"learning_rate": 4.174283973187081e-05,
"loss": 0.1783,
"step": 1360
},
{
"epoch": 2.5,
"learning_rate": 4.164127564493195e-05,
"loss": 0.1725,
"step": 1370
},
{
"epoch": 2.52,
"learning_rate": 4.153971155799309e-05,
"loss": 0.1401,
"step": 1380
},
{
"epoch": 2.54,
"learning_rate": 4.143814747105424e-05,
"loss": 0.1597,
"step": 1390
},
{
"epoch": 2.56,
"learning_rate": 4.133658338411538e-05,
"loss": 0.1399,
"step": 1400
},
{
"epoch": 2.58,
"learning_rate": 4.123501929717652e-05,
"loss": 0.1735,
"step": 1410
},
{
"epoch": 2.6,
"learning_rate": 4.113345521023766e-05,
"loss": 0.173,
"step": 1420
},
{
"epoch": 2.61,
"learning_rate": 4.10318911232988e-05,
"loss": 0.1778,
"step": 1430
},
{
"epoch": 2.63,
"learning_rate": 4.0930327036359944e-05,
"loss": 0.1637,
"step": 1440
},
{
"epoch": 2.65,
"learning_rate": 4.0828762949421084e-05,
"loss": 0.1724,
"step": 1450
},
{
"epoch": 2.67,
"learning_rate": 4.072719886248223e-05,
"loss": 0.161,
"step": 1460
},
{
"epoch": 2.69,
"learning_rate": 4.062563477554337e-05,
"loss": 0.1609,
"step": 1470
},
{
"epoch": 2.71,
"learning_rate": 4.0524070688604506e-05,
"loss": 0.1719,
"step": 1480
},
{
"epoch": 2.72,
"learning_rate": 4.0422506601665654e-05,
"loss": 0.1711,
"step": 1490
},
{
"epoch": 2.74,
"learning_rate": 4.0320942514726794e-05,
"loss": 0.1687,
"step": 1500
},
{
"epoch": 2.76,
"learning_rate": 4.0219378427787935e-05,
"loss": 0.1996,
"step": 1510
},
{
"epoch": 2.78,
"learning_rate": 4.011781434084908e-05,
"loss": 0.144,
"step": 1520
},
{
"epoch": 2.8,
"learning_rate": 4.001625025391022e-05,
"loss": 0.1861,
"step": 1530
},
{
"epoch": 2.82,
"learning_rate": 3.991468616697136e-05,
"loss": 0.1492,
"step": 1540
},
{
"epoch": 2.83,
"learning_rate": 3.98131220800325e-05,
"loss": 0.1399,
"step": 1550
},
{
"epoch": 2.85,
"learning_rate": 3.9711557993093645e-05,
"loss": 0.1488,
"step": 1560
},
{
"epoch": 2.87,
"learning_rate": 3.9609993906154786e-05,
"loss": 0.2207,
"step": 1570
},
{
"epoch": 2.89,
"learning_rate": 3.9508429819215927e-05,
"loss": 0.1679,
"step": 1580
},
{
"epoch": 2.91,
"learning_rate": 3.9406865732277074e-05,
"loss": 0.159,
"step": 1590
},
{
"epoch": 2.93,
"learning_rate": 3.930530164533821e-05,
"loss": 0.1415,
"step": 1600
},
{
"epoch": 2.94,
"learning_rate": 3.920373755839935e-05,
"loss": 0.1436,
"step": 1610
},
{
"epoch": 2.96,
"learning_rate": 3.9102173471460496e-05,
"loss": 0.1503,
"step": 1620
},
{
"epoch": 2.98,
"learning_rate": 3.900060938452164e-05,
"loss": 0.2036,
"step": 1630
},
{
"epoch": 3.0,
"learning_rate": 3.889904529758278e-05,
"loss": 0.1722,
"step": 1640
},
{
"epoch": 3.0,
"eval_accuracy": 0.9671333333333333,
"eval_loss": 0.08246280997991562,
"eval_runtime": 107.3836,
"eval_samples_per_second": 279.372,
"eval_steps_per_second": 8.735,
"step": 1641
},
{
"epoch": 3.02,
"learning_rate": 3.879748121064392e-05,
"loss": 0.1468,
"step": 1650
},
{
"epoch": 3.03,
"learning_rate": 3.869591712370506e-05,
"loss": 0.1678,
"step": 1660
},
{
"epoch": 3.05,
"learning_rate": 3.85943530367662e-05,
"loss": 0.1453,
"step": 1670
},
{
"epoch": 3.07,
"learning_rate": 3.849278894982734e-05,
"loss": 0.1664,
"step": 1680
},
{
"epoch": 3.09,
"learning_rate": 3.839122486288849e-05,
"loss": 0.1452,
"step": 1690
},
{
"epoch": 3.11,
"learning_rate": 3.828966077594963e-05,
"loss": 0.1661,
"step": 1700
},
{
"epoch": 3.13,
"learning_rate": 3.818809668901077e-05,
"loss": 0.1367,
"step": 1710
},
{
"epoch": 3.14,
"learning_rate": 3.808653260207191e-05,
"loss": 0.1345,
"step": 1720
},
{
"epoch": 3.16,
"learning_rate": 3.798496851513305e-05,
"loss": 0.1143,
"step": 1730
},
{
"epoch": 3.18,
"learning_rate": 3.788340442819419e-05,
"loss": 0.1928,
"step": 1740
},
{
"epoch": 3.2,
"learning_rate": 3.778184034125533e-05,
"loss": 0.152,
"step": 1750
},
{
"epoch": 3.22,
"learning_rate": 3.768027625431648e-05,
"loss": 0.1454,
"step": 1760
},
{
"epoch": 3.24,
"learning_rate": 3.757871216737762e-05,
"loss": 0.1523,
"step": 1770
},
{
"epoch": 3.25,
"learning_rate": 3.7477148080438754e-05,
"loss": 0.1212,
"step": 1780
},
{
"epoch": 3.27,
"learning_rate": 3.73755839934999e-05,
"loss": 0.159,
"step": 1790
},
{
"epoch": 3.29,
"learning_rate": 3.727401990656104e-05,
"loss": 0.1403,
"step": 1800
},
{
"epoch": 3.31,
"learning_rate": 3.717245581962218e-05,
"loss": 0.1345,
"step": 1810
},
{
"epoch": 3.33,
"learning_rate": 3.707089173268333e-05,
"loss": 0.1669,
"step": 1820
},
{
"epoch": 3.35,
"learning_rate": 3.6969327645744464e-05,
"loss": 0.1526,
"step": 1830
},
{
"epoch": 3.36,
"learning_rate": 3.6867763558805604e-05,
"loss": 0.1423,
"step": 1840
},
{
"epoch": 3.38,
"learning_rate": 3.676619947186675e-05,
"loss": 0.1443,
"step": 1850
},
{
"epoch": 3.4,
"learning_rate": 3.666463538492789e-05,
"loss": 0.1467,
"step": 1860
},
{
"epoch": 3.42,
"learning_rate": 3.656307129798903e-05,
"loss": 0.146,
"step": 1870
},
{
"epoch": 3.44,
"learning_rate": 3.6461507211050174e-05,
"loss": 0.1397,
"step": 1880
},
{
"epoch": 3.46,
"learning_rate": 3.6359943124111314e-05,
"loss": 0.1533,
"step": 1890
},
{
"epoch": 3.47,
"learning_rate": 3.6258379037172455e-05,
"loss": 0.1379,
"step": 1900
},
{
"epoch": 3.49,
"learning_rate": 3.6156814950233596e-05,
"loss": 0.1572,
"step": 1910
},
{
"epoch": 3.51,
"learning_rate": 3.605525086329474e-05,
"loss": 0.1483,
"step": 1920
},
{
"epoch": 3.53,
"learning_rate": 3.5953686776355884e-05,
"loss": 0.1603,
"step": 1930
},
{
"epoch": 3.55,
"learning_rate": 3.5852122689417025e-05,
"loss": 0.1434,
"step": 1940
},
{
"epoch": 3.56,
"learning_rate": 3.5750558602478165e-05,
"loss": 0.1655,
"step": 1950
},
{
"epoch": 3.58,
"learning_rate": 3.5648994515539306e-05,
"loss": 0.1564,
"step": 1960
},
{
"epoch": 3.6,
"learning_rate": 3.554743042860045e-05,
"loss": 0.1335,
"step": 1970
},
{
"epoch": 3.62,
"learning_rate": 3.544586634166159e-05,
"loss": 0.1419,
"step": 1980
},
{
"epoch": 3.64,
"learning_rate": 3.5344302254722735e-05,
"loss": 0.1464,
"step": 1990
},
{
"epoch": 3.66,
"learning_rate": 3.5242738167783875e-05,
"loss": 0.1423,
"step": 2000
},
{
"epoch": 3.67,
"learning_rate": 3.5141174080845016e-05,
"loss": 0.1425,
"step": 2010
},
{
"epoch": 3.69,
"learning_rate": 3.503960999390616e-05,
"loss": 0.1429,
"step": 2020
},
{
"epoch": 3.71,
"learning_rate": 3.49380459069673e-05,
"loss": 0.1237,
"step": 2030
},
{
"epoch": 3.73,
"learning_rate": 3.483648182002844e-05,
"loss": 0.1571,
"step": 2040
},
{
"epoch": 3.75,
"learning_rate": 3.4734917733089586e-05,
"loss": 0.1504,
"step": 2050
},
{
"epoch": 3.77,
"learning_rate": 3.4633353646150726e-05,
"loss": 0.1423,
"step": 2060
},
{
"epoch": 3.78,
"learning_rate": 3.453178955921186e-05,
"loss": 0.1373,
"step": 2070
},
{
"epoch": 3.8,
"learning_rate": 3.4430225472273e-05,
"loss": 0.1387,
"step": 2080
},
{
"epoch": 3.82,
"learning_rate": 3.432866138533415e-05,
"loss": 0.1628,
"step": 2090
},
{
"epoch": 3.84,
"learning_rate": 3.422709729839529e-05,
"loss": 0.134,
"step": 2100
},
{
"epoch": 3.86,
"learning_rate": 3.412553321145643e-05,
"loss": 0.1636,
"step": 2110
},
{
"epoch": 3.88,
"learning_rate": 3.402396912451758e-05,
"loss": 0.1548,
"step": 2120
},
{
"epoch": 3.89,
"learning_rate": 3.392240503757871e-05,
"loss": 0.1643,
"step": 2130
},
{
"epoch": 3.91,
"learning_rate": 3.382084095063985e-05,
"loss": 0.1427,
"step": 2140
},
{
"epoch": 3.93,
"learning_rate": 3.3719276863701e-05,
"loss": 0.1171,
"step": 2150
},
{
"epoch": 3.95,
"learning_rate": 3.361771277676214e-05,
"loss": 0.1536,
"step": 2160
},
{
"epoch": 3.97,
"learning_rate": 3.351614868982328e-05,
"loss": 0.1432,
"step": 2170
},
{
"epoch": 3.99,
"learning_rate": 3.341458460288442e-05,
"loss": 0.1424,
"step": 2180
},
{
"epoch": 4.0,
"eval_accuracy": 0.9686333333333333,
"eval_loss": 0.08514226227998734,
"eval_runtime": 107.4467,
"eval_samples_per_second": 279.208,
"eval_steps_per_second": 8.73,
"step": 2188
},
{
"epoch": 4.0,
"learning_rate": 3.331302051594556e-05,
"loss": 0.1166,
"step": 2190
},
{
"epoch": 4.02,
"learning_rate": 3.32114564290067e-05,
"loss": 0.1505,
"step": 2200
},
{
"epoch": 4.04,
"learning_rate": 3.310989234206784e-05,
"loss": 0.125,
"step": 2210
},
{
"epoch": 4.06,
"learning_rate": 3.300832825512899e-05,
"loss": 0.1363,
"step": 2220
},
{
"epoch": 4.08,
"learning_rate": 3.290676416819013e-05,
"loss": 0.1539,
"step": 2230
},
{
"epoch": 4.1,
"learning_rate": 3.280520008125127e-05,
"loss": 0.1323,
"step": 2240
},
{
"epoch": 4.11,
"learning_rate": 3.270363599431241e-05,
"loss": 0.1549,
"step": 2250
},
{
"epoch": 4.13,
"learning_rate": 3.260207190737355e-05,
"loss": 0.1708,
"step": 2260
},
{
"epoch": 4.15,
"learning_rate": 3.2500507820434694e-05,
"loss": 0.1367,
"step": 2270
},
{
"epoch": 4.17,
"learning_rate": 3.239894373349584e-05,
"loss": 0.1474,
"step": 2280
},
{
"epoch": 4.19,
"learning_rate": 3.229737964655698e-05,
"loss": 0.1296,
"step": 2290
},
{
"epoch": 4.2,
"learning_rate": 3.219581555961812e-05,
"loss": 0.1349,
"step": 2300
},
{
"epoch": 4.22,
"learning_rate": 3.2094251472679257e-05,
"loss": 0.1239,
"step": 2310
},
{
"epoch": 4.24,
"learning_rate": 3.1992687385740404e-05,
"loss": 0.1555,
"step": 2320
},
{
"epoch": 4.26,
"learning_rate": 3.1891123298801545e-05,
"loss": 0.1417,
"step": 2330
},
{
"epoch": 4.28,
"learning_rate": 3.1789559211862685e-05,
"loss": 0.1362,
"step": 2340
},
{
"epoch": 4.3,
"learning_rate": 3.168799512492383e-05,
"loss": 0.1335,
"step": 2350
},
{
"epoch": 4.31,
"learning_rate": 3.1586431037984973e-05,
"loss": 0.145,
"step": 2360
},
{
"epoch": 4.33,
"learning_rate": 3.148486695104611e-05,
"loss": 0.1235,
"step": 2370
},
{
"epoch": 4.35,
"learning_rate": 3.1383302864107255e-05,
"loss": 0.1101,
"step": 2380
},
{
"epoch": 4.37,
"learning_rate": 3.1281738777168395e-05,
"loss": 0.1134,
"step": 2390
},
{
"epoch": 4.39,
"learning_rate": 3.1180174690229536e-05,
"loss": 0.1345,
"step": 2400
},
{
"epoch": 4.41,
"learning_rate": 3.107861060329068e-05,
"loss": 0.1466,
"step": 2410
},
{
"epoch": 4.42,
"learning_rate": 3.0977046516351824e-05,
"loss": 0.1239,
"step": 2420
},
{
"epoch": 4.44,
"learning_rate": 3.087548242941296e-05,
"loss": 0.1315,
"step": 2430
},
{
"epoch": 4.46,
"learning_rate": 3.07739183424741e-05,
"loss": 0.1165,
"step": 2440
},
{
"epoch": 4.48,
"learning_rate": 3.0672354255535246e-05,
"loss": 0.1401,
"step": 2450
},
{
"epoch": 4.5,
"learning_rate": 3.057079016859639e-05,
"loss": 0.1428,
"step": 2460
},
{
"epoch": 4.52,
"learning_rate": 3.0469226081657524e-05,
"loss": 0.1198,
"step": 2470
},
{
"epoch": 4.53,
"learning_rate": 3.036766199471867e-05,
"loss": 0.1192,
"step": 2480
},
{
"epoch": 4.55,
"learning_rate": 3.0266097907779812e-05,
"loss": 0.1212,
"step": 2490
},
{
"epoch": 4.57,
"learning_rate": 3.016453382084095e-05,
"loss": 0.1148,
"step": 2500
},
{
"epoch": 4.59,
"learning_rate": 3.006296973390209e-05,
"loss": 0.1472,
"step": 2510
},
{
"epoch": 4.61,
"learning_rate": 2.9961405646963238e-05,
"loss": 0.1403,
"step": 2520
},
{
"epoch": 4.63,
"learning_rate": 2.9859841560024375e-05,
"loss": 0.1353,
"step": 2530
},
{
"epoch": 4.64,
"learning_rate": 2.9758277473085516e-05,
"loss": 0.15,
"step": 2540
},
{
"epoch": 4.66,
"learning_rate": 2.9656713386146663e-05,
"loss": 0.1106,
"step": 2550
},
{
"epoch": 4.68,
"learning_rate": 2.95551492992078e-05,
"loss": 0.1476,
"step": 2560
},
{
"epoch": 4.7,
"learning_rate": 2.945358521226894e-05,
"loss": 0.141,
"step": 2570
},
{
"epoch": 4.72,
"learning_rate": 2.935202112533009e-05,
"loss": 0.1299,
"step": 2580
},
{
"epoch": 4.73,
"learning_rate": 2.9250457038391226e-05,
"loss": 0.1546,
"step": 2590
},
{
"epoch": 4.75,
"learning_rate": 2.9148892951452366e-05,
"loss": 0.175,
"step": 2600
},
{
"epoch": 4.77,
"learning_rate": 2.9047328864513507e-05,
"loss": 0.1388,
"step": 2610
},
{
"epoch": 4.79,
"learning_rate": 2.894576477757465e-05,
"loss": 0.1279,
"step": 2620
},
{
"epoch": 4.81,
"learning_rate": 2.8844200690635792e-05,
"loss": 0.135,
"step": 2630
},
{
"epoch": 4.83,
"learning_rate": 2.8742636603696933e-05,
"loss": 0.101,
"step": 2640
},
{
"epoch": 4.84,
"learning_rate": 2.8641072516758077e-05,
"loss": 0.1344,
"step": 2650
},
{
"epoch": 4.86,
"learning_rate": 2.8539508429819217e-05,
"loss": 0.1122,
"step": 2660
},
{
"epoch": 4.88,
"learning_rate": 2.8437944342880358e-05,
"loss": 0.1043,
"step": 2670
},
{
"epoch": 4.9,
"learning_rate": 2.8336380255941502e-05,
"loss": 0.1296,
"step": 2680
},
{
"epoch": 4.92,
"learning_rate": 2.8234816169002643e-05,
"loss": 0.1195,
"step": 2690
},
{
"epoch": 4.94,
"learning_rate": 2.8133252082063783e-05,
"loss": 0.117,
"step": 2700
},
{
"epoch": 4.95,
"learning_rate": 2.8031687995124927e-05,
"loss": 0.1098,
"step": 2710
},
{
"epoch": 4.97,
"learning_rate": 2.7930123908186068e-05,
"loss": 0.1168,
"step": 2720
},
{
"epoch": 4.99,
"learning_rate": 2.782855982124721e-05,
"loss": 0.1244,
"step": 2730
},
{
"epoch": 5.0,
"eval_accuracy": 0.9732666666666666,
"eval_loss": 0.07139687240123749,
"eval_runtime": 107.2208,
"eval_samples_per_second": 279.796,
"eval_steps_per_second": 8.748,
"step": 2735
},
{
"epoch": 5.01,
"learning_rate": 2.7726995734308346e-05,
"loss": 0.1521,
"step": 2740
},
{
"epoch": 5.03,
"learning_rate": 2.7625431647369494e-05,
"loss": 0.1246,
"step": 2750
},
{
"epoch": 5.05,
"learning_rate": 2.7523867560430634e-05,
"loss": 0.1046,
"step": 2760
},
{
"epoch": 5.06,
"learning_rate": 2.742230347349177e-05,
"loss": 0.1186,
"step": 2770
},
{
"epoch": 5.08,
"learning_rate": 2.732073938655292e-05,
"loss": 0.1086,
"step": 2780
},
{
"epoch": 5.1,
"learning_rate": 2.721917529961406e-05,
"loss": 0.1158,
"step": 2790
},
{
"epoch": 5.12,
"learning_rate": 2.7117611212675197e-05,
"loss": 0.1199,
"step": 2800
},
{
"epoch": 5.14,
"learning_rate": 2.7016047125736344e-05,
"loss": 0.1027,
"step": 2810
},
{
"epoch": 5.16,
"learning_rate": 2.6914483038797485e-05,
"loss": 0.1379,
"step": 2820
},
{
"epoch": 5.17,
"learning_rate": 2.6812918951858622e-05,
"loss": 0.1285,
"step": 2830
},
{
"epoch": 5.19,
"learning_rate": 2.6711354864919763e-05,
"loss": 0.1068,
"step": 2840
},
{
"epoch": 5.21,
"learning_rate": 2.6609790777980907e-05,
"loss": 0.1201,
"step": 2850
},
{
"epoch": 5.23,
"learning_rate": 2.6508226691042048e-05,
"loss": 0.1081,
"step": 2860
},
{
"epoch": 5.25,
"learning_rate": 2.640666260410319e-05,
"loss": 0.1378,
"step": 2870
},
{
"epoch": 5.27,
"learning_rate": 2.6305098517164332e-05,
"loss": 0.1165,
"step": 2880
},
{
"epoch": 5.28,
"learning_rate": 2.6203534430225473e-05,
"loss": 0.0983,
"step": 2890
},
{
"epoch": 5.3,
"learning_rate": 2.6101970343286614e-05,
"loss": 0.1398,
"step": 2900
},
{
"epoch": 5.32,
"learning_rate": 2.6000406256347758e-05,
"loss": 0.0956,
"step": 2910
},
{
"epoch": 5.34,
"learning_rate": 2.58988421694089e-05,
"loss": 0.1129,
"step": 2920
},
{
"epoch": 5.36,
"learning_rate": 2.579727808247004e-05,
"loss": 0.1347,
"step": 2930
},
{
"epoch": 5.37,
"learning_rate": 2.569571399553118e-05,
"loss": 0.1196,
"step": 2940
},
{
"epoch": 5.39,
"learning_rate": 2.5594149908592324e-05,
"loss": 0.119,
"step": 2950
},
{
"epoch": 5.41,
"learning_rate": 2.5492585821653465e-05,
"loss": 0.1259,
"step": 2960
},
{
"epoch": 5.43,
"learning_rate": 2.5391021734714605e-05,
"loss": 0.1212,
"step": 2970
},
{
"epoch": 5.45,
"learning_rate": 2.528945764777575e-05,
"loss": 0.1481,
"step": 2980
},
{
"epoch": 5.47,
"learning_rate": 2.518789356083689e-05,
"loss": 0.0944,
"step": 2990
},
{
"epoch": 5.48,
"learning_rate": 2.508632947389803e-05,
"loss": 0.1074,
"step": 3000
},
{
"epoch": 5.5,
"learning_rate": 2.498476538695917e-05,
"loss": 0.1107,
"step": 3010
},
{
"epoch": 5.52,
"learning_rate": 2.4883201300020315e-05,
"loss": 0.1103,
"step": 3020
},
{
"epoch": 5.54,
"learning_rate": 2.4781637213081456e-05,
"loss": 0.128,
"step": 3030
},
{
"epoch": 5.56,
"learning_rate": 2.4680073126142597e-05,
"loss": 0.0964,
"step": 3040
},
{
"epoch": 5.58,
"learning_rate": 2.457850903920374e-05,
"loss": 0.1367,
"step": 3050
},
{
"epoch": 5.59,
"learning_rate": 2.4476944952264878e-05,
"loss": 0.1317,
"step": 3060
},
{
"epoch": 5.61,
"learning_rate": 2.4375380865326022e-05,
"loss": 0.116,
"step": 3070
},
{
"epoch": 5.63,
"learning_rate": 2.4273816778387163e-05,
"loss": 0.1198,
"step": 3080
},
{
"epoch": 5.65,
"learning_rate": 2.4172252691448303e-05,
"loss": 0.1302,
"step": 3090
},
{
"epoch": 5.67,
"learning_rate": 2.4070688604509447e-05,
"loss": 0.1217,
"step": 3100
},
{
"epoch": 5.69,
"learning_rate": 2.3969124517570588e-05,
"loss": 0.1108,
"step": 3110
},
{
"epoch": 5.7,
"learning_rate": 2.386756043063173e-05,
"loss": 0.1093,
"step": 3120
},
{
"epoch": 5.72,
"learning_rate": 2.376599634369287e-05,
"loss": 0.1178,
"step": 3130
},
{
"epoch": 5.74,
"learning_rate": 2.3664432256754014e-05,
"loss": 0.1315,
"step": 3140
},
{
"epoch": 5.76,
"learning_rate": 2.3562868169815154e-05,
"loss": 0.1138,
"step": 3150
},
{
"epoch": 5.78,
"learning_rate": 2.3461304082876295e-05,
"loss": 0.1099,
"step": 3160
},
{
"epoch": 5.8,
"learning_rate": 2.335973999593744e-05,
"loss": 0.1269,
"step": 3170
},
{
"epoch": 5.81,
"learning_rate": 2.3258175908998576e-05,
"loss": 0.1086,
"step": 3180
},
{
"epoch": 5.83,
"learning_rate": 2.315661182205972e-05,
"loss": 0.1093,
"step": 3190
},
{
"epoch": 5.85,
"learning_rate": 2.3055047735120864e-05,
"loss": 0.1157,
"step": 3200
},
{
"epoch": 5.87,
"learning_rate": 2.2953483648182e-05,
"loss": 0.1169,
"step": 3210
},
{
"epoch": 5.89,
"learning_rate": 2.2851919561243146e-05,
"loss": 0.1128,
"step": 3220
},
{
"epoch": 5.9,
"learning_rate": 2.2750355474304286e-05,
"loss": 0.1142,
"step": 3230
},
{
"epoch": 5.92,
"learning_rate": 2.2648791387365427e-05,
"loss": 0.1058,
"step": 3240
},
{
"epoch": 5.94,
"learning_rate": 2.254722730042657e-05,
"loss": 0.1145,
"step": 3250
},
{
"epoch": 5.96,
"learning_rate": 2.2445663213487712e-05,
"loss": 0.1027,
"step": 3260
},
{
"epoch": 5.98,
"learning_rate": 2.2344099126548852e-05,
"loss": 0.1139,
"step": 3270
},
{
"epoch": 6.0,
"learning_rate": 2.2242535039609997e-05,
"loss": 0.1089,
"step": 3280
},
{
"epoch": 6.0,
"eval_accuracy": 0.9733666666666667,
"eval_loss": 0.07119759172201157,
"eval_runtime": 106.9967,
"eval_samples_per_second": 280.382,
"eval_steps_per_second": 8.767,
"step": 3282
},
{
"epoch": 6.01,
"learning_rate": 2.2140970952671137e-05,
"loss": 0.096,
"step": 3290
},
{
"epoch": 6.03,
"learning_rate": 2.2039406865732278e-05,
"loss": 0.1201,
"step": 3300
},
{
"epoch": 6.05,
"learning_rate": 2.193784277879342e-05,
"loss": 0.1199,
"step": 3310
},
{
"epoch": 6.07,
"learning_rate": 2.1836278691854563e-05,
"loss": 0.1293,
"step": 3320
},
{
"epoch": 6.09,
"learning_rate": 2.1734714604915703e-05,
"loss": 0.1067,
"step": 3330
},
{
"epoch": 6.11,
"learning_rate": 2.1633150517976844e-05,
"loss": 0.1032,
"step": 3340
},
{
"epoch": 6.12,
"learning_rate": 2.1531586431037988e-05,
"loss": 0.1162,
"step": 3350
},
{
"epoch": 6.14,
"learning_rate": 2.1430022344099125e-05,
"loss": 0.105,
"step": 3360
},
{
"epoch": 6.16,
"learning_rate": 2.132845825716027e-05,
"loss": 0.1064,
"step": 3370
},
{
"epoch": 6.18,
"learning_rate": 2.1226894170221413e-05,
"loss": 0.1013,
"step": 3380
},
{
"epoch": 6.2,
"learning_rate": 2.112533008328255e-05,
"loss": 0.1113,
"step": 3390
},
{
"epoch": 6.22,
"learning_rate": 2.1023765996343695e-05,
"loss": 0.1065,
"step": 3400
},
{
"epoch": 6.23,
"learning_rate": 2.0922201909404835e-05,
"loss": 0.1014,
"step": 3410
},
{
"epoch": 6.25,
"learning_rate": 2.0820637822465976e-05,
"loss": 0.1169,
"step": 3420
},
{
"epoch": 6.27,
"learning_rate": 2.071907373552712e-05,
"loss": 0.1086,
"step": 3430
},
{
"epoch": 6.29,
"learning_rate": 2.061750964858826e-05,
"loss": 0.125,
"step": 3440
},
{
"epoch": 6.31,
"learning_rate": 2.05159455616494e-05,
"loss": 0.089,
"step": 3450
},
{
"epoch": 6.33,
"learning_rate": 2.0414381474710542e-05,
"loss": 0.1056,
"step": 3460
},
{
"epoch": 6.34,
"learning_rate": 2.0312817387771686e-05,
"loss": 0.0981,
"step": 3470
},
{
"epoch": 6.36,
"learning_rate": 2.0211253300832827e-05,
"loss": 0.1178,
"step": 3480
},
{
"epoch": 6.38,
"learning_rate": 2.0109689213893968e-05,
"loss": 0.1065,
"step": 3490
},
{
"epoch": 6.4,
"learning_rate": 2.000812512695511e-05,
"loss": 0.1219,
"step": 3500
},
{
"epoch": 6.42,
"learning_rate": 1.990656104001625e-05,
"loss": 0.1114,
"step": 3510
},
{
"epoch": 6.44,
"learning_rate": 1.9804996953077393e-05,
"loss": 0.1034,
"step": 3520
},
{
"epoch": 6.45,
"learning_rate": 1.9703432866138537e-05,
"loss": 0.1159,
"step": 3530
},
{
"epoch": 6.47,
"learning_rate": 1.9601868779199674e-05,
"loss": 0.0968,
"step": 3540
},
{
"epoch": 6.49,
"learning_rate": 1.950030469226082e-05,
"loss": 0.108,
"step": 3550
},
{
"epoch": 6.51,
"learning_rate": 1.939874060532196e-05,
"loss": 0.1239,
"step": 3560
},
{
"epoch": 6.53,
"learning_rate": 1.92971765183831e-05,
"loss": 0.0928,
"step": 3570
},
{
"epoch": 6.54,
"learning_rate": 1.9195612431444244e-05,
"loss": 0.099,
"step": 3580
},
{
"epoch": 6.56,
"learning_rate": 1.9094048344505384e-05,
"loss": 0.1039,
"step": 3590
},
{
"epoch": 6.58,
"learning_rate": 1.8992484257566525e-05,
"loss": 0.0904,
"step": 3600
},
{
"epoch": 6.6,
"learning_rate": 1.8890920170627666e-05,
"loss": 0.1183,
"step": 3610
},
{
"epoch": 6.62,
"learning_rate": 1.878935608368881e-05,
"loss": 0.1198,
"step": 3620
},
{
"epoch": 6.64,
"learning_rate": 1.868779199674995e-05,
"loss": 0.1213,
"step": 3630
},
{
"epoch": 6.65,
"learning_rate": 1.858622790981109e-05,
"loss": 0.0993,
"step": 3640
},
{
"epoch": 6.67,
"learning_rate": 1.8484663822872232e-05,
"loss": 0.1091,
"step": 3650
},
{
"epoch": 6.69,
"learning_rate": 1.8383099735933376e-05,
"loss": 0.1043,
"step": 3660
},
{
"epoch": 6.71,
"learning_rate": 1.8281535648994517e-05,
"loss": 0.1065,
"step": 3670
},
{
"epoch": 6.73,
"learning_rate": 1.8179971562055657e-05,
"loss": 0.1055,
"step": 3680
},
{
"epoch": 6.75,
"learning_rate": 1.8078407475116798e-05,
"loss": 0.1151,
"step": 3690
},
{
"epoch": 6.76,
"learning_rate": 1.7976843388177942e-05,
"loss": 0.0938,
"step": 3700
},
{
"epoch": 6.78,
"learning_rate": 1.7875279301239083e-05,
"loss": 0.113,
"step": 3710
},
{
"epoch": 6.8,
"learning_rate": 1.7773715214300223e-05,
"loss": 0.1068,
"step": 3720
},
{
"epoch": 6.82,
"learning_rate": 1.7672151127361367e-05,
"loss": 0.0978,
"step": 3730
},
{
"epoch": 6.84,
"learning_rate": 1.7570587040422508e-05,
"loss": 0.1129,
"step": 3740
},
{
"epoch": 6.86,
"learning_rate": 1.746902295348365e-05,
"loss": 0.1069,
"step": 3750
},
{
"epoch": 6.87,
"learning_rate": 1.7367458866544793e-05,
"loss": 0.111,
"step": 3760
},
{
"epoch": 6.89,
"learning_rate": 1.726589477960593e-05,
"loss": 0.0973,
"step": 3770
},
{
"epoch": 6.91,
"learning_rate": 1.7164330692667074e-05,
"loss": 0.1002,
"step": 3780
},
{
"epoch": 6.93,
"learning_rate": 1.7062766605728215e-05,
"loss": 0.1095,
"step": 3790
},
{
"epoch": 6.95,
"learning_rate": 1.6961202518789355e-05,
"loss": 0.1023,
"step": 3800
},
{
"epoch": 6.97,
"learning_rate": 1.68596384318505e-05,
"loss": 0.0972,
"step": 3810
},
{
"epoch": 6.98,
"learning_rate": 1.675807434491164e-05,
"loss": 0.1047,
"step": 3820
},
{
"epoch": 7.0,
"eval_accuracy": 0.9833333333333333,
"eval_loss": 0.04612118750810623,
"eval_runtime": 107.2386,
"eval_samples_per_second": 279.75,
"eval_steps_per_second": 8.747,
"step": 3829
},
{
"epoch": 7.0,
"learning_rate": 1.665651025797278e-05,
"loss": 0.1033,
"step": 3830
},
{
"epoch": 7.02,
"learning_rate": 1.655494617103392e-05,
"loss": 0.1108,
"step": 3840
},
{
"epoch": 7.04,
"learning_rate": 1.6453382084095066e-05,
"loss": 0.1146,
"step": 3850
},
{
"epoch": 7.06,
"learning_rate": 1.6351817997156206e-05,
"loss": 0.0947,
"step": 3860
},
{
"epoch": 7.07,
"learning_rate": 1.6250253910217347e-05,
"loss": 0.101,
"step": 3870
},
{
"epoch": 7.09,
"learning_rate": 1.614868982327849e-05,
"loss": 0.114,
"step": 3880
},
{
"epoch": 7.11,
"learning_rate": 1.6047125736339628e-05,
"loss": 0.0816,
"step": 3890
},
{
"epoch": 7.13,
"learning_rate": 1.5945561649400772e-05,
"loss": 0.1092,
"step": 3900
},
{
"epoch": 7.15,
"learning_rate": 1.5843997562461916e-05,
"loss": 0.1109,
"step": 3910
},
{
"epoch": 7.17,
"learning_rate": 1.5742433475523054e-05,
"loss": 0.1223,
"step": 3920
},
{
"epoch": 7.18,
"learning_rate": 1.5640869388584198e-05,
"loss": 0.0923,
"step": 3930
},
{
"epoch": 7.2,
"learning_rate": 1.553930530164534e-05,
"loss": 0.0975,
"step": 3940
},
{
"epoch": 7.22,
"learning_rate": 1.543774121470648e-05,
"loss": 0.1179,
"step": 3950
},
{
"epoch": 7.24,
"learning_rate": 1.5336177127767623e-05,
"loss": 0.1035,
"step": 3960
},
{
"epoch": 7.26,
"learning_rate": 1.5234613040828762e-05,
"loss": 0.1164,
"step": 3970
},
{
"epoch": 7.28,
"learning_rate": 1.5133048953889906e-05,
"loss": 0.0975,
"step": 3980
},
{
"epoch": 7.29,
"learning_rate": 1.5031484866951045e-05,
"loss": 0.1078,
"step": 3990
},
{
"epoch": 7.31,
"learning_rate": 1.4929920780012188e-05,
"loss": 0.1012,
"step": 4000
},
{
"epoch": 7.33,
"learning_rate": 1.4828356693073332e-05,
"loss": 0.0958,
"step": 4010
},
{
"epoch": 7.35,
"learning_rate": 1.472679260613447e-05,
"loss": 0.1068,
"step": 4020
},
{
"epoch": 7.37,
"learning_rate": 1.4625228519195613e-05,
"loss": 0.0858,
"step": 4030
},
{
"epoch": 7.39,
"learning_rate": 1.4523664432256754e-05,
"loss": 0.0975,
"step": 4040
},
{
"epoch": 7.4,
"learning_rate": 1.4422100345317896e-05,
"loss": 0.097,
"step": 4050
},
{
"epoch": 7.42,
"learning_rate": 1.4320536258379038e-05,
"loss": 0.0943,
"step": 4060
},
{
"epoch": 7.44,
"learning_rate": 1.4218972171440179e-05,
"loss": 0.0925,
"step": 4070
},
{
"epoch": 7.46,
"learning_rate": 1.4117408084501321e-05,
"loss": 0.109,
"step": 4080
},
{
"epoch": 7.48,
"learning_rate": 1.4015843997562464e-05,
"loss": 0.0927,
"step": 4090
},
{
"epoch": 7.5,
"learning_rate": 1.3914279910623604e-05,
"loss": 0.1082,
"step": 4100
},
{
"epoch": 7.51,
"learning_rate": 1.3812715823684747e-05,
"loss": 0.1114,
"step": 4110
},
{
"epoch": 7.53,
"learning_rate": 1.3711151736745886e-05,
"loss": 0.1156,
"step": 4120
},
{
"epoch": 7.55,
"learning_rate": 1.360958764980703e-05,
"loss": 0.0884,
"step": 4130
},
{
"epoch": 7.57,
"learning_rate": 1.3508023562868172e-05,
"loss": 0.0813,
"step": 4140
},
{
"epoch": 7.59,
"learning_rate": 1.3406459475929311e-05,
"loss": 0.0929,
"step": 4150
},
{
"epoch": 7.61,
"learning_rate": 1.3304895388990453e-05,
"loss": 0.0768,
"step": 4160
},
{
"epoch": 7.62,
"learning_rate": 1.3203331302051594e-05,
"loss": 0.0982,
"step": 4170
},
{
"epoch": 7.64,
"learning_rate": 1.3101767215112737e-05,
"loss": 0.0945,
"step": 4180
},
{
"epoch": 7.66,
"learning_rate": 1.3000203128173879e-05,
"loss": 0.1056,
"step": 4190
},
{
"epoch": 7.68,
"learning_rate": 1.289863904123502e-05,
"loss": 0.1006,
"step": 4200
},
{
"epoch": 7.7,
"learning_rate": 1.2797074954296162e-05,
"loss": 0.0994,
"step": 4210
},
{
"epoch": 7.71,
"learning_rate": 1.2695510867357303e-05,
"loss": 0.1097,
"step": 4220
},
{
"epoch": 7.73,
"learning_rate": 1.2593946780418445e-05,
"loss": 0.0899,
"step": 4230
},
{
"epoch": 7.75,
"learning_rate": 1.2492382693479586e-05,
"loss": 0.0872,
"step": 4240
},
{
"epoch": 7.77,
"learning_rate": 1.2390818606540728e-05,
"loss": 0.1199,
"step": 4250
},
{
"epoch": 7.79,
"learning_rate": 1.228925451960187e-05,
"loss": 0.1142,
"step": 4260
},
{
"epoch": 7.81,
"learning_rate": 1.2187690432663011e-05,
"loss": 0.1281,
"step": 4270
},
{
"epoch": 7.82,
"learning_rate": 1.2086126345724152e-05,
"loss": 0.1163,
"step": 4280
},
{
"epoch": 7.84,
"learning_rate": 1.1984562258785294e-05,
"loss": 0.0965,
"step": 4290
},
{
"epoch": 7.86,
"learning_rate": 1.1882998171846435e-05,
"loss": 0.1009,
"step": 4300
},
{
"epoch": 7.88,
"learning_rate": 1.1781434084907577e-05,
"loss": 0.0951,
"step": 4310
},
{
"epoch": 7.9,
"learning_rate": 1.167986999796872e-05,
"loss": 0.093,
"step": 4320
},
{
"epoch": 7.92,
"learning_rate": 1.157830591102986e-05,
"loss": 0.0951,
"step": 4330
},
{
"epoch": 7.93,
"learning_rate": 1.1476741824091e-05,
"loss": 0.0957,
"step": 4340
},
{
"epoch": 7.95,
"learning_rate": 1.1375177737152143e-05,
"loss": 0.09,
"step": 4350
},
{
"epoch": 7.97,
"learning_rate": 1.1273613650213286e-05,
"loss": 0.1188,
"step": 4360
},
{
"epoch": 7.99,
"learning_rate": 1.1172049563274426e-05,
"loss": 0.1079,
"step": 4370
},
{
"epoch": 8.0,
"eval_accuracy": 0.9829,
"eval_loss": 0.045376840978860855,
"eval_runtime": 107.845,
"eval_samples_per_second": 278.177,
"eval_steps_per_second": 8.698,
"step": 4376
},
{
"epoch": 8.01,
"learning_rate": 1.1070485476335569e-05,
"loss": 0.0779,
"step": 4380
},
{
"epoch": 8.03,
"learning_rate": 1.096892138939671e-05,
"loss": 0.1101,
"step": 4390
},
{
"epoch": 8.04,
"learning_rate": 1.0867357302457852e-05,
"loss": 0.0848,
"step": 4400
},
{
"epoch": 8.06,
"learning_rate": 1.0765793215518994e-05,
"loss": 0.0859,
"step": 4410
},
{
"epoch": 8.08,
"learning_rate": 1.0664229128580135e-05,
"loss": 0.0805,
"step": 4420
},
{
"epoch": 8.1,
"learning_rate": 1.0562665041641275e-05,
"loss": 0.0838,
"step": 4430
},
{
"epoch": 8.12,
"learning_rate": 1.0461100954702418e-05,
"loss": 0.1039,
"step": 4440
},
{
"epoch": 8.14,
"learning_rate": 1.035953686776356e-05,
"loss": 0.0895,
"step": 4450
},
{
"epoch": 8.15,
"learning_rate": 1.02579727808247e-05,
"loss": 0.095,
"step": 4460
},
{
"epoch": 8.17,
"learning_rate": 1.0156408693885843e-05,
"loss": 0.0947,
"step": 4470
},
{
"epoch": 8.19,
"learning_rate": 1.0054844606946984e-05,
"loss": 0.0817,
"step": 4480
},
{
"epoch": 8.21,
"learning_rate": 9.953280520008124e-06,
"loss": 0.102,
"step": 4490
},
{
"epoch": 8.23,
"learning_rate": 9.851716433069269e-06,
"loss": 0.0909,
"step": 4500
},
{
"epoch": 8.24,
"learning_rate": 9.75015234613041e-06,
"loss": 0.0908,
"step": 4510
},
{
"epoch": 8.26,
"learning_rate": 9.64858825919155e-06,
"loss": 0.1009,
"step": 4520
},
{
"epoch": 8.28,
"learning_rate": 9.547024172252692e-06,
"loss": 0.0984,
"step": 4530
},
{
"epoch": 8.3,
"learning_rate": 9.445460085313833e-06,
"loss": 0.0914,
"step": 4540
},
{
"epoch": 8.32,
"learning_rate": 9.343895998374975e-06,
"loss": 0.0986,
"step": 4550
},
{
"epoch": 8.34,
"learning_rate": 9.242331911436116e-06,
"loss": 0.0975,
"step": 4560
},
{
"epoch": 8.35,
"learning_rate": 9.140767824497258e-06,
"loss": 0.1058,
"step": 4570
},
{
"epoch": 8.37,
"learning_rate": 9.039203737558399e-06,
"loss": 0.0891,
"step": 4580
},
{
"epoch": 8.39,
"learning_rate": 8.937639650619541e-06,
"loss": 0.0834,
"step": 4590
},
{
"epoch": 8.41,
"learning_rate": 8.836075563680684e-06,
"loss": 0.0894,
"step": 4600
},
{
"epoch": 8.43,
"learning_rate": 8.734511476741824e-06,
"loss": 0.102,
"step": 4610
},
{
"epoch": 8.45,
"learning_rate": 8.632947389802965e-06,
"loss": 0.0703,
"step": 4620
},
{
"epoch": 8.46,
"learning_rate": 8.531383302864107e-06,
"loss": 0.0821,
"step": 4630
},
{
"epoch": 8.48,
"learning_rate": 8.42981921592525e-06,
"loss": 0.093,
"step": 4640
},
{
"epoch": 8.5,
"learning_rate": 8.32825512898639e-06,
"loss": 0.0901,
"step": 4650
},
{
"epoch": 8.52,
"learning_rate": 8.226691042047533e-06,
"loss": 0.0958,
"step": 4660
},
{
"epoch": 8.54,
"learning_rate": 8.125126955108673e-06,
"loss": 0.0978,
"step": 4670
},
{
"epoch": 8.56,
"learning_rate": 8.023562868169814e-06,
"loss": 0.1037,
"step": 4680
},
{
"epoch": 8.57,
"learning_rate": 7.921998781230958e-06,
"loss": 0.0704,
"step": 4690
},
{
"epoch": 8.59,
"learning_rate": 7.820434694292099e-06,
"loss": 0.097,
"step": 4700
},
{
"epoch": 8.61,
"learning_rate": 7.71887060735324e-06,
"loss": 0.0957,
"step": 4710
},
{
"epoch": 8.63,
"learning_rate": 7.617306520414381e-06,
"loss": 0.0784,
"step": 4720
},
{
"epoch": 8.65,
"learning_rate": 7.5157424334755226e-06,
"loss": 0.0959,
"step": 4730
},
{
"epoch": 8.67,
"learning_rate": 7.414178346536666e-06,
"loss": 0.1128,
"step": 4740
},
{
"epoch": 8.68,
"learning_rate": 7.3126142595978065e-06,
"loss": 0.0889,
"step": 4750
},
{
"epoch": 8.7,
"learning_rate": 7.211050172658948e-06,
"loss": 0.0939,
"step": 4760
},
{
"epoch": 8.72,
"learning_rate": 7.1094860857200895e-06,
"loss": 0.0994,
"step": 4770
},
{
"epoch": 8.74,
"learning_rate": 7.007921998781232e-06,
"loss": 0.1026,
"step": 4780
},
{
"epoch": 8.76,
"learning_rate": 6.906357911842373e-06,
"loss": 0.0855,
"step": 4790
},
{
"epoch": 8.78,
"learning_rate": 6.804793824903515e-06,
"loss": 0.097,
"step": 4800
},
{
"epoch": 8.79,
"learning_rate": 6.7032297379646556e-06,
"loss": 0.0877,
"step": 4810
},
{
"epoch": 8.81,
"learning_rate": 6.601665651025797e-06,
"loss": 0.1014,
"step": 4820
},
{
"epoch": 8.83,
"learning_rate": 6.5001015640869394e-06,
"loss": 0.0919,
"step": 4830
},
{
"epoch": 8.85,
"learning_rate": 6.398537477148081e-06,
"loss": 0.1037,
"step": 4840
},
{
"epoch": 8.87,
"learning_rate": 6.2969733902092225e-06,
"loss": 0.0964,
"step": 4850
},
{
"epoch": 8.88,
"learning_rate": 6.195409303270364e-06,
"loss": 0.0892,
"step": 4860
},
{
"epoch": 8.9,
"learning_rate": 6.0938452163315055e-06,
"loss": 0.1053,
"step": 4870
},
{
"epoch": 8.92,
"learning_rate": 5.992281129392647e-06,
"loss": 0.0824,
"step": 4880
},
{
"epoch": 8.94,
"learning_rate": 5.8907170424537886e-06,
"loss": 0.0837,
"step": 4890
},
{
"epoch": 8.96,
"learning_rate": 5.78915295551493e-06,
"loss": 0.0976,
"step": 4900
},
{
"epoch": 8.98,
"learning_rate": 5.687588868576072e-06,
"loss": 0.0895,
"step": 4910
},
{
"epoch": 8.99,
"learning_rate": 5.586024781637213e-06,
"loss": 0.0805,
"step": 4920
},
{
"epoch": 9.0,
"eval_accuracy": 0.9790333333333333,
"eval_loss": 0.05767909437417984,
"eval_runtime": 107.4183,
"eval_samples_per_second": 279.282,
"eval_steps_per_second": 8.732,
"step": 4923
},
{
"epoch": 9.01,
"learning_rate": 5.484460694698355e-06,
"loss": 0.084,
"step": 4930
},
{
"epoch": 9.03,
"learning_rate": 5.382896607759497e-06,
"loss": 0.083,
"step": 4940
},
{
"epoch": 9.05,
"learning_rate": 5.281332520820638e-06,
"loss": 0.0871,
"step": 4950
},
{
"epoch": 9.07,
"learning_rate": 5.17976843388178e-06,
"loss": 0.0793,
"step": 4960
},
{
"epoch": 9.09,
"learning_rate": 5.0782043469429216e-06,
"loss": 0.0813,
"step": 4970
},
{
"epoch": 9.1,
"learning_rate": 4.976640260004062e-06,
"loss": 0.0886,
"step": 4980
},
{
"epoch": 9.12,
"learning_rate": 4.875076173065205e-06,
"loss": 0.1115,
"step": 4990
},
{
"epoch": 9.14,
"learning_rate": 4.773512086126346e-06,
"loss": 0.0812,
"step": 5000
},
{
"epoch": 9.16,
"learning_rate": 4.671947999187488e-06,
"loss": 0.0768,
"step": 5010
},
{
"epoch": 9.18,
"learning_rate": 4.570383912248629e-06,
"loss": 0.0835,
"step": 5020
},
{
"epoch": 9.2,
"learning_rate": 4.468819825309771e-06,
"loss": 0.0817,
"step": 5030
},
{
"epoch": 9.21,
"learning_rate": 4.367255738370912e-06,
"loss": 0.0866,
"step": 5040
},
{
"epoch": 9.23,
"learning_rate": 4.265691651432054e-06,
"loss": 0.0742,
"step": 5050
},
{
"epoch": 9.25,
"learning_rate": 4.164127564493195e-06,
"loss": 0.0755,
"step": 5060
},
{
"epoch": 9.27,
"learning_rate": 4.062563477554337e-06,
"loss": 0.0804,
"step": 5070
},
{
"epoch": 9.29,
"learning_rate": 3.960999390615479e-06,
"loss": 0.0847,
"step": 5080
},
{
"epoch": 9.31,
"learning_rate": 3.85943530367662e-06,
"loss": 0.0821,
"step": 5090
},
{
"epoch": 9.32,
"learning_rate": 3.7578712167377613e-06,
"loss": 0.0707,
"step": 5100
},
{
"epoch": 9.34,
"learning_rate": 3.6563071297989032e-06,
"loss": 0.0957,
"step": 5110
},
{
"epoch": 9.36,
"learning_rate": 3.5547430428600447e-06,
"loss": 0.0795,
"step": 5120
},
{
"epoch": 9.38,
"learning_rate": 3.4531789559211867e-06,
"loss": 0.0887,
"step": 5130
},
{
"epoch": 9.4,
"learning_rate": 3.3516148689823278e-06,
"loss": 0.0752,
"step": 5140
},
{
"epoch": 9.41,
"learning_rate": 3.2500507820434697e-06,
"loss": 0.0869,
"step": 5150
},
{
"epoch": 9.43,
"learning_rate": 3.1484866951046112e-06,
"loss": 0.0769,
"step": 5160
},
{
"epoch": 9.45,
"learning_rate": 3.0469226081657528e-06,
"loss": 0.0965,
"step": 5170
},
{
"epoch": 9.47,
"learning_rate": 2.9453585212268943e-06,
"loss": 0.0832,
"step": 5180
},
{
"epoch": 9.49,
"learning_rate": 2.843794434288036e-06,
"loss": 0.1018,
"step": 5190
},
{
"epoch": 9.51,
"learning_rate": 2.7422303473491773e-06,
"loss": 0.0818,
"step": 5200
},
{
"epoch": 9.52,
"learning_rate": 2.640666260410319e-06,
"loss": 0.084,
"step": 5210
},
{
"epoch": 9.54,
"learning_rate": 2.5391021734714608e-06,
"loss": 0.0711,
"step": 5220
},
{
"epoch": 9.56,
"learning_rate": 2.4375380865326023e-06,
"loss": 0.0786,
"step": 5230
},
{
"epoch": 9.58,
"learning_rate": 2.335973999593744e-06,
"loss": 0.0875,
"step": 5240
},
{
"epoch": 9.6,
"learning_rate": 2.2344099126548853e-06,
"loss": 0.0948,
"step": 5250
},
{
"epoch": 9.62,
"learning_rate": 2.132845825716027e-06,
"loss": 0.0863,
"step": 5260
},
{
"epoch": 9.63,
"learning_rate": 2.0312817387771684e-06,
"loss": 0.0563,
"step": 5270
},
{
"epoch": 9.65,
"learning_rate": 1.92971765183831e-06,
"loss": 0.0836,
"step": 5280
},
{
"epoch": 9.67,
"learning_rate": 1.8281535648994516e-06,
"loss": 0.0869,
"step": 5290
},
{
"epoch": 9.69,
"learning_rate": 1.7265894779605933e-06,
"loss": 0.0961,
"step": 5300
},
{
"epoch": 9.71,
"learning_rate": 1.6250253910217349e-06,
"loss": 0.0811,
"step": 5310
},
{
"epoch": 9.73,
"learning_rate": 1.5234613040828764e-06,
"loss": 0.092,
"step": 5320
},
{
"epoch": 9.74,
"learning_rate": 1.421897217144018e-06,
"loss": 0.059,
"step": 5330
},
{
"epoch": 9.76,
"learning_rate": 1.3203331302051594e-06,
"loss": 0.0866,
"step": 5340
},
{
"epoch": 9.78,
"learning_rate": 1.2187690432663011e-06,
"loss": 0.075,
"step": 5350
},
{
"epoch": 9.8,
"learning_rate": 1.1172049563274427e-06,
"loss": 0.0818,
"step": 5360
},
{
"epoch": 9.82,
"learning_rate": 1.0156408693885842e-06,
"loss": 0.086,
"step": 5370
},
{
"epoch": 9.84,
"learning_rate": 9.140767824497258e-07,
"loss": 0.0767,
"step": 5380
},
{
"epoch": 9.85,
"learning_rate": 8.125126955108674e-07,
"loss": 0.0817,
"step": 5390
},
{
"epoch": 9.87,
"learning_rate": 7.10948608572009e-07,
"loss": 0.0831,
"step": 5400
},
{
"epoch": 9.89,
"learning_rate": 6.093845216331506e-07,
"loss": 0.0717,
"step": 5410
},
{
"epoch": 9.91,
"learning_rate": 5.078204346942921e-07,
"loss": 0.097,
"step": 5420
},
{
"epoch": 9.93,
"learning_rate": 4.062563477554337e-07,
"loss": 0.0781,
"step": 5430
},
{
"epoch": 9.95,
"learning_rate": 3.046922608165753e-07,
"loss": 0.0792,
"step": 5440
},
{
"epoch": 9.96,
"learning_rate": 2.0312817387771686e-07,
"loss": 0.0827,
"step": 5450
},
{
"epoch": 9.98,
"learning_rate": 1.0156408693885843e-07,
"loss": 0.1004,
"step": 5460
},
{
"epoch": 10.0,
"learning_rate": 0.0,
"loss": 0.0778,
"step": 5470
},
{
"epoch": 10.0,
"eval_accuracy": 0.9807,
"eval_loss": 0.05392194166779518,
"eval_runtime": 108.2183,
"eval_samples_per_second": 277.217,
"eval_steps_per_second": 8.668,
"step": 5470
},
{
"epoch": 10.0,
"step": 5470,
"total_flos": 1.73991922900992e+19,
"train_loss": 0.15263754843139998,
"train_runtime": 7208.4995,
"train_samples_per_second": 97.108,
"train_steps_per_second": 0.759
}
],
"max_steps": 5470,
"num_train_epochs": 10,
"total_flos": 1.73991922900992e+19,
"trial_name": null,
"trial_params": null
}