ghostemane / trainer_state.json
AlekseyKorshuk's picture
huggingartists
2922f4d
raw
history blame contribute delete
No virus
81.9 kB
{
"best_metric": null,
"best_model_checkpoint": null,
"epoch": 60.0,
"global_step": 3360,
"is_hyper_param_search": false,
"is_local_process_zero": true,
"is_world_process_zero": true,
"log_history": [
{
"epoch": 0.09,
"learning_rate": 0.00013451892828543385,
"loss": 3.9488,
"step": 5
},
{
"epoch": 0.18,
"learning_rate": 0.00012668528006706028,
"loss": 3.8298,
"step": 10
},
{
"epoch": 0.27,
"learning_rate": 0.00011431137524750748,
"loss": 3.7557,
"step": 15
},
{
"epoch": 0.36,
"learning_rate": 9.836442450346448e-05,
"loss": 3.5311,
"step": 20
},
{
"epoch": 0.45,
"learning_rate": 8.009092691870492e-05,
"loss": 3.3526,
"step": 25
},
{
"epoch": 0.54,
"learning_rate": 6.0919236939313083e-05,
"loss": 3.5934,
"step": 30
},
{
"epoch": 0.62,
"learning_rate": 4.2347916539754844e-05,
"loss": 3.3983,
"step": 35
},
{
"epoch": 0.71,
"learning_rate": 2.5828599592490882e-05,
"loss": 3.6609,
"step": 40
},
{
"epoch": 0.8,
"learning_rate": 1.2652524389394753e-05,
"loss": 3.4898,
"step": 45
},
{
"epoch": 0.89,
"learning_rate": 3.849603540845984e-06,
"loss": 3.5749,
"step": 50
},
{
"epoch": 0.98,
"learning_rate": 1.0792048977778093e-07,
"loss": 3.4588,
"step": 55
},
{
"epoch": 1.07,
"learning_rate": 1.7199452243268996e-06,
"loss": 3.2536,
"step": 60
},
{
"epoch": 1.16,
"learning_rate": 8.559673257059505e-06,
"loss": 3.2439,
"step": 65
},
{
"epoch": 1.25,
"learning_rate": 2.009247481060283e-05,
"loss": 3.2968,
"step": 70
},
{
"epoch": 1.34,
"learning_rate": 3.541688434458052e-05,
"loss": 3.4346,
"step": 75
},
{
"epoch": 1.43,
"learning_rate": 5.333506393059682e-05,
"loss": 3.3563,
"step": 80
},
{
"epoch": 1.52,
"learning_rate": 7.244643268047132e-05,
"loss": 3.2696,
"step": 85
},
{
"epoch": 1.61,
"learning_rate": 9.125714365012444e-05,
"loss": 3.4046,
"step": 90
},
{
"epoch": 1.7,
"learning_rate": 0.00010829685091793463,
"loss": 3.4708,
"step": 95
},
{
"epoch": 1.79,
"learning_rate": 0.00012223363969730684,
"loss": 3.2387,
"step": 100
},
{
"epoch": 1.88,
"learning_rate": 0.00013197813593027427,
"loss": 3.3163,
"step": 105
},
{
"epoch": 1.96,
"learning_rate": 0.00013676865759867644,
"loss": 3.2581,
"step": 110
},
{
"epoch": 2.05,
"learning_rate": 0.000136230751870351,
"loss": 3.1941,
"step": 115
},
{
"epoch": 2.14,
"learning_rate": 0.00013040646433810595,
"loss": 3.0392,
"step": 120
},
{
"epoch": 2.23,
"learning_rate": 0.00011975105251098516,
"loss": 3.0188,
"step": 125
},
{
"epoch": 2.32,
"learning_rate": 0.00010509740044895205,
"loss": 3.1013,
"step": 130
},
{
"epoch": 2.41,
"learning_rate": 8.759091608374473e-05,
"loss": 3.0585,
"step": 135
},
{
"epoch": 2.5,
"learning_rate": 6.860000000000001e-05,
"loss": 3.1302,
"step": 140
},
{
"epoch": 2.59,
"learning_rate": 4.9609083916255386e-05,
"loss": 3.2358,
"step": 145
},
{
"epoch": 2.68,
"learning_rate": 3.210259955104798e-05,
"loss": 3.119,
"step": 150
},
{
"epoch": 2.77,
"learning_rate": 1.744894748901483e-05,
"loss": 2.9946,
"step": 155
},
{
"epoch": 2.86,
"learning_rate": 6.793535661894062e-06,
"loss": 3.0184,
"step": 160
},
{
"epoch": 2.95,
"learning_rate": 9.692481296490106e-07,
"loss": 2.9798,
"step": 165
},
{
"epoch": 3.04,
"learning_rate": 4.313424013235498e-07,
"loss": 3.1282,
"step": 170
},
{
"epoch": 3.12,
"learning_rate": 5.22186406972573e-06,
"loss": 3.0772,
"step": 175
},
{
"epoch": 3.21,
"learning_rate": 1.496636030269314e-05,
"loss": 2.8216,
"step": 180
},
{
"epoch": 3.3,
"learning_rate": 2.890314908206528e-05,
"loss": 2.7665,
"step": 185
},
{
"epoch": 3.39,
"learning_rate": 4.594285634987545e-05,
"loss": 3.0073,
"step": 190
},
{
"epoch": 3.48,
"learning_rate": 6.475356731952864e-05,
"loss": 3.0372,
"step": 195
},
{
"epoch": 3.57,
"learning_rate": 8.386493606940314e-05,
"loss": 2.807,
"step": 200
},
{
"epoch": 3.66,
"learning_rate": 0.0001017831156554194,
"loss": 3.1058,
"step": 205
},
{
"epoch": 3.75,
"learning_rate": 0.0001171075251893971,
"loss": 2.961,
"step": 210
},
{
"epoch": 3.84,
"learning_rate": 0.0001286403267429405,
"loss": 3.1032,
"step": 215
},
{
"epoch": 3.93,
"learning_rate": 0.0001354800547756731,
"loss": 2.7667,
"step": 220
},
{
"epoch": 4.02,
"learning_rate": 0.00013709207951022223,
"loss": 3.0024,
"step": 225
},
{
"epoch": 4.11,
"learning_rate": 0.00013335039645915404,
"loss": 2.8538,
"step": 230
},
{
"epoch": 4.2,
"learning_rate": 0.00012454747561060531,
"loss": 2.8202,
"step": 235
},
{
"epoch": 4.29,
"learning_rate": 0.00011137140040750914,
"loss": 2.6845,
"step": 240
},
{
"epoch": 4.38,
"learning_rate": 9.485208346024522e-05,
"loss": 2.6865,
"step": 245
},
{
"epoch": 4.46,
"learning_rate": 7.62807630606869e-05,
"loss": 2.8686,
"step": 250
},
{
"epoch": 4.55,
"learning_rate": 5.710907308129509e-05,
"loss": 2.9936,
"step": 255
},
{
"epoch": 4.64,
"learning_rate": 3.883557549653544e-05,
"loss": 2.5979,
"step": 260
},
{
"epoch": 4.73,
"learning_rate": 2.2888624752492583e-05,
"loss": 2.7179,
"step": 265
},
{
"epoch": 4.82,
"learning_rate": 1.0514719932939762e-05,
"loss": 2.9387,
"step": 270
},
{
"epoch": 4.91,
"learning_rate": 2.681071714566175e-06,
"loss": 2.6822,
"step": 275
},
{
"epoch": 5.0,
"learning_rate": 0.0,
"loss": 2.7684,
"step": 280
},
{
"epoch": 5.09,
"learning_rate": 2.6810717145661523e-06,
"loss": 2.6722,
"step": 285
},
{
"epoch": 5.18,
"learning_rate": 1.0514719932939649e-05,
"loss": 2.597,
"step": 290
},
{
"epoch": 5.27,
"learning_rate": 2.2888624752492607e-05,
"loss": 2.7343,
"step": 295
},
{
"epoch": 5.36,
"learning_rate": 3.8835575496535365e-05,
"loss": 2.567,
"step": 300
},
{
"epoch": 5.45,
"learning_rate": 5.7109073081294886e-05,
"loss": 2.6375,
"step": 305
},
{
"epoch": 5.54,
"learning_rate": 7.628076306068694e-05,
"loss": 2.734,
"step": 310
},
{
"epoch": 5.62,
"learning_rate": 9.485208346024515e-05,
"loss": 2.6448,
"step": 315
},
{
"epoch": 5.71,
"learning_rate": 0.00011137140040750908,
"loss": 2.6255,
"step": 320
},
{
"epoch": 5.8,
"learning_rate": 0.0001245474756106052,
"loss": 2.6455,
"step": 325
},
{
"epoch": 5.89,
"learning_rate": 0.00013335039645915407,
"loss": 2.5969,
"step": 330
},
{
"epoch": 5.98,
"learning_rate": 0.00013709207951022223,
"loss": 2.6923,
"step": 335
},
{
"epoch": 6.07,
"learning_rate": 0.00013548005477567314,
"loss": 2.3761,
"step": 340
},
{
"epoch": 6.16,
"learning_rate": 0.00012864032674294047,
"loss": 2.4563,
"step": 345
},
{
"epoch": 6.25,
"learning_rate": 0.00011710752518939715,
"loss": 2.4791,
"step": 350
},
{
"epoch": 6.34,
"learning_rate": 0.00010178311565541947,
"loss": 2.446,
"step": 355
},
{
"epoch": 6.43,
"learning_rate": 8.386493606940322e-05,
"loss": 2.5515,
"step": 360
},
{
"epoch": 6.52,
"learning_rate": 6.475356731952872e-05,
"loss": 2.5469,
"step": 365
},
{
"epoch": 6.61,
"learning_rate": 4.594285634987565e-05,
"loss": 2.6391,
"step": 370
},
{
"epoch": 6.7,
"learning_rate": 2.890314908206545e-05,
"loss": 2.36,
"step": 375
},
{
"epoch": 6.79,
"learning_rate": 1.496636030269327e-05,
"loss": 2.4806,
"step": 380
},
{
"epoch": 6.88,
"learning_rate": 5.221864069725715e-06,
"loss": 2.6083,
"step": 385
},
{
"epoch": 6.96,
"learning_rate": 4.3134240132355735e-07,
"loss": 2.6457,
"step": 390
},
{
"epoch": 7.05,
"learning_rate": 9.692481296490106e-07,
"loss": 2.4165,
"step": 395
},
{
"epoch": 7.14,
"learning_rate": 6.793535661894024e-06,
"loss": 2.301,
"step": 400
},
{
"epoch": 7.23,
"learning_rate": 1.744894748901478e-05,
"loss": 2.4478,
"step": 405
},
{
"epoch": 7.32,
"learning_rate": 3.2102599551047805e-05,
"loss": 2.3692,
"step": 410
},
{
"epoch": 7.41,
"learning_rate": 4.960908391625518e-05,
"loss": 2.3269,
"step": 415
},
{
"epoch": 7.5,
"learning_rate": 6.859999999999982e-05,
"loss": 2.1706,
"step": 420
},
{
"epoch": 7.59,
"learning_rate": 8.759091608374469e-05,
"loss": 2.3618,
"step": 425
},
{
"epoch": 7.68,
"learning_rate": 0.00010509740044895209,
"loss": 2.284,
"step": 430
},
{
"epoch": 7.77,
"learning_rate": 0.00011975105251098514,
"loss": 2.3587,
"step": 435
},
{
"epoch": 7.86,
"learning_rate": 0.00013040646433810593,
"loss": 2.4467,
"step": 440
},
{
"epoch": 7.95,
"learning_rate": 0.000136230751870351,
"loss": 2.5326,
"step": 445
},
{
"epoch": 8.04,
"learning_rate": 0.00013676865759867642,
"loss": 2.3045,
"step": 450
},
{
"epoch": 8.12,
"learning_rate": 0.00013197813593027432,
"loss": 2.1819,
"step": 455
},
{
"epoch": 8.21,
"learning_rate": 0.00012223363969730697,
"loss": 2.2893,
"step": 460
},
{
"epoch": 8.3,
"learning_rate": 0.00010829685091793466,
"loss": 2.3117,
"step": 465
},
{
"epoch": 8.39,
"learning_rate": 9.12571436501247e-05,
"loss": 2.272,
"step": 470
},
{
"epoch": 8.48,
"learning_rate": 7.24464326804714e-05,
"loss": 2.3461,
"step": 475
},
{
"epoch": 8.57,
"learning_rate": 5.33350639305969e-05,
"loss": 2.1348,
"step": 480
},
{
"epoch": 8.66,
"learning_rate": 3.541688434458043e-05,
"loss": 2.2985,
"step": 485
},
{
"epoch": 8.75,
"learning_rate": 2.0092474810602934e-05,
"loss": 2.077,
"step": 490
},
{
"epoch": 8.84,
"learning_rate": 8.559673257059573e-06,
"loss": 2.2565,
"step": 495
},
{
"epoch": 8.93,
"learning_rate": 1.719945224326892e-06,
"loss": 2.1992,
"step": 500
},
{
"epoch": 9.02,
"learning_rate": 1.0792048977777332e-07,
"loss": 2.1455,
"step": 505
},
{
"epoch": 9.11,
"learning_rate": 3.849603540845977e-06,
"loss": 2.1314,
"step": 510
},
{
"epoch": 9.2,
"learning_rate": 1.2652524389394722e-05,
"loss": 1.9046,
"step": 515
},
{
"epoch": 9.29,
"learning_rate": 2.582859959249101e-05,
"loss": 2.0235,
"step": 520
},
{
"epoch": 9.38,
"learning_rate": 4.234791653975475e-05,
"loss": 2.0746,
"step": 525
},
{
"epoch": 9.46,
"learning_rate": 6.091923693931295e-05,
"loss": 2.0545,
"step": 530
},
{
"epoch": 9.55,
"learning_rate": 8.0090926918705e-05,
"loss": 2.0609,
"step": 535
},
{
"epoch": 9.64,
"learning_rate": 9.83644245034643e-05,
"loss": 2.1458,
"step": 540
},
{
"epoch": 9.73,
"learning_rate": 0.00011431137524750748,
"loss": 1.9214,
"step": 545
},
{
"epoch": 9.82,
"learning_rate": 0.00012668528006706028,
"loss": 2.2293,
"step": 550
},
{
"epoch": 9.91,
"learning_rate": 0.00013451892828543387,
"loss": 2.2592,
"step": 555
},
{
"epoch": 10.0,
"learning_rate": 0.0001372,
"loss": 2.1707,
"step": 560
},
{
"epoch": 10.09,
"learning_rate": 0.00013451892828543393,
"loss": 1.7839,
"step": 565
},
{
"epoch": 10.18,
"learning_rate": 0.0001266852800670604,
"loss": 1.931,
"step": 570
},
{
"epoch": 10.27,
"learning_rate": 0.00011431137524750779,
"loss": 1.981,
"step": 575
},
{
"epoch": 10.36,
"learning_rate": 9.836442450346467e-05,
"loss": 1.9186,
"step": 580
},
{
"epoch": 10.45,
"learning_rate": 8.009092691870492e-05,
"loss": 1.9439,
"step": 585
},
{
"epoch": 10.54,
"learning_rate": 6.0919236939312867e-05,
"loss": 1.7981,
"step": 590
},
{
"epoch": 10.62,
"learning_rate": 4.23479165397549e-05,
"loss": 1.8503,
"step": 595
},
{
"epoch": 10.71,
"learning_rate": 2.5828599592491143e-05,
"loss": 2.1472,
"step": 600
},
{
"epoch": 10.8,
"learning_rate": 1.265252438939482e-05,
"loss": 1.8879,
"step": 605
},
{
"epoch": 10.89,
"learning_rate": 3.849603540846114e-06,
"loss": 2.0268,
"step": 610
},
{
"epoch": 10.98,
"learning_rate": 1.0792048977779616e-07,
"loss": 2.0845,
"step": 615
},
{
"epoch": 11.07,
"learning_rate": 1.7199452243269073e-06,
"loss": 2.1026,
"step": 620
},
{
"epoch": 11.16,
"learning_rate": 8.559673257059612e-06,
"loss": 1.8359,
"step": 625
},
{
"epoch": 11.25,
"learning_rate": 2.0092474810602812e-05,
"loss": 1.6721,
"step": 630
},
{
"epoch": 11.34,
"learning_rate": 3.541688434458027e-05,
"loss": 1.6882,
"step": 635
},
{
"epoch": 11.43,
"learning_rate": 5.333506393059674e-05,
"loss": 1.7527,
"step": 640
},
{
"epoch": 11.52,
"learning_rate": 7.244643268047099e-05,
"loss": 1.7121,
"step": 645
},
{
"epoch": 11.61,
"learning_rate": 9.125714365012432e-05,
"loss": 1.8074,
"step": 650
},
{
"epoch": 11.7,
"learning_rate": 0.00010829685091793471,
"loss": 1.8594,
"step": 655
},
{
"epoch": 11.79,
"learning_rate": 0.000122233639697307,
"loss": 1.8073,
"step": 660
},
{
"epoch": 11.88,
"learning_rate": 0.00013197813593027427,
"loss": 1.8472,
"step": 665
},
{
"epoch": 11.96,
"learning_rate": 0.00013676865759867642,
"loss": 1.8812,
"step": 670
},
{
"epoch": 12.05,
"learning_rate": 0.000136230751870351,
"loss": 1.5812,
"step": 675
},
{
"epoch": 12.14,
"learning_rate": 0.0001304064643381061,
"loss": 1.6567,
"step": 680
},
{
"epoch": 12.23,
"learning_rate": 0.00011975105251098525,
"loss": 1.6487,
"step": 685
},
{
"epoch": 12.32,
"learning_rate": 0.000105097400448952,
"loss": 1.698,
"step": 690
},
{
"epoch": 12.41,
"learning_rate": 8.759091608374439e-05,
"loss": 1.7092,
"step": 695
},
{
"epoch": 12.5,
"learning_rate": 6.859999999999999e-05,
"loss": 1.5692,
"step": 700
},
{
"epoch": 12.59,
"learning_rate": 4.960908391625558e-05,
"loss": 1.7959,
"step": 705
},
{
"epoch": 12.68,
"learning_rate": 3.210259955104795e-05,
"loss": 1.769,
"step": 710
},
{
"epoch": 12.77,
"learning_rate": 1.7448947489015055e-05,
"loss": 1.672,
"step": 715
},
{
"epoch": 12.86,
"learning_rate": 6.793535661894092e-06,
"loss": 1.6729,
"step": 720
},
{
"epoch": 12.95,
"learning_rate": 9.692481296490868e-07,
"loss": 1.6378,
"step": 725
},
{
"epoch": 13.04,
"learning_rate": 4.3134240132354215e-07,
"loss": 1.5157,
"step": 730
},
{
"epoch": 13.12,
"learning_rate": 5.221864069725745e-06,
"loss": 1.5695,
"step": 735
},
{
"epoch": 13.21,
"learning_rate": 1.496636030269301e-05,
"loss": 1.4221,
"step": 740
},
{
"epoch": 13.3,
"learning_rate": 2.890314908206531e-05,
"loss": 1.5325,
"step": 745
},
{
"epoch": 13.39,
"learning_rate": 4.5942856349875256e-05,
"loss": 1.5675,
"step": 750
},
{
"epoch": 13.48,
"learning_rate": 6.475356731952856e-05,
"loss": 1.4491,
"step": 755
},
{
"epoch": 13.57,
"learning_rate": 8.386493606940281e-05,
"loss": 1.4567,
"step": 760
},
{
"epoch": 13.66,
"learning_rate": 0.00010178311565541931,
"loss": 1.4152,
"step": 765
},
{
"epoch": 13.75,
"learning_rate": 0.00011710752518939722,
"loss": 1.6904,
"step": 770
},
{
"epoch": 13.84,
"learning_rate": 0.00012864032674294042,
"loss": 1.7226,
"step": 775
},
{
"epoch": 13.93,
"learning_rate": 0.0001354800547756731,
"loss": 1.6131,
"step": 780
},
{
"epoch": 14.02,
"learning_rate": 0.00013709207951022223,
"loss": 1.4874,
"step": 785
},
{
"epoch": 14.11,
"learning_rate": 0.00013335039645915404,
"loss": 1.321,
"step": 790
},
{
"epoch": 14.2,
"learning_rate": 0.00012454747561060542,
"loss": 1.5378,
"step": 795
},
{
"epoch": 14.29,
"learning_rate": 0.00011137140040750922,
"loss": 1.5117,
"step": 800
},
{
"epoch": 14.38,
"learning_rate": 9.485208346024507e-05,
"loss": 1.3334,
"step": 805
},
{
"epoch": 14.46,
"learning_rate": 7.62807630606871e-05,
"loss": 1.418,
"step": 810
},
{
"epoch": 14.55,
"learning_rate": 5.710907308129505e-05,
"loss": 1.4658,
"step": 815
},
{
"epoch": 14.64,
"learning_rate": 3.883557549653573e-05,
"loss": 1.6222,
"step": 820
},
{
"epoch": 14.73,
"learning_rate": 2.2888624752492553e-05,
"loss": 1.4192,
"step": 825
},
{
"epoch": 14.82,
"learning_rate": 1.0514719932939869e-05,
"loss": 1.346,
"step": 830
},
{
"epoch": 14.91,
"learning_rate": 2.681071714566198e-06,
"loss": 1.3806,
"step": 835
},
{
"epoch": 15.0,
"learning_rate": 0.0,
"loss": 1.384,
"step": 840
},
{
"epoch": 15.09,
"learning_rate": 2.6810717145661294e-06,
"loss": 1.2256,
"step": 845
},
{
"epoch": 15.18,
"learning_rate": 1.0514719932939732e-05,
"loss": 1.3474,
"step": 850
},
{
"epoch": 15.27,
"learning_rate": 2.2888624752492363e-05,
"loss": 1.3215,
"step": 855
},
{
"epoch": 15.36,
"learning_rate": 3.88355754965355e-05,
"loss": 1.2651,
"step": 860
},
{
"epoch": 15.45,
"learning_rate": 5.710907308129481e-05,
"loss": 1.3224,
"step": 865
},
{
"epoch": 15.54,
"learning_rate": 7.628076306068686e-05,
"loss": 1.1728,
"step": 870
},
{
"epoch": 15.62,
"learning_rate": 9.485208346024484e-05,
"loss": 1.3578,
"step": 875
},
{
"epoch": 15.71,
"learning_rate": 0.00011137140040750902,
"loss": 1.1825,
"step": 880
},
{
"epoch": 15.8,
"learning_rate": 0.0001245474756106053,
"loss": 1.3246,
"step": 885
},
{
"epoch": 15.89,
"learning_rate": 0.00013335039645915393,
"loss": 1.332,
"step": 890
},
{
"epoch": 15.98,
"learning_rate": 0.00013709207951022223,
"loss": 1.388,
"step": 895
},
{
"epoch": 16.07,
"learning_rate": 0.00013548005477567304,
"loss": 1.2832,
"step": 900
},
{
"epoch": 16.16,
"learning_rate": 0.00012864032674294074,
"loss": 1.2149,
"step": 905
},
{
"epoch": 16.25,
"learning_rate": 0.00011710752518939739,
"loss": 1.0402,
"step": 910
},
{
"epoch": 16.34,
"learning_rate": 0.00010178311565541954,
"loss": 1.2199,
"step": 915
},
{
"epoch": 16.43,
"learning_rate": 8.386493606940354e-05,
"loss": 1.2836,
"step": 920
},
{
"epoch": 16.52,
"learning_rate": 6.47535673195288e-05,
"loss": 1.2179,
"step": 925
},
{
"epoch": 16.61,
"learning_rate": 4.59428563498755e-05,
"loss": 1.2299,
"step": 930
},
{
"epoch": 16.7,
"learning_rate": 2.8903149082065114e-05,
"loss": 1.072,
"step": 935
},
{
"epoch": 16.79,
"learning_rate": 1.4966360302693468e-05,
"loss": 1.2192,
"step": 940
},
{
"epoch": 16.88,
"learning_rate": 5.221864069725844e-06,
"loss": 1.2081,
"step": 945
},
{
"epoch": 16.96,
"learning_rate": 4.3134240132356497e-07,
"loss": 1.3213,
"step": 950
},
{
"epoch": 17.05,
"learning_rate": 9.692481296489572e-07,
"loss": 1.0468,
"step": 955
},
{
"epoch": 17.14,
"learning_rate": 6.793535661893986e-06,
"loss": 1.1071,
"step": 960
},
{
"epoch": 17.23,
"learning_rate": 1.7448947489014885e-05,
"loss": 1.0509,
"step": 965
},
{
"epoch": 17.32,
"learning_rate": 3.210259955104815e-05,
"loss": 0.96,
"step": 970
},
{
"epoch": 17.41,
"learning_rate": 4.9609083916254864e-05,
"loss": 1.1523,
"step": 975
},
{
"epoch": 17.5,
"learning_rate": 6.859999999999973e-05,
"loss": 0.9922,
"step": 980
},
{
"epoch": 17.59,
"learning_rate": 8.759091608374462e-05,
"loss": 1.1601,
"step": 985
},
{
"epoch": 17.68,
"learning_rate": 0.0001050974004489518,
"loss": 0.9979,
"step": 990
},
{
"epoch": 17.77,
"learning_rate": 0.00011975105251098509,
"loss": 1.1048,
"step": 995
},
{
"epoch": 17.86,
"learning_rate": 0.00013040646433810598,
"loss": 1.0999,
"step": 1000
},
{
"epoch": 17.95,
"learning_rate": 0.00013623075187035104,
"loss": 1.2753,
"step": 1005
},
{
"epoch": 18.04,
"learning_rate": 0.0001367686575986765,
"loss": 1.0046,
"step": 1010
},
{
"epoch": 18.12,
"learning_rate": 0.00013197813593027435,
"loss": 0.9683,
"step": 1015
},
{
"epoch": 18.21,
"learning_rate": 0.00012223363969730686,
"loss": 1.0153,
"step": 1020
},
{
"epoch": 18.3,
"learning_rate": 0.00010829685091793493,
"loss": 0.9562,
"step": 1025
},
{
"epoch": 18.39,
"learning_rate": 9.125714365012455e-05,
"loss": 1.0145,
"step": 1030
},
{
"epoch": 18.48,
"learning_rate": 7.244643268047124e-05,
"loss": 1.0414,
"step": 1035
},
{
"epoch": 18.57,
"learning_rate": 5.333506393059651e-05,
"loss": 1.0547,
"step": 1040
},
{
"epoch": 18.66,
"learning_rate": 3.541688434458093e-05,
"loss": 1.038,
"step": 1045
},
{
"epoch": 18.75,
"learning_rate": 2.0092474810602995e-05,
"loss": 0.9887,
"step": 1050
},
{
"epoch": 18.84,
"learning_rate": 8.559673257059497e-06,
"loss": 1.0939,
"step": 1055
},
{
"epoch": 18.93,
"learning_rate": 1.7199452243269606e-06,
"loss": 1.0271,
"step": 1060
},
{
"epoch": 19.02,
"learning_rate": 1.0792048977778093e-07,
"loss": 0.8933,
"step": 1065
},
{
"epoch": 19.11,
"learning_rate": 3.84960354084603e-06,
"loss": 0.9411,
"step": 1070
},
{
"epoch": 19.2,
"learning_rate": 1.2652524389394958e-05,
"loss": 0.8984,
"step": 1075
},
{
"epoch": 19.29,
"learning_rate": 2.5828599592490564e-05,
"loss": 0.8203,
"step": 1080
},
{
"epoch": 19.38,
"learning_rate": 4.234791653975466e-05,
"loss": 0.8312,
"step": 1085
},
{
"epoch": 19.46,
"learning_rate": 6.09192369393131e-05,
"loss": 0.7923,
"step": 1090
},
{
"epoch": 19.55,
"learning_rate": 8.009092691870466e-05,
"loss": 0.8649,
"step": 1095
},
{
"epoch": 19.64,
"learning_rate": 9.836442450346445e-05,
"loss": 0.939,
"step": 1100
},
{
"epoch": 19.73,
"learning_rate": 0.0001143113752475076,
"loss": 0.8501,
"step": 1105
},
{
"epoch": 19.82,
"learning_rate": 0.0001266852800670605,
"loss": 0.9123,
"step": 1110
},
{
"epoch": 19.91,
"learning_rate": 0.0001345189282854337,
"loss": 0.9598,
"step": 1115
},
{
"epoch": 20.0,
"learning_rate": 0.0001372,
"loss": 0.9292,
"step": 1120
},
{
"epoch": 20.09,
"learning_rate": 0.00013451892828543382,
"loss": 0.8035,
"step": 1125
},
{
"epoch": 20.18,
"learning_rate": 0.00012668528006706069,
"loss": 0.792,
"step": 1130
},
{
"epoch": 20.27,
"learning_rate": 0.00011431137524750785,
"loss": 0.8557,
"step": 1135
},
{
"epoch": 20.36,
"learning_rate": 9.836442450346476e-05,
"loss": 0.8628,
"step": 1140
},
{
"epoch": 20.45,
"learning_rate": 8.0090926918705e-05,
"loss": 0.8219,
"step": 1145
},
{
"epoch": 20.54,
"learning_rate": 6.091923693931392e-05,
"loss": 0.8509,
"step": 1150
},
{
"epoch": 20.62,
"learning_rate": 4.234791653975543e-05,
"loss": 0.8817,
"step": 1155
},
{
"epoch": 20.71,
"learning_rate": 2.5828599592491204e-05,
"loss": 0.8186,
"step": 1160
},
{
"epoch": 20.8,
"learning_rate": 1.2652524389394875e-05,
"loss": 0.8184,
"step": 1165
},
{
"epoch": 20.89,
"learning_rate": 3.849603540845984e-06,
"loss": 0.8315,
"step": 1170
},
{
"epoch": 20.98,
"learning_rate": 1.0792048977777332e-07,
"loss": 0.8831,
"step": 1175
},
{
"epoch": 21.07,
"learning_rate": 1.7199452243269987e-06,
"loss": 0.737,
"step": 1180
},
{
"epoch": 21.16,
"learning_rate": 8.559673257059337e-06,
"loss": 0.776,
"step": 1185
},
{
"epoch": 21.25,
"learning_rate": 2.009247481060276e-05,
"loss": 0.6944,
"step": 1190
},
{
"epoch": 21.34,
"learning_rate": 3.541688434458063e-05,
"loss": 0.6937,
"step": 1195
},
{
"epoch": 21.43,
"learning_rate": 5.333506393059618e-05,
"loss": 0.7423,
"step": 1200
},
{
"epoch": 21.52,
"learning_rate": 7.24464326804709e-05,
"loss": 0.7534,
"step": 1205
},
{
"epoch": 21.61,
"learning_rate": 9.125714365012422e-05,
"loss": 0.6734,
"step": 1210
},
{
"epoch": 21.7,
"learning_rate": 0.00010829685091793466,
"loss": 0.7151,
"step": 1215
},
{
"epoch": 21.79,
"learning_rate": 0.00012223363969730635,
"loss": 0.8225,
"step": 1220
},
{
"epoch": 21.88,
"learning_rate": 0.00013197813593027405,
"loss": 0.6728,
"step": 1225
},
{
"epoch": 21.96,
"learning_rate": 0.0001367686575986764,
"loss": 0.7781,
"step": 1230
},
{
"epoch": 22.05,
"learning_rate": 0.00013623075187035101,
"loss": 0.724,
"step": 1235
},
{
"epoch": 22.14,
"learning_rate": 0.00013040646433810593,
"loss": 0.5813,
"step": 1240
},
{
"epoch": 22.23,
"learning_rate": 0.00011975105251098498,
"loss": 0.7065,
"step": 1245
},
{
"epoch": 22.32,
"learning_rate": 0.00010509740044895168,
"loss": 0.6825,
"step": 1250
},
{
"epoch": 22.41,
"learning_rate": 8.759091608374493e-05,
"loss": 0.7155,
"step": 1255
},
{
"epoch": 22.5,
"learning_rate": 6.860000000000005e-05,
"loss": 0.7597,
"step": 1260
},
{
"epoch": 22.59,
"learning_rate": 4.9609083916255196e-05,
"loss": 0.749,
"step": 1265
},
{
"epoch": 22.68,
"learning_rate": 3.2102599551048435e-05,
"loss": 0.7066,
"step": 1270
},
{
"epoch": 22.77,
"learning_rate": 1.7448947489015106e-05,
"loss": 0.6666,
"step": 1275
},
{
"epoch": 22.86,
"learning_rate": 6.7935356618941304e-06,
"loss": 0.6573,
"step": 1280
},
{
"epoch": 22.95,
"learning_rate": 9.692481296490182e-07,
"loss": 0.7112,
"step": 1285
},
{
"epoch": 23.04,
"learning_rate": 4.313424013234736e-07,
"loss": 0.5373,
"step": 1290
},
{
"epoch": 23.12,
"learning_rate": 5.221864069725524e-06,
"loss": 0.5479,
"step": 1295
},
{
"epoch": 23.21,
"learning_rate": 1.4966360302692958e-05,
"loss": 0.5882,
"step": 1300
},
{
"epoch": 23.3,
"learning_rate": 2.8903149082065243e-05,
"loss": 0.5555,
"step": 1305
},
{
"epoch": 23.39,
"learning_rate": 4.5942856349875636e-05,
"loss": 0.5804,
"step": 1310
},
{
"epoch": 23.48,
"learning_rate": 6.475356731952897e-05,
"loss": 0.5679,
"step": 1315
},
{
"epoch": 23.57,
"learning_rate": 8.386493606940368e-05,
"loss": 0.6024,
"step": 1320
},
{
"epoch": 23.66,
"learning_rate": 0.00010178311565541925,
"loss": 0.552,
"step": 1325
},
{
"epoch": 23.75,
"learning_rate": 0.00011710752518939715,
"loss": 0.6404,
"step": 1330
},
{
"epoch": 23.84,
"learning_rate": 0.0001286403267429406,
"loss": 0.581,
"step": 1335
},
{
"epoch": 23.93,
"learning_rate": 0.00013548005477567298,
"loss": 0.6956,
"step": 1340
},
{
"epoch": 24.02,
"learning_rate": 0.00013709207951022223,
"loss": 0.6156,
"step": 1345
},
{
"epoch": 24.11,
"learning_rate": 0.00013335039645915407,
"loss": 0.5472,
"step": 1350
},
{
"epoch": 24.2,
"learning_rate": 0.0001245474756106052,
"loss": 0.5691,
"step": 1355
},
{
"epoch": 24.29,
"learning_rate": 0.00011137140040750965,
"loss": 0.5632,
"step": 1360
},
{
"epoch": 24.38,
"learning_rate": 9.485208346024561e-05,
"loss": 0.6406,
"step": 1365
},
{
"epoch": 24.46,
"learning_rate": 7.628076306068718e-05,
"loss": 0.5618,
"step": 1370
},
{
"epoch": 24.55,
"learning_rate": 5.710907308129514e-05,
"loss": 0.5619,
"step": 1375
},
{
"epoch": 24.64,
"learning_rate": 3.8835575496535365e-05,
"loss": 0.5453,
"step": 1380
},
{
"epoch": 24.73,
"learning_rate": 2.288862475249225e-05,
"loss": 0.6172,
"step": 1385
},
{
"epoch": 24.82,
"learning_rate": 1.0514719932939396e-05,
"loss": 0.5714,
"step": 1390
},
{
"epoch": 24.91,
"learning_rate": 2.681071714566221e-06,
"loss": 0.5045,
"step": 1395
},
{
"epoch": 25.0,
"learning_rate": 0.0,
"loss": 0.5848,
"step": 1400
},
{
"epoch": 25.09,
"learning_rate": 2.681071714566236e-06,
"loss": 0.4364,
"step": 1405
},
{
"epoch": 25.18,
"learning_rate": 1.0514719932939435e-05,
"loss": 0.4791,
"step": 1410
},
{
"epoch": 25.27,
"learning_rate": 2.2888624752492302e-05,
"loss": 0.4854,
"step": 1415
},
{
"epoch": 25.36,
"learning_rate": 3.8835575496535426e-05,
"loss": 0.3908,
"step": 1420
},
{
"epoch": 25.45,
"learning_rate": 5.7109073081295205e-05,
"loss": 0.4823,
"step": 1425
},
{
"epoch": 25.54,
"learning_rate": 7.628076306068627e-05,
"loss": 0.478,
"step": 1430
},
{
"epoch": 25.62,
"learning_rate": 9.485208346024477e-05,
"loss": 0.4768,
"step": 1435
},
{
"epoch": 25.71,
"learning_rate": 0.00011137140040750896,
"loss": 0.4795,
"step": 1440
},
{
"epoch": 25.8,
"learning_rate": 0.0001245474756106047,
"loss": 0.4712,
"step": 1445
},
{
"epoch": 25.89,
"learning_rate": 0.00013335039645915377,
"loss": 0.5162,
"step": 1450
},
{
"epoch": 25.98,
"learning_rate": 0.0001370920795102222,
"loss": 0.6066,
"step": 1455
},
{
"epoch": 26.07,
"learning_rate": 0.0001354800547756732,
"loss": 0.4381,
"step": 1460
},
{
"epoch": 26.16,
"learning_rate": 0.00012864032674294058,
"loss": 0.4356,
"step": 1465
},
{
"epoch": 26.25,
"learning_rate": 0.0001171075251893971,
"loss": 0.4588,
"step": 1470
},
{
"epoch": 26.34,
"learning_rate": 0.00010178311565541919,
"loss": 0.4378,
"step": 1475
},
{
"epoch": 26.43,
"learning_rate": 8.386493606940363e-05,
"loss": 0.4972,
"step": 1480
},
{
"epoch": 26.52,
"learning_rate": 6.47535673195289e-05,
"loss": 0.5317,
"step": 1485
},
{
"epoch": 26.61,
"learning_rate": 4.5942856349875575e-05,
"loss": 0.4449,
"step": 1490
},
{
"epoch": 26.7,
"learning_rate": 2.8903149082065182e-05,
"loss": 0.5294,
"step": 1495
},
{
"epoch": 26.79,
"learning_rate": 1.496636030269352e-05,
"loss": 0.4735,
"step": 1500
},
{
"epoch": 26.88,
"learning_rate": 5.221864069725874e-06,
"loss": 0.4522,
"step": 1505
},
{
"epoch": 26.96,
"learning_rate": 4.313424013235802e-07,
"loss": 0.4696,
"step": 1510
},
{
"epoch": 27.05,
"learning_rate": 9.69248129648866e-07,
"loss": 0.3814,
"step": 1515
},
{
"epoch": 27.14,
"learning_rate": 6.793535661893734e-06,
"loss": 0.3927,
"step": 1520
},
{
"epoch": 27.23,
"learning_rate": 1.7448947489014506e-05,
"loss": 0.4072,
"step": 1525
},
{
"epoch": 27.32,
"learning_rate": 3.210259955104767e-05,
"loss": 0.4177,
"step": 1530
},
{
"epoch": 27.41,
"learning_rate": 4.9609083916255264e-05,
"loss": 0.3721,
"step": 1535
},
{
"epoch": 27.5,
"learning_rate": 6.860000000000014e-05,
"loss": 0.3918,
"step": 1540
},
{
"epoch": 27.59,
"learning_rate": 8.7590916083745e-05,
"loss": 0.391,
"step": 1545
},
{
"epoch": 27.68,
"learning_rate": 0.00010509740044895174,
"loss": 0.368,
"step": 1550
},
{
"epoch": 27.77,
"learning_rate": 0.00011975105251098503,
"loss": 0.4225,
"step": 1555
},
{
"epoch": 27.86,
"learning_rate": 0.00013040646433810595,
"loss": 0.3877,
"step": 1560
},
{
"epoch": 27.95,
"learning_rate": 0.00013623075187035101,
"loss": 0.4493,
"step": 1565
},
{
"epoch": 28.04,
"learning_rate": 0.00013676865759867652,
"loss": 0.4004,
"step": 1570
},
{
"epoch": 28.12,
"learning_rate": 0.0001319781359302744,
"loss": 0.3739,
"step": 1575
},
{
"epoch": 28.21,
"learning_rate": 0.00012223363969730692,
"loss": 0.3801,
"step": 1580
},
{
"epoch": 28.3,
"learning_rate": 0.00010829685091793539,
"loss": 0.4067,
"step": 1585
},
{
"epoch": 28.39,
"learning_rate": 9.125714365012509e-05,
"loss": 0.3734,
"step": 1590
},
{
"epoch": 28.48,
"learning_rate": 7.244643268047182e-05,
"loss": 0.3794,
"step": 1595
},
{
"epoch": 28.57,
"learning_rate": 5.3335063930597066e-05,
"loss": 0.4073,
"step": 1600
},
{
"epoch": 28.66,
"learning_rate": 3.541688434458058e-05,
"loss": 0.4003,
"step": 1605
},
{
"epoch": 28.75,
"learning_rate": 2.0092474810602707e-05,
"loss": 0.3867,
"step": 1610
},
{
"epoch": 28.84,
"learning_rate": 8.559673257059307e-06,
"loss": 0.368,
"step": 1615
},
{
"epoch": 28.93,
"learning_rate": 1.7199452243269835e-06,
"loss": 0.3861,
"step": 1620
},
{
"epoch": 29.02,
"learning_rate": 1.0792048977777332e-07,
"loss": 0.3113,
"step": 1625
},
{
"epoch": 29.11,
"learning_rate": 3.849603540846007e-06,
"loss": 0.2884,
"step": 1630
},
{
"epoch": 29.2,
"learning_rate": 1.2652524389394912e-05,
"loss": 0.341,
"step": 1635
},
{
"epoch": 29.29,
"learning_rate": 2.5828599592490496e-05,
"loss": 0.2955,
"step": 1640
},
{
"epoch": 29.38,
"learning_rate": 4.234791653975459e-05,
"loss": 0.282,
"step": 1645
},
{
"epoch": 29.46,
"learning_rate": 6.0919236939313016e-05,
"loss": 0.2748,
"step": 1650
},
{
"epoch": 29.55,
"learning_rate": 8.009092691870409e-05,
"loss": 0.3101,
"step": 1655
},
{
"epoch": 29.64,
"learning_rate": 9.836442450346394e-05,
"loss": 0.334,
"step": 1660
},
{
"epoch": 29.73,
"learning_rate": 0.00011431137524750716,
"loss": 0.3008,
"step": 1665
},
{
"epoch": 29.82,
"learning_rate": 0.0001266852800670602,
"loss": 0.3448,
"step": 1670
},
{
"epoch": 29.91,
"learning_rate": 0.00013451892828543358,
"loss": 0.3626,
"step": 1675
},
{
"epoch": 30.0,
"learning_rate": 0.0001372,
"loss": 0.3804,
"step": 1680
},
{
"epoch": 30.09,
"learning_rate": 0.0001345189282854337,
"loss": 0.2925,
"step": 1685
},
{
"epoch": 30.18,
"learning_rate": 0.00012668528006706047,
"loss": 0.3217,
"step": 1690
},
{
"epoch": 30.27,
"learning_rate": 0.00011431137524750754,
"loss": 0.3431,
"step": 1695
},
{
"epoch": 30.36,
"learning_rate": 9.83644245034644e-05,
"loss": 0.2936,
"step": 1700
},
{
"epoch": 30.45,
"learning_rate": 8.009092691870459e-05,
"loss": 0.3465,
"step": 1705
},
{
"epoch": 30.54,
"learning_rate": 6.091923693931352e-05,
"loss": 0.3303,
"step": 1710
},
{
"epoch": 30.62,
"learning_rate": 4.234791653975505e-05,
"loss": 0.3262,
"step": 1715
},
{
"epoch": 30.71,
"learning_rate": 2.582859959249089e-05,
"loss": 0.3351,
"step": 1720
},
{
"epoch": 30.8,
"learning_rate": 1.2652524389395202e-05,
"loss": 0.3175,
"step": 1725
},
{
"epoch": 30.89,
"learning_rate": 3.849603540846175e-06,
"loss": 0.2689,
"step": 1730
},
{
"epoch": 30.98,
"learning_rate": 1.0792048977780377e-07,
"loss": 0.3047,
"step": 1735
},
{
"epoch": 31.07,
"learning_rate": 1.7199452243268694e-06,
"loss": 0.2869,
"step": 1740
},
{
"epoch": 31.16,
"learning_rate": 8.559673257059063e-06,
"loss": 0.2604,
"step": 1745
},
{
"epoch": 31.25,
"learning_rate": 2.0092474810602348e-05,
"loss": 0.2769,
"step": 1750
},
{
"epoch": 31.34,
"learning_rate": 3.541688434458014e-05,
"loss": 0.2263,
"step": 1755
},
{
"epoch": 31.43,
"learning_rate": 5.333506393059658e-05,
"loss": 0.2346,
"step": 1760
},
{
"epoch": 31.52,
"learning_rate": 7.244643268047132e-05,
"loss": 0.2531,
"step": 1765
},
{
"epoch": 31.61,
"learning_rate": 9.125714365012463e-05,
"loss": 0.26,
"step": 1770
},
{
"epoch": 31.7,
"learning_rate": 0.00010829685091793499,
"loss": 0.2807,
"step": 1775
},
{
"epoch": 31.79,
"learning_rate": 0.00012223363969730662,
"loss": 0.2235,
"step": 1780
},
{
"epoch": 31.88,
"learning_rate": 0.0001319781359302742,
"loss": 0.2793,
"step": 1785
},
{
"epoch": 31.96,
"learning_rate": 0.00013676865759867644,
"loss": 0.2978,
"step": 1790
},
{
"epoch": 32.05,
"learning_rate": 0.00013623075187035093,
"loss": 0.274,
"step": 1795
},
{
"epoch": 32.14,
"learning_rate": 0.00013040646433810576,
"loss": 0.2887,
"step": 1800
},
{
"epoch": 32.23,
"learning_rate": 0.00011975105251098601,
"loss": 0.2648,
"step": 1805
},
{
"epoch": 32.32,
"learning_rate": 0.00010509740044895298,
"loss": 0.2468,
"step": 1810
},
{
"epoch": 32.41,
"learning_rate": 8.759091608374549e-05,
"loss": 0.2918,
"step": 1815
},
{
"epoch": 32.5,
"learning_rate": 6.860000000000064e-05,
"loss": 0.2799,
"step": 1820
},
{
"epoch": 32.59,
"learning_rate": 4.9609083916255745e-05,
"loss": 0.2562,
"step": 1825
},
{
"epoch": 32.68,
"learning_rate": 3.2102599551048096e-05,
"loss": 0.2663,
"step": 1830
},
{
"epoch": 32.77,
"learning_rate": 1.744894748901484e-05,
"loss": 0.2774,
"step": 1835
},
{
"epoch": 32.86,
"learning_rate": 6.793535661894382e-06,
"loss": 0.2487,
"step": 1840
},
{
"epoch": 32.95,
"learning_rate": 9.692481296491097e-07,
"loss": 0.2686,
"step": 1845
},
{
"epoch": 33.04,
"learning_rate": 4.313424013235193e-07,
"loss": 0.2175,
"step": 1850
},
{
"epoch": 33.12,
"learning_rate": 5.221864069725684e-06,
"loss": 0.218,
"step": 1855
},
{
"epoch": 33.21,
"learning_rate": 1.4966360302693209e-05,
"loss": 0.216,
"step": 1860
},
{
"epoch": 33.3,
"learning_rate": 2.890314908206557e-05,
"loss": 0.2596,
"step": 1865
},
{
"epoch": 33.39,
"learning_rate": 4.5942856349876015e-05,
"loss": 0.2206,
"step": 1870
},
{
"epoch": 33.48,
"learning_rate": 6.475356731952742e-05,
"loss": 0.1793,
"step": 1875
},
{
"epoch": 33.57,
"learning_rate": 8.386493606940219e-05,
"loss": 0.2108,
"step": 1880
},
{
"epoch": 33.66,
"learning_rate": 0.00010178311565541875,
"loss": 0.2287,
"step": 1885
},
{
"epoch": 33.75,
"learning_rate": 0.00011710752518939675,
"loss": 0.1961,
"step": 1890
},
{
"epoch": 33.84,
"learning_rate": 0.00012864032674294034,
"loss": 0.2072,
"step": 1895
},
{
"epoch": 33.93,
"learning_rate": 0.00013548005477567306,
"loss": 0.2405,
"step": 1900
},
{
"epoch": 34.02,
"learning_rate": 0.00013709207951022223,
"loss": 0.2522,
"step": 1905
},
{
"epoch": 34.11,
"learning_rate": 0.00013335039645915423,
"loss": 0.2533,
"step": 1910
},
{
"epoch": 34.2,
"learning_rate": 0.00012454747561060553,
"loss": 0.2087,
"step": 1915
},
{
"epoch": 34.29,
"learning_rate": 0.00011137140040750936,
"loss": 0.2236,
"step": 1920
},
{
"epoch": 34.38,
"learning_rate": 9.485208346024522e-05,
"loss": 0.2418,
"step": 1925
},
{
"epoch": 34.46,
"learning_rate": 7.628076306068678e-05,
"loss": 0.2321,
"step": 1930
},
{
"epoch": 34.55,
"learning_rate": 5.710907308129474e-05,
"loss": 0.2204,
"step": 1935
},
{
"epoch": 34.64,
"learning_rate": 3.8835575496535006e-05,
"loss": 0.253,
"step": 1940
},
{
"epoch": 34.73,
"learning_rate": 2.2888624752493407e-05,
"loss": 0.1896,
"step": 1945
},
{
"epoch": 34.82,
"learning_rate": 1.051471993294022e-05,
"loss": 0.2327,
"step": 1950
},
{
"epoch": 34.91,
"learning_rate": 2.6810717145663806e-06,
"loss": 0.1881,
"step": 1955
},
{
"epoch": 35.0,
"learning_rate": 0.0,
"loss": 0.2131,
"step": 1960
},
{
"epoch": 35.09,
"learning_rate": 2.6810717145660837e-06,
"loss": 0.2035,
"step": 1965
},
{
"epoch": 35.18,
"learning_rate": 1.0514719932939649e-05,
"loss": 0.1686,
"step": 1970
},
{
"epoch": 35.27,
"learning_rate": 2.28886247524926e-05,
"loss": 0.1647,
"step": 1975
},
{
"epoch": 35.36,
"learning_rate": 3.883557549653492e-05,
"loss": 0.1727,
"step": 1980
},
{
"epoch": 35.45,
"learning_rate": 5.710907308129464e-05,
"loss": 0.1767,
"step": 1985
},
{
"epoch": 35.54,
"learning_rate": 7.62807630606867e-05,
"loss": 0.1769,
"step": 1990
},
{
"epoch": 35.62,
"learning_rate": 9.485208346024514e-05,
"loss": 0.1662,
"step": 1995
},
{
"epoch": 35.71,
"learning_rate": 0.00011137140040750926,
"loss": 0.1953,
"step": 2000
},
{
"epoch": 35.8,
"learning_rate": 0.00012454747561060548,
"loss": 0.186,
"step": 2005
},
{
"epoch": 35.89,
"learning_rate": 0.0001333503964591542,
"loss": 0.2162,
"step": 2010
},
{
"epoch": 35.98,
"learning_rate": 0.00013709207951022217,
"loss": 0.215,
"step": 2015
},
{
"epoch": 36.07,
"learning_rate": 0.0001354800547756733,
"loss": 0.2198,
"step": 2020
},
{
"epoch": 36.16,
"learning_rate": 0.00012864032674294085,
"loss": 0.189,
"step": 2025
},
{
"epoch": 36.25,
"learning_rate": 0.00011710752518939751,
"loss": 0.2029,
"step": 2030
},
{
"epoch": 36.34,
"learning_rate": 0.0001017831156554197,
"loss": 0.1957,
"step": 2035
},
{
"epoch": 36.43,
"learning_rate": 8.386493606940322e-05,
"loss": 0.2062,
"step": 2040
},
{
"epoch": 36.52,
"learning_rate": 6.47535673195285e-05,
"loss": 0.2046,
"step": 2045
},
{
"epoch": 36.61,
"learning_rate": 4.594285634987612e-05,
"loss": 0.1769,
"step": 2050
},
{
"epoch": 36.7,
"learning_rate": 2.8903149082065656e-05,
"loss": 0.193,
"step": 2055
},
{
"epoch": 36.79,
"learning_rate": 1.496636030269327e-05,
"loss": 0.1815,
"step": 2060
},
{
"epoch": 36.88,
"learning_rate": 5.2218640697257225e-06,
"loss": 0.1812,
"step": 2065
},
{
"epoch": 36.96,
"learning_rate": 4.3134240132353453e-07,
"loss": 0.177,
"step": 2070
},
{
"epoch": 37.05,
"learning_rate": 9.692481296490944e-07,
"loss": 0.1544,
"step": 2075
},
{
"epoch": 37.14,
"learning_rate": 6.793535661894336e-06,
"loss": 0.1606,
"step": 2080
},
{
"epoch": 37.23,
"learning_rate": 1.7448947489014123e-05,
"loss": 0.1656,
"step": 2085
},
{
"epoch": 37.32,
"learning_rate": 3.210259955104718e-05,
"loss": 0.1505,
"step": 2090
},
{
"epoch": 37.41,
"learning_rate": 4.9609083916254715e-05,
"loss": 0.1481,
"step": 2095
},
{
"epoch": 37.5,
"learning_rate": 6.859999999999957e-05,
"loss": 0.1534,
"step": 2100
},
{
"epoch": 37.59,
"learning_rate": 8.759091608374445e-05,
"loss": 0.1499,
"step": 2105
},
{
"epoch": 37.68,
"learning_rate": 0.00010509740044895207,
"loss": 0.1818,
"step": 2110
},
{
"epoch": 37.77,
"learning_rate": 0.0001197510525109853,
"loss": 0.1553,
"step": 2115
},
{
"epoch": 37.86,
"learning_rate": 0.0001304064643381057,
"loss": 0.1696,
"step": 2120
},
{
"epoch": 37.95,
"learning_rate": 0.0001362307518703509,
"loss": 0.1848,
"step": 2125
},
{
"epoch": 38.04,
"learning_rate": 0.00013676865759867644,
"loss": 0.1581,
"step": 2130
},
{
"epoch": 38.12,
"learning_rate": 0.00013197813593027424,
"loss": 0.1639,
"step": 2135
},
{
"epoch": 38.21,
"learning_rate": 0.00012223363969730668,
"loss": 0.1736,
"step": 2140
},
{
"epoch": 38.3,
"learning_rate": 0.00010829685091793427,
"loss": 0.1716,
"step": 2145
},
{
"epoch": 38.39,
"learning_rate": 9.125714365012379e-05,
"loss": 0.1616,
"step": 2150
},
{
"epoch": 38.48,
"learning_rate": 7.244643268047237e-05,
"loss": 0.1674,
"step": 2155
},
{
"epoch": 38.57,
"learning_rate": 5.333506393059762e-05,
"loss": 0.181,
"step": 2160
},
{
"epoch": 38.66,
"learning_rate": 3.541688434458107e-05,
"loss": 0.1737,
"step": 2165
},
{
"epoch": 38.75,
"learning_rate": 2.009247481060311e-05,
"loss": 0.1692,
"step": 2170
},
{
"epoch": 38.84,
"learning_rate": 8.559673257059581e-06,
"loss": 0.1796,
"step": 2175
},
{
"epoch": 38.93,
"learning_rate": 1.719945224326892e-06,
"loss": 0.149,
"step": 2180
},
{
"epoch": 39.02,
"learning_rate": 1.0792048977779616e-07,
"loss": 0.1672,
"step": 2185
},
{
"epoch": 39.11,
"learning_rate": 3.849603540845817e-06,
"loss": 0.1322,
"step": 2190
},
{
"epoch": 39.2,
"learning_rate": 1.2652524389394578e-05,
"loss": 0.134,
"step": 2195
},
{
"epoch": 39.29,
"learning_rate": 2.5828599592490815e-05,
"loss": 0.1382,
"step": 2200
},
{
"epoch": 39.38,
"learning_rate": 4.234791653975496e-05,
"loss": 0.131,
"step": 2205
},
{
"epoch": 39.46,
"learning_rate": 6.0919236939313415e-05,
"loss": 0.1396,
"step": 2210
},
{
"epoch": 39.55,
"learning_rate": 8.009092691870546e-05,
"loss": 0.1475,
"step": 2215
},
{
"epoch": 39.64,
"learning_rate": 9.836442450346518e-05,
"loss": 0.1374,
"step": 2220
},
{
"epoch": 39.73,
"learning_rate": 0.00011431137524750674,
"loss": 0.1528,
"step": 2225
},
{
"epoch": 39.82,
"learning_rate": 0.0001266852800670599,
"loss": 0.1468,
"step": 2230
},
{
"epoch": 39.91,
"learning_rate": 0.00013451892828543368,
"loss": 0.1589,
"step": 2235
},
{
"epoch": 40.0,
"learning_rate": 0.0001372,
"loss": 0.153,
"step": 2240
},
{
"epoch": 40.09,
"learning_rate": 0.00013451892828543387,
"loss": 0.1516,
"step": 2245
},
{
"epoch": 40.18,
"learning_rate": 0.00012668528006706025,
"loss": 0.1432,
"step": 2250
},
{
"epoch": 40.27,
"learning_rate": 0.00011431137524750726,
"loss": 0.1714,
"step": 2255
},
{
"epoch": 40.36,
"learning_rate": 9.836442450346578e-05,
"loss": 0.1401,
"step": 2260
},
{
"epoch": 40.45,
"learning_rate": 8.009092691870612e-05,
"loss": 0.1529,
"step": 2265
},
{
"epoch": 40.54,
"learning_rate": 6.0919236939314086e-05,
"loss": 0.1557,
"step": 2270
},
{
"epoch": 40.62,
"learning_rate": 4.234791653975558e-05,
"loss": 0.1562,
"step": 2275
},
{
"epoch": 40.71,
"learning_rate": 2.582859959249134e-05,
"loss": 0.1449,
"step": 2280
},
{
"epoch": 40.8,
"learning_rate": 1.2652524389394402e-05,
"loss": 0.1595,
"step": 2285
},
{
"epoch": 40.89,
"learning_rate": 3.849603540846038e-06,
"loss": 0.1353,
"step": 2290
},
{
"epoch": 40.98,
"learning_rate": 1.0792048977783424e-07,
"loss": 0.1344,
"step": 2295
},
{
"epoch": 41.07,
"learning_rate": 1.7199452243265265e-06,
"loss": 0.1224,
"step": 2300
},
{
"epoch": 41.16,
"learning_rate": 8.559673257059253e-06,
"loss": 0.1157,
"step": 2305
},
{
"epoch": 41.25,
"learning_rate": 2.0092474810601945e-05,
"loss": 0.1231,
"step": 2310
},
{
"epoch": 41.34,
"learning_rate": 3.541688434458049e-05,
"loss": 0.1233,
"step": 2315
},
{
"epoch": 41.43,
"learning_rate": 5.3335063930596016e-05,
"loss": 0.1129,
"step": 2320
},
{
"epoch": 41.52,
"learning_rate": 7.244643268047171e-05,
"loss": 0.1297,
"step": 2325
},
{
"epoch": 41.61,
"learning_rate": 9.125714365012407e-05,
"loss": 0.1117,
"step": 2330
},
{
"epoch": 41.7,
"learning_rate": 0.00010829685091793371,
"loss": 0.117,
"step": 2335
},
{
"epoch": 41.79,
"learning_rate": 0.00012223363969730686,
"loss": 0.1357,
"step": 2340
},
{
"epoch": 41.88,
"learning_rate": 0.00013197813593027397,
"loss": 0.1191,
"step": 2345
},
{
"epoch": 41.96,
"learning_rate": 0.00013676865759867647,
"loss": 0.1514,
"step": 2350
},
{
"epoch": 42.05,
"learning_rate": 0.00013623075187035104,
"loss": 0.1221,
"step": 2355
},
{
"epoch": 42.14,
"learning_rate": 0.00013040646433810557,
"loss": 0.1356,
"step": 2360
},
{
"epoch": 42.23,
"learning_rate": 0.0001197510525109864,
"loss": 0.1469,
"step": 2365
},
{
"epoch": 42.32,
"learning_rate": 0.00010509740044895266,
"loss": 0.1364,
"step": 2370
},
{
"epoch": 42.41,
"learning_rate": 8.759091608374603e-05,
"loss": 0.146,
"step": 2375
},
{
"epoch": 42.5,
"learning_rate": 6.860000000000023e-05,
"loss": 0.1439,
"step": 2380
},
{
"epoch": 42.59,
"learning_rate": 4.960908391625629e-05,
"loss": 0.1339,
"step": 2385
},
{
"epoch": 42.68,
"learning_rate": 3.210259955104775e-05,
"loss": 0.1282,
"step": 2390
},
{
"epoch": 42.77,
"learning_rate": 1.744894748901522e-05,
"loss": 0.1341,
"step": 2395
},
{
"epoch": 42.86,
"learning_rate": 6.793535661894625e-06,
"loss": 0.1195,
"step": 2400
},
{
"epoch": 42.95,
"learning_rate": 9.69248129649041e-07,
"loss": 0.128,
"step": 2405
},
{
"epoch": 43.04,
"learning_rate": 4.3134240132345835e-07,
"loss": 0.1227,
"step": 2410
},
{
"epoch": 43.12,
"learning_rate": 5.221864069725836e-06,
"loss": 0.1054,
"step": 2415
},
{
"epoch": 43.21,
"learning_rate": 1.4966360302692852e-05,
"loss": 0.113,
"step": 2420
},
{
"epoch": 43.3,
"learning_rate": 2.8903149082065897e-05,
"loss": 0.1135,
"step": 2425
},
{
"epoch": 43.39,
"learning_rate": 4.594285634987549e-05,
"loss": 0.0998,
"step": 2430
},
{
"epoch": 43.48,
"learning_rate": 6.475356731952782e-05,
"loss": 0.1042,
"step": 2435
},
{
"epoch": 43.57,
"learning_rate": 8.386493606940162e-05,
"loss": 0.1103,
"step": 2440
},
{
"epoch": 43.66,
"learning_rate": 0.0001017831156554191,
"loss": 0.0937,
"step": 2445
},
{
"epoch": 43.75,
"learning_rate": 0.00011710752518939633,
"loss": 0.1037,
"step": 2450
},
{
"epoch": 43.84,
"learning_rate": 0.0001286403267429405,
"loss": 0.1108,
"step": 2455
},
{
"epoch": 43.93,
"learning_rate": 0.00013548005477567295,
"loss": 0.1088,
"step": 2460
},
{
"epoch": 44.02,
"learning_rate": 0.0001370920795102222,
"loss": 0.132,
"step": 2465
},
{
"epoch": 44.11,
"learning_rate": 0.00013335039645915412,
"loss": 0.1115,
"step": 2470
},
{
"epoch": 44.2,
"learning_rate": 0.00012454747561060588,
"loss": 0.1322,
"step": 2475
},
{
"epoch": 44.29,
"learning_rate": 0.00011137140040750903,
"loss": 0.1218,
"step": 2480
},
{
"epoch": 44.38,
"learning_rate": 9.485208346024576e-05,
"loss": 0.1266,
"step": 2485
},
{
"epoch": 44.46,
"learning_rate": 7.628076306068638e-05,
"loss": 0.1337,
"step": 2490
},
{
"epoch": 44.55,
"learning_rate": 5.7109073081295306e-05,
"loss": 0.1423,
"step": 2495
},
{
"epoch": 44.64,
"learning_rate": 3.883557549653465e-05,
"loss": 0.1254,
"step": 2500
},
{
"epoch": 44.73,
"learning_rate": 2.2888624752493833e-05,
"loss": 0.1102,
"step": 2505
},
{
"epoch": 44.82,
"learning_rate": 1.0514719932940006e-05,
"loss": 0.1057,
"step": 2510
},
{
"epoch": 44.91,
"learning_rate": 2.6810717145665407e-06,
"loss": 0.1127,
"step": 2515
},
{
"epoch": 45.0,
"learning_rate": 0.0,
"loss": 0.1,
"step": 2520
},
{
"epoch": 45.09,
"learning_rate": 2.6810717145659236e-06,
"loss": 0.0923,
"step": 2525
},
{
"epoch": 45.18,
"learning_rate": 1.0514719932939862e-05,
"loss": 0.1096,
"step": 2530
},
{
"epoch": 45.27,
"learning_rate": 2.288862475249218e-05,
"loss": 0.1037,
"step": 2535
},
{
"epoch": 45.36,
"learning_rate": 3.88355754965344e-05,
"loss": 0.0967,
"step": 2540
},
{
"epoch": 45.45,
"learning_rate": 5.7109073081295035e-05,
"loss": 0.0873,
"step": 2545
},
{
"epoch": 45.54,
"learning_rate": 7.628076306068611e-05,
"loss": 0.1035,
"step": 2550
},
{
"epoch": 45.62,
"learning_rate": 9.485208346024552e-05,
"loss": 0.0923,
"step": 2555
},
{
"epoch": 45.71,
"learning_rate": 0.00011137140040750883,
"loss": 0.0814,
"step": 2560
},
{
"epoch": 45.8,
"learning_rate": 0.0001245474756106057,
"loss": 0.0889,
"step": 2565
},
{
"epoch": 45.89,
"learning_rate": 0.00013335039645915404,
"loss": 0.1126,
"step": 2570
},
{
"epoch": 45.98,
"learning_rate": 0.00013709207951022217,
"loss": 0.1126,
"step": 2575
},
{
"epoch": 46.07,
"learning_rate": 0.00013548005477567344,
"loss": 0.115,
"step": 2580
},
{
"epoch": 46.16,
"learning_rate": 0.00012864032674294066,
"loss": 0.1107,
"step": 2585
},
{
"epoch": 46.25,
"learning_rate": 0.00011710752518939791,
"loss": 0.1185,
"step": 2590
},
{
"epoch": 46.34,
"learning_rate": 0.00010178311565541935,
"loss": 0.0988,
"step": 2595
},
{
"epoch": 46.43,
"learning_rate": 8.386493606940379e-05,
"loss": 0.106,
"step": 2600
},
{
"epoch": 46.52,
"learning_rate": 6.475356731952809e-05,
"loss": 0.1046,
"step": 2605
},
{
"epoch": 46.61,
"learning_rate": 4.594285634987574e-05,
"loss": 0.1119,
"step": 2610
},
{
"epoch": 46.7,
"learning_rate": 2.890314908206612e-05,
"loss": 0.1081,
"step": 2615
},
{
"epoch": 46.79,
"learning_rate": 1.496636030269302e-05,
"loss": 0.0968,
"step": 2620
},
{
"epoch": 46.88,
"learning_rate": 5.221864069725935e-06,
"loss": 0.1116,
"step": 2625
},
{
"epoch": 46.96,
"learning_rate": 4.3134240132348884e-07,
"loss": 0.1083,
"step": 2630
},
{
"epoch": 47.05,
"learning_rate": 9.692481296489953e-07,
"loss": 0.0932,
"step": 2635
},
{
"epoch": 47.14,
"learning_rate": 6.793535661894512e-06,
"loss": 0.0854,
"step": 2640
},
{
"epoch": 47.23,
"learning_rate": 1.7448947489013744e-05,
"loss": 0.0843,
"step": 2645
},
{
"epoch": 47.32,
"learning_rate": 3.210259955104752e-05,
"loss": 0.0891,
"step": 2650
},
{
"epoch": 47.41,
"learning_rate": 4.960908391625416e-05,
"loss": 0.0791,
"step": 2655
},
{
"epoch": 47.5,
"learning_rate": 6.859999999999997e-05,
"loss": 0.09,
"step": 2660
},
{
"epoch": 47.59,
"learning_rate": 8.75909160837439e-05,
"loss": 0.0897,
"step": 2665
},
{
"epoch": 47.68,
"learning_rate": 0.00010509740044895241,
"loss": 0.0856,
"step": 2670
},
{
"epoch": 47.77,
"learning_rate": 0.00011975105251098491,
"loss": 0.0932,
"step": 2675
},
{
"epoch": 47.86,
"learning_rate": 0.00013040646433810547,
"loss": 0.0998,
"step": 2680
},
{
"epoch": 47.95,
"learning_rate": 0.000136230751870351,
"loss": 0.1009,
"step": 2685
},
{
"epoch": 48.04,
"learning_rate": 0.00013676865759867652,
"loss": 0.0975,
"step": 2690
},
{
"epoch": 48.12,
"learning_rate": 0.00013197813593027408,
"loss": 0.1072,
"step": 2695
},
{
"epoch": 48.21,
"learning_rate": 0.00012223363969730703,
"loss": 0.105,
"step": 2700
},
{
"epoch": 48.3,
"learning_rate": 0.00010829685091793393,
"loss": 0.102,
"step": 2705
},
{
"epoch": 48.39,
"learning_rate": 9.125714365012433e-05,
"loss": 0.0928,
"step": 2710
},
{
"epoch": 48.48,
"learning_rate": 7.244643268047198e-05,
"loss": 0.117,
"step": 2715
},
{
"epoch": 48.57,
"learning_rate": 5.333506393059818e-05,
"loss": 0.0985,
"step": 2720
},
{
"epoch": 48.66,
"learning_rate": 3.541688434458072e-05,
"loss": 0.1006,
"step": 2725
},
{
"epoch": 48.75,
"learning_rate": 2.0092474810603514e-05,
"loss": 0.0912,
"step": 2730
},
{
"epoch": 48.84,
"learning_rate": 8.559673257059383e-06,
"loss": 0.0824,
"step": 2735
},
{
"epoch": 48.93,
"learning_rate": 1.7199452243270216e-06,
"loss": 0.0914,
"step": 2740
},
{
"epoch": 49.02,
"learning_rate": 1.07920489777819e-07,
"loss": 0.103,
"step": 2745
},
{
"epoch": 49.11,
"learning_rate": 3.849603540845946e-06,
"loss": 0.0818,
"step": 2750
},
{
"epoch": 49.2,
"learning_rate": 1.265252438939425e-05,
"loss": 0.083,
"step": 2755
},
{
"epoch": 49.29,
"learning_rate": 2.5828599592491126e-05,
"loss": 0.0802,
"step": 2760
},
{
"epoch": 49.38,
"learning_rate": 4.2347916539754424e-05,
"loss": 0.0776,
"step": 2765
},
{
"epoch": 49.46,
"learning_rate": 6.091923693931382e-05,
"loss": 0.0897,
"step": 2770
},
{
"epoch": 49.55,
"learning_rate": 8.00909269187049e-05,
"loss": 0.0723,
"step": 2775
},
{
"epoch": 49.64,
"learning_rate": 9.836442450346554e-05,
"loss": 0.0913,
"step": 2780
},
{
"epoch": 49.73,
"learning_rate": 0.00011431137524750631,
"loss": 0.0869,
"step": 2785
},
{
"epoch": 49.82,
"learning_rate": 0.00012668528006706012,
"loss": 0.0839,
"step": 2790
},
{
"epoch": 49.91,
"learning_rate": 0.00013451892828543352,
"loss": 0.0867,
"step": 2795
},
{
"epoch": 50.0,
"learning_rate": 0.0001372,
"loss": 0.0961,
"step": 2800
},
{
"epoch": 50.09,
"learning_rate": 0.00013451892828543404,
"loss": 0.0873,
"step": 2805
},
{
"epoch": 50.18,
"learning_rate": 0.00012668528006706004,
"loss": 0.0834,
"step": 2810
},
{
"epoch": 50.27,
"learning_rate": 0.00011431137524750768,
"loss": 0.0971,
"step": 2815
},
{
"epoch": 50.36,
"learning_rate": 9.836442450346542e-05,
"loss": 0.0968,
"step": 2820
},
{
"epoch": 50.45,
"learning_rate": 8.009092691870668e-05,
"loss": 0.0985,
"step": 2825
},
{
"epoch": 50.54,
"learning_rate": 6.0919236939313686e-05,
"loss": 0.092,
"step": 2830
},
{
"epoch": 50.62,
"learning_rate": 4.2347916539756105e-05,
"loss": 0.0898,
"step": 2835
},
{
"epoch": 50.71,
"learning_rate": 2.582859959249102e-05,
"loss": 0.0918,
"step": 2840
},
{
"epoch": 50.8,
"learning_rate": 1.2652524389395302e-05,
"loss": 0.0936,
"step": 2845
},
{
"epoch": 50.89,
"learning_rate": 3.849603540845908e-06,
"loss": 0.0775,
"step": 2850
},
{
"epoch": 50.98,
"learning_rate": 1.0792048977781139e-07,
"loss": 0.086,
"step": 2855
},
{
"epoch": 51.07,
"learning_rate": 1.719945224326618e-06,
"loss": 0.071,
"step": 2860
},
{
"epoch": 51.16,
"learning_rate": 8.559673257059451e-06,
"loss": 0.0789,
"step": 2865
},
{
"epoch": 51.25,
"learning_rate": 2.0092474810602236e-05,
"loss": 0.0844,
"step": 2870
},
{
"epoch": 51.34,
"learning_rate": 3.541688434458084e-05,
"loss": 0.0717,
"step": 2875
},
{
"epoch": 51.43,
"learning_rate": 5.333506393059641e-05,
"loss": 0.0714,
"step": 2880
},
{
"epoch": 51.52,
"learning_rate": 7.244643268047212e-05,
"loss": 0.0735,
"step": 2885
},
{
"epoch": 51.61,
"learning_rate": 9.125714365012261e-05,
"loss": 0.0773,
"step": 2890
},
{
"epoch": 51.7,
"learning_rate": 0.00010829685091793405,
"loss": 0.0718,
"step": 2895
},
{
"epoch": 51.79,
"learning_rate": 0.0001222336396973059,
"loss": 0.0782,
"step": 2900
},
{
"epoch": 51.88,
"learning_rate": 0.00013197813593027413,
"loss": 0.0685,
"step": 2905
},
{
"epoch": 51.96,
"learning_rate": 0.00013676865759867633,
"loss": 0.0792,
"step": 2910
},
{
"epoch": 52.05,
"learning_rate": 0.000136230751870351,
"loss": 0.076,
"step": 2915
},
{
"epoch": 52.14,
"learning_rate": 0.00013040646433810625,
"loss": 0.0802,
"step": 2920
},
{
"epoch": 52.23,
"learning_rate": 0.00011975105251098613,
"loss": 0.0753,
"step": 2925
},
{
"epoch": 52.32,
"learning_rate": 0.0001050974004489523,
"loss": 0.0935,
"step": 2930
},
{
"epoch": 52.41,
"learning_rate": 8.759091608374565e-05,
"loss": 0.0924,
"step": 2935
},
{
"epoch": 52.5,
"learning_rate": 6.859999999999984e-05,
"loss": 0.0846,
"step": 2940
},
{
"epoch": 52.59,
"learning_rate": 4.96090839162559e-05,
"loss": 0.0729,
"step": 2945
},
{
"epoch": 52.68,
"learning_rate": 3.210259955104741e-05,
"loss": 0.0807,
"step": 2950
},
{
"epoch": 52.77,
"learning_rate": 1.7448947489014953e-05,
"loss": 0.0955,
"step": 2955
},
{
"epoch": 52.86,
"learning_rate": 6.793535661894451e-06,
"loss": 0.0765,
"step": 2960
},
{
"epoch": 52.95,
"learning_rate": 9.692481296493e-07,
"loss": 0.0808,
"step": 2965
},
{
"epoch": 53.04,
"learning_rate": 4.3134240132350404e-07,
"loss": 0.0749,
"step": 2970
},
{
"epoch": 53.12,
"learning_rate": 5.221864069725242e-06,
"loss": 0.0756,
"step": 2975
},
{
"epoch": 53.21,
"learning_rate": 1.4966360302693102e-05,
"loss": 0.0691,
"step": 2980
},
{
"epoch": 53.3,
"learning_rate": 2.8903149082064643e-05,
"loss": 0.0656,
"step": 2985
},
{
"epoch": 53.39,
"learning_rate": 4.5942856349875866e-05,
"loss": 0.0701,
"step": 2990
},
{
"epoch": 53.48,
"learning_rate": 6.475356731952822e-05,
"loss": 0.0626,
"step": 2995
},
{
"epoch": 53.57,
"learning_rate": 8.386493606940203e-05,
"loss": 0.066,
"step": 3000
},
{
"epoch": 53.66,
"learning_rate": 0.00010178311565541946,
"loss": 0.0758,
"step": 3005
},
{
"epoch": 53.75,
"learning_rate": 0.00011710752518939663,
"loss": 0.0757,
"step": 3010
},
{
"epoch": 53.84,
"learning_rate": 0.00012864032674294072,
"loss": 0.0673,
"step": 3015
},
{
"epoch": 53.93,
"learning_rate": 0.00013548005477567304,
"loss": 0.0775,
"step": 3020
},
{
"epoch": 54.02,
"learning_rate": 0.00013709207951022217,
"loss": 0.0768,
"step": 3025
},
{
"epoch": 54.11,
"learning_rate": 0.00013335039645915464,
"loss": 0.0761,
"step": 3030
},
{
"epoch": 54.2,
"learning_rate": 0.00012454747561060564,
"loss": 0.0757,
"step": 3035
},
{
"epoch": 54.29,
"learning_rate": 0.00011137140040751024,
"loss": 0.0755,
"step": 3040
},
{
"epoch": 54.38,
"learning_rate": 9.48520834602454e-05,
"loss": 0.0803,
"step": 3045
},
{
"epoch": 54.46,
"learning_rate": 7.628076306068793e-05,
"loss": 0.0809,
"step": 3050
},
{
"epoch": 54.55,
"learning_rate": 5.71090730812949e-05,
"loss": 0.0821,
"step": 3055
},
{
"epoch": 54.64,
"learning_rate": 3.8835575496536036e-05,
"loss": 0.0792,
"step": 3060
},
{
"epoch": 54.73,
"learning_rate": 2.288862475249353e-05,
"loss": 0.073,
"step": 3065
},
{
"epoch": 54.82,
"learning_rate": 1.0514719932939786e-05,
"loss": 0.0754,
"step": 3070
},
{
"epoch": 54.91,
"learning_rate": 2.6810717145664263e-06,
"loss": 0.0837,
"step": 3075
},
{
"epoch": 55.0,
"learning_rate": 0.0,
"loss": 0.0772,
"step": 3080
},
{
"epoch": 55.09,
"learning_rate": 2.681071714566038e-06,
"loss": 0.0661,
"step": 3085
},
{
"epoch": 55.18,
"learning_rate": 1.0514719932940075e-05,
"loss": 0.0746,
"step": 3090
},
{
"epoch": 55.27,
"learning_rate": 2.2888624752492478e-05,
"loss": 0.0666,
"step": 3095
},
{
"epoch": 55.36,
"learning_rate": 3.883557549653476e-05,
"loss": 0.0662,
"step": 3100
},
{
"epoch": 55.45,
"learning_rate": 5.710907308129351e-05,
"loss": 0.0673,
"step": 3105
},
{
"epoch": 55.54,
"learning_rate": 7.628076306068652e-05,
"loss": 0.0648,
"step": 3110
},
{
"epoch": 55.62,
"learning_rate": 9.485208346024408e-05,
"loss": 0.0616,
"step": 3115
},
{
"epoch": 55.71,
"learning_rate": 0.00011137140040750914,
"loss": 0.066,
"step": 3120
},
{
"epoch": 55.8,
"learning_rate": 0.00012454747561060483,
"loss": 0.0587,
"step": 3125
},
{
"epoch": 55.89,
"learning_rate": 0.00013335039645915415,
"loss": 0.0635,
"step": 3130
},
{
"epoch": 55.98,
"learning_rate": 0.0001370920795102222,
"loss": 0.0683,
"step": 3135
},
{
"epoch": 56.07,
"learning_rate": 0.00013548005477567333,
"loss": 0.0734,
"step": 3140
},
{
"epoch": 56.16,
"learning_rate": 0.00012864032674294044,
"loss": 0.0734,
"step": 3145
},
{
"epoch": 56.25,
"learning_rate": 0.00011710752518939762,
"loss": 0.0714,
"step": 3150
},
{
"epoch": 56.34,
"learning_rate": 0.00010178311565541898,
"loss": 0.0763,
"step": 3155
},
{
"epoch": 56.43,
"learning_rate": 8.38649360694034e-05,
"loss": 0.076,
"step": 3160
},
{
"epoch": 56.52,
"learning_rate": 6.47535673195277e-05,
"loss": 0.065,
"step": 3165
},
{
"epoch": 56.61,
"learning_rate": 4.594285634987719e-05,
"loss": 0.0772,
"step": 3170
},
{
"epoch": 56.7,
"learning_rate": 2.8903149082065792e-05,
"loss": 0.0719,
"step": 3175
},
{
"epoch": 56.79,
"learning_rate": 1.4966360302693987e-05,
"loss": 0.0661,
"step": 3180
},
{
"epoch": 56.88,
"learning_rate": 5.2218640697257835e-06,
"loss": 0.0754,
"step": 3185
},
{
"epoch": 56.96,
"learning_rate": 4.313424013236564e-07,
"loss": 0.0747,
"step": 3190
},
{
"epoch": 57.05,
"learning_rate": 9.69248129649064e-07,
"loss": 0.067,
"step": 3195
},
{
"epoch": 57.14,
"learning_rate": 6.793535661893841e-06,
"loss": 0.0616,
"step": 3200
},
{
"epoch": 57.23,
"learning_rate": 1.744894748901401e-05,
"loss": 0.0606,
"step": 3205
},
{
"epoch": 57.32,
"learning_rate": 3.2102599551047865e-05,
"loss": 0.0688,
"step": 3210
},
{
"epoch": 57.41,
"learning_rate": 4.9609083916254546e-05,
"loss": 0.0625,
"step": 3215
},
{
"epoch": 57.5,
"learning_rate": 6.860000000000037e-05,
"loss": 0.0574,
"step": 3220
},
{
"epoch": 57.59,
"learning_rate": 8.75909160837443e-05,
"loss": 0.0555,
"step": 3225
},
{
"epoch": 57.68,
"learning_rate": 0.00010509740044895277,
"loss": 0.0597,
"step": 3230
},
{
"epoch": 57.77,
"learning_rate": 0.00011975105251098518,
"loss": 0.0566,
"step": 3235
},
{
"epoch": 57.86,
"learning_rate": 0.00013040646433810563,
"loss": 0.0627,
"step": 3240
},
{
"epoch": 57.95,
"learning_rate": 0.00013623075187035074,
"loss": 0.0736,
"step": 3245
},
{
"epoch": 58.04,
"learning_rate": 0.00013676865759867647,
"loss": 0.07,
"step": 3250
},
{
"epoch": 58.12,
"learning_rate": 0.00013197813593027467,
"loss": 0.0627,
"step": 3255
},
{
"epoch": 58.21,
"learning_rate": 0.00012223363969730676,
"loss": 0.0709,
"step": 3260
},
{
"epoch": 58.3,
"learning_rate": 0.0001082968509179352,
"loss": 0.0699,
"step": 3265
},
{
"epoch": 58.39,
"learning_rate": 9.125714365012395e-05,
"loss": 0.0791,
"step": 3270
},
{
"epoch": 58.48,
"learning_rate": 7.244643268047157e-05,
"loss": 0.0732,
"step": 3275
},
{
"epoch": 58.57,
"learning_rate": 5.333506393059779e-05,
"loss": 0.0646,
"step": 3280
},
{
"epoch": 58.66,
"learning_rate": 3.541688434458037e-05,
"loss": 0.0675,
"step": 3285
},
{
"epoch": 58.75,
"learning_rate": 2.0092474810603233e-05,
"loss": 0.0586,
"step": 3290
},
{
"epoch": 58.84,
"learning_rate": 8.559673257059193e-06,
"loss": 0.0643,
"step": 3295
},
{
"epoch": 58.93,
"learning_rate": 1.7199452243269301e-06,
"loss": 0.0693,
"step": 3300
},
{
"epoch": 59.02,
"learning_rate": 1.0792048977784186e-07,
"loss": 0.0652,
"step": 3305
},
{
"epoch": 59.11,
"learning_rate": 3.849603540845436e-06,
"loss": 0.058,
"step": 3310
},
{
"epoch": 59.2,
"learning_rate": 1.2652524389394478e-05,
"loss": 0.0575,
"step": 3315
},
{
"epoch": 59.29,
"learning_rate": 2.5828599592489917e-05,
"loss": 0.0572,
"step": 3320
},
{
"epoch": 59.38,
"learning_rate": 4.234791653975481e-05,
"loss": 0.0532,
"step": 3325
},
{
"epoch": 59.46,
"learning_rate": 6.0919236939312284e-05,
"loss": 0.057,
"step": 3330
},
{
"epoch": 59.55,
"learning_rate": 8.00909269187053e-05,
"loss": 0.0541,
"step": 3335
},
{
"epoch": 59.64,
"learning_rate": 9.836442450346415e-05,
"loss": 0.0572,
"step": 3340
},
{
"epoch": 59.73,
"learning_rate": 0.00011431137524750662,
"loss": 0.0581,
"step": 3345
},
{
"epoch": 59.82,
"learning_rate": 0.00012668528006705928,
"loss": 0.0531,
"step": 3350
},
{
"epoch": 59.91,
"learning_rate": 0.00013451892828543363,
"loss": 0.0639,
"step": 3355
},
{
"epoch": 60.0,
"learning_rate": 0.0001372,
"loss": 0.0604,
"step": 3360
}
],
"max_steps": 3360,
"num_train_epochs": 60,
"total_flos": 3472571105280000.0,
"trial_name": null,
"trial_params": null
}