{
"best_metric": null,
"best_model_checkpoint": null,
"epoch": 0.9997658627955982,
"eval_steps": 500,
"global_step": 2135,
"is_hyper_param_search": false,
"is_local_process_zero": true,
"is_world_process_zero": true,
"log_history": [
{
"epoch": 0.0,
"learning_rate": 9.345794392523364e-07,
"loss": 1.2941,
"step": 1
},
{
"epoch": 0.0,
"learning_rate": 4.6728971962616825e-06,
"loss": 1.2342,
"step": 5
},
{
"epoch": 0.0,
"learning_rate": 9.345794392523365e-06,
"loss": 1.3088,
"step": 10
},
{
"epoch": 0.01,
"learning_rate": 1.4018691588785047e-05,
"loss": 1.2922,
"step": 15
},
{
"epoch": 0.01,
"learning_rate": 1.869158878504673e-05,
"loss": 1.2423,
"step": 20
},
{
"epoch": 0.01,
"learning_rate": 2.3364485981308414e-05,
"loss": 1.1562,
"step": 25
},
{
"epoch": 0.01,
"learning_rate": 2.8037383177570094e-05,
"loss": 1.1215,
"step": 30
},
{
"epoch": 0.02,
"learning_rate": 3.2710280373831774e-05,
"loss": 1.1269,
"step": 35
},
{
"epoch": 0.02,
"learning_rate": 3.738317757009346e-05,
"loss": 1.091,
"step": 40
},
{
"epoch": 0.02,
"learning_rate": 4.205607476635514e-05,
"loss": 1.0565,
"step": 45
},
{
"epoch": 0.02,
"learning_rate": 4.672897196261683e-05,
"loss": 1.0861,
"step": 50
},
{
"epoch": 0.03,
"learning_rate": 5.14018691588785e-05,
"loss": 1.0887,
"step": 55
},
{
"epoch": 0.03,
"learning_rate": 5.607476635514019e-05,
"loss": 1.0357,
"step": 60
},
{
"epoch": 0.03,
"learning_rate": 6.074766355140187e-05,
"loss": 1.0388,
"step": 65
},
{
"epoch": 0.03,
"learning_rate": 6.542056074766355e-05,
"loss": 1.0332,
"step": 70
},
{
"epoch": 0.04,
"learning_rate": 7.009345794392523e-05,
"loss": 1.0051,
"step": 75
},
{
"epoch": 0.04,
"learning_rate": 7.476635514018692e-05,
"loss": 1.0493,
"step": 80
},
{
"epoch": 0.04,
"learning_rate": 7.94392523364486e-05,
"loss": 1.0037,
"step": 85
},
{
"epoch": 0.04,
"learning_rate": 8.411214953271028e-05,
"loss": 1.0064,
"step": 90
},
{
"epoch": 0.04,
"learning_rate": 8.878504672897197e-05,
"loss": 1.0123,
"step": 95
},
{
"epoch": 0.05,
"learning_rate": 9.345794392523365e-05,
"loss": 0.996,
"step": 100
},
{
"epoch": 0.05,
"learning_rate": 9.813084112149533e-05,
"loss": 1.0225,
"step": 105
},
{
"epoch": 0.05,
"learning_rate": 0.000102803738317757,
"loss": 1.0023,
"step": 110
},
{
"epoch": 0.05,
"learning_rate": 0.0001074766355140187,
"loss": 1.0279,
"step": 115
},
{
"epoch": 0.06,
"learning_rate": 0.00011214953271028037,
"loss": 0.9937,
"step": 120
},
{
"epoch": 0.06,
"learning_rate": 0.00011682242990654206,
"loss": 1.0253,
"step": 125
},
{
"epoch": 0.06,
"learning_rate": 0.00012149532710280373,
"loss": 0.9973,
"step": 130
},
{
"epoch": 0.06,
"learning_rate": 0.00012616822429906542,
"loss": 1.0207,
"step": 135
},
{
"epoch": 0.07,
"learning_rate": 0.0001308411214953271,
"loss": 0.9923,
"step": 140
},
{
"epoch": 0.07,
"learning_rate": 0.0001355140186915888,
"loss": 1.0081,
"step": 145
},
{
"epoch": 0.07,
"learning_rate": 0.00014018691588785047,
"loss": 0.9948,
"step": 150
},
{
"epoch": 0.07,
"learning_rate": 0.00014485981308411217,
"loss": 1.0017,
"step": 155
},
{
"epoch": 0.07,
"learning_rate": 0.00014953271028037384,
"loss": 0.9957,
"step": 160
},
{
"epoch": 0.08,
"learning_rate": 0.00015420560747663551,
"loss": 0.9777,
"step": 165
},
{
"epoch": 0.08,
"learning_rate": 0.0001588785046728972,
"loss": 1.0305,
"step": 170
},
{
"epoch": 0.08,
"learning_rate": 0.0001635514018691589,
"loss": 1.0018,
"step": 175
},
{
"epoch": 0.08,
"learning_rate": 0.00016822429906542056,
"loss": 0.9955,
"step": 180
},
{
"epoch": 0.09,
"learning_rate": 0.00017289719626168226,
"loss": 1.0194,
"step": 185
},
{
"epoch": 0.09,
"learning_rate": 0.00017757009345794393,
"loss": 1.0042,
"step": 190
},
{
"epoch": 0.09,
"learning_rate": 0.00018224299065420564,
"loss": 0.9807,
"step": 195
},
{
"epoch": 0.09,
"learning_rate": 0.0001869158878504673,
"loss": 0.9986,
"step": 200
},
{
"epoch": 0.1,
"learning_rate": 0.00019158878504672898,
"loss": 1.0198,
"step": 205
},
{
"epoch": 0.1,
"learning_rate": 0.00019626168224299065,
"loss": 1.0206,
"step": 210
},
{
"epoch": 0.1,
"learning_rate": 0.00019999986627428667,
"loss": 1.0095,
"step": 215
},
{
"epoch": 0.1,
"learning_rate": 0.00019999518591187355,
"loss": 0.9929,
"step": 220
},
{
"epoch": 0.11,
"learning_rate": 0.00019998381962144118,
"loss": 1.0345,
"step": 225
},
{
"epoch": 0.11,
"learning_rate": 0.00019996576816296808,
"loss": 0.9902,
"step": 230
},
{
"epoch": 0.11,
"learning_rate": 0.00019994103274341996,
"loss": 0.9747,
"step": 235
},
{
"epoch": 0.11,
"learning_rate": 0.00019990961501666874,
"loss": 1.0159,
"step": 240
},
{
"epoch": 0.11,
"learning_rate": 0.00019987151708338215,
"loss": 0.9714,
"step": 245
},
{
"epoch": 0.12,
"learning_rate": 0.00019982674149088321,
"loss": 0.9779,
"step": 250
},
{
"epoch": 0.12,
"learning_rate": 0.00019977529123297992,
"loss": 1.0151,
"step": 255
},
{
"epoch": 0.12,
"learning_rate": 0.00019971716974976513,
"loss": 1.0168,
"step": 260
},
{
"epoch": 0.12,
"learning_rate": 0.00019965238092738643,
"loss": 0.9975,
"step": 265
},
{
"epoch": 0.13,
"learning_rate": 0.0001995809290977864,
"loss": 0.977,
"step": 270
},
{
"epoch": 0.13,
"learning_rate": 0.00019950281903841294,
"loss": 0.9709,
"step": 275
},
{
"epoch": 0.13,
"learning_rate": 0.00019941805597189978,
"loss": 1.0065,
"step": 280
},
{
"epoch": 0.13,
"learning_rate": 0.00019932664556571744,
"loss": 0.9941,
"step": 285
},
{
"epoch": 0.14,
"learning_rate": 0.00019922859393179404,
"loss": 0.9921,
"step": 290
},
{
"epoch": 0.14,
"learning_rate": 0.0001991239076261069,
"loss": 0.9867,
"step": 295
},
{
"epoch": 0.14,
"learning_rate": 0.00019901259364824402,
"loss": 0.9502,
"step": 300
},
{
"epoch": 0.14,
"learning_rate": 0.00019889465944093607,
"loss": 0.9917,
"step": 305
},
{
"epoch": 0.15,
"learning_rate": 0.00019877011288955897,
"loss": 0.9956,
"step": 310
},
{
"epoch": 0.15,
"learning_rate": 0.00019863896232160636,
"loss": 0.9845,
"step": 315
},
{
"epoch": 0.15,
"learning_rate": 0.00019850121650613294,
"loss": 0.9825,
"step": 320
},
{
"epoch": 0.15,
"learning_rate": 0.00019835688465316817,
"loss": 0.992,
"step": 325
},
{
"epoch": 0.15,
"learning_rate": 0.00019820597641310043,
"loss": 0.9658,
"step": 330
},
{
"epoch": 0.16,
"learning_rate": 0.00019804850187603177,
"loss": 0.9612,
"step": 335
},
{
"epoch": 0.16,
"learning_rate": 0.00019788447157110324,
"loss": 0.9886,
"step": 340
},
{
"epoch": 0.16,
"learning_rate": 0.000197713896465791,
"loss": 0.9608,
"step": 345
},
{
"epoch": 0.16,
"learning_rate": 0.00019753678796517282,
"loss": 0.9718,
"step": 350
},
{
"epoch": 0.17,
"learning_rate": 0.00019735315791116564,
"loss": 0.9944,
"step": 355
},
{
"epoch": 0.17,
"learning_rate": 0.00019716301858173382,
"loss": 0.9956,
"step": 360
},
{
"epoch": 0.17,
"learning_rate": 0.00019696638269006803,
"loss": 0.9601,
"step": 365
},
{
"epoch": 0.17,
"learning_rate": 0.0001967632633837354,
"loss": 0.9937,
"step": 370
},
{
"epoch": 0.18,
"learning_rate": 0.0001965536742438003,
"loss": 0.9991,
"step": 375
},
{
"epoch": 0.18,
"learning_rate": 0.00019633762928391647,
"loss": 0.9942,
"step": 380
},
{
"epoch": 0.18,
"learning_rate": 0.00019611514294938984,
"loss": 0.9772,
"step": 385
},
{
"epoch": 0.18,
"learning_rate": 0.00019588623011621267,
"loss": 0.9883,
"step": 390
},
{
"epoch": 0.18,
"learning_rate": 0.00019565090609006904,
"loss": 0.9767,
"step": 395
},
{
"epoch": 0.19,
"learning_rate": 0.00019540918660531146,
"loss": 0.9879,
"step": 400
},
{
"epoch": 0.19,
"learning_rate": 0.0001951610878239087,
"loss": 0.9539,
"step": 405
},
{
"epoch": 0.19,
"learning_rate": 0.00019490662633436535,
"loss": 0.9673,
"step": 410
},
{
"epoch": 0.19,
"learning_rate": 0.0001946458191506124,
"loss": 0.9867,
"step": 415
},
{
"epoch": 0.2,
"learning_rate": 0.00019437868371087,
"loss": 1.0239,
"step": 420
},
{
"epoch": 0.2,
"learning_rate": 0.0001941052378764812,
"loss": 0.9925,
"step": 425
},
{
"epoch": 0.2,
"learning_rate": 0.00019382549993071784,
"loss": 0.9917,
"step": 430
},
{
"epoch": 0.2,
"learning_rate": 0.00019353948857755803,
"loss": 0.9818,
"step": 435
},
{
"epoch": 0.21,
"learning_rate": 0.00019324722294043558,
"loss": 0.9829,
"step": 440
},
{
"epoch": 0.21,
"learning_rate": 0.00019294872256096142,
"loss": 0.9639,
"step": 445
},
{
"epoch": 0.21,
"learning_rate": 0.0001926440073976168,
"loss": 0.9901,
"step": 450
},
{
"epoch": 0.21,
"learning_rate": 0.00019233309782441907,
"loss": 0.9943,
"step": 455
},
{
"epoch": 0.22,
"learning_rate": 0.00019201601462955921,
"loss": 0.9851,
"step": 460
},
{
"epoch": 0.22,
"learning_rate": 0.00019169277901401213,
"loss": 0.9705,
"step": 465
},
{
"epoch": 0.22,
"learning_rate": 0.00019136341259011876,
"loss": 1.0047,
"step": 470
},
{
"epoch": 0.22,
"learning_rate": 0.00019102793738014133,
"loss": 0.9877,
"step": 475
},
{
"epoch": 0.22,
"learning_rate": 0.00019068637581479072,
"loss": 0.9798,
"step": 480
},
{
"epoch": 0.23,
"learning_rate": 0.00019033875073172678,
"loss": 0.9738,
"step": 485
},
{
"epoch": 0.23,
"learning_rate": 0.00018998508537403127,
"loss": 0.9755,
"step": 490
},
{
"epoch": 0.23,
"learning_rate": 0.0001896254033886538,
"loss": 0.9878,
"step": 495
},
{
"epoch": 0.23,
"learning_rate": 0.0001892597288248308,
"loss": 0.9731,
"step": 500
},
{
"epoch": 0.24,
"learning_rate": 0.0001888880861324774,
"loss": 0.9863,
"step": 505
},
{
"epoch": 0.24,
"learning_rate": 0.00018851050016055275,
"loss": 0.9749,
"step": 510
},
{
"epoch": 0.24,
"learning_rate": 0.0001881269961553986,
"loss": 0.9991,
"step": 515
},
{
"epoch": 0.24,
"learning_rate": 0.00018773759975905098,
"loss": 0.9769,
"step": 520
},
{
"epoch": 0.25,
"learning_rate": 0.00018734233700752617,
"loss": 0.9751,
"step": 525
},
{
"epoch": 0.25,
"learning_rate": 0.00018694123432907945,
"loss": 0.9783,
"step": 530
},
{
"epoch": 0.25,
"learning_rate": 0.00018653431854243828,
"loss": 0.9968,
"step": 535
},
{
"epoch": 0.25,
"learning_rate": 0.00018612161685500905,
"loss": 0.988,
"step": 540
},
{
"epoch": 0.26,
"learning_rate": 0.00018570315686105787,
"loss": 0.9802,
"step": 545
},
{
"epoch": 0.26,
"learning_rate": 0.00018527896653986576,
"loss": 0.9667,
"step": 550
},
{
"epoch": 0.26,
"learning_rate": 0.0001848490742538576,
"loss": 0.9736,
"step": 555
},
{
"epoch": 0.26,
"learning_rate": 0.00018441350874670595,
"loss": 0.9777,
"step": 560
},
{
"epoch": 0.26,
"learning_rate": 0.0001839722991414091,
"loss": 0.9931,
"step": 565
},
{
"epoch": 0.27,
"learning_rate": 0.00018352547493834384,
"loss": 0.9518,
"step": 570
},
{
"epoch": 0.27,
"learning_rate": 0.00018307306601329302,
"loss": 0.9664,
"step": 575
},
{
"epoch": 0.27,
"learning_rate": 0.00018261510261544792,
"loss": 0.9788,
"step": 580
},
{
"epoch": 0.27,
"learning_rate": 0.00018215161536538582,
"loss": 0.9746,
"step": 585
},
{
"epoch": 0.28,
"learning_rate": 0.00018168263525302248,
"loss": 0.9983,
"step": 590
},
{
"epoch": 0.28,
"learning_rate": 0.0001812081936355402,
"loss": 0.9968,
"step": 595
},
{
"epoch": 0.28,
"learning_rate": 0.0001807283222352912,
"loss": 0.9762,
"step": 600
},
{
"epoch": 0.28,
"learning_rate": 0.00018024305313767646,
"loss": 0.9725,
"step": 605
},
{
"epoch": 0.29,
"learning_rate": 0.00017975241878900059,
"loss": 0.9516,
"step": 610
},
{
"epoch": 0.29,
"learning_rate": 0.00017925645199430215,
"loss": 0.9665,
"step": 615
},
{
"epoch": 0.29,
"learning_rate": 0.0001787551859151606,
"loss": 0.989,
"step": 620
},
{
"epoch": 0.29,
"learning_rate": 0.00017824865406747856,
"loss": 0.994,
"step": 625
},
{
"epoch": 0.3,
"learning_rate": 0.0001777368903192412,
"loss": 0.99,
"step": 630
},
{
"epoch": 0.3,
"learning_rate": 0.00017721992888825166,
"loss": 0.9745,
"step": 635
},
{
"epoch": 0.3,
"learning_rate": 0.00017669780433984312,
"loss": 0.9517,
"step": 640
},
{
"epoch": 0.3,
"learning_rate": 0.00017617055158456761,
"loss": 0.9651,
"step": 645
},
{
"epoch": 0.3,
"learning_rate": 0.00017563820587586205,
"loss": 0.9391,
"step": 650
},
{
"epoch": 0.31,
"learning_rate": 0.00017510080280769082,
"loss": 0.9695,
"step": 655
},
{
"epoch": 0.31,
"learning_rate": 0.00017455837831216605,
"loss": 0.9776,
"step": 660
},
{
"epoch": 0.31,
"learning_rate": 0.000174010968657145,
"loss": 0.979,
"step": 665
},
{
"epoch": 0.31,
"learning_rate": 0.00017345861044380524,
"loss": 0.971,
"step": 670
},
{
"epoch": 0.32,
"learning_rate": 0.0001729013406041972,
"loss": 0.9787,
"step": 675
},
{
"epoch": 0.32,
"learning_rate": 0.00017233919639877504,
"loss": 0.9693,
"step": 680
},
{
"epoch": 0.32,
"learning_rate": 0.00017177221541390513,
"loss": 0.9923,
"step": 685
},
{
"epoch": 0.32,
"learning_rate": 0.00017120043555935298,
"loss": 0.9755,
"step": 690
},
{
"epoch": 0.33,
"learning_rate": 0.00017062389506574855,
"loss": 0.9742,
"step": 695
},
{
"epoch": 0.33,
"learning_rate": 0.0001700426324820301,
"loss": 0.9506,
"step": 700
},
{
"epoch": 0.33,
"learning_rate": 0.00016945668667286647,
"loss": 0.976,
"step": 705
},
{
"epoch": 0.33,
"learning_rate": 0.00016886609681605877,
"loss": 0.9565,
"step": 710
},
{
"epoch": 0.33,
"learning_rate": 0.00016827090239992072,
"loss": 0.9903,
"step": 715
},
{
"epoch": 0.34,
"learning_rate": 0.0001676711432206384,
"loss": 0.9854,
"step": 720
},
{
"epoch": 0.34,
"learning_rate": 0.00016706685937960933,
"loss": 0.9799,
"step": 725
},
{
"epoch": 0.34,
"learning_rate": 0.00016645809128076127,
"loss": 0.9656,
"step": 730
},
{
"epoch": 0.34,
"learning_rate": 0.00016584487962785055,
"loss": 0.9596,
"step": 735
},
{
"epoch": 0.35,
"learning_rate": 0.00016522726542174073,
"loss": 0.9823,
"step": 740
},
{
"epoch": 0.35,
"learning_rate": 0.0001646052899576611,
"loss": 0.9761,
"step": 745
},
{
"epoch": 0.35,
"learning_rate": 0.00016397899482244549,
"loss": 0.9734,
"step": 750
},
{
"epoch": 0.35,
"learning_rate": 0.00016334842189175174,
"loss": 0.9688,
"step": 755
},
{
"epoch": 0.36,
"learning_rate": 0.00016271361332726185,
"loss": 0.9618,
"step": 760
},
{
"epoch": 0.36,
"learning_rate": 0.0001620746115738628,
"loss": 0.9534,
"step": 765
},
{
"epoch": 0.36,
"learning_rate": 0.0001614314593568087,
"loss": 0.9783,
"step": 770
},
{
"epoch": 0.36,
"learning_rate": 0.00016078419967886402,
"loss": 0.9581,
"step": 775
},
{
"epoch": 0.37,
"learning_rate": 0.00016013287581742835,
"loss": 0.9617,
"step": 780
},
{
"epoch": 0.37,
"learning_rate": 0.00015947753132164276,
"loss": 0.9949,
"step": 785
},
{
"epoch": 0.37,
"learning_rate": 0.0001588182100094779,
"loss": 0.9943,
"step": 790
},
{
"epoch": 0.37,
"learning_rate": 0.0001581549559648044,
"loss": 0.971,
"step": 795
},
{
"epoch": 0.37,
"learning_rate": 0.0001574878135344451,
"loss": 0.9698,
"step": 800
},
{
"epoch": 0.38,
"learning_rate": 0.00015681682732521015,
"loss": 0.9898,
"step": 805
},
{
"epoch": 0.38,
"learning_rate": 0.0001561420422009143,
"loss": 0.9851,
"step": 810
},
{
"epoch": 0.38,
"learning_rate": 0.00015546350327937724,
"loss": 0.9624,
"step": 815
},
{
"epoch": 0.38,
"learning_rate": 0.00015478125592940692,
"loss": 0.9884,
"step": 820
},
{
"epoch": 0.39,
"learning_rate": 0.00015409534576776618,
"loss": 0.9695,
"step": 825
},
{
"epoch": 0.39,
"learning_rate": 0.00015340581865612245,
"loss": 0.9812,
"step": 830
},
{
"epoch": 0.39,
"learning_rate": 0.00015271272069798157,
"loss": 0.9434,
"step": 835
},
{
"epoch": 0.39,
"learning_rate": 0.00015201609823560505,
"loss": 0.9697,
"step": 840
},
{
"epoch": 0.4,
"learning_rate": 0.0001513159978469117,
"loss": 1.0019,
"step": 845
},
{
"epoch": 0.4,
"learning_rate": 0.00015061246634236297,
"loss": 0.9909,
"step": 850
},
{
"epoch": 0.4,
"learning_rate": 0.0001499055507618334,
"loss": 0.9861,
"step": 855
},
{
"epoch": 0.4,
"learning_rate": 0.00014919529837146528,
"loss": 0.9814,
"step": 860
},
{
"epoch": 0.41,
"learning_rate": 0.00014848175666050842,
"loss": 0.9709,
"step": 865
},
{
"epoch": 0.41,
"learning_rate": 0.00014776497333814477,
"loss": 0.9609,
"step": 870
},
{
"epoch": 0.41,
"learning_rate": 0.0001470449963302984,
"loss": 0.9776,
"step": 875
},
{
"epoch": 0.41,
"learning_rate": 0.00014632187377643143,
"loss": 0.9853,
"step": 880
},
{
"epoch": 0.41,
"learning_rate": 0.0001455956540263248,
"loss": 0.9712,
"step": 885
},
{
"epoch": 0.42,
"learning_rate": 0.0001448663856368459,
"loss": 1.0023,
"step": 890
},
{
"epoch": 0.42,
"learning_rate": 0.00014413411736870164,
"loss": 0.9827,
"step": 895
},
{
"epoch": 0.42,
"learning_rate": 0.00014339889818317843,
"loss": 0.9673,
"step": 900
},
{
"epoch": 0.42,
"learning_rate": 0.0001426607772388683,
"loss": 0.9884,
"step": 905
},
{
"epoch": 0.43,
"learning_rate": 0.00014191980388838217,
"loss": 0.9492,
"step": 910
},
{
"epoch": 0.43,
"learning_rate": 0.00014117602767504996,
"loss": 0.9896,
"step": 915
},
{
"epoch": 0.43,
"learning_rate": 0.00014042949832960799,
"loss": 0.9707,
"step": 920
},
{
"epoch": 0.43,
"learning_rate": 0.0001396802657668739,
"loss": 0.9692,
"step": 925
},
{
"epoch": 0.44,
"learning_rate": 0.00013892838008240917,
"loss": 0.9834,
"step": 930
},
{
"epoch": 0.44,
"learning_rate": 0.00013817389154916962,
"loss": 0.9739,
"step": 935
},
{
"epoch": 0.44,
"learning_rate": 0.00013741685061414405,
"loss": 0.9597,
"step": 940
},
{
"epoch": 0.44,
"learning_rate": 0.0001366573078949813,
"loss": 0.9908,
"step": 945
},
{
"epoch": 0.44,
"learning_rate": 0.0001358953141766056,
"loss": 0.9623,
"step": 950
},
{
"epoch": 0.45,
"learning_rate": 0.0001351309204078212,
"loss": 0.9579,
"step": 955
},
{
"epoch": 0.45,
"learning_rate": 0.00013436417769790578,
"loss": 0.986,
"step": 960
},
{
"epoch": 0.45,
"learning_rate": 0.00013359513731319293,
"loss": 0.9642,
"step": 965
},
{
"epoch": 0.45,
"learning_rate": 0.00013282385067364466,
"loss": 0.9835,
"step": 970
},
{
"epoch": 0.46,
"learning_rate": 0.00013205036934941318,
"loss": 0.9459,
"step": 975
},
{
"epoch": 0.46,
"learning_rate": 0.0001312747450573927,
"loss": 0.9774,
"step": 980
},
{
"epoch": 0.46,
"learning_rate": 0.00013049702965776174,
"loss": 0.9404,
"step": 985
},
{
"epoch": 0.46,
"learning_rate": 0.00012971727515051536,
"loss": 0.9711,
"step": 990
},
{
"epoch": 0.47,
"learning_rate": 0.00012893553367198862,
"loss": 0.9812,
"step": 995
},
{
"epoch": 0.47,
"learning_rate": 0.0001281518574913703,
"loss": 0.9789,
"step": 1000
},
{
"epoch": 0.47,
"learning_rate": 0.0001273662990072083,
"loss": 0.9795,
"step": 1005
},
{
"epoch": 0.47,
"learning_rate": 0.00012657891074390593,
"loss": 0.9789,
"step": 1010
},
{
"epoch": 0.48,
"learning_rate": 0.00012578974534821022,
"loss": 0.9839,
"step": 1015
},
{
"epoch": 0.48,
"learning_rate": 0.0001249988555856916,
"loss": 0.9711,
"step": 1020
},
{
"epoch": 0.48,
"learning_rate": 0.0001242062943372161,
"loss": 0.9792,
"step": 1025
},
{
"epoch": 0.48,
"learning_rate": 0.0001234121145954094,
"loss": 0.965,
"step": 1030
},
{
"epoch": 0.48,
"learning_rate": 0.00012261636946111367,
"loss": 0.9506,
"step": 1035
},
{
"epoch": 0.49,
"learning_rate": 0.00012181911213983719,
"loss": 0.9823,
"step": 1040
},
{
"epoch": 0.49,
"learning_rate": 0.00012102039593819681,
"loss": 0.9525,
"step": 1045
},
{
"epoch": 0.49,
"learning_rate": 0.00012022027426035378,
"loss": 0.9684,
"step": 1050
},
{
"epoch": 0.49,
"learning_rate": 0.00011941880060444297,
"loss": 0.9582,
"step": 1055
},
{
"epoch": 0.5,
"learning_rate": 0.00011861602855899594,
"loss": 0.9762,
"step": 1060
},
{
"epoch": 0.5,
"learning_rate": 0.00011781201179935777,
"loss": 0.9555,
"step": 1065
},
{
"epoch": 0.5,
"learning_rate": 0.00011700680408409825,
"loss": 0.9698,
"step": 1070
},
{
"epoch": 0.5,
"learning_rate": 0.00011620045925141747,
"loss": 0.9931,
"step": 1075
},
{
"epoch": 0.51,
"learning_rate": 0.00011539303121554599,
"loss": 0.9601,
"step": 1080
},
{
"epoch": 0.51,
"learning_rate": 0.00011458457396314009,
"loss": 0.9903,
"step": 1085
},
{
"epoch": 0.51,
"learning_rate": 0.00011377514154967195,
"loss": 0.9675,
"step": 1090
},
{
"epoch": 0.51,
"learning_rate": 0.00011296478809581552,
"loss": 0.9869,
"step": 1095
},
{
"epoch": 0.52,
"learning_rate": 0.00011215356778382788,
"loss": 0.9511,
"step": 1100
},
{
"epoch": 0.52,
"learning_rate": 0.00011134153485392632,
"loss": 0.975,
"step": 1105
},
{
"epoch": 0.52,
"learning_rate": 0.00011052874360066177,
"loss": 0.9483,
"step": 1110
},
{
"epoch": 0.52,
"learning_rate": 0.0001097152483692886,
"loss": 0.9664,
"step": 1115
},
{
"epoch": 0.52,
"learning_rate": 0.00010890110355213092,
"loss": 0.9699,
"step": 1120
},
{
"epoch": 0.53,
"learning_rate": 0.00010808636358494584,
"loss": 0.9845,
"step": 1125
},
{
"epoch": 0.53,
"learning_rate": 0.00010727108294328352,
"loss": 0.9556,
"step": 1130
},
{
"epoch": 0.53,
"learning_rate": 0.00010645531613884508,
"loss": 0.9756,
"step": 1135
},
{
"epoch": 0.53,
"learning_rate": 0.00010563911771583767,
"loss": 0.9569,
"step": 1140
},
{
"epoch": 0.54,
"learning_rate": 0.00010482254224732757,
"loss": 0.9798,
"step": 1145
},
{
"epoch": 0.54,
"learning_rate": 0.0001040056443315912,
"loss": 0.9716,
"step": 1150
},
{
"epoch": 0.54,
"learning_rate": 0.0001031884785884647,
"loss": 0.978,
"step": 1155
},
{
"epoch": 0.54,
"learning_rate": 0.00010237109965569171,
"loss": 0.9673,
"step": 1160
},
{
"epoch": 0.55,
"learning_rate": 0.00010155356218527036,
"loss": 0.959,
"step": 1165
},
{
"epoch": 0.55,
"learning_rate": 0.0001007359208397989,
"loss": 0.9847,
"step": 1170
},
{
"epoch": 0.55,
"learning_rate": 9.991823028882101e-05,
"loss": 0.9547,
"step": 1175
},
{
"epoch": 0.55,
"learning_rate": 9.91005452051704e-05,
"loss": 0.9719,
"step": 1180
},
{
"epoch": 0.55,
"learning_rate": 9.828292026131506e-05,
"loss": 0.9806,
"step": 1185
},
{
"epoch": 0.56,
"learning_rate": 9.74654101257021e-05,
"loss": 0.9824,
"step": 1190
},
{
"epoch": 0.56,
"learning_rate": 9.664806945910209e-05,
"loss": 0.9558,
"step": 1195
},
{
"epoch": 0.56,
"learning_rate": 9.583095291095453e-05,
"loss": 0.946,
"step": 1200
},
{
"epoch": 0.56,
"learning_rate": 9.50141151157139e-05,
"loss": 0.981,
"step": 1205
},
{
"epoch": 0.57,
"learning_rate": 9.419761068919636e-05,
"loss": 0.959,
"step": 1210
},
{
"epoch": 0.57,
"learning_rate": 9.338149422492843e-05,
"loss": 0.9484,
"step": 1215
},
{
"epoch": 0.57,
"learning_rate": 9.256582029049634e-05,
"loss": 0.9838,
"step": 1220
},
{
"epoch": 0.57,
"learning_rate": 9.175064342389769e-05,
"loss": 0.9326,
"step": 1225
},
{
"epoch": 0.58,
"learning_rate": 9.093601812989503e-05,
"loss": 0.9819,
"step": 1230
},
{
"epoch": 0.58,
"learning_rate": 9.01219988763712e-05,
"loss": 0.9508,
"step": 1235
},
{
"epoch": 0.58,
"learning_rate": 8.93086400906877e-05,
"loss": 0.9967,
"step": 1240
},
{
"epoch": 0.58,
"learning_rate": 8.849599615604558e-05,
"loss": 0.973,
"step": 1245
},
{
"epoch": 0.59,
"learning_rate": 8.768412140784913e-05,
"loss": 0.9597,
"step": 1250
},
{
"epoch": 0.59,
"learning_rate": 8.687307013007288e-05,
"loss": 0.9484,
"step": 1255
},
{
"epoch": 0.59,
"learning_rate": 8.60628965516321e-05,
"loss": 0.972,
"step": 1260
},
{
"epoch": 0.59,
"learning_rate": 8.525365484275686e-05,
"loss": 0.9596,
"step": 1265
},
{
"epoch": 0.59,
"learning_rate": 8.444539911137016e-05,
"loss": 0.9844,
"step": 1270
},
{
"epoch": 0.6,
"learning_rate": 8.363818339946998e-05,
"loss": 0.9506,
"step": 1275
},
{
"epoch": 0.6,
"learning_rate": 8.283206167951608e-05,
"loss": 0.961,
"step": 1280
},
{
"epoch": 0.6,
"learning_rate": 8.202708785082121e-05,
"loss": 0.957,
"step": 1285
},
{
"epoch": 0.6,
"learning_rate": 8.122331573594713e-05,
"loss": 0.9625,
"step": 1290
},
{
"epoch": 0.61,
"learning_rate": 8.042079907710615e-05,
"loss": 0.9641,
"step": 1295
},
{
"epoch": 0.61,
"learning_rate": 7.961959153256751e-05,
"loss": 0.9851,
"step": 1300
},
{
"epoch": 0.61,
"learning_rate": 7.881974667306988e-05,
"loss": 0.9625,
"step": 1305
},
{
"epoch": 0.61,
"learning_rate": 7.802131797823943e-05,
"loss": 0.9433,
"step": 1310
},
{
"epoch": 0.62,
"learning_rate": 7.722435883301398e-05,
"loss": 0.9651,
"step": 1315
},
{
"epoch": 0.62,
"learning_rate": 7.64289225240735e-05,
"loss": 0.9537,
"step": 1320
},
{
"epoch": 0.62,
"learning_rate": 7.56350622362775e-05,
"loss": 0.9622,
"step": 1325
},
{
"epoch": 0.62,
"learning_rate": 7.484283104910862e-05,
"loss": 0.9533,
"step": 1330
},
{
"epoch": 0.63,
"learning_rate": 7.405228193312394e-05,
"loss": 0.9717,
"step": 1335
},
{
"epoch": 0.63,
"learning_rate": 7.326346774641287e-05,
"loss": 0.9514,
"step": 1340
},
{
"epoch": 0.63,
"learning_rate": 7.247644123106315e-05,
"loss": 0.9715,
"step": 1345
},
{
"epoch": 0.63,
"learning_rate": 7.169125500963446e-05,
"loss": 0.9503,
"step": 1350
},
{
"epoch": 0.63,
"learning_rate": 7.090796158163974e-05,
"loss": 0.9575,
"step": 1355
},
{
"epoch": 0.64,
"learning_rate": 7.0126613320035e-05,
"loss": 0.9616,
"step": 1360
},
{
"epoch": 0.64,
"learning_rate": 6.934726246771768e-05,
"loss": 0.9473,
"step": 1365
},
{
"epoch": 0.64,
"learning_rate": 6.85699611340333e-05,
"loss": 0.9474,
"step": 1370
},
{
"epoch": 0.64,
"learning_rate": 6.779476129129158e-05,
"loss": 0.9823,
"step": 1375
},
{
"epoch": 0.65,
"learning_rate": 6.702171477129121e-05,
"loss": 0.9705,
"step": 1380
},
{
"epoch": 0.65,
"learning_rate": 6.625087326185435e-05,
"loss": 0.9633,
"step": 1385
},
{
"epoch": 0.65,
"learning_rate": 6.548228830337071e-05,
"loss": 0.9623,
"step": 1390
},
{
"epoch": 0.65,
"learning_rate": 6.47160112853513e-05,
"loss": 0.9751,
"step": 1395
},
{
"epoch": 0.66,
"learning_rate": 6.395209344299243e-05,
"loss": 0.9896,
"step": 1400
},
{
"epoch": 0.66,
"learning_rate": 6.319058585375015e-05,
"loss": 0.972,
"step": 1405
},
{
"epoch": 0.66,
"learning_rate": 6.243153943392483e-05,
"loss": 0.9388,
"step": 1410
},
{
"epoch": 0.66,
"learning_rate": 6.167500493525706e-05,
"loss": 0.9666,
"step": 1415
},
{
"epoch": 0.66,
"learning_rate": 6.092103294153395e-05,
"loss": 0.9605,
"step": 1420
},
{
"epoch": 0.67,
"learning_rate": 6.016967386520713e-05,
"loss": 0.9732,
"step": 1425
},
{
"epoch": 0.67,
"learning_rate": 5.9420977944022194e-05,
"loss": 0.9875,
"step": 1430
},
{
"epoch": 0.67,
"learning_rate": 5.867499523765942e-05,
"loss": 0.9626,
"step": 1435
},
{
"epoch": 0.67,
"learning_rate": 5.7931775624386696e-05,
"loss": 0.974,
"step": 1440
},
{
"epoch": 0.68,
"learning_rate": 5.719136879772476e-05,
"loss": 0.9973,
"step": 1445
},
{
"epoch": 0.68,
"learning_rate": 5.645382426312431e-05,
"loss": 0.9448,
"step": 1450
},
{
"epoch": 0.68,
"learning_rate": 5.571919133465605e-05,
"loss": 0.9357,
"step": 1455
},
{
"epoch": 0.68,
"learning_rate": 5.498751913171348e-05,
"loss": 0.9627,
"step": 1460
},
{
"epoch": 0.69,
"learning_rate": 5.425885657572843e-05,
"loss": 0.9628,
"step": 1465
},
{
"epoch": 0.69,
"learning_rate": 5.3533252386900445e-05,
"loss": 0.9533,
"step": 1470
},
{
"epoch": 0.69,
"learning_rate": 5.281075508093889e-05,
"loss": 0.9504,
"step": 1475
},
{
"epoch": 0.69,
"learning_rate": 5.209141296581903e-05,
"loss": 0.9753,
"step": 1480
},
{
"epoch": 0.7,
"learning_rate": 5.137527413855252e-05,
"loss": 0.987,
"step": 1485
},
{
"epoch": 0.7,
"learning_rate": 5.066238648197081e-05,
"loss": 0.9582,
"step": 1490
},
{
"epoch": 0.7,
"learning_rate": 4.9952797661524254e-05,
"loss": 0.976,
"step": 1495
},
{
"epoch": 0.7,
"learning_rate": 4.9246555122094664e-05,
"loss": 0.9696,
"step": 1500
},
{
"epoch": 0.7,
"learning_rate": 4.8543706084823015e-05,
"loss": 0.9874,
"step": 1505
},
{
"epoch": 0.71,
"learning_rate": 4.784429754395252e-05,
"loss": 0.9728,
"step": 1510
},
{
"epoch": 0.71,
"learning_rate": 4.714837626368594e-05,
"loss": 0.9651,
"step": 1515
},
{
"epoch": 0.71,
"learning_rate": 4.6455988775059186e-05,
"loss": 0.9734,
"step": 1520
},
{
"epoch": 0.71,
"learning_rate": 4.576718137283016e-05,
"loss": 0.954,
"step": 1525
},
{
"epoch": 0.72,
"learning_rate": 4.508200011238295e-05,
"loss": 0.9678,
"step": 1530
},
{
"epoch": 0.72,
"learning_rate": 4.440049080664904e-05,
"loss": 0.9681,
"step": 1535
},
{
"epoch": 0.72,
"learning_rate": 4.372269902304363e-05,
"loss": 0.9529,
"step": 1540
},
{
"epoch": 0.72,
"learning_rate": 4.304867008041921e-05,
"loss": 0.9926,
"step": 1545
},
{
"epoch": 0.73,
"learning_rate": 4.237844904603529e-05,
"loss": 0.9557,
"step": 1550
},
{
"epoch": 0.73,
"learning_rate": 4.17120807325451e-05,
"loss": 0.9683,
"step": 1555
},
{
"epoch": 0.73,
"learning_rate": 4.104960969499937e-05,
"loss": 0.9552,
"step": 1560
},
{
"epoch": 0.73,
"learning_rate": 4.0391080227867176e-05,
"loss": 0.9812,
"step": 1565
},
{
"epoch": 0.74,
"learning_rate": 3.973653636207437e-05,
"loss": 0.9612,
"step": 1570
},
{
"epoch": 0.74,
"learning_rate": 3.908602186205954e-05,
"loss": 1.0019,
"step": 1575
},
{
"epoch": 0.74,
"learning_rate": 3.8439580222847784e-05,
"loss": 0.9753,
"step": 1580
},
{
"epoch": 0.74,
"learning_rate": 3.7797254667142576e-05,
"loss": 0.9622,
"step": 1585
},
{
"epoch": 0.74,
"learning_rate": 3.715908814243575e-05,
"loss": 0.9594,
"step": 1590
},
{
"epoch": 0.75,
"learning_rate": 3.6525123318135936e-05,
"loss": 0.9475,
"step": 1595
},
{
"epoch": 0.75,
"learning_rate": 3.5895402582715565e-05,
"loss": 0.9623,
"step": 1600
},
{
"epoch": 0.75,
"learning_rate": 3.526996804087669e-05,
"loss": 0.9484,
"step": 1605
},
{
"epoch": 0.75,
"learning_rate": 3.464886151073574e-05,
"loss": 0.9975,
"step": 1610
},
{
"epoch": 0.76,
"learning_rate": 3.4032124521027484e-05,
"loss": 0.9708,
"step": 1615
},
{
"epoch": 0.76,
"learning_rate": 3.341979830832829e-05,
"loss": 0.9635,
"step": 1620
},
{
"epoch": 0.76,
"learning_rate": 3.281192381429894e-05,
"loss": 0.9694,
"step": 1625
},
{
"epoch": 0.76,
"learning_rate": 3.220854168294719e-05,
"loss": 0.9649,
"step": 1630
},
{
"epoch": 0.77,
"learning_rate": 3.160969225791024e-05,
"loss": 0.9435,
"step": 1635
},
{
"epoch": 0.77,
"learning_rate": 3.1015415579757077e-05,
"loss": 0.9615,
"step": 1640
},
{
"epoch": 0.77,
"learning_rate": 3.042575138331162e-05,
"loss": 0.9532,
"step": 1645
},
{
"epoch": 0.77,
"learning_rate": 2.9840739094995572e-05,
"loss": 0.9491,
"step": 1650
},
{
"epoch": 0.77,
"learning_rate": 2.9260417830192477e-05,
"loss": 0.9546,
"step": 1655
},
{
"epoch": 0.78,
"learning_rate": 2.868482639063238e-05,
"loss": 0.9773,
"step": 1660
},
{
"epoch": 0.78,
"learning_rate": 2.811400326179724e-05,
"loss": 0.9671,
"step": 1665
},
{
"epoch": 0.78,
"learning_rate": 2.7547986610348075e-05,
"loss": 0.9685,
"step": 1670
},
{
"epoch": 0.78,
"learning_rate": 2.698681428157257e-05,
"loss": 0.9811,
"step": 1675
},
{
"epoch": 0.79,
"learning_rate": 2.6430523796855024e-05,
"loss": 0.9796,
"step": 1680
},
{
"epoch": 0.79,
"learning_rate": 2.5879152351167503e-05,
"loss": 0.9623,
"step": 1685
},
{
"epoch": 0.79,
"learning_rate": 2.5332736810582623e-05,
"loss": 0.9719,
"step": 1690
},
{
"epoch": 0.79,
"learning_rate": 2.4791313709809073e-05,
"loss": 0.9519,
"step": 1695
},
{
"epoch": 0.8,
"learning_rate": 2.4254919249748277e-05,
"loss": 0.9573,
"step": 1700
},
{
"epoch": 0.8,
"learning_rate": 2.37235892950743e-05,
"loss": 0.9777,
"step": 1705
},
{
"epoch": 0.8,
"learning_rate": 2.3197359371835802e-05,
"loss": 0.9863,
"step": 1710
},
{
"epoch": 0.8,
"learning_rate": 2.2676264665080414e-05,
"loss": 0.9409,
"step": 1715
},
{
"epoch": 0.81,
"learning_rate": 2.216034001650249e-05,
"loss": 0.9693,
"step": 1720
},
{
"epoch": 0.81,
"learning_rate": 2.164961992211333e-05,
"loss": 0.9698,
"step": 1725
},
{
"epoch": 0.81,
"learning_rate": 2.114413852993471e-05,
"loss": 0.9479,
"step": 1730
},
{
"epoch": 0.81,
"learning_rate": 2.06439296377158e-05,
"loss": 0.9407,
"step": 1735
},
{
"epoch": 0.81,
"learning_rate": 2.014902669067308e-05,
"loss": 0.9805,
"step": 1740
},
{
"epoch": 0.82,
"learning_rate": 1.9659462779254377e-05,
"loss": 0.9551,
"step": 1745
},
{
"epoch": 0.82,
"learning_rate": 1.9175270636926256e-05,
"loss": 0.9601,
"step": 1750
},
{
"epoch": 0.82,
"learning_rate": 1.8696482637985335e-05,
"loss": 0.958,
"step": 1755
},
{
"epoch": 0.82,
"learning_rate": 1.8223130795393726e-05,
"loss": 0.9462,
"step": 1760
},
{
"epoch": 0.83,
"learning_rate": 1.7755246758638545e-05,
"loss": 0.9494,
"step": 1765
},
{
"epoch": 0.83,
"learning_rate": 1.7292861811615723e-05,
"loss": 0.9417,
"step": 1770
},
{
"epoch": 0.83,
"learning_rate": 1.6836006870538334e-05,
"loss": 0.9737,
"step": 1775
},
{
"epoch": 0.83,
"learning_rate": 1.6384712481869426e-05,
"loss": 0.9911,
"step": 1780
},
{
"epoch": 0.84,
"learning_rate": 1.593900882027961e-05,
"loss": 0.9737,
"step": 1785
},
{
"epoch": 0.84,
"learning_rate": 1.549892568662952e-05,
"loss": 0.9626,
"step": 1790
},
{
"epoch": 0.84,
"learning_rate": 1.5064492505977234e-05,
"loss": 0.9629,
"step": 1795
},
{
"epoch": 0.84,
"learning_rate": 1.4635738325610893e-05,
"loss": 0.9623,
"step": 1800
},
{
"epoch": 0.85,
"learning_rate": 1.4212691813106427e-05,
"loss": 1.0034,
"step": 1805
},
{
"epoch": 0.85,
"learning_rate": 1.3795381254410877e-05,
"loss": 0.9382,
"step": 1810
},
{
"epoch": 0.85,
"learning_rate": 1.3383834551951058e-05,
"loss": 0.989,
"step": 1815
},
{
"epoch": 0.85,
"learning_rate": 1.297807922276798e-05,
"loss": 0.9668,
"step": 1820
},
{
"epoch": 0.85,
"learning_rate": 1.2578142396676883e-05,
"loss": 0.9663,
"step": 1825
},
{
"epoch": 0.86,
"learning_rate": 1.2184050814453452e-05,
"loss": 0.9796,
"step": 1830
},
{
"epoch": 0.86,
"learning_rate": 1.179583082604573e-05,
"loss": 0.9683,
"step": 1835
},
{
"epoch": 0.86,
"learning_rate": 1.1413508388812233e-05,
"loss": 0.9543,
"step": 1840
},
{
"epoch": 0.86,
"learning_rate": 1.1037109065786633e-05,
"loss": 0.941,
"step": 1845
},
{
"epoch": 0.87,
"learning_rate": 1.0666658023968213e-05,
"loss": 0.9636,
"step": 1850
},
{
"epoch": 0.87,
"learning_rate": 1.0302180032639452e-05,
"loss": 0.9812,
"step": 1855
},
{
"epoch": 0.87,
"learning_rate": 9.943699461709687e-06,
"loss": 0.968,
"step": 1860
},
{
"epoch": 0.87,
"learning_rate": 9.591240280085667e-06,
"loss": 0.9652,
"step": 1865
},
{
"epoch": 0.88,
"learning_rate": 9.244826054069167e-06,
"loss": 0.9856,
"step": 1870
},
{
"epoch": 0.88,
"learning_rate": 8.904479945780963e-06,
"loss": 0.9597,
"step": 1875
},
{
"epoch": 0.88,
"learning_rate": 8.570224711612385e-06,
"loss": 0.9672,
"step": 1880
},
{
"epoch": 0.88,
"learning_rate": 8.242082700703768e-06,
"loss": 0.9622,
"step": 1885
},
{
"epoch": 0.89,
"learning_rate": 7.920075853449903e-06,
"loss": 0.9564,
"step": 1890
},
{
"epoch": 0.89,
"learning_rate": 7.6042257000333735e-06,
"loss": 0.955,
"step": 1895
},
{
"epoch": 0.89,
"learning_rate": 7.29455335898469e-06,
"loss": 0.927,
"step": 1900
},
{
"epoch": 0.89,
"learning_rate": 6.991079535770484e-06,
"loss": 0.9741,
"step": 1905
},
{
"epoch": 0.89,
"learning_rate": 6.693824521408953e-06,
"loss": 0.9765,
"step": 1910
},
{
"epoch": 0.9,
"learning_rate": 6.402808191113196e-06,
"loss": 0.957,
"step": 1915
},
{
"epoch": 0.9,
"learning_rate": 6.118050002962316e-06,
"loss": 0.9557,
"step": 1920
},
{
"epoch": 0.9,
"learning_rate": 5.839568996600386e-06,
"loss": 0.9751,
"step": 1925
},
{
"epoch": 0.9,
"learning_rate": 5.567383791963421e-06,
"loss": 0.9656,
"step": 1930
},
{
"epoch": 0.91,
"learning_rate": 5.301512588034386e-06,
"loss": 0.969,
"step": 1935
},
{
"epoch": 0.91,
"learning_rate": 5.041973161626401e-06,
"loss": 0.9519,
"step": 1940
},
{
"epoch": 0.91,
"learning_rate": 4.788782866194108e-06,
"loss": 0.9769,
"step": 1945
},
{
"epoch": 0.91,
"learning_rate": 4.541958630673382e-06,
"loss": 0.9643,
"step": 1950
},
{
"epoch": 0.92,
"learning_rate": 4.3015169583494275e-06,
"loss": 0.9698,
"step": 1955
},
{
"epoch": 0.92,
"learning_rate": 4.067473925753318e-06,
"loss": 0.9661,
"step": 1960
},
{
"epoch": 0.92,
"learning_rate": 3.839845181587098e-06,
"loss": 0.9689,
"step": 1965
},
{
"epoch": 0.92,
"learning_rate": 3.6186459456774456e-06,
"loss": 0.9756,
"step": 1970
},
{
"epoch": 0.92,
"learning_rate": 3.4038910079580597e-06,
"loss": 0.9428,
"step": 1975
},
{
"epoch": 0.93,
"learning_rate": 3.195594727480733e-06,
"loss": 0.9622,
"step": 1980
},
{
"epoch": 0.93,
"learning_rate": 2.993771031455328e-06,
"loss": 0.9581,
"step": 1985
},
{
"epoch": 0.93,
"learning_rate": 2.7984334143185242e-06,
"loss": 0.9504,
"step": 1990
},
{
"epoch": 0.93,
"learning_rate": 2.609594936831561e-06,
"loss": 0.939,
"step": 1995
},
{
"epoch": 0.94,
"learning_rate": 2.4272682252069135e-06,
"loss": 0.963,
"step": 2000
},
{
"epoch": 0.94,
"learning_rate": 2.251465470264191e-06,
"loss": 0.9797,
"step": 2005
},
{
"epoch": 0.94,
"learning_rate": 2.0821984266149232e-06,
"loss": 0.9382,
"step": 2010
},
{
"epoch": 0.94,
"learning_rate": 1.9194784118766407e-06,
"loss": 0.9696,
"step": 2015
},
{
"epoch": 0.95,
"learning_rate": 1.763316305916174e-06,
"loss": 0.9668,
"step": 2020
},
{
"epoch": 0.95,
"learning_rate": 1.6137225501221654e-06,
"loss": 0.9905,
"step": 2025
},
{
"epoch": 0.95,
"learning_rate": 1.4707071467069733e-06,
"loss": 0.9455,
"step": 2030
},
{
"epoch": 0.95,
"learning_rate": 1.3342796580378736e-06,
"loss": 0.9529,
"step": 2035
},
{
"epoch": 0.96,
"learning_rate": 1.204449205997671e-06,
"loss": 0.9581,
"step": 2040
},
{
"epoch": 0.96,
"learning_rate": 1.0812244713748642e-06,
"loss": 0.9824,
"step": 2045
},
{
"epoch": 0.96,
"learning_rate": 9.64613693283123e-07,
"loss": 0.9632,
"step": 2050
},
{
"epoch": 0.96,
"learning_rate": 8.546246686105041e-07,
"loss": 0.9774,
"step": 2055
},
{
"epoch": 0.96,
"learning_rate": 7.512647514980486e-07,
"loss": 0.9723,
"step": 2060
},
{
"epoch": 0.97,
"learning_rate": 6.545408528481178e-07,
"loss": 0.9807,
"step": 2065
},
{
"epoch": 0.97,
"learning_rate": 5.644594398622971e-07,
"loss": 0.959,
"step": 2070
},
{
"epoch": 0.97,
"learning_rate": 4.810265356089638e-07,
"loss": 0.9432,
"step": 2075
},
{
"epoch": 0.97,
"learning_rate": 4.042477186205873e-07,
"loss": 0.9648,
"step": 2080
},
{
"epoch": 0.98,
"learning_rate": 3.341281225207604e-07,
"loss": 0.9755,
"step": 2085
},
{
"epoch": 0.98,
"learning_rate": 2.706724356808965e-07,
"loss": 0.9684,
"step": 2090
},
{
"epoch": 0.98,
"learning_rate": 2.1388490090680223e-07,
"loss": 0.9425,
"step": 2095
},
{
"epoch": 0.98,
"learning_rate": 1.637693151549602e-07,
"loss": 0.979,
"step": 2100
},
{
"epoch": 0.99,
"learning_rate": 1.203290292786763e-07,
"loss": 0.9589,
"step": 2105
},
{
"epoch": 0.99,
"learning_rate": 8.356694780401463e-08,
"loss": 0.9717,
"step": 2110
},
{
"epoch": 0.99,
"learning_rate": 5.3485528735619516e-08,
"loss": 0.9703,
"step": 2115
},
{
"epoch": 0.99,
"learning_rate": 3.008678339234683e-08,
"loss": 0.9435,
"step": 2120
},
{
"epoch": 1.0,
"learning_rate": 1.3372276272771712e-08,
"loss": 0.9864,
"step": 2125
},
{
"epoch": 1.0,
"learning_rate": 3.3431249506166163e-09,
"loss": 0.9747,
"step": 2130
},
{
"epoch": 1.0,
"learning_rate": 0.0,
"loss": 0.961,
"step": 2135
},
{
"epoch": 1.0,
"eval_loss": 0.9693147540092468,
"eval_runtime": 359.0834,
"eval_samples_per_second": 42.118,
"eval_steps_per_second": 0.66,
"step": 2135
},
{
"epoch": 1.0,
"step": 2135,
"total_flos": 1.2009452294891373e+19,
"train_loss": 0.9789489164285414,
"train_runtime": 11596.772,
"train_samples_per_second": 11.784,
"train_steps_per_second": 0.184
}
],
"logging_steps": 5,
"max_steps": 2135,
"num_input_tokens_seen": 0,
"num_train_epochs": 1,
"save_steps": 100,
"total_flos": 1.2009452294891373e+19,
"train_batch_size": 4,
"trial_name": null,
"trial_params": null
}