VIT_AI_image_detector / trainer_state.json
mmanikanta's picture
End of training
dc72c30
raw
history blame
No virus
52.3 kB
{
"best_metric": 0.9901,
"best_model_checkpoint": "VIT_AI_image_detector/checkpoint-4218",
"epoch": 2.9994666666666667,
"global_step": 4218,
"is_hyper_param_search": false,
"is_local_process_zero": true,
"is_world_process_zero": true,
"log_history": [
{
"epoch": 0.01,
"learning_rate": 1.1848341232227488e-06,
"loss": 0.6878,
"step": 10
},
{
"epoch": 0.01,
"learning_rate": 2.3696682464454976e-06,
"loss": 0.6907,
"step": 20
},
{
"epoch": 0.02,
"learning_rate": 3.5545023696682464e-06,
"loss": 0.6836,
"step": 30
},
{
"epoch": 0.03,
"learning_rate": 4.739336492890995e-06,
"loss": 0.6766,
"step": 40
},
{
"epoch": 0.04,
"learning_rate": 5.924170616113745e-06,
"loss": 0.6704,
"step": 50
},
{
"epoch": 0.04,
"learning_rate": 7.109004739336493e-06,
"loss": 0.6556,
"step": 60
},
{
"epoch": 0.05,
"learning_rate": 8.293838862559241e-06,
"loss": 0.6506,
"step": 70
},
{
"epoch": 0.06,
"learning_rate": 9.47867298578199e-06,
"loss": 0.6276,
"step": 80
},
{
"epoch": 0.06,
"learning_rate": 1.066350710900474e-05,
"loss": 0.5957,
"step": 90
},
{
"epoch": 0.07,
"learning_rate": 1.184834123222749e-05,
"loss": 0.5524,
"step": 100
},
{
"epoch": 0.08,
"learning_rate": 1.3033175355450238e-05,
"loss": 0.4691,
"step": 110
},
{
"epoch": 0.09,
"learning_rate": 1.4218009478672985e-05,
"loss": 0.4064,
"step": 120
},
{
"epoch": 0.09,
"learning_rate": 1.5402843601895736e-05,
"loss": 0.3614,
"step": 130
},
{
"epoch": 0.1,
"learning_rate": 1.6587677725118483e-05,
"loss": 0.3137,
"step": 140
},
{
"epoch": 0.11,
"learning_rate": 1.7772511848341233e-05,
"loss": 0.3232,
"step": 150
},
{
"epoch": 0.11,
"learning_rate": 1.895734597156398e-05,
"loss": 0.3303,
"step": 160
},
{
"epoch": 0.12,
"learning_rate": 2.014218009478673e-05,
"loss": 0.2869,
"step": 170
},
{
"epoch": 0.13,
"learning_rate": 2.132701421800948e-05,
"loss": 0.2903,
"step": 180
},
{
"epoch": 0.14,
"learning_rate": 2.251184834123223e-05,
"loss": 0.2726,
"step": 190
},
{
"epoch": 0.14,
"learning_rate": 2.369668246445498e-05,
"loss": 0.2885,
"step": 200
},
{
"epoch": 0.15,
"learning_rate": 2.4881516587677726e-05,
"loss": 0.2507,
"step": 210
},
{
"epoch": 0.16,
"learning_rate": 2.6066350710900477e-05,
"loss": 0.2451,
"step": 220
},
{
"epoch": 0.16,
"learning_rate": 2.7251184834123224e-05,
"loss": 0.239,
"step": 230
},
{
"epoch": 0.17,
"learning_rate": 2.843601895734597e-05,
"loss": 0.2621,
"step": 240
},
{
"epoch": 0.18,
"learning_rate": 2.962085308056872e-05,
"loss": 0.2509,
"step": 250
},
{
"epoch": 0.18,
"learning_rate": 3.080568720379147e-05,
"loss": 0.2168,
"step": 260
},
{
"epoch": 0.19,
"learning_rate": 3.1990521327014215e-05,
"loss": 0.2547,
"step": 270
},
{
"epoch": 0.2,
"learning_rate": 3.3175355450236966e-05,
"loss": 0.1996,
"step": 280
},
{
"epoch": 0.21,
"learning_rate": 3.4360189573459716e-05,
"loss": 0.2337,
"step": 290
},
{
"epoch": 0.21,
"learning_rate": 3.554502369668247e-05,
"loss": 0.2903,
"step": 300
},
{
"epoch": 0.22,
"learning_rate": 3.672985781990522e-05,
"loss": 0.2271,
"step": 310
},
{
"epoch": 0.23,
"learning_rate": 3.791469194312796e-05,
"loss": 0.2384,
"step": 320
},
{
"epoch": 0.23,
"learning_rate": 3.909952606635071e-05,
"loss": 0.2323,
"step": 330
},
{
"epoch": 0.24,
"learning_rate": 4.028436018957346e-05,
"loss": 0.208,
"step": 340
},
{
"epoch": 0.25,
"learning_rate": 4.146919431279621e-05,
"loss": 0.2671,
"step": 350
},
{
"epoch": 0.26,
"learning_rate": 4.265402843601896e-05,
"loss": 0.2282,
"step": 360
},
{
"epoch": 0.26,
"learning_rate": 4.383886255924171e-05,
"loss": 0.1993,
"step": 370
},
{
"epoch": 0.27,
"learning_rate": 4.502369668246446e-05,
"loss": 0.225,
"step": 380
},
{
"epoch": 0.28,
"learning_rate": 4.620853080568721e-05,
"loss": 0.2417,
"step": 390
},
{
"epoch": 0.28,
"learning_rate": 4.739336492890996e-05,
"loss": 0.1922,
"step": 400
},
{
"epoch": 0.29,
"learning_rate": 4.857819905213271e-05,
"loss": 0.208,
"step": 410
},
{
"epoch": 0.3,
"learning_rate": 4.976303317535545e-05,
"loss": 0.2107,
"step": 420
},
{
"epoch": 0.31,
"learning_rate": 4.989462592202318e-05,
"loss": 0.22,
"step": 430
},
{
"epoch": 0.31,
"learning_rate": 4.976290832455216e-05,
"loss": 0.2069,
"step": 440
},
{
"epoch": 0.32,
"learning_rate": 4.963119072708114e-05,
"loss": 0.1807,
"step": 450
},
{
"epoch": 0.33,
"learning_rate": 4.949947312961012e-05,
"loss": 0.1716,
"step": 460
},
{
"epoch": 0.33,
"learning_rate": 4.93677555321391e-05,
"loss": 0.2296,
"step": 470
},
{
"epoch": 0.34,
"learning_rate": 4.923603793466807e-05,
"loss": 0.2679,
"step": 480
},
{
"epoch": 0.35,
"learning_rate": 4.910432033719705e-05,
"loss": 0.2065,
"step": 490
},
{
"epoch": 0.36,
"learning_rate": 4.8972602739726034e-05,
"loss": 0.1917,
"step": 500
},
{
"epoch": 0.36,
"learning_rate": 4.8840885142255003e-05,
"loss": 0.2167,
"step": 510
},
{
"epoch": 0.37,
"learning_rate": 4.8709167544783986e-05,
"loss": 0.1904,
"step": 520
},
{
"epoch": 0.38,
"learning_rate": 4.857744994731296e-05,
"loss": 0.1902,
"step": 530
},
{
"epoch": 0.38,
"learning_rate": 4.8445732349841945e-05,
"loss": 0.2011,
"step": 540
},
{
"epoch": 0.39,
"learning_rate": 4.831401475237092e-05,
"loss": 0.2673,
"step": 550
},
{
"epoch": 0.4,
"learning_rate": 4.8182297154899896e-05,
"loss": 0.232,
"step": 560
},
{
"epoch": 0.41,
"learning_rate": 4.805057955742888e-05,
"loss": 0.2093,
"step": 570
},
{
"epoch": 0.41,
"learning_rate": 4.791886195995785e-05,
"loss": 0.182,
"step": 580
},
{
"epoch": 0.42,
"learning_rate": 4.778714436248683e-05,
"loss": 0.194,
"step": 590
},
{
"epoch": 0.43,
"learning_rate": 4.7655426765015806e-05,
"loss": 0.2216,
"step": 600
},
{
"epoch": 0.43,
"learning_rate": 4.752370916754479e-05,
"loss": 0.1712,
"step": 610
},
{
"epoch": 0.44,
"learning_rate": 4.7391991570073765e-05,
"loss": 0.207,
"step": 620
},
{
"epoch": 0.45,
"learning_rate": 4.726027397260274e-05,
"loss": 0.1605,
"step": 630
},
{
"epoch": 0.46,
"learning_rate": 4.712855637513172e-05,
"loss": 0.1532,
"step": 640
},
{
"epoch": 0.46,
"learning_rate": 4.699683877766069e-05,
"loss": 0.2005,
"step": 650
},
{
"epoch": 0.47,
"learning_rate": 4.6865121180189675e-05,
"loss": 0.1555,
"step": 660
},
{
"epoch": 0.48,
"learning_rate": 4.673340358271866e-05,
"loss": 0.1744,
"step": 670
},
{
"epoch": 0.48,
"learning_rate": 4.6601685985247633e-05,
"loss": 0.1753,
"step": 680
},
{
"epoch": 0.49,
"learning_rate": 4.646996838777661e-05,
"loss": 0.1799,
"step": 690
},
{
"epoch": 0.5,
"learning_rate": 4.6338250790305585e-05,
"loss": 0.1929,
"step": 700
},
{
"epoch": 0.5,
"learning_rate": 4.620653319283457e-05,
"loss": 0.246,
"step": 710
},
{
"epoch": 0.51,
"learning_rate": 4.6074815595363544e-05,
"loss": 0.1788,
"step": 720
},
{
"epoch": 0.52,
"learning_rate": 4.594309799789252e-05,
"loss": 0.1918,
"step": 730
},
{
"epoch": 0.53,
"learning_rate": 4.58113804004215e-05,
"loss": 0.1817,
"step": 740
},
{
"epoch": 0.53,
"learning_rate": 4.567966280295047e-05,
"loss": 0.1969,
"step": 750
},
{
"epoch": 0.54,
"learning_rate": 4.5547945205479454e-05,
"loss": 0.1572,
"step": 760
},
{
"epoch": 0.55,
"learning_rate": 4.541622760800843e-05,
"loss": 0.1323,
"step": 770
},
{
"epoch": 0.55,
"learning_rate": 4.528451001053741e-05,
"loss": 0.1533,
"step": 780
},
{
"epoch": 0.56,
"learning_rate": 4.515279241306639e-05,
"loss": 0.1593,
"step": 790
},
{
"epoch": 0.57,
"learning_rate": 4.5021074815595364e-05,
"loss": 0.2544,
"step": 800
},
{
"epoch": 0.58,
"learning_rate": 4.488935721812435e-05,
"loss": 0.2024,
"step": 810
},
{
"epoch": 0.58,
"learning_rate": 4.4757639620653316e-05,
"loss": 0.1798,
"step": 820
},
{
"epoch": 0.59,
"learning_rate": 4.46259220231823e-05,
"loss": 0.1406,
"step": 830
},
{
"epoch": 0.6,
"learning_rate": 4.449420442571128e-05,
"loss": 0.1335,
"step": 840
},
{
"epoch": 0.6,
"learning_rate": 4.436248682824026e-05,
"loss": 0.1829,
"step": 850
},
{
"epoch": 0.61,
"learning_rate": 4.423076923076923e-05,
"loss": 0.1353,
"step": 860
},
{
"epoch": 0.62,
"learning_rate": 4.409905163329821e-05,
"loss": 0.1598,
"step": 870
},
{
"epoch": 0.63,
"learning_rate": 4.396733403582719e-05,
"loss": 0.1518,
"step": 880
},
{
"epoch": 0.63,
"learning_rate": 4.383561643835617e-05,
"loss": 0.1554,
"step": 890
},
{
"epoch": 0.64,
"learning_rate": 4.370389884088514e-05,
"loss": 0.1387,
"step": 900
},
{
"epoch": 0.65,
"learning_rate": 4.3572181243414125e-05,
"loss": 0.1769,
"step": 910
},
{
"epoch": 0.65,
"learning_rate": 4.34404636459431e-05,
"loss": 0.1697,
"step": 920
},
{
"epoch": 0.66,
"learning_rate": 4.330874604847208e-05,
"loss": 0.1711,
"step": 930
},
{
"epoch": 0.67,
"learning_rate": 4.317702845100105e-05,
"loss": 0.1842,
"step": 940
},
{
"epoch": 0.68,
"learning_rate": 4.3045310853530036e-05,
"loss": 0.1413,
"step": 950
},
{
"epoch": 0.68,
"learning_rate": 4.291359325605901e-05,
"loss": 0.1799,
"step": 960
},
{
"epoch": 0.69,
"learning_rate": 4.278187565858799e-05,
"loss": 0.1479,
"step": 970
},
{
"epoch": 0.7,
"learning_rate": 4.265015806111697e-05,
"loss": 0.1873,
"step": 980
},
{
"epoch": 0.7,
"learning_rate": 4.2518440463645946e-05,
"loss": 0.1741,
"step": 990
},
{
"epoch": 0.71,
"learning_rate": 4.238672286617492e-05,
"loss": 0.1262,
"step": 1000
},
{
"epoch": 0.72,
"learning_rate": 4.2255005268703904e-05,
"loss": 0.1856,
"step": 1010
},
{
"epoch": 0.73,
"learning_rate": 4.212328767123288e-05,
"loss": 0.1391,
"step": 1020
},
{
"epoch": 0.73,
"learning_rate": 4.1991570073761856e-05,
"loss": 0.1748,
"step": 1030
},
{
"epoch": 0.74,
"learning_rate": 4.185985247629083e-05,
"loss": 0.1404,
"step": 1040
},
{
"epoch": 0.75,
"learning_rate": 4.1728134878819814e-05,
"loss": 0.1151,
"step": 1050
},
{
"epoch": 0.75,
"learning_rate": 4.159641728134879e-05,
"loss": 0.192,
"step": 1060
},
{
"epoch": 0.76,
"learning_rate": 4.1464699683877766e-05,
"loss": 0.1731,
"step": 1070
},
{
"epoch": 0.77,
"learning_rate": 4.133298208640675e-05,
"loss": 0.1551,
"step": 1080
},
{
"epoch": 0.78,
"learning_rate": 4.1201264488935725e-05,
"loss": 0.1545,
"step": 1090
},
{
"epoch": 0.78,
"learning_rate": 4.10695468914647e-05,
"loss": 0.1238,
"step": 1100
},
{
"epoch": 0.79,
"learning_rate": 4.0937829293993676e-05,
"loss": 0.1036,
"step": 1110
},
{
"epoch": 0.8,
"learning_rate": 4.080611169652266e-05,
"loss": 0.173,
"step": 1120
},
{
"epoch": 0.8,
"learning_rate": 4.0674394099051635e-05,
"loss": 0.123,
"step": 1130
},
{
"epoch": 0.81,
"learning_rate": 4.054267650158061e-05,
"loss": 0.1379,
"step": 1140
},
{
"epoch": 0.82,
"learning_rate": 4.041095890410959e-05,
"loss": 0.1605,
"step": 1150
},
{
"epoch": 0.82,
"learning_rate": 4.027924130663857e-05,
"loss": 0.1423,
"step": 1160
},
{
"epoch": 0.83,
"learning_rate": 4.0147523709167545e-05,
"loss": 0.1369,
"step": 1170
},
{
"epoch": 0.84,
"learning_rate": 4.001580611169653e-05,
"loss": 0.129,
"step": 1180
},
{
"epoch": 0.85,
"learning_rate": 3.98840885142255e-05,
"loss": 0.1163,
"step": 1190
},
{
"epoch": 0.85,
"learning_rate": 3.975237091675448e-05,
"loss": 0.1269,
"step": 1200
},
{
"epoch": 0.86,
"learning_rate": 3.9620653319283455e-05,
"loss": 0.1513,
"step": 1210
},
{
"epoch": 0.87,
"learning_rate": 3.948893572181244e-05,
"loss": 0.1355,
"step": 1220
},
{
"epoch": 0.87,
"learning_rate": 3.9357218124341414e-05,
"loss": 0.1308,
"step": 1230
},
{
"epoch": 0.88,
"learning_rate": 3.922550052687039e-05,
"loss": 0.1474,
"step": 1240
},
{
"epoch": 0.89,
"learning_rate": 3.909378292939937e-05,
"loss": 0.1599,
"step": 1250
},
{
"epoch": 0.9,
"learning_rate": 3.896206533192835e-05,
"loss": 0.1383,
"step": 1260
},
{
"epoch": 0.9,
"learning_rate": 3.8830347734457324e-05,
"loss": 0.1245,
"step": 1270
},
{
"epoch": 0.91,
"learning_rate": 3.86986301369863e-05,
"loss": 0.1458,
"step": 1280
},
{
"epoch": 0.92,
"learning_rate": 3.856691253951528e-05,
"loss": 0.1361,
"step": 1290
},
{
"epoch": 0.92,
"learning_rate": 3.843519494204426e-05,
"loss": 0.1498,
"step": 1300
},
{
"epoch": 0.93,
"learning_rate": 3.8303477344573234e-05,
"loss": 0.1605,
"step": 1310
},
{
"epoch": 0.94,
"learning_rate": 3.8171759747102217e-05,
"loss": 0.1485,
"step": 1320
},
{
"epoch": 0.95,
"learning_rate": 3.804004214963119e-05,
"loss": 0.1245,
"step": 1330
},
{
"epoch": 0.95,
"learning_rate": 3.790832455216017e-05,
"loss": 0.149,
"step": 1340
},
{
"epoch": 0.96,
"learning_rate": 3.777660695468915e-05,
"loss": 0.1138,
"step": 1350
},
{
"epoch": 0.97,
"learning_rate": 3.764488935721813e-05,
"loss": 0.1203,
"step": 1360
},
{
"epoch": 0.97,
"learning_rate": 3.751317175974711e-05,
"loss": 0.1144,
"step": 1370
},
{
"epoch": 0.98,
"learning_rate": 3.738145416227608e-05,
"loss": 0.1451,
"step": 1380
},
{
"epoch": 0.99,
"learning_rate": 3.724973656480506e-05,
"loss": 0.1256,
"step": 1390
},
{
"epoch": 1.0,
"learning_rate": 3.711801896733404e-05,
"loss": 0.1624,
"step": 1400
},
{
"epoch": 1.0,
"eval_accuracy": 0.9799,
"eval_loss": 0.05523204430937767,
"eval_runtime": 133.9464,
"eval_samples_per_second": 74.657,
"eval_steps_per_second": 4.666,
"step": 1406
},
{
"epoch": 1.0,
"learning_rate": 3.698630136986301e-05,
"loss": 0.1724,
"step": 1410
},
{
"epoch": 1.01,
"learning_rate": 3.6854583772391995e-05,
"loss": 0.1486,
"step": 1420
},
{
"epoch": 1.02,
"learning_rate": 3.672286617492097e-05,
"loss": 0.1642,
"step": 1430
},
{
"epoch": 1.02,
"learning_rate": 3.6591148577449954e-05,
"loss": 0.1472,
"step": 1440
},
{
"epoch": 1.03,
"learning_rate": 3.645943097997892e-05,
"loss": 0.118,
"step": 1450
},
{
"epoch": 1.04,
"learning_rate": 3.6327713382507905e-05,
"loss": 0.1535,
"step": 1460
},
{
"epoch": 1.05,
"learning_rate": 3.619599578503688e-05,
"loss": 0.1396,
"step": 1470
},
{
"epoch": 1.05,
"learning_rate": 3.606427818756586e-05,
"loss": 0.1082,
"step": 1480
},
{
"epoch": 1.06,
"learning_rate": 3.593256059009484e-05,
"loss": 0.0991,
"step": 1490
},
{
"epoch": 1.07,
"learning_rate": 3.5800842992623816e-05,
"loss": 0.1091,
"step": 1500
},
{
"epoch": 1.07,
"learning_rate": 3.566912539515279e-05,
"loss": 0.1155,
"step": 1510
},
{
"epoch": 1.08,
"learning_rate": 3.5537407797681774e-05,
"loss": 0.1383,
"step": 1520
},
{
"epoch": 1.09,
"learning_rate": 3.540569020021075e-05,
"loss": 0.0968,
"step": 1530
},
{
"epoch": 1.1,
"learning_rate": 3.527397260273973e-05,
"loss": 0.1082,
"step": 1540
},
{
"epoch": 1.1,
"learning_rate": 3.51422550052687e-05,
"loss": 0.1382,
"step": 1550
},
{
"epoch": 1.11,
"learning_rate": 3.5010537407797684e-05,
"loss": 0.1374,
"step": 1560
},
{
"epoch": 1.12,
"learning_rate": 3.487881981032666e-05,
"loss": 0.1468,
"step": 1570
},
{
"epoch": 1.12,
"learning_rate": 3.4747102212855636e-05,
"loss": 0.1012,
"step": 1580
},
{
"epoch": 1.13,
"learning_rate": 3.461538461538462e-05,
"loss": 0.1253,
"step": 1590
},
{
"epoch": 1.14,
"learning_rate": 3.4483667017913594e-05,
"loss": 0.1533,
"step": 1600
},
{
"epoch": 1.14,
"learning_rate": 3.435194942044258e-05,
"loss": 0.1365,
"step": 1610
},
{
"epoch": 1.15,
"learning_rate": 3.4220231822971546e-05,
"loss": 0.1451,
"step": 1620
},
{
"epoch": 1.16,
"learning_rate": 3.408851422550053e-05,
"loss": 0.1192,
"step": 1630
},
{
"epoch": 1.17,
"learning_rate": 3.3956796628029505e-05,
"loss": 0.1206,
"step": 1640
},
{
"epoch": 1.17,
"learning_rate": 3.382507903055848e-05,
"loss": 0.1222,
"step": 1650
},
{
"epoch": 1.18,
"learning_rate": 3.369336143308746e-05,
"loss": 0.1068,
"step": 1660
},
{
"epoch": 1.19,
"learning_rate": 3.356164383561644e-05,
"loss": 0.1541,
"step": 1670
},
{
"epoch": 1.19,
"learning_rate": 3.342992623814542e-05,
"loss": 0.1448,
"step": 1680
},
{
"epoch": 1.2,
"learning_rate": 3.32982086406744e-05,
"loss": 0.1205,
"step": 1690
},
{
"epoch": 1.21,
"learning_rate": 3.316649104320337e-05,
"loss": 0.1337,
"step": 1700
},
{
"epoch": 1.22,
"learning_rate": 3.3034773445732356e-05,
"loss": 0.1059,
"step": 1710
},
{
"epoch": 1.22,
"learning_rate": 3.2903055848261325e-05,
"loss": 0.1046,
"step": 1720
},
{
"epoch": 1.23,
"learning_rate": 3.277133825079031e-05,
"loss": 0.1184,
"step": 1730
},
{
"epoch": 1.24,
"learning_rate": 3.2639620653319283e-05,
"loss": 0.1183,
"step": 1740
},
{
"epoch": 1.24,
"learning_rate": 3.2507903055848266e-05,
"loss": 0.0955,
"step": 1750
},
{
"epoch": 1.25,
"learning_rate": 3.237618545837724e-05,
"loss": 0.1096,
"step": 1760
},
{
"epoch": 1.26,
"learning_rate": 3.224446786090622e-05,
"loss": 0.0932,
"step": 1770
},
{
"epoch": 1.27,
"learning_rate": 3.21127502634352e-05,
"loss": 0.1468,
"step": 1780
},
{
"epoch": 1.27,
"learning_rate": 3.198103266596417e-05,
"loss": 0.1202,
"step": 1790
},
{
"epoch": 1.28,
"learning_rate": 3.184931506849315e-05,
"loss": 0.1458,
"step": 1800
},
{
"epoch": 1.29,
"learning_rate": 3.171759747102213e-05,
"loss": 0.1065,
"step": 1810
},
{
"epoch": 1.29,
"learning_rate": 3.1585879873551104e-05,
"loss": 0.1114,
"step": 1820
},
{
"epoch": 1.3,
"learning_rate": 3.1454162276080086e-05,
"loss": 0.1282,
"step": 1830
},
{
"epoch": 1.31,
"learning_rate": 3.132244467860906e-05,
"loss": 0.1271,
"step": 1840
},
{
"epoch": 1.32,
"learning_rate": 3.1190727081138045e-05,
"loss": 0.111,
"step": 1850
},
{
"epoch": 1.32,
"learning_rate": 3.105900948366702e-05,
"loss": 0.1371,
"step": 1860
},
{
"epoch": 1.33,
"learning_rate": 3.0927291886195997e-05,
"loss": 0.1111,
"step": 1870
},
{
"epoch": 1.34,
"learning_rate": 3.079557428872498e-05,
"loss": 0.1484,
"step": 1880
},
{
"epoch": 1.34,
"learning_rate": 3.066385669125395e-05,
"loss": 0.1045,
"step": 1890
},
{
"epoch": 1.35,
"learning_rate": 3.053213909378293e-05,
"loss": 0.1128,
"step": 1900
},
{
"epoch": 1.36,
"learning_rate": 3.040042149631191e-05,
"loss": 0.1391,
"step": 1910
},
{
"epoch": 1.37,
"learning_rate": 3.026870389884089e-05,
"loss": 0.1183,
"step": 1920
},
{
"epoch": 1.37,
"learning_rate": 3.0136986301369862e-05,
"loss": 0.0963,
"step": 1930
},
{
"epoch": 1.38,
"learning_rate": 3.000526870389884e-05,
"loss": 0.0996,
"step": 1940
},
{
"epoch": 1.39,
"learning_rate": 2.987355110642782e-05,
"loss": 0.106,
"step": 1950
},
{
"epoch": 1.39,
"learning_rate": 2.9741833508956796e-05,
"loss": 0.138,
"step": 1960
},
{
"epoch": 1.4,
"learning_rate": 2.9610115911485775e-05,
"loss": 0.1299,
"step": 1970
},
{
"epoch": 1.41,
"learning_rate": 2.9478398314014755e-05,
"loss": 0.1163,
"step": 1980
},
{
"epoch": 1.42,
"learning_rate": 2.9346680716543734e-05,
"loss": 0.0867,
"step": 1990
},
{
"epoch": 1.42,
"learning_rate": 2.921496311907271e-05,
"loss": 0.1192,
"step": 2000
},
{
"epoch": 1.43,
"learning_rate": 2.908324552160169e-05,
"loss": 0.0969,
"step": 2010
},
{
"epoch": 1.44,
"learning_rate": 2.8951527924130668e-05,
"loss": 0.1388,
"step": 2020
},
{
"epoch": 1.44,
"learning_rate": 2.881981032665964e-05,
"loss": 0.1036,
"step": 2030
},
{
"epoch": 1.45,
"learning_rate": 2.868809272918862e-05,
"loss": 0.09,
"step": 2040
},
{
"epoch": 1.46,
"learning_rate": 2.85563751317176e-05,
"loss": 0.1042,
"step": 2050
},
{
"epoch": 1.46,
"learning_rate": 2.842465753424658e-05,
"loss": 0.1187,
"step": 2060
},
{
"epoch": 1.47,
"learning_rate": 2.8292939936775554e-05,
"loss": 0.1013,
"step": 2070
},
{
"epoch": 1.48,
"learning_rate": 2.8161222339304533e-05,
"loss": 0.1275,
"step": 2080
},
{
"epoch": 1.49,
"learning_rate": 2.8029504741833513e-05,
"loss": 0.0892,
"step": 2090
},
{
"epoch": 1.49,
"learning_rate": 2.7897787144362485e-05,
"loss": 0.1066,
"step": 2100
},
{
"epoch": 1.5,
"learning_rate": 2.7766069546891464e-05,
"loss": 0.096,
"step": 2110
},
{
"epoch": 1.51,
"learning_rate": 2.7634351949420444e-05,
"loss": 0.1263,
"step": 2120
},
{
"epoch": 1.51,
"learning_rate": 2.7502634351949423e-05,
"loss": 0.118,
"step": 2130
},
{
"epoch": 1.52,
"learning_rate": 2.73709167544784e-05,
"loss": 0.1079,
"step": 2140
},
{
"epoch": 1.53,
"learning_rate": 2.7239199157007378e-05,
"loss": 0.1137,
"step": 2150
},
{
"epoch": 1.54,
"learning_rate": 2.7107481559536357e-05,
"loss": 0.1092,
"step": 2160
},
{
"epoch": 1.54,
"learning_rate": 2.6975763962065333e-05,
"loss": 0.1195,
"step": 2170
},
{
"epoch": 1.55,
"learning_rate": 2.6844046364594312e-05,
"loss": 0.1131,
"step": 2180
},
{
"epoch": 1.56,
"learning_rate": 2.671232876712329e-05,
"loss": 0.1363,
"step": 2190
},
{
"epoch": 1.56,
"learning_rate": 2.6580611169652264e-05,
"loss": 0.13,
"step": 2200
},
{
"epoch": 1.57,
"learning_rate": 2.6448893572181243e-05,
"loss": 0.1152,
"step": 2210
},
{
"epoch": 1.58,
"learning_rate": 2.6317175974710222e-05,
"loss": 0.1155,
"step": 2220
},
{
"epoch": 1.59,
"learning_rate": 2.61854583772392e-05,
"loss": 0.1159,
"step": 2230
},
{
"epoch": 1.59,
"learning_rate": 2.6053740779768177e-05,
"loss": 0.1146,
"step": 2240
},
{
"epoch": 1.6,
"learning_rate": 2.5922023182297157e-05,
"loss": 0.1176,
"step": 2250
},
{
"epoch": 1.61,
"learning_rate": 2.5790305584826136e-05,
"loss": 0.1327,
"step": 2260
},
{
"epoch": 1.61,
"learning_rate": 2.565858798735511e-05,
"loss": 0.0938,
"step": 2270
},
{
"epoch": 1.62,
"learning_rate": 2.5526870389884088e-05,
"loss": 0.1234,
"step": 2280
},
{
"epoch": 1.63,
"learning_rate": 2.5395152792413067e-05,
"loss": 0.1329,
"step": 2290
},
{
"epoch": 1.64,
"learning_rate": 2.5263435194942046e-05,
"loss": 0.1093,
"step": 2300
},
{
"epoch": 1.64,
"learning_rate": 2.5131717597471022e-05,
"loss": 0.1051,
"step": 2310
},
{
"epoch": 1.65,
"learning_rate": 2.5e-05,
"loss": 0.1022,
"step": 2320
},
{
"epoch": 1.66,
"learning_rate": 2.4868282402528977e-05,
"loss": 0.1071,
"step": 2330
},
{
"epoch": 1.66,
"learning_rate": 2.4736564805057956e-05,
"loss": 0.1097,
"step": 2340
},
{
"epoch": 1.67,
"learning_rate": 2.4604847207586936e-05,
"loss": 0.0796,
"step": 2350
},
{
"epoch": 1.68,
"learning_rate": 2.4473129610115915e-05,
"loss": 0.0795,
"step": 2360
},
{
"epoch": 1.69,
"learning_rate": 2.434141201264489e-05,
"loss": 0.0999,
"step": 2370
},
{
"epoch": 1.69,
"learning_rate": 2.420969441517387e-05,
"loss": 0.1133,
"step": 2380
},
{
"epoch": 1.7,
"learning_rate": 2.4077976817702846e-05,
"loss": 0.0935,
"step": 2390
},
{
"epoch": 1.71,
"learning_rate": 2.394625922023182e-05,
"loss": 0.0992,
"step": 2400
},
{
"epoch": 1.71,
"learning_rate": 2.3814541622760804e-05,
"loss": 0.1119,
"step": 2410
},
{
"epoch": 1.72,
"learning_rate": 2.368282402528978e-05,
"loss": 0.1198,
"step": 2420
},
{
"epoch": 1.73,
"learning_rate": 2.355110642781876e-05,
"loss": 0.0977,
"step": 2430
},
{
"epoch": 1.74,
"learning_rate": 2.3419388830347735e-05,
"loss": 0.1244,
"step": 2440
},
{
"epoch": 1.74,
"learning_rate": 2.328767123287671e-05,
"loss": 0.1181,
"step": 2450
},
{
"epoch": 1.75,
"learning_rate": 2.315595363540569e-05,
"loss": 0.1042,
"step": 2460
},
{
"epoch": 1.76,
"learning_rate": 2.302423603793467e-05,
"loss": 0.1323,
"step": 2470
},
{
"epoch": 1.76,
"learning_rate": 2.289251844046365e-05,
"loss": 0.1283,
"step": 2480
},
{
"epoch": 1.77,
"learning_rate": 2.2760800842992624e-05,
"loss": 0.1031,
"step": 2490
},
{
"epoch": 1.78,
"learning_rate": 2.2629083245521604e-05,
"loss": 0.0992,
"step": 2500
},
{
"epoch": 1.78,
"learning_rate": 2.249736564805058e-05,
"loss": 0.083,
"step": 2510
},
{
"epoch": 1.79,
"learning_rate": 2.236564805057956e-05,
"loss": 0.1079,
"step": 2520
},
{
"epoch": 1.8,
"learning_rate": 2.2233930453108538e-05,
"loss": 0.0877,
"step": 2530
},
{
"epoch": 1.81,
"learning_rate": 2.2102212855637514e-05,
"loss": 0.1187,
"step": 2540
},
{
"epoch": 1.81,
"learning_rate": 2.1970495258166493e-05,
"loss": 0.0835,
"step": 2550
},
{
"epoch": 1.82,
"learning_rate": 2.183877766069547e-05,
"loss": 0.0926,
"step": 2560
},
{
"epoch": 1.83,
"learning_rate": 2.1707060063224448e-05,
"loss": 0.1383,
"step": 2570
},
{
"epoch": 1.83,
"learning_rate": 2.1575342465753427e-05,
"loss": 0.1143,
"step": 2580
},
{
"epoch": 1.84,
"learning_rate": 2.1443624868282403e-05,
"loss": 0.1197,
"step": 2590
},
{
"epoch": 1.85,
"learning_rate": 2.1311907270811383e-05,
"loss": 0.0983,
"step": 2600
},
{
"epoch": 1.86,
"learning_rate": 2.118018967334036e-05,
"loss": 0.1105,
"step": 2610
},
{
"epoch": 1.86,
"learning_rate": 2.1048472075869338e-05,
"loss": 0.0993,
"step": 2620
},
{
"epoch": 1.87,
"learning_rate": 2.0916754478398313e-05,
"loss": 0.1156,
"step": 2630
},
{
"epoch": 1.88,
"learning_rate": 2.0785036880927293e-05,
"loss": 0.1111,
"step": 2640
},
{
"epoch": 1.88,
"learning_rate": 2.0653319283456272e-05,
"loss": 0.0834,
"step": 2650
},
{
"epoch": 1.89,
"learning_rate": 2.0521601685985248e-05,
"loss": 0.0993,
"step": 2660
},
{
"epoch": 1.9,
"learning_rate": 2.0389884088514227e-05,
"loss": 0.0803,
"step": 2670
},
{
"epoch": 1.91,
"learning_rate": 2.0258166491043203e-05,
"loss": 0.1206,
"step": 2680
},
{
"epoch": 1.91,
"learning_rate": 2.0126448893572182e-05,
"loss": 0.09,
"step": 2690
},
{
"epoch": 1.92,
"learning_rate": 1.999473129610116e-05,
"loss": 0.0954,
"step": 2700
},
{
"epoch": 1.93,
"learning_rate": 1.9863013698630137e-05,
"loss": 0.1062,
"step": 2710
},
{
"epoch": 1.93,
"learning_rate": 1.9731296101159116e-05,
"loss": 0.1117,
"step": 2720
},
{
"epoch": 1.94,
"learning_rate": 1.9599578503688092e-05,
"loss": 0.1496,
"step": 2730
},
{
"epoch": 1.95,
"learning_rate": 1.946786090621707e-05,
"loss": 0.0816,
"step": 2740
},
{
"epoch": 1.96,
"learning_rate": 1.933614330874605e-05,
"loss": 0.1257,
"step": 2750
},
{
"epoch": 1.96,
"learning_rate": 1.9204425711275027e-05,
"loss": 0.1006,
"step": 2760
},
{
"epoch": 1.97,
"learning_rate": 1.9072708113804006e-05,
"loss": 0.0912,
"step": 2770
},
{
"epoch": 1.98,
"learning_rate": 1.894099051633298e-05,
"loss": 0.0774,
"step": 2780
},
{
"epoch": 1.98,
"learning_rate": 1.880927291886196e-05,
"loss": 0.0785,
"step": 2790
},
{
"epoch": 1.99,
"learning_rate": 1.8677555321390937e-05,
"loss": 0.063,
"step": 2800
},
{
"epoch": 2.0,
"learning_rate": 1.8545837723919916e-05,
"loss": 0.1376,
"step": 2810
},
{
"epoch": 2.0,
"eval_accuracy": 0.9895,
"eval_loss": 0.03449298441410065,
"eval_runtime": 133.6612,
"eval_samples_per_second": 74.816,
"eval_steps_per_second": 4.676,
"step": 2812
},
{
"epoch": 2.01,
"learning_rate": 1.8414120126448895e-05,
"loss": 0.0691,
"step": 2820
},
{
"epoch": 2.01,
"learning_rate": 1.828240252897787e-05,
"loss": 0.0683,
"step": 2830
},
{
"epoch": 2.02,
"learning_rate": 1.815068493150685e-05,
"loss": 0.0809,
"step": 2840
},
{
"epoch": 2.03,
"learning_rate": 1.8018967334035826e-05,
"loss": 0.0951,
"step": 2850
},
{
"epoch": 2.03,
"learning_rate": 1.7887249736564805e-05,
"loss": 0.0883,
"step": 2860
},
{
"epoch": 2.04,
"learning_rate": 1.7755532139093785e-05,
"loss": 0.0633,
"step": 2870
},
{
"epoch": 2.05,
"learning_rate": 1.7623814541622764e-05,
"loss": 0.0828,
"step": 2880
},
{
"epoch": 2.06,
"learning_rate": 1.749209694415174e-05,
"loss": 0.0937,
"step": 2890
},
{
"epoch": 2.06,
"learning_rate": 1.7360379346680716e-05,
"loss": 0.1109,
"step": 2900
},
{
"epoch": 2.07,
"learning_rate": 1.7228661749209695e-05,
"loss": 0.081,
"step": 2910
},
{
"epoch": 2.08,
"learning_rate": 1.7096944151738674e-05,
"loss": 0.0974,
"step": 2920
},
{
"epoch": 2.08,
"learning_rate": 1.6965226554267653e-05,
"loss": 0.0826,
"step": 2930
},
{
"epoch": 2.09,
"learning_rate": 1.683350895679663e-05,
"loss": 0.0593,
"step": 2940
},
{
"epoch": 2.1,
"learning_rate": 1.6701791359325605e-05,
"loss": 0.073,
"step": 2950
},
{
"epoch": 2.1,
"learning_rate": 1.6570073761854584e-05,
"loss": 0.1086,
"step": 2960
},
{
"epoch": 2.11,
"learning_rate": 1.643835616438356e-05,
"loss": 0.0972,
"step": 2970
},
{
"epoch": 2.12,
"learning_rate": 1.630663856691254e-05,
"loss": 0.0953,
"step": 2980
},
{
"epoch": 2.13,
"learning_rate": 1.617492096944152e-05,
"loss": 0.0819,
"step": 2990
},
{
"epoch": 2.13,
"learning_rate": 1.6043203371970498e-05,
"loss": 0.0631,
"step": 3000
},
{
"epoch": 2.14,
"learning_rate": 1.5911485774499474e-05,
"loss": 0.1025,
"step": 3010
},
{
"epoch": 2.15,
"learning_rate": 1.577976817702845e-05,
"loss": 0.0983,
"step": 3020
},
{
"epoch": 2.15,
"learning_rate": 1.564805057955743e-05,
"loss": 0.1099,
"step": 3030
},
{
"epoch": 2.16,
"learning_rate": 1.5516332982086408e-05,
"loss": 0.0981,
"step": 3040
},
{
"epoch": 2.17,
"learning_rate": 1.5384615384615387e-05,
"loss": 0.0666,
"step": 3050
},
{
"epoch": 2.18,
"learning_rate": 1.5252897787144363e-05,
"loss": 0.0664,
"step": 3060
},
{
"epoch": 2.18,
"learning_rate": 1.5121180189673342e-05,
"loss": 0.0918,
"step": 3070
},
{
"epoch": 2.19,
"learning_rate": 1.498946259220232e-05,
"loss": 0.0968,
"step": 3080
},
{
"epoch": 2.2,
"learning_rate": 1.4857744994731296e-05,
"loss": 0.0992,
"step": 3090
},
{
"epoch": 2.2,
"learning_rate": 1.4726027397260275e-05,
"loss": 0.0951,
"step": 3100
},
{
"epoch": 2.21,
"learning_rate": 1.4594309799789252e-05,
"loss": 0.0966,
"step": 3110
},
{
"epoch": 2.22,
"learning_rate": 1.4462592202318232e-05,
"loss": 0.0805,
"step": 3120
},
{
"epoch": 2.23,
"learning_rate": 1.4330874604847208e-05,
"loss": 0.0969,
"step": 3130
},
{
"epoch": 2.23,
"learning_rate": 1.4199157007376185e-05,
"loss": 0.0749,
"step": 3140
},
{
"epoch": 2.24,
"learning_rate": 1.4067439409905164e-05,
"loss": 0.0712,
"step": 3150
},
{
"epoch": 2.25,
"learning_rate": 1.3935721812434142e-05,
"loss": 0.068,
"step": 3160
},
{
"epoch": 2.25,
"learning_rate": 1.3804004214963121e-05,
"loss": 0.1018,
"step": 3170
},
{
"epoch": 2.26,
"learning_rate": 1.3672286617492097e-05,
"loss": 0.0862,
"step": 3180
},
{
"epoch": 2.27,
"learning_rate": 1.3540569020021076e-05,
"loss": 0.0855,
"step": 3190
},
{
"epoch": 2.28,
"learning_rate": 1.3408851422550054e-05,
"loss": 0.0979,
"step": 3200
},
{
"epoch": 2.28,
"learning_rate": 1.327713382507903e-05,
"loss": 0.0842,
"step": 3210
},
{
"epoch": 2.29,
"learning_rate": 1.3145416227608009e-05,
"loss": 0.0732,
"step": 3220
},
{
"epoch": 2.3,
"learning_rate": 1.3013698630136986e-05,
"loss": 0.0867,
"step": 3230
},
{
"epoch": 2.3,
"learning_rate": 1.2881981032665966e-05,
"loss": 0.0735,
"step": 3240
},
{
"epoch": 2.31,
"learning_rate": 1.2750263435194943e-05,
"loss": 0.0785,
"step": 3250
},
{
"epoch": 2.32,
"learning_rate": 1.2618545837723922e-05,
"loss": 0.0638,
"step": 3260
},
{
"epoch": 2.33,
"learning_rate": 1.2486828240252898e-05,
"loss": 0.0971,
"step": 3270
},
{
"epoch": 2.33,
"learning_rate": 1.2355110642781877e-05,
"loss": 0.093,
"step": 3280
},
{
"epoch": 2.34,
"learning_rate": 1.2223393045310855e-05,
"loss": 0.0736,
"step": 3290
},
{
"epoch": 2.35,
"learning_rate": 1.209167544783983e-05,
"loss": 0.0776,
"step": 3300
},
{
"epoch": 2.35,
"learning_rate": 1.195995785036881e-05,
"loss": 0.0852,
"step": 3310
},
{
"epoch": 2.36,
"learning_rate": 1.1828240252897788e-05,
"loss": 0.1035,
"step": 3320
},
{
"epoch": 2.37,
"learning_rate": 1.1696522655426765e-05,
"loss": 0.0769,
"step": 3330
},
{
"epoch": 2.38,
"learning_rate": 1.1564805057955744e-05,
"loss": 0.0953,
"step": 3340
},
{
"epoch": 2.38,
"learning_rate": 1.1433087460484722e-05,
"loss": 0.0646,
"step": 3350
},
{
"epoch": 2.39,
"learning_rate": 1.1301369863013698e-05,
"loss": 0.0785,
"step": 3360
},
{
"epoch": 2.4,
"learning_rate": 1.1169652265542677e-05,
"loss": 0.0738,
"step": 3370
},
{
"epoch": 2.4,
"learning_rate": 1.1037934668071655e-05,
"loss": 0.0638,
"step": 3380
},
{
"epoch": 2.41,
"learning_rate": 1.0906217070600632e-05,
"loss": 0.0948,
"step": 3390
},
{
"epoch": 2.42,
"learning_rate": 1.0774499473129611e-05,
"loss": 0.1331,
"step": 3400
},
{
"epoch": 2.42,
"learning_rate": 1.0642781875658589e-05,
"loss": 0.1084,
"step": 3410
},
{
"epoch": 2.43,
"learning_rate": 1.0511064278187566e-05,
"loss": 0.0919,
"step": 3420
},
{
"epoch": 2.44,
"learning_rate": 1.0379346680716544e-05,
"loss": 0.0666,
"step": 3430
},
{
"epoch": 2.45,
"learning_rate": 1.0247629083245521e-05,
"loss": 0.0911,
"step": 3440
},
{
"epoch": 2.45,
"learning_rate": 1.01159114857745e-05,
"loss": 0.0706,
"step": 3450
},
{
"epoch": 2.46,
"learning_rate": 9.984193888303478e-06,
"loss": 0.1086,
"step": 3460
},
{
"epoch": 2.47,
"learning_rate": 9.852476290832456e-06,
"loss": 0.0729,
"step": 3470
},
{
"epoch": 2.47,
"learning_rate": 9.720758693361433e-06,
"loss": 0.1097,
"step": 3480
},
{
"epoch": 2.48,
"learning_rate": 9.589041095890411e-06,
"loss": 0.0737,
"step": 3490
},
{
"epoch": 2.49,
"learning_rate": 9.457323498419388e-06,
"loss": 0.0839,
"step": 3500
},
{
"epoch": 2.5,
"learning_rate": 9.325605900948368e-06,
"loss": 0.1214,
"step": 3510
},
{
"epoch": 2.5,
"learning_rate": 9.193888303477345e-06,
"loss": 0.1008,
"step": 3520
},
{
"epoch": 2.51,
"learning_rate": 9.062170706006323e-06,
"loss": 0.0774,
"step": 3530
},
{
"epoch": 2.52,
"learning_rate": 8.930453108535302e-06,
"loss": 0.0631,
"step": 3540
},
{
"epoch": 2.52,
"learning_rate": 8.798735511064278e-06,
"loss": 0.0705,
"step": 3550
},
{
"epoch": 2.53,
"learning_rate": 8.667017913593255e-06,
"loss": 0.1071,
"step": 3560
},
{
"epoch": 2.54,
"learning_rate": 8.535300316122235e-06,
"loss": 0.0645,
"step": 3570
},
{
"epoch": 2.55,
"learning_rate": 8.403582718651212e-06,
"loss": 0.0713,
"step": 3580
},
{
"epoch": 2.55,
"learning_rate": 8.27186512118019e-06,
"loss": 0.0961,
"step": 3590
},
{
"epoch": 2.56,
"learning_rate": 8.140147523709169e-06,
"loss": 0.0805,
"step": 3600
},
{
"epoch": 2.57,
"learning_rate": 8.008429926238145e-06,
"loss": 0.0885,
"step": 3610
},
{
"epoch": 2.57,
"learning_rate": 7.876712328767124e-06,
"loss": 0.0623,
"step": 3620
},
{
"epoch": 2.58,
"learning_rate": 7.744994731296102e-06,
"loss": 0.1018,
"step": 3630
},
{
"epoch": 2.59,
"learning_rate": 7.613277133825079e-06,
"loss": 0.0997,
"step": 3640
},
{
"epoch": 2.6,
"learning_rate": 7.4815595363540575e-06,
"loss": 0.0845,
"step": 3650
},
{
"epoch": 2.6,
"learning_rate": 7.349841938883036e-06,
"loss": 0.0419,
"step": 3660
},
{
"epoch": 2.61,
"learning_rate": 7.218124341412013e-06,
"loss": 0.0906,
"step": 3670
},
{
"epoch": 2.62,
"learning_rate": 7.08640674394099e-06,
"loss": 0.0784,
"step": 3680
},
{
"epoch": 2.62,
"learning_rate": 6.9546891464699685e-06,
"loss": 0.0744,
"step": 3690
},
{
"epoch": 2.63,
"learning_rate": 6.822971548998947e-06,
"loss": 0.0621,
"step": 3700
},
{
"epoch": 2.64,
"learning_rate": 6.6912539515279244e-06,
"loss": 0.0877,
"step": 3710
},
{
"epoch": 2.65,
"learning_rate": 6.559536354056903e-06,
"loss": 0.0639,
"step": 3720
},
{
"epoch": 2.65,
"learning_rate": 6.42781875658588e-06,
"loss": 0.0977,
"step": 3730
},
{
"epoch": 2.66,
"learning_rate": 6.296101159114858e-06,
"loss": 0.0833,
"step": 3740
},
{
"epoch": 2.67,
"learning_rate": 6.1643835616438354e-06,
"loss": 0.0595,
"step": 3750
},
{
"epoch": 2.67,
"learning_rate": 6.032665964172814e-06,
"loss": 0.0697,
"step": 3760
},
{
"epoch": 2.68,
"learning_rate": 5.900948366701791e-06,
"loss": 0.0869,
"step": 3770
},
{
"epoch": 2.69,
"learning_rate": 5.76923076923077e-06,
"loss": 0.0671,
"step": 3780
},
{
"epoch": 2.7,
"learning_rate": 5.637513171759747e-06,
"loss": 0.0699,
"step": 3790
},
{
"epoch": 2.7,
"learning_rate": 5.505795574288726e-06,
"loss": 0.0711,
"step": 3800
},
{
"epoch": 2.71,
"learning_rate": 5.374077976817703e-06,
"loss": 0.0665,
"step": 3810
},
{
"epoch": 2.72,
"learning_rate": 5.242360379346681e-06,
"loss": 0.0904,
"step": 3820
},
{
"epoch": 2.72,
"learning_rate": 5.110642781875659e-06,
"loss": 0.0876,
"step": 3830
},
{
"epoch": 2.73,
"learning_rate": 4.978925184404637e-06,
"loss": 0.0891,
"step": 3840
},
{
"epoch": 2.74,
"learning_rate": 4.847207586933614e-06,
"loss": 0.0819,
"step": 3850
},
{
"epoch": 2.74,
"learning_rate": 4.715489989462593e-06,
"loss": 0.0673,
"step": 3860
},
{
"epoch": 2.75,
"learning_rate": 4.58377239199157e-06,
"loss": 0.0519,
"step": 3870
},
{
"epoch": 2.76,
"learning_rate": 4.452054794520548e-06,
"loss": 0.087,
"step": 3880
},
{
"epoch": 2.77,
"learning_rate": 4.320337197049526e-06,
"loss": 0.0651,
"step": 3890
},
{
"epoch": 2.77,
"learning_rate": 4.188619599578504e-06,
"loss": 0.0633,
"step": 3900
},
{
"epoch": 2.78,
"learning_rate": 4.056902002107481e-06,
"loss": 0.0846,
"step": 3910
},
{
"epoch": 2.79,
"learning_rate": 3.92518440463646e-06,
"loss": 0.0721,
"step": 3920
},
{
"epoch": 2.79,
"learning_rate": 3.7934668071654376e-06,
"loss": 0.1049,
"step": 3930
},
{
"epoch": 2.8,
"learning_rate": 3.661749209694415e-06,
"loss": 0.1,
"step": 3940
},
{
"epoch": 2.81,
"learning_rate": 3.530031612223393e-06,
"loss": 0.086,
"step": 3950
},
{
"epoch": 2.82,
"learning_rate": 3.398314014752371e-06,
"loss": 0.0802,
"step": 3960
},
{
"epoch": 2.82,
"learning_rate": 3.2665964172813486e-06,
"loss": 0.0789,
"step": 3970
},
{
"epoch": 2.83,
"learning_rate": 3.1348788198103265e-06,
"loss": 0.0612,
"step": 3980
},
{
"epoch": 2.84,
"learning_rate": 3.003161222339305e-06,
"loss": 0.0697,
"step": 3990
},
{
"epoch": 2.84,
"learning_rate": 2.8714436248682825e-06,
"loss": 0.0862,
"step": 4000
},
{
"epoch": 2.85,
"learning_rate": 2.7397260273972604e-06,
"loss": 0.0541,
"step": 4010
},
{
"epoch": 2.86,
"learning_rate": 2.6080084299262384e-06,
"loss": 0.0559,
"step": 4020
},
{
"epoch": 2.87,
"learning_rate": 2.476290832455216e-06,
"loss": 0.0611,
"step": 4030
},
{
"epoch": 2.87,
"learning_rate": 2.3445732349841943e-06,
"loss": 0.0657,
"step": 4040
},
{
"epoch": 2.88,
"learning_rate": 2.212855637513172e-06,
"loss": 0.049,
"step": 4050
},
{
"epoch": 2.89,
"learning_rate": 2.08113804004215e-06,
"loss": 0.0733,
"step": 4060
},
{
"epoch": 2.89,
"learning_rate": 1.949420442571128e-06,
"loss": 0.0608,
"step": 4070
},
{
"epoch": 2.9,
"learning_rate": 1.8177028451001056e-06,
"loss": 0.0718,
"step": 4080
},
{
"epoch": 2.91,
"learning_rate": 1.6859852476290833e-06,
"loss": 0.0728,
"step": 4090
},
{
"epoch": 2.92,
"learning_rate": 1.554267650158061e-06,
"loss": 0.0627,
"step": 4100
},
{
"epoch": 2.92,
"learning_rate": 1.422550052687039e-06,
"loss": 0.0953,
"step": 4110
},
{
"epoch": 2.93,
"learning_rate": 1.290832455216017e-06,
"loss": 0.0794,
"step": 4120
},
{
"epoch": 2.94,
"learning_rate": 1.1591148577449948e-06,
"loss": 0.0844,
"step": 4130
},
{
"epoch": 2.94,
"learning_rate": 1.0273972602739725e-06,
"loss": 0.0601,
"step": 4140
},
{
"epoch": 2.95,
"learning_rate": 8.956796628029505e-07,
"loss": 0.0775,
"step": 4150
},
{
"epoch": 2.96,
"learning_rate": 7.639620653319284e-07,
"loss": 0.0666,
"step": 4160
},
{
"epoch": 2.97,
"learning_rate": 6.322444678609063e-07,
"loss": 0.0676,
"step": 4170
},
{
"epoch": 2.97,
"learning_rate": 5.00526870389884e-07,
"loss": 0.1339,
"step": 4180
},
{
"epoch": 2.98,
"learning_rate": 3.6880927291886196e-07,
"loss": 0.0614,
"step": 4190
},
{
"epoch": 2.99,
"learning_rate": 2.3709167544783985e-07,
"loss": 0.092,
"step": 4200
},
{
"epoch": 2.99,
"learning_rate": 1.0537407797681771e-07,
"loss": 0.086,
"step": 4210
},
{
"epoch": 3.0,
"eval_accuracy": 0.9901,
"eval_loss": 0.03231855481863022,
"eval_runtime": 133.5993,
"eval_samples_per_second": 74.851,
"eval_steps_per_second": 4.678,
"step": 4218
},
{
"epoch": 3.0,
"step": 4218,
"total_flos": 2.091911758043834e+19,
"train_loss": 0.13970869080807824,
"train_runtime": 10165.3024,
"train_samples_per_second": 26.561,
"train_steps_per_second": 0.415
}
],
"max_steps": 4218,
"num_train_epochs": 3,
"total_flos": 2.091911758043834e+19,
"trial_name": null,
"trial_params": null
}