Files
SmolLM2-MagpieUltra-v2/trainer_state.json

4426 lines
101 KiB
JSON
Raw Normal View History

{
"best_metric": null,
"best_model_checkpoint": null,
"epoch": 1.0,
"eval_steps": 500,
"global_step": 3125,
"is_hyper_param_search": false,
"is_local_process_zero": true,
"is_world_process_zero": true,
"log_history": [
{
"epoch": 0.0016,
"grad_norm": 64.79862964893701,
"learning_rate": 4.792332268370606e-06,
"loss": 3.7986,
"step": 5
},
{
"epoch": 0.0032,
"grad_norm": 29.290033796462424,
"learning_rate": 9.584664536741212e-06,
"loss": 3.304,
"step": 10
},
{
"epoch": 0.0048,
"grad_norm": 10.895694829851312,
"learning_rate": 1.4376996805111818e-05,
"loss": 2.3345,
"step": 15
},
{
"epoch": 0.0064,
"grad_norm": 4.22950366323865,
"learning_rate": 1.9169329073482425e-05,
"loss": 1.6672,
"step": 20
},
{
"epoch": 0.008,
"grad_norm": 1.6360137415435663,
"learning_rate": 2.3961661341853032e-05,
"loss": 1.281,
"step": 25
},
{
"epoch": 0.0096,
"grad_norm": 0.8534883766043228,
"learning_rate": 2.8753993610223637e-05,
"loss": 1.1186,
"step": 30
},
{
"epoch": 0.0112,
"grad_norm": 0.42541063066843565,
"learning_rate": 3.3546325878594245e-05,
"loss": 1.0106,
"step": 35
},
{
"epoch": 0.0128,
"grad_norm": 0.33403768880643325,
"learning_rate": 3.833865814696485e-05,
"loss": 1.0092,
"step": 40
},
{
"epoch": 0.0144,
"grad_norm": 0.2385548037475669,
"learning_rate": 4.313099041533546e-05,
"loss": 0.9504,
"step": 45
},
{
"epoch": 0.016,
"grad_norm": 0.21711175160204815,
"learning_rate": 4.7923322683706065e-05,
"loss": 0.9146,
"step": 50
},
{
"epoch": 0.0176,
"grad_norm": 0.19324897304220315,
"learning_rate": 5.2715654952076676e-05,
"loss": 0.9051,
"step": 55
},
{
"epoch": 0.0192,
"grad_norm": 0.16941355387097437,
"learning_rate": 5.7507987220447274e-05,
"loss": 0.9291,
"step": 60
},
{
"epoch": 0.0208,
"grad_norm": 0.1520711327338434,
"learning_rate": 6.230031948881788e-05,
"loss": 0.8873,
"step": 65
},
{
"epoch": 0.0224,
"grad_norm": 0.15291874280710463,
"learning_rate": 6.709265175718849e-05,
"loss": 0.8896,
"step": 70
},
{
"epoch": 0.024,
"grad_norm": 0.14977669003307942,
"learning_rate": 7.18849840255591e-05,
"loss": 0.8982,
"step": 75
},
{
"epoch": 0.0256,
"grad_norm": 0.14407867303158503,
"learning_rate": 7.66773162939297e-05,
"loss": 0.8826,
"step": 80
},
{
"epoch": 0.0272,
"grad_norm": 0.127640372929375,
"learning_rate": 8.14696485623003e-05,
"loss": 0.8738,
"step": 85
},
{
"epoch": 0.0288,
"grad_norm": 0.12916578728919045,
"learning_rate": 8.626198083067092e-05,
"loss": 0.8659,
"step": 90
},
{
"epoch": 0.0304,
"grad_norm": 0.09674048749471525,
"learning_rate": 9.105431309904153e-05,
"loss": 0.8423,
"step": 95
},
{
"epoch": 0.032,
"grad_norm": 0.09282410939453126,
"learning_rate": 9.584664536741213e-05,
"loss": 0.8458,
"step": 100
},
{
"epoch": 0.0336,
"grad_norm": 0.10134540252174387,
"learning_rate": 0.00010063897763578275,
"loss": 0.8346,
"step": 105
},
{
"epoch": 0.0352,
"grad_norm": 0.07389448471081606,
"learning_rate": 0.00010543130990415335,
"loss": 0.8338,
"step": 110
},
{
"epoch": 0.0368,
"grad_norm": 0.08447006505296106,
"learning_rate": 0.00011022364217252396,
"loss": 0.8563,
"step": 115
},
{
"epoch": 0.0384,
"grad_norm": 0.06958293644640014,
"learning_rate": 0.00011501597444089455,
"loss": 0.8441,
"step": 120
},
{
"epoch": 0.04,
"grad_norm": 0.0688195545868785,
"learning_rate": 0.00011980830670926518,
"loss": 0.8388,
"step": 125
},
{
"epoch": 0.0416,
"grad_norm": 0.06613022535500636,
"learning_rate": 0.00012460063897763577,
"loss": 0.8388,
"step": 130
},
{
"epoch": 0.0432,
"grad_norm": 0.05884003981608602,
"learning_rate": 0.00012939297124600637,
"loss": 0.849,
"step": 135
},
{
"epoch": 0.0448,
"grad_norm": 0.0675868331017935,
"learning_rate": 0.00013418530351437698,
"loss": 0.8157,
"step": 140
},
{
"epoch": 0.0464,
"grad_norm": 0.06520369147774627,
"learning_rate": 0.0001389776357827476,
"loss": 0.8377,
"step": 145
},
{
"epoch": 0.048,
"grad_norm": 0.05591417905435732,
"learning_rate": 0.0001437699680511182,
"loss": 0.8346,
"step": 150
},
{
"epoch": 0.0496,
"grad_norm": 0.06885094447020612,
"learning_rate": 0.0001485623003194888,
"loss": 0.8325,
"step": 155
},
{
"epoch": 0.0512,
"grad_norm": 0.07693195506697653,
"learning_rate": 0.0001533546325878594,
"loss": 0.8261,
"step": 160
},
{
"epoch": 0.0528,
"grad_norm": 0.06096756973358526,
"learning_rate": 0.00015814696485623,
"loss": 0.8248,
"step": 165
},
{
"epoch": 0.0544,
"grad_norm": 0.06475575830743056,
"learning_rate": 0.0001629392971246006,
"loss": 0.855,
"step": 170
},
{
"epoch": 0.056,
"grad_norm": 0.056612739195628536,
"learning_rate": 0.00016773162939297124,
"loss": 0.821,
"step": 175
},
{
"epoch": 0.0576,
"grad_norm": 0.06071836821069472,
"learning_rate": 0.00017252396166134184,
"loss": 0.8293,
"step": 180
},
{
"epoch": 0.0592,
"grad_norm": 0.06751182828614498,
"learning_rate": 0.00017731629392971245,
"loss": 0.8155,
"step": 185
},
{
"epoch": 0.0608,
"grad_norm": 0.05955983363606016,
"learning_rate": 0.00018210862619808305,
"loss": 0.7971,
"step": 190
},
{
"epoch": 0.0624,
"grad_norm": 0.07052819972972657,
"learning_rate": 0.00018690095846645365,
"loss": 0.8288,
"step": 195
},
{
"epoch": 0.064,
"grad_norm": 0.05923622995995164,
"learning_rate": 0.00019169329073482426,
"loss": 0.8184,
"step": 200
},
{
"epoch": 0.0656,
"grad_norm": 0.06963646035027679,
"learning_rate": 0.00019648562300319486,
"loss": 0.8449,
"step": 205
},
{
"epoch": 0.0672,
"grad_norm": 0.06860014085323075,
"learning_rate": 0.0002012779552715655,
"loss": 0.8341,
"step": 210
},
{
"epoch": 0.0688,
"grad_norm": 0.06941082488577938,
"learning_rate": 0.0002060702875399361,
"loss": 0.8383,
"step": 215
},
{
"epoch": 0.0704,
"grad_norm": 0.06616271548096953,
"learning_rate": 0.0002108626198083067,
"loss": 0.804,
"step": 220
},
{
"epoch": 0.072,
"grad_norm": 0.07107556899925888,
"learning_rate": 0.0002156549520766773,
"loss": 0.8156,
"step": 225
},
{
"epoch": 0.0736,
"grad_norm": 0.06995785055398655,
"learning_rate": 0.0002204472843450479,
"loss": 0.8049,
"step": 230
},
{
"epoch": 0.0752,
"grad_norm": 0.059688190721088996,
"learning_rate": 0.00022523961661341852,
"loss": 0.8209,
"step": 235
},
{
"epoch": 0.0768,
"grad_norm": 0.06833286026364968,
"learning_rate": 0.0002300319488817891,
"loss": 0.8233,
"step": 240
},
{
"epoch": 0.0784,
"grad_norm": 0.08297292157685465,
"learning_rate": 0.0002348242811501597,
"loss": 0.8359,
"step": 245
},
{
"epoch": 0.08,
"grad_norm": 0.06316799845757914,
"learning_rate": 0.00023961661341853036,
"loss": 0.831,
"step": 250
},
{
"epoch": 0.0816,
"grad_norm": 0.07179318857392636,
"learning_rate": 0.00024440894568690096,
"loss": 0.8359,
"step": 255
},
{
"epoch": 0.0832,
"grad_norm": 0.06293057603954971,
"learning_rate": 0.00024920127795527154,
"loss": 0.8242,
"step": 260
},
{
"epoch": 0.0848,
"grad_norm": 0.06494498769023198,
"learning_rate": 0.00025399361022364217,
"loss": 0.8131,
"step": 265
},
{
"epoch": 0.0864,
"grad_norm": 0.06254880531963027,
"learning_rate": 0.00025878594249201275,
"loss": 0.8133,
"step": 270
},
{
"epoch": 0.088,
"grad_norm": 0.05791128690870581,
"learning_rate": 0.0002635782747603834,
"loss": 0.7985,
"step": 275
},
{
"epoch": 0.0896,
"grad_norm": 0.05968336480331785,
"learning_rate": 0.00026837060702875396,
"loss": 0.8259,
"step": 280
},
{
"epoch": 0.0912,
"grad_norm": 0.05507468875480531,
"learning_rate": 0.00027316293929712453,
"loss": 0.8044,
"step": 285
},
{
"epoch": 0.0928,
"grad_norm": 0.07304020634733059,
"learning_rate": 0.0002779552715654952,
"loss": 0.8182,
"step": 290
},
{
"epoch": 0.0944,
"grad_norm": 0.06792046663527374,
"learning_rate": 0.0002827476038338658,
"loss": 0.8342,
"step": 295
},
{
"epoch": 0.096,
"grad_norm": 0.060175207345708116,
"learning_rate": 0.0002875399361022364,
"loss": 0.8122,
"step": 300
},
{
"epoch": 0.0976,
"grad_norm": 0.07531435295140049,
"learning_rate": 0.000292332268370607,
"loss": 0.822,
"step": 305
},
{
"epoch": 0.0992,
"grad_norm": 0.05753958870761837,
"learning_rate": 0.0002971246006389776,
"loss": 0.8364,
"step": 310
},
{
"epoch": 0.1008,
"grad_norm": 0.05265101912795107,
"learning_rate": 0.00029999962555314543,
"loss": 0.8159,
"step": 315
},
{
"epoch": 0.1024,
"grad_norm": 0.04957141033828283,
"learning_rate": 0.00029999541304750175,
"loss": 0.8005,
"step": 320
},
{
"epoch": 0.104,
"grad_norm": 0.0509283015803962,
"learning_rate": 0.00029998652010953066,
"loss": 0.8268,
"step": 325
},
{
"epoch": 0.1056,
"grad_norm": 0.06390145591389865,
"learning_rate": 0.000299972947016726,
"loss": 0.8203,
"step": 330
},
{
"epoch": 0.1072,
"grad_norm": 0.05898681941656638,
"learning_rate": 0.00029995469419262033,
"loss": 0.8276,
"step": 335
},
{
"epoch": 0.1088,
"grad_norm": 0.06512724796221968,
"learning_rate": 0.00029993176220677174,
"loss": 0.8122,
"step": 340
},
{
"epoch": 0.1104,
"grad_norm": 0.05551305751170922,
"learning_rate": 0.0002999041517747462,
"loss": 0.8018,
"step": 345
},
{
"epoch": 0.112,
"grad_norm": 0.04874158533755733,
"learning_rate": 0.0002998718637580951,
"loss": 0.8192,
"step": 350
},
{
"epoch": 0.1136,
"grad_norm": 0.05538536507227949,
"learning_rate": 0.00029983489916432846,
"loss": 0.8135,
"step": 355
},
{
"epoch": 0.1152,
"grad_norm": 0.06949832683320262,
"learning_rate": 0.00029979325914688344,
"loss": 0.8094,
"step": 360
},
{
"epoch": 0.1168,
"grad_norm": 0.06096787363526232,
"learning_rate": 0.0002997469450050883,
"loss": 0.8196,
"step": 365
},
{
"epoch": 0.1184,
"grad_norm": 0.05591490066156486,
"learning_rate": 0.00029969595818412183,
"loss": 0.8093,
"step": 370
},
{
"epoch": 0.12,
"grad_norm": 0.0551286129224898,
"learning_rate": 0.0002996403002749686,
"loss": 0.8057,
"step": 375
},
{
"epoch": 0.1216,
"grad_norm": 0.050504199679135255,
"learning_rate": 0.0002995799730143689,
"loss": 0.8253,
"step": 380
},
{
"epoch": 0.1232,
"grad_norm": 0.054113844131886295,
"learning_rate": 0.0002995149782847646,
"loss": 0.7952,
"step": 385
},
{
"epoch": 0.1248,
"grad_norm": 0.059361929440663486,
"learning_rate": 0.0002994453181142407,
"loss": 0.8148,
"step": 390
},
{
"epoch": 0.1264,
"grad_norm": 0.05484005641635372,
"learning_rate": 0.00029937099467646167,
"loss": 0.8025,
"step": 395
},
{
"epoch": 0.128,
"grad_norm": 0.046303100449141,
"learning_rate": 0.000299292010290604,
"loss": 0.8043,
"step": 400
},
{
"epoch": 0.1296,
"grad_norm": 0.0527166773177604,
"learning_rate": 0.0002992083674212835,
"loss": 0.8115,
"step": 405
},
{
"epoch": 0.1312,
"grad_norm": 0.04940671723136873,
"learning_rate": 0.0002991200686784785,
"loss": 0.7842,
"step": 410
},
{
"epoch": 0.1328,
"grad_norm": 0.05682145984208437,
"learning_rate": 0.0002990271168174487,
"loss": 0.8118,
"step": 415
},
{
"epoch": 0.1344,
"grad_norm": 0.04926281134667431,
"learning_rate": 0.00029892951473864865,
"loss": 0.7986,
"step": 420
},
{
"epoch": 0.136,
"grad_norm": 0.05053759860453711,
"learning_rate": 0.0002988272654876376,
"loss": 0.8055,
"step": 425
},
{
"epoch": 0.1376,
"grad_norm": 0.05587322290343626,
"learning_rate": 0.00029872037225498453,
"loss": 0.8104,
"step": 430
},
{
"epoch": 0.1392,
"grad_norm": 0.04968986730979809,
"learning_rate": 0.0002986088383761684,
"loss": 0.8106,
"step": 435
},
{
"epoch": 0.1408,
"grad_norm": 0.05709843408250463,
"learning_rate": 0.0002984926673314741,
"loss": 0.7998,
"step": 440
},
{
"epoch": 0.1424,
"grad_norm": 0.05115846157311485,
"learning_rate": 0.0002983718627458838,
"loss": 0.7888,
"step": 445
},
{
"epoch": 0.144,
"grad_norm": 0.04934224261471247,
"learning_rate": 0.0002982464283889642,
"loss": 0.7981,
"step": 450
},
{
"epoch": 0.1456,
"grad_norm": 0.046214903923842345,
"learning_rate": 0.0002981163681747483,
"loss": 0.7939,
"step": 455
},
{
"epoch": 0.1472,
"grad_norm": 0.04829372890505013,
"learning_rate": 0.0002979816861616138,
"loss": 0.8021,
"step": 460
},
{
"epoch": 0.1488,
"grad_norm": 0.052778074611499734,
"learning_rate": 0.00029784238655215626,
"loss": 0.8089,
"step": 465
},
{
"epoch": 0.1504,
"grad_norm": 0.05413941613875908,
"learning_rate": 0.0002976984736930578,
"loss": 0.8022,
"step": 470
},
{
"epoch": 0.152,
"grad_norm": 0.05129352790614527,
"learning_rate": 0.0002975499520749518,
"loss": 0.8,
"step": 475
},
{
"epoch": 0.1536,
"grad_norm": 0.05179202869475361,
"learning_rate": 0.00029739682633228245,
"loss": 0.809,
"step": 480
},
{
"epoch": 0.1552,
"grad_norm": 0.04635356891083487,
"learning_rate": 0.0002972391012431605,
"loss": 0.807,
"step": 485
},
{
"epoch": 0.1568,
"grad_norm": 0.04999190557107593,
"learning_rate": 0.00029707678172921366,
"loss": 0.7911,
"step": 490
},
{
"epoch": 0.1584,
"grad_norm": 0.05456560092214985,
"learning_rate": 0.0002969098728554336,
"loss": 0.7817,
"step": 495
},
{
"epoch": 0.16,
"grad_norm": 0.05317881843455394,
"learning_rate": 0.00029673837983001754,
"loss": 0.7886,
"step": 500
},
{
"epoch": 0.1616,
"grad_norm": 0.054628625112347194,
"learning_rate": 0.0002965623080042057,
"loss": 0.7797,
"step": 505
},
{
"epoch": 0.1632,
"grad_norm": 0.0454818393425224,
"learning_rate": 0.00029638166287211453,
"loss": 0.7794,
"step": 510
},
{
"epoch": 0.1648,
"grad_norm": 0.045903405195861856,
"learning_rate": 0.00029619645007056527,
"loss": 0.7823,
"step": 515
},
{
"epoch": 0.1664,
"grad_norm": 0.05778313013653165,
"learning_rate": 0.0002960066753789077,
"loss": 0.7938,
"step": 520
},
{
"epoch": 0.168,
"grad_norm": 0.04794213698716533,
"learning_rate": 0.00029581234471884045,
"loss": 0.7921,
"step": 525
},
{
"epoch": 0.1696,
"grad_norm": 0.04682149018802875,
"learning_rate": 0.00029561346415422554,
"loss": 0.7728,
"step": 530
},
{
"epoch": 0.1712,
"grad_norm": 0.047089674218094714,
"learning_rate": 0.0002954100398908995,
"loss": 0.7966,
"step": 535
},
{
"epoch": 0.1728,
"grad_norm": 0.044859883579268366,
"learning_rate": 0.00029520207827647997,
"loss": 0.788,
"step": 540
},
{
"epoch": 0.1744,
"grad_norm": 0.05495628492120463,
"learning_rate": 0.0002949895858001669,
"loss": 0.7903,
"step": 545
},
{
"epoch": 0.176,
"grad_norm": 0.06013853184360737,
"learning_rate": 0.0002947725690925409,
"loss": 0.7977,
"step": 550
},
{
"epoch": 0.1776,
"grad_norm": 0.048624933325734164,
"learning_rate": 0.00029455103492535573,
"loss": 0.7928,
"step": 555
},
{
"epoch": 0.1792,
"grad_norm": 0.047300655523156064,
"learning_rate": 0.00029432499021132737,
"loss": 0.7906,
"step": 560
},
{
"epoch": 0.1808,
"grad_norm": 0.056115287002102256,
"learning_rate": 0.00029409444200391807,
"loss": 0.786,
"step": 565
},
{
"epoch": 0.1824,
"grad_norm": 0.049117514914560706,
"learning_rate": 0.0002938593974971163,
"loss": 0.7908,
"step": 570
},
{
"epoch": 0.184,
"grad_norm": 0.04940006503376019,
"learning_rate": 0.00029361986402521255,
"loss": 0.7874,
"step": 575
},
{
"epoch": 0.1856,
"grad_norm": 0.05373210996094955,
"learning_rate": 0.00029337584906256994,
"loss": 0.7984,
"step": 580
},
{
"epoch": 0.1872,
"grad_norm": 0.05474306098461321,
"learning_rate": 0.0002931273602233915,
"loss": 0.7884,
"step": 585
},
{
"epoch": 0.1888,
"grad_norm": 0.14465646121684841,
"learning_rate": 0.0002928744052614823,
"loss": 0.7983,
"step": 590
},
{
"epoch": 0.1904,
"grad_norm": 0.5255096935790569,
"learning_rate": 0.00029261699207000757,
"loss": 0.8383,
"step": 595
},
{
"epoch": 0.192,
"grad_norm": 3.0948754803086005,
"learning_rate": 0.0002923551286812465,
"loss": 1.0175,
"step": 600
},
{
"epoch": 0.1936,
"grad_norm": 3.615101942230243,
"learning_rate": 0.00029208882326634125,
"loss": 1.4932,
"step": 605
},
{
"epoch": 0.1952,
"grad_norm": 0.14089470273338237,
"learning_rate": 0.0002918180841350427,
"loss": 1.0401,
"step": 610
},
{
"epoch": 0.1968,
"grad_norm": 0.3373287784802123,
"learning_rate": 0.00029154291973545007,
"loss": 0.9607,
"step": 615
},
{
"epoch": 0.1984,
"grad_norm": 0.1033440858245417,
"learning_rate": 0.0002912633386537485,
"loss": 0.9787,
"step": 620
},
{
"epoch": 0.2,
"grad_norm": 0.06639475073647456,
"learning_rate": 0.00029097934961394025,
"loss": 0.9303,
"step": 625
},
{
"epoch": 0.2016,
"grad_norm": 0.05862768804117062,
"learning_rate": 0.0002906909614775729,
"loss": 0.8954,
"step": 630
},
{
"epoch": 0.2032,
"grad_norm": 0.05119833574003903,
"learning_rate": 0.00029039818324346254,
"loss": 0.8801,
"step": 635
},
{
"epoch": 0.2048,
"grad_norm": 0.04790825972385907,
"learning_rate": 0.00029010102404741344,
"loss": 0.8483,
"step": 640
},
{
"epoch": 0.2064,
"grad_norm": 0.04898413486706179,
"learning_rate": 0.00028979949316193244,
"loss": 0.8365,
"step": 645
},
{
"epoch": 0.208,
"grad_norm": 0.044373887795511005,
"learning_rate": 0.00028949359999593996,
"loss": 0.8366,
"step": 650
},
{
"epoch": 0.2096,
"grad_norm": 0.04575881648687471,
"learning_rate": 0.0002891833540944764,
"loss": 0.8458,
"step": 655
},
{
"epoch": 0.2112,
"grad_norm": 0.04638167285807861,
"learning_rate": 0.000288868765138404,
"loss": 0.8321,
"step": 660
},
{
"epoch": 0.2128,
"grad_norm": 0.07135297419164925,
"learning_rate": 0.00028854984294410507,
"loss": 0.8134,
"step": 665
},
{
"epoch": 0.2144,
"grad_norm": 0.04281324262990822,
"learning_rate": 0.00028822659746317566,
"loss": 0.799,
"step": 670
},
{
"epoch": 0.216,
"grad_norm": 0.04481987085327319,
"learning_rate": 0.00028789903878211477,
"loss": 0.8338,
"step": 675
},
{
"epoch": 0.2176,
"grad_norm": 0.048657304997143574,
"learning_rate": 0.00028756717712201,
"loss": 0.8061,
"step": 680
},
{
"epoch": 0.2192,
"grad_norm": 0.05132910171873023,
"learning_rate": 0.00028723102283821823,
"loss": 0.8016,
"step": 685
},
{
"epoch": 0.2208,
"grad_norm": 0.043704814768955186,
"learning_rate": 0.0002868905864200428,
"loss": 0.7902,
"step": 690
},
{
"epoch": 0.2224,
"grad_norm": 0.051516499601557166,
"learning_rate": 0.0002865458784904059,
"loss": 0.8192,
"step": 695
},
{
"epoch": 0.224,
"grad_norm": 0.05476128143933315,
"learning_rate": 0.0002861969098055174,
"loss": 0.7807,
"step": 700
},
{
"epoch": 0.2256,
"grad_norm": 0.04701129985299958,
"learning_rate": 0.0002858436912545391,
"loss": 0.787,
"step": 705
},
{
"epoch": 0.2272,
"grad_norm": 0.045976584602508,
"learning_rate": 0.0002854862338592448,
"loss": 0.8211,
"step": 710
},
{
"epoch": 0.2288,
"grad_norm": 0.04421088788080479,
"learning_rate": 0.0002851245487736766,
"loss": 0.8096,
"step": 715
},
{
"epoch": 0.2304,
"grad_norm": 0.04437240484817772,
"learning_rate": 0.0002847586472837968,
"loss": 0.8009,
"step": 720
},
{
"epoch": 0.232,
"grad_norm": 0.04595607719220013,
"learning_rate": 0.00028438854080713557,
"loss": 0.8053,
"step": 725
},
{
"epoch": 0.2336,
"grad_norm": 0.043767356252426635,
"learning_rate": 0.0002840142408924348,
"loss": 0.8005,
"step": 730
},
{
"epoch": 0.2352,
"grad_norm": 0.041648966315012526,
"learning_rate": 0.00028363575921928793,
"loss": 0.813,
"step": 735
},
{
"epoch": 0.2368,
"grad_norm": 0.04396054654683414,
"learning_rate": 0.000283253107597775,
"loss": 0.8083,
"step": 740
},
{
"epoch": 0.2384,
"grad_norm": 0.04481454329880548,
"learning_rate": 0.0002828662979680947,
"loss": 0.7762,
"step": 745
},
{
"epoch": 0.24,
"grad_norm": 0.04442483114968266,
"learning_rate": 0.0002824753424001914,
"loss": 0.8076,
"step": 750
},
{
"epoch": 0.2416,
"grad_norm": 0.04511753997848536,
"learning_rate": 0.00028208025309337865,
"loss": 0.7927,
"step": 755
},
{
"epoch": 0.2432,
"grad_norm": 0.046354138137126584,
"learning_rate": 0.00028168104237595863,
"loss": 0.7683,
"step": 760
},
{
"epoch": 0.2448,
"grad_norm": 0.04082380158391817,
"learning_rate": 0.0002812777227048371,
"loss": 0.7677,
"step": 765
},
{
"epoch": 0.2464,
"grad_norm": 0.049309971949502214,
"learning_rate": 0.00028087030666513525,
"loss": 0.7988,
"step": 770
},
{
"epoch": 0.248,
"grad_norm": 0.045440684885208,
"learning_rate": 0.0002804588069697964,
"loss": 0.7662,
"step": 775
},
{
"epoch": 0.2496,
"grad_norm": 0.0444532237833008,
"learning_rate": 0.00028004323645918974,
"loss": 0.7842,
"step": 780
},
{
"epoch": 0.2512,
"grad_norm": 0.040657171713024576,
"learning_rate": 0.00027962360810070955,
"loss": 0.7839,
"step": 785
},
{
"epoch": 0.2528,
"grad_norm": 0.0540110179307627,
"learning_rate": 0.00027919993498837047,
"loss": 0.7896,
"step": 790
},
{
"epoch": 0.2544,
"grad_norm": 0.04259787691620654,
"learning_rate": 0.00027877223034239896,
"loss": 0.7793,
"step": 795
},
{
"epoch": 0.256,
"grad_norm": 0.04457294144966481,
"learning_rate": 0.00027834050750882074,
"loss": 0.7855,
"step": 800
},
{
"epoch": 0.2576,
"grad_norm": 0.04410681383697115,
"learning_rate": 0.0002779047799590445,
"loss": 0.7695,
"step": 805
},
{
"epoch": 0.2592,
"grad_norm": 0.053791843672986804,
"learning_rate": 0.00027746506128944147,
"loss": 0.7757,
"step": 810
},
{
"epoch": 0.2608,
"grad_norm": 0.0486575710394306,
"learning_rate": 0.00027702136522092107,
"loss": 0.7818,
"step": 815
},
{
"epoch": 0.2624,
"grad_norm": 0.04125809722919823,
"learning_rate": 0.0002765737055985028,
"loss": 0.8047,
"step": 820
},
{
"epoch": 0.264,
"grad_norm": 0.047411942700633564,
"learning_rate": 0.00027612209639088427,
"loss": 0.7578,
"step": 825
},
{
"epoch": 0.2656,
"grad_norm": 0.050940775491989096,
"learning_rate": 0.0002756665516900053,
"loss": 0.7681,
"step": 830
},
{
"epoch": 0.2672,
"grad_norm": 0.04082369288507536,
"learning_rate": 0.00027520708571060823,
"loss": 0.7556,
"step": 835
},
{
"epoch": 0.2688,
"grad_norm": 0.03911502648937811,
"learning_rate": 0.0002747437127897943,
"loss": 0.7508,
"step": 840
},
{
"epoch": 0.2704,
"grad_norm": 0.044224505916356134,
"learning_rate": 0.0002742764473865763,
"loss": 0.7954,
"step": 845
},
{
"epoch": 0.272,
"grad_norm": 0.04247036421357291,
"learning_rate": 0.0002738053040814274,
"loss": 0.7607,
"step": 850
},
{
"epoch": 0.2736,
"grad_norm": 0.041899587734798266,
"learning_rate": 0.00027333029757582624,
"loss": 0.7707,
"step": 855
},
{
"epoch": 0.2752,
"grad_norm": 0.041672158877511176,
"learning_rate": 0.00027285144269179816,
"loss": 0.7702,
"step": 860
},
{
"epoch": 0.2768,
"grad_norm": 0.04155845099801199,
"learning_rate": 0.0002723687543714525,
"loss": 0.7893,
"step": 865
},
{
"epoch": 0.2784,
"grad_norm": 0.04108348014143511,
"learning_rate": 0.0002718822476765167,
"loss": 0.7614,
"step": 870
},
{
"epoch": 0.28,
"grad_norm": 0.045256463749299106,
"learning_rate": 0.00027139193778786603,
"loss": 0.7833,
"step": 875
},
{
"epoch": 0.2816,
"grad_norm": 0.04253172171776479,
"learning_rate": 0.0002708978400050501,
"loss": 0.7706,
"step": 880
},
{
"epoch": 0.2832,
"grad_norm": 0.0500050312669766,
"learning_rate": 0.0002703999697458152,
"loss": 0.7865,
"step": 885
},
{
"epoch": 0.2848,
"grad_norm": 0.04149213458472267,
"learning_rate": 0.0002698983425456235,
"loss": 0.7554,
"step": 890
},
{
"epoch": 0.2864,
"grad_norm": 0.04595218793063625,
"learning_rate": 0.00026939297405716807,
"loss": 0.7734,
"step": 895
},
{
"epoch": 0.288,
"grad_norm": 0.04481528857577919,
"learning_rate": 0.00026888388004988456,
"loss": 0.746,
"step": 900
},
{
"epoch": 0.2896,
"grad_norm": 0.043477674071316814,
"learning_rate": 0.00026837107640945905,
"loss": 0.7704,
"step": 905
},
{
"epoch": 0.2912,
"grad_norm": 0.04814185600999023,
"learning_rate": 0.00026785457913733237,
"loss": 0.7662,
"step": 910
},
{
"epoch": 0.2928,
"grad_norm": 0.0468699434302569,
"learning_rate": 0.000267334404350201,
"loss": 0.772,
"step": 915
},
{
"epoch": 0.2944,
"grad_norm": 0.03923205014558229,
"learning_rate": 0.00026681056827951386,
"loss": 0.7396,
"step": 920
},
{
"epoch": 0.296,
"grad_norm": 0.03969515280288691,
"learning_rate": 0.000266283087270966,
"loss": 0.7742,
"step": 925
},
{
"epoch": 0.2976,
"grad_norm": 0.04612291434466239,
"learning_rate": 0.00026575197778398843,
"loss": 0.7713,
"step": 930
},
{
"epoch": 0.2992,
"grad_norm": 0.04710609947249134,
"learning_rate": 0.0002652172563912348,
"loss": 0.774,
"step": 935
},
{
"epoch": 0.3008,
"grad_norm": 0.04627823939147109,
"learning_rate": 0.00026467893977806387,
"loss": 0.759,
"step": 940
},
{
"epoch": 0.3024,
"grad_norm": 0.04215962416625985,
"learning_rate": 0.0002641370447420192,
"loss": 0.766,
"step": 945
},
{
"epoch": 0.304,
"grad_norm": 0.04876220916142212,
"learning_rate": 0.00026359158819230484,
"loss": 0.7699,
"step": 950
},
{
"epoch": 0.3056,
"grad_norm": 0.0465570007489398,
"learning_rate": 0.00026304258714925766,
"loss": 0.7792,
"step": 955
},
{
"epoch": 0.3072,
"grad_norm": 0.0437908623508308,
"learning_rate": 0.00026249005874381655,
"loss": 0.7648,
"step": 960
},
{
"epoch": 0.3088,
"grad_norm": 0.04319239474141502,
"learning_rate": 0.0002619340202169873,
"loss": 0.7651,
"step": 965
},
{
"epoch": 0.3104,
"grad_norm": 0.039590732151859404,
"learning_rate": 0.00026137448891930514,
"loss": 0.7563,
"step": 970
},
{
"epoch": 0.312,
"grad_norm": 0.04478476030600327,
"learning_rate": 0.0002608114823102932,
"loss": 0.7739,
"step": 975
},
{
"epoch": 0.3136,
"grad_norm": 0.044680126096864815,
"learning_rate": 0.0002602450179579176,
"loss": 0.7384,
"step": 980
},
{
"epoch": 0.3152,
"grad_norm": 0.04607582062714833,
"learning_rate": 0.0002596751135380392,
"loss": 0.7589,
"step": 985
},
{
"epoch": 0.3168,
"grad_norm": 0.04652451296184953,
"learning_rate": 0.00025910178683386247,
"loss": 0.7667,
"step": 990
},
{
"epoch": 0.3184,
"grad_norm": 0.038899699811397556,
"learning_rate": 0.0002585250557353799,
"loss": 0.7372,
"step": 995
},
{
"epoch": 0.32,
"grad_norm": 0.042642226425794946,
"learning_rate": 0.0002579449382388144,
"loss": 0.7485,
"step": 1000
},
{
"epoch": 0.3216,
"grad_norm": 0.039833966566896535,
"learning_rate": 0.00025736145244605746,
"loss": 0.7707,
"step": 1005
},
{
"epoch": 0.3232,
"grad_norm": 0.04531367629526789,
"learning_rate": 0.00025677461656410417,
"loss": 0.7553,
"step": 1010
},
{
"epoch": 0.3248,
"grad_norm": 0.04349819611665461,
"learning_rate": 0.00025618444890448543,
"loss": 0.7738,
"step": 1015
},
{
"epoch": 0.3264,
"grad_norm": 0.05023483178830991,
"learning_rate": 0.0002555909678826964,
"loss": 0.7796,
"step": 1020
},
{
"epoch": 0.328,
"grad_norm": 0.046817340356191714,
"learning_rate": 0.0002549941920176217,
"loss": 0.7387,
"step": 1025
},
{
"epoch": 0.3296,
"grad_norm": 6.598378786173098,
"learning_rate": 0.00025439413993095784,
"loss": 0.7527,
"step": 1030
},
{
"epoch": 0.3312,
"grad_norm": 0.06539168468511167,
"learning_rate": 0.00025379083034663194,
"loss": 0.7542,
"step": 1035
},
{
"epoch": 0.3328,
"grad_norm": 0.0411591198615963,
"learning_rate": 0.0002531842820902175,
"loss": 0.748,
"step": 1040
},
{
"epoch": 0.3344,
"grad_norm": 0.04659529885950376,
"learning_rate": 0.0002525745140883471,
"loss": 0.7702,
"step": 1045
},
{
"epoch": 0.336,
"grad_norm": 0.0505797168335258,
"learning_rate": 0.0002519615453681218,
"loss": 0.7619,
"step": 1050
},
{
"epoch": 0.3376,
"grad_norm": 0.04830566943349081,
"learning_rate": 0.00025134539505651716,
"loss": 0.762,
"step": 1055
},
{
"epoch": 0.3392,
"grad_norm": 0.04952600716886585,
"learning_rate": 0.00025072608237978664,
"loss": 0.7534,
"step": 1060
},
{
"epoch": 0.3408,
"grad_norm": 0.044007033813835646,
"learning_rate": 0.00025010362666286185,
"loss": 0.7692,
"step": 1065
},
{
"epoch": 0.3424,
"grad_norm": 0.04527316097454219,
"learning_rate": 0.00024947804732874896,
"loss": 0.7636,
"step": 1070
},
{
"epoch": 0.344,
"grad_norm": 0.045960440573293695,
"learning_rate": 0.0002488493638979233,
"loss": 0.7522,
"step": 1075
},
{
"epoch": 0.3456,
"grad_norm": 0.04427468398129268,
"learning_rate": 0.00024821759598771964,
"loss": 0.7541,
"step": 1080
},
{
"epoch": 0.3472,
"grad_norm": 0.04994994624897907,
"learning_rate": 0.0002475827633117206,
"loss": 0.7604,
"step": 1085
},
{
"epoch": 0.3488,
"grad_norm": 0.04346019358697389,
"learning_rate": 0.00024694488567914106,
"loss": 0.7466,
"step": 1090
},
{
"epoch": 0.3504,
"grad_norm": 0.045596836164609016,
"learning_rate": 0.0002463039829942104,
"loss": 0.7568,
"step": 1095
},
{
"epoch": 0.352,
"grad_norm": 0.04088907171766439,
"learning_rate": 0.0002456600752555511,
"loss": 0.7616,
"step": 1100
},
{
"epoch": 0.3536,
"grad_norm": 0.041758722037058514,
"learning_rate": 0.0002450131825555548,
"loss": 0.7501,
"step": 1105
},
{
"epoch": 0.3552,
"grad_norm": 0.050631334358167975,
"learning_rate": 0.00024436332507975553,
"loss": 0.7438,
"step": 1110
},
{
"epoch": 0.3568,
"grad_norm": 0.07414558397207001,
"learning_rate": 0.00024371052310619967,
"loss": 0.7289,
"step": 1115
},
{
"epoch": 0.3584,
"grad_norm": 0.043628197381851856,
"learning_rate": 0.00024305479700481317,
"loss": 0.7394,
"step": 1120
},
{
"epoch": 0.36,
"grad_norm": 0.0446052211482763,
"learning_rate": 0.000242396167236766,
"loss": 0.759,
"step": 1125
},
{
"epoch": 0.3616,
"grad_norm": 0.04481759941780531,
"learning_rate": 0.0002417346543538337,
"loss": 0.7491,
"step": 1130
},
{
"epoch": 0.3632,
"grad_norm": 0.0413461889977132,
"learning_rate": 0.00024107027899775596,
"loss": 0.7632,
"step": 1135
},
{
"epoch": 0.3648,
"grad_norm": 0.038375490339053624,
"learning_rate": 0.00024040306189959283,
"loss": 0.7524,
"step": 1140
},
{
"epoch": 0.3664,
"grad_norm": 0.04671975131634136,
"learning_rate": 0.0002397330238790774,
"loss": 0.746,
"step": 1145
},
{
"epoch": 0.368,
"grad_norm": 0.048330776029605456,
"learning_rate": 0.00023906018584396641,
"loss": 0.7619,
"step": 1150
},
{
"epoch": 0.3696,
"grad_norm": 0.04143069989457122,
"learning_rate": 0.00023838456878938786,
"loss": 0.7612,
"step": 1155
},
{
"epoch": 0.3712,
"grad_norm": 0.04377555777582751,
"learning_rate": 0.00023770619379718573,
"loss": 0.7658,
"step": 1160
},
{
"epoch": 0.3728,
"grad_norm": 0.049064434439728785,
"learning_rate": 0.00023702508203526235,
"loss": 0.7527,
"step": 1165
},
{
"epoch": 0.3744,
"grad_norm": 0.03784065311529772,
"learning_rate": 0.0002363412547569177,
"loss": 0.7452,
"step": 1170
},
{
"epoch": 0.376,
"grad_norm": 0.03980251791841271,
"learning_rate": 0.00023565473330018622,
"loss": 0.7385,
"step": 1175
},
{
"epoch": 0.3776,
"grad_norm": 0.04493262099167099,
"learning_rate": 0.00023496553908717115,
"loss": 0.7625,
"step": 1180
},
{
"epoch": 0.3792,
"grad_norm": 0.04314266993057437,
"learning_rate": 0.00023427369362337598,
"loss": 0.7492,
"step": 1185
},
{
"epoch": 0.3808,
"grad_norm": 0.04248357043924128,
"learning_rate": 0.00023357921849703335,
"loss": 0.7411,
"step": 1190
},
{
"epoch": 0.3824,
"grad_norm": 0.040379817198323655,
"learning_rate": 0.0002328821353784315,
"loss": 0.7321,
"step": 1195
},
{
"epoch": 0.384,
"grad_norm": 0.0360033809207297,
"learning_rate": 0.000232182466019238,
"loss": 0.7212,
"step": 1200
},
{
"epoch": 0.3856,
"grad_norm": 0.04200778920904238,
"learning_rate": 0.00023148023225182102,
"loss": 0.7527,
"step": 1205
},
{
"epoch": 0.3872,
"grad_norm": 0.046085792309150775,
"learning_rate": 0.00023077545598856815,
"loss": 0.7442,
"step": 1210
},
{
"epoch": 0.3888,
"grad_norm": 0.04277270461232854,
"learning_rate": 0.00023006815922120262,
"loss": 0.7189,
"step": 1215
},
{
"epoch": 0.3904,
"grad_norm": 0.0405134146379952,
"learning_rate": 0.00022935836402009706,
"loss": 0.7714,
"step": 1220
},
{
"epoch": 0.392,
"grad_norm": 0.04173101382827032,
"learning_rate": 0.00022864609253358474,
"loss": 0.7415,
"step": 1225
},
{
"epoch": 0.3936,
"grad_norm": 0.03865851171630709,
"learning_rate": 0.00022793136698726863,
"loss": 0.7428,
"step": 1230
},
{
"epoch": 0.3952,
"grad_norm": 0.03910224183500891,
"learning_rate": 0.00022721420968332766,
"loss": 0.7493,
"step": 1235
},
{
"epoch": 0.3968,
"grad_norm": 0.040305577555079186,
"learning_rate": 0.00022649464299982116,
"loss": 0.752,
"step": 1240
},
{
"epoch": 0.3984,
"grad_norm": 0.03909306971976583,
"learning_rate": 0.00022577268938999006,
"loss": 0.7216,
"step": 1245
},
{
"epoch": 0.4,
"grad_norm": 0.043954857050380644,
"learning_rate": 0.00022504837138155666,
"loss": 0.7335,
"step": 1250
},
{
"epoch": 0.4016,
"grad_norm": 0.04130228180090928,
"learning_rate": 0.0002243217115760217,
"loss": 0.7484,
"step": 1255
},
{
"epoch": 0.4032,
"grad_norm": 0.051231296902632124,
"learning_rate": 0.0002235927326479588,
"loss": 0.7453,
"step": 1260
},
{
"epoch": 0.4048,
"grad_norm": 0.048719573133536864,
"learning_rate": 0.00022286145734430713,
"loss": 0.7223,
"step": 1265
},
{
"epoch": 0.4064,
"grad_norm": 0.04017818677971695,
"learning_rate": 0.00022212790848366164,
"loss": 0.7282,
"step": 1270
},
{
"epoch": 0.408,
"grad_norm": 0.040992304875421104,
"learning_rate": 0.00022139210895556104,
"loss": 0.7559,
"step": 1275
},
{
"epoch": 0.4096,
"grad_norm": 0.04633597226041717,
"learning_rate": 0.00022065408171977326,
"loss": 0.7519,
"step": 1280
},
{
"epoch": 0.4112,
"grad_norm": 0.04261151707795407,
"learning_rate": 0.00021991384980557958,
"loss": 0.7467,
"step": 1285
},
{
"epoch": 0.4128,
"grad_norm": 0.04197091400228777,
"learning_rate": 0.0002191714363110554,
"loss": 0.7589,
"step": 1290
},
{
"epoch": 0.4144,
"grad_norm": 0.043035522590453124,
"learning_rate": 0.00021842686440235002,
"loss": 0.7537,
"step": 1295
},
{
"epoch": 0.416,
"grad_norm": 0.04395896783293326,
"learning_rate": 0.00021768015731296345,
"loss": 0.7481,
"step": 1300
},
{
"epoch": 0.4176,
"grad_norm": 0.043457994496385745,
"learning_rate": 0.00021693133834302145,
"loss": 0.7316,
"step": 1305
},
{
"epoch": 0.4192,
"grad_norm": 0.04478552802534991,
"learning_rate": 0.00021618043085854872,
"loss": 0.7404,
"step": 1310
},
{
"epoch": 0.4208,
"grad_norm": 0.0387795040439355,
"learning_rate": 0.00021542745829073958,
"loss": 0.7389,
"step": 1315
},
{
"epoch": 0.4224,
"grad_norm": 0.041546417532758506,
"learning_rate": 0.00021467244413522673,
"loss": 0.7416,
"step": 1320
},
{
"epoch": 0.424,
"grad_norm": 0.044665087388519445,
"learning_rate": 0.00021391541195134843,
"loss": 0.7459,
"step": 1325
},
{
"epoch": 0.4256,
"grad_norm": 0.04379214190156243,
"learning_rate": 0.0002131563853614131,
"loss": 0.7452,
"step": 1330
},
{
"epoch": 0.4272,
"grad_norm": 0.04238125297996064,
"learning_rate": 0.00021239538804996216,
"loss": 0.7539,
"step": 1335
},
{
"epoch": 0.4288,
"grad_norm": 0.038993894170660494,
"learning_rate": 0.00021163244376303132,
"loss": 0.7628,
"step": 1340
},
{
"epoch": 0.4304,
"grad_norm": 0.043983536037091105,
"learning_rate": 0.0002108675763074092,
"loss": 0.7467,
"step": 1345
},
{
"epoch": 0.432,
"grad_norm": 0.040167490443743736,
"learning_rate": 0.00021010080954989482,
"loss": 0.7438,
"step": 1350
},
{
"epoch": 0.4336,
"grad_norm": 0.03826596993091824,
"learning_rate": 0.00020933216741655258,
"loss": 0.7376,
"step": 1355
},
{
"epoch": 0.4352,
"grad_norm": 0.04237309509280299,
"learning_rate": 0.00020856167389196584,
"loss": 0.7297,
"step": 1360
},
{
"epoch": 0.4368,
"grad_norm": 0.03933319137399159,
"learning_rate": 0.0002077893530184885,
"loss": 0.7301,
"step": 1365
},
{
"epoch": 0.4384,
"grad_norm": 0.0412391306717837,
"learning_rate": 0.0002070152288954947,
"loss": 0.7404,
"step": 1370
},
{
"epoch": 0.44,
"grad_norm": 0.040127156770831986,
"learning_rate": 0.00020623932567862693,
"loss": 0.7268,
"step": 1375
},
{
"epoch": 0.4416,
"grad_norm": 0.044175415048005075,
"learning_rate": 0.0002054616675790423,
"loss": 0.7528,
"step": 1380
},
{
"epoch": 0.4432,
"grad_norm": 0.0403339831078633,
"learning_rate": 0.0002046822788626568,
"loss": 0.7392,
"step": 1385
},
{
"epoch": 0.4448,
"grad_norm": 0.03981323331482141,
"learning_rate": 0.00020390118384938842,
"loss": 0.7355,
"step": 1390
},
{
"epoch": 0.4464,
"grad_norm": 0.03936124576111004,
"learning_rate": 0.00020311840691239822,
"loss": 0.7415,
"step": 1395
},
{
"epoch": 0.448,
"grad_norm": 0.0474856170739699,
"learning_rate": 0.0002023339724773297,
"loss": 0.7372,
"step": 1400
},
{
"epoch": 0.4496,
"grad_norm": 0.0457480090375028,
"learning_rate": 0.00020154790502154653,
"loss": 0.7306,
"step": 1405
},
{
"epoch": 0.4512,
"grad_norm": 0.043907783218970334,
"learning_rate": 0.00020076022907336902,
"loss": 0.747,
"step": 1410
},
{
"epoch": 0.4528,
"grad_norm": 0.04296393850125629,
"learning_rate": 0.00019997096921130862,
"loss": 0.723,
"step": 1415
},
{
"epoch": 0.4544,
"grad_norm": 0.04697046178845861,
"learning_rate": 0.00019918015006330087,
"loss": 0.7437,
"step": 1420
},
{
"epoch": 0.456,
"grad_norm": 0.040119672728064056,
"learning_rate": 0.0001983877963059372,
"loss": 0.7397,
"step": 1425
},
{
"epoch": 0.4576,
"grad_norm": 0.046656663085348056,
"learning_rate": 0.00019759393266369443,
"loss": 0.7462,
"step": 1430
},
{
"epoch": 0.4592,
"grad_norm": 0.047948921068881094,
"learning_rate": 0.00019679858390816385,
"loss": 0.7375,
"step": 1435
},
{
"epoch": 0.4608,
"grad_norm": 0.04612317860998999,
"learning_rate": 0.000196001774857278,
"loss": 0.7282,
"step": 1440
},
{
"epoch": 0.4624,
"grad_norm": 0.041194594175399085,
"learning_rate": 0.00019520353037453598,
"loss": 0.7276,
"step": 1445
},
{
"epoch": 0.464,
"grad_norm": 0.048425503869563465,
"learning_rate": 0.00019440387536822807,
"loss": 0.7572,
"step": 1450
},
{
"epoch": 0.4656,
"grad_norm": 0.04213193734970952,
"learning_rate": 0.00019360283479065833,
"loss": 0.7349,
"step": 1455
},
{
"epoch": 0.4672,
"grad_norm": 0.04899128871136098,
"learning_rate": 0.00019280043363736579,
"loss": 0.734,
"step": 1460
},
{
"epoch": 0.4688,
"grad_norm": 0.04326757561876585,
"learning_rate": 0.0001919966969463448,
"loss": 0.7297,
"step": 1465
},
{
"epoch": 0.4704,
"grad_norm": 0.04024630860939001,
"learning_rate": 0.0001911916497972635,
"loss": 0.7204,
"step": 1470
},
{
"epoch": 0.472,
"grad_norm": 0.04008125998287669,
"learning_rate": 0.0001903853173106815,
"loss": 0.7317,
"step": 1475
},
{
"epoch": 0.4736,
"grad_norm": 0.03777929299122506,
"learning_rate": 0.00018957772464726574,
"loss": 0.732,
"step": 1480
},
{
"epoch": 0.4752,
"grad_norm": 0.04340084226893865,
"learning_rate": 0.00018876889700700556,
"loss": 0.7211,
"step": 1485
},
{
"epoch": 0.4768,
"grad_norm": 0.0420179308735609,
"learning_rate": 0.00018795885962842628,
"loss": 0.7399,
"step": 1490
},
{
"epoch": 0.4784,
"grad_norm": 0.044123021371490996,
"learning_rate": 0.0001871476377878018,
"loss": 0.7383,
"step": 1495
},
{
"epoch": 0.48,
"grad_norm": 0.040049404005910644,
"learning_rate": 0.00018633525679836568,
"loss": 0.7517,
"step": 1500
},
{
"epoch": 0.4816,
"grad_norm": 0.039647217074159906,
"learning_rate": 0.00018552174200952133,
"loss": 0.7312,
"step": 1505
},
{
"epoch": 0.4832,
"grad_norm": 0.04013028022885121,
"learning_rate": 0.00018470711880605122,
"loss": 0.729,
"step": 1510
},
{
"epoch": 0.4848,
"grad_norm": 0.04367909174578893,
"learning_rate": 0.00018389141260732444,
"loss": 0.7304,
"step": 1515
},
{
"epoch": 0.4864,
"grad_norm": 0.04061130668006248,
"learning_rate": 0.00018307464886650377,
"loss": 0.7197,
"step": 1520
},
{
"epoch": 0.488,
"grad_norm": 0.03900764609449006,
"learning_rate": 0.00018225685306975134,
"loss": 0.7274,
"step": 1525
},
{
"epoch": 0.4896,
"grad_norm": 0.044192934458913945,
"learning_rate": 0.00018143805073543343,
"loss": 0.7257,
"step": 1530
},
{
"epoch": 0.4912,
"grad_norm": 0.03866868780491527,
"learning_rate": 0.00018061826741332405,
"loss": 0.7209,
"step": 1535
},
{
"epoch": 0.4928,
"grad_norm": 0.03872685480769255,
"learning_rate": 0.0001797975286838079,
"loss": 0.7224,
"step": 1540
},
{
"epoch": 0.4944,
"grad_norm": 0.04005309906829608,
"learning_rate": 0.00017897586015708198,
"loss": 0.7302,
"step": 1545
},
{
"epoch": 0.496,
"grad_norm": 0.04231012970632499,
"learning_rate": 0.0001781532874723566,
"loss": 0.7443,
"step": 1550
},
{
"epoch": 0.4976,
"grad_norm": 0.04312701702411617,
"learning_rate": 0.0001773298362970552,
"loss": 0.7231,
"step": 1555
},
{
"epoch": 0.4992,
"grad_norm": 0.041854731258554205,
"learning_rate": 0.00017650553232601354,
"loss": 0.738,
"step": 1560
},
{
"epoch": 0.5008,
"grad_norm": 0.0423243123135773,
"learning_rate": 0.00017568040128067782,
"loss": 0.7481,
"step": 1565
},
{
"epoch": 0.5024,
"grad_norm": 0.041282692211946664,
"learning_rate": 0.00017485446890830225,
"loss": 0.7263,
"step": 1570
},
{
"epoch": 0.504,
"grad_norm": 0.037317198782344456,
"learning_rate": 0.00017402776098114539,
"loss": 0.7184,
"step": 1575
},
{
"epoch": 0.5056,
"grad_norm": 0.040461410318691195,
"learning_rate": 0.00017320030329566615,
"loss": 0.7199,
"step": 1580
},
{
"epoch": 0.5072,
"grad_norm": 0.03915242982171056,
"learning_rate": 0.0001723721216717188,
"loss": 0.7238,
"step": 1585
},
{
"epoch": 0.5088,
"grad_norm": 0.035306430185746605,
"learning_rate": 0.00017154324195174718,
"loss": 0.7186,
"step": 1590
},
{
"epoch": 0.5104,
"grad_norm": 0.04328294482977798,
"learning_rate": 0.00017071368999997848,
"loss": 0.7187,
"step": 1595
},
{
"epoch": 0.512,
"grad_norm": 0.040552285825681135,
"learning_rate": 0.00016988349170161607,
"loss": 0.7271,
"step": 1600
},
{
"epoch": 0.5136,
"grad_norm": 0.03792304405548647,
"learning_rate": 0.0001690526729620318,
"loss": 0.7401,
"step": 1605
},
{
"epoch": 0.5152,
"grad_norm": 0.03739907062711767,
"learning_rate": 0.00016822125970595772,
"loss": 0.7298,
"step": 1610
},
{
"epoch": 0.5168,
"grad_norm": 0.04045585582372931,
"learning_rate": 0.00016738927787667706,
"loss": 0.7357,
"step": 1615
},
{
"epoch": 0.5184,
"grad_norm": 0.039086636300895435,
"learning_rate": 0.00016655675343521463,
"loss": 0.7216,
"step": 1620
},
{
"epoch": 0.52,
"grad_norm": 0.0378622792604208,
"learning_rate": 0.00016572371235952697,
"loss": 0.725,
"step": 1625
},
{
"epoch": 0.5216,
"grad_norm": 0.03849206417818264,
"learning_rate": 0.0001648901806436914,
"loss": 0.7315,
"step": 1630
},
{
"epoch": 0.5232,
"grad_norm": 0.03949752544512463,
"learning_rate": 0.00016405618429709531,
"loss": 0.7093,
"step": 1635
},
{
"epoch": 0.5248,
"grad_norm": 0.044062583731413654,
"learning_rate": 0.00016322174934362422,
"loss": 0.7385,
"step": 1640
},
{
"epoch": 0.5264,
"grad_norm": 0.04089556591693187,
"learning_rate": 0.00016238690182084986,
"loss": 0.7046,
"step": 1645
},
{
"epoch": 0.528,
"grad_norm": 0.04222802674498491,
"learning_rate": 0.0001615516677792177,
"loss": 0.7289,
"step": 1650
},
{
"epoch": 0.5296,
"grad_norm": 0.03761221440074904,
"learning_rate": 0.00016071607328123425,
"loss": 0.7258,
"step": 1655
},
{
"epoch": 0.5312,
"grad_norm": 0.04005046134119886,
"learning_rate": 0.0001598801444006534,
"loss": 0.7424,
"step": 1660
},
{
"epoch": 0.5328,
"grad_norm": 0.040509948741415,
"learning_rate": 0.0001590439072216632,
"loss": 0.7041,
"step": 1665
},
{
"epoch": 0.5344,
"grad_norm": 0.03810512707022326,
"learning_rate": 0.0001582073878380718,
"loss": 0.7185,
"step": 1670
},
{
"epoch": 0.536,
"grad_norm": 0.04280979793390455,
"learning_rate": 0.00015737061235249309,
"loss": 0.7029,
"step": 1675
},
{
"epoch": 0.5376,
"grad_norm": 0.04248308027092899,
"learning_rate": 0.0001565336068755325,
"loss": 0.735,
"step": 1680
},
{
"epoch": 0.5392,
"grad_norm": 0.03928332103104811,
"learning_rate": 0.00015569639752497196,
"loss": 0.7102,
"step": 1685
},
{
"epoch": 0.5408,
"grad_norm": 0.03919558675753705,
"learning_rate": 0.00015485901042495504,
"loss": 0.7291,
"step": 1690
},
{
"epoch": 0.5424,
"grad_norm": 0.0391348168167258,
"learning_rate": 0.00015402147170517184,
"loss": 0.7166,
"step": 1695
},
{
"epoch": 0.544,
"grad_norm": 0.03714108757450143,
"learning_rate": 0.00015318380750004352,
"loss": 0.7272,
"step": 1700
},
{
"epoch": 0.5456,
"grad_norm": 0.03568601370768205,
"learning_rate": 0.00015234604394790687,
"loss": 0.7361,
"step": 1705
},
{
"epoch": 0.5472,
"grad_norm": 0.0386630719265334,
"learning_rate": 0.00015150820719019874,
"loss": 0.725,
"step": 1710
},
{
"epoch": 0.5488,
"grad_norm": 0.03993919583787361,
"learning_rate": 0.00015067032337064018,
"loss": 0.7225,
"step": 1715
},
{
"epoch": 0.5504,
"grad_norm": 0.04091268785555554,
"learning_rate": 0.00014983241863442086,
"loss": 0.7355,
"step": 1720
},
{
"epoch": 0.552,
"grad_norm": 0.03823212279742222,
"learning_rate": 0.00014899451912738307,
"loss": 0.7124,
"step": 1725
},
{
"epoch": 0.5536,
"grad_norm": 0.03775761218114033,
"learning_rate": 0.00014815665099520588,
"loss": 0.6985,
"step": 1730
},
{
"epoch": 0.5552,
"grad_norm": 0.04149923045436972,
"learning_rate": 0.00014731884038258946,
"loss": 0.7099,
"step": 1735
},
{
"epoch": 0.5568,
"grad_norm": 0.04014086331686014,
"learning_rate": 0.00014648111343243907,
"loss": 0.7364,
"step": 1740
},
{
"epoch": 0.5584,
"grad_norm": 0.039575959854944824,
"learning_rate": 0.00014564349628504937,
"loss": 0.7114,
"step": 1745
},
{
"epoch": 0.56,
"grad_norm": 0.036851853297479614,
"learning_rate": 0.00014480601507728885,
"loss": 0.6964,
"step": 1750
},
{
"epoch": 0.5616,
"grad_norm": 0.041213192635326586,
"learning_rate": 0.00014396869594178403,
"loss": 0.7139,
"step": 1755
},
{
"epoch": 0.5632,
"grad_norm": 0.04352972854945724,
"learning_rate": 0.0001431315650061042,
"loss": 0.7267,
"step": 1760
},
{
"epoch": 0.5648,
"grad_norm": 0.04497516779493022,
"learning_rate": 0.00014229464839194618,
"loss": 0.7189,
"step": 1765
},
{
"epoch": 0.5664,
"grad_norm": 0.039398507397311704,
"learning_rate": 0.00014145797221431903,
"loss": 0.7236,
"step": 1770
},
{
"epoch": 0.568,
"grad_norm": 0.040805259594060624,
"learning_rate": 0.0001406215625807293,
"loss": 0.7167,
"step": 1775
},
{
"epoch": 0.5696,
"grad_norm": 0.04047421525047546,
"learning_rate": 0.00013978544559036638,
"loss": 0.7137,
"step": 1780
},
{
"epoch": 0.5712,
"grad_norm": 0.03919630804845936,
"learning_rate": 0.00013894964733328797,
"loss": 0.719,
"step": 1785
},
{
"epoch": 0.5728,
"grad_norm": 0.03685077424222423,
"learning_rate": 0.00013811419388960622,
"loss": 0.7278,
"step": 1790
},
{
"epoch": 0.5744,
"grad_norm": 0.044753528752375836,
"learning_rate": 0.00013727911132867365,
"loss": 0.7095,
"step": 1795
},
{
"epoch": 0.576,
"grad_norm": 0.03991640163138795,
"learning_rate": 0.00013644442570826982,
"loss": 0.723,
"step": 1800
},
{
"epoch": 0.5776,
"grad_norm": 0.040087229919929294,
"learning_rate": 0.0001356101630737883,
"loss": 0.7127,
"step": 1805
},
{
"epoch": 0.5792,
"grad_norm": 0.04266162698298314,
"learning_rate": 0.0001347763494574239,
"loss": 0.7279,
"step": 1810
},
{
"epoch": 0.5808,
"grad_norm": 0.037795236144364724,
"learning_rate": 0.0001339430108773602,
"loss": 0.7191,
"step": 1815
},
{
"epoch": 0.5824,
"grad_norm": 0.0387693849128071,
"learning_rate": 0.00013311017333695796,
"loss": 0.7243,
"step": 1820
},
{
"epoch": 0.584,
"grad_norm": 0.03774693738753151,
"learning_rate": 0.00013227786282394349,
"loss": 0.7038,
"step": 1825
},
{
"epoch": 0.5856,
"grad_norm": 0.03812021420863255,
"learning_rate": 0.00013144610530959784,
"loss": 0.7161,
"step": 1830
},
{
"epoch": 0.5872,
"grad_norm": 0.03834520692315359,
"learning_rate": 0.00013061492674794646,
"loss": 0.7187,
"step": 1835
},
{
"epoch": 0.5888,
"grad_norm": 0.03678653982482293,
"learning_rate": 0.00012978435307494916,
"loss": 0.7103,
"step": 1840
},
{
"epoch": 0.5904,
"grad_norm": 0.039024408400889526,
"learning_rate": 0.00012895441020769095,
"loss": 0.7081,
"step": 1845
},
{
"epoch": 0.592,
"grad_norm": 0.03946176567463419,
"learning_rate": 0.00012812512404357328,
"loss": 0.7297,
"step": 1850
},
{
"epoch": 0.5936,
"grad_norm": 0.03966986239112185,
"learning_rate": 0.0001272965204595059,
"loss": 0.7224,
"step": 1855
},
{
"epoch": 0.5952,
"grad_norm": 0.04045671877501302,
"learning_rate": 0.00012646862531109945,
"loss": 0.7138,
"step": 1860
},
{
"epoch": 0.5968,
"grad_norm": 0.04200376635921014,
"learning_rate": 0.0001256414644318588,
"loss": 0.7178,
"step": 1865
},
{
"epoch": 0.5984,
"grad_norm": 0.03978046841085274,
"learning_rate": 0.00012481506363237654,
"loss": 0.7067,
"step": 1870
},
{
"epoch": 0.6,
"grad_norm": 0.04238382651336426,
"learning_rate": 0.00012398944869952816,
"loss": 0.7274,
"step": 1875
},
{
"epoch": 0.6016,
"grad_norm": 0.038817886095841726,
"learning_rate": 0.00012316464539566683,
"loss": 0.7122,
"step": 1880
},
{
"epoch": 0.6032,
"grad_norm": 0.03594793213572876,
"learning_rate": 0.0001223406794578199,
"loss": 0.6961,
"step": 1885
},
{
"epoch": 0.6048,
"grad_norm": 0.03697364543172882,
"learning_rate": 0.00012151757659688571,
"loss": 0.7113,
"step": 1890
},
{
"epoch": 0.6064,
"grad_norm": 0.037004779496744064,
"learning_rate": 0.00012069536249683134,
"loss": 0.7131,
"step": 1895
},
{
"epoch": 0.608,
"grad_norm": 0.0396435179197239,
"learning_rate": 0.00011987406281389094,
"loss": 0.6998,
"step": 1900
},
{
"epoch": 0.6096,
"grad_norm": 0.03869430143018022,
"learning_rate": 0.00011905370317576543,
"loss": 0.703,
"step": 1905
},
{
"epoch": 0.6112,
"grad_norm": 0.040681377456567816,
"learning_rate": 0.00011823430918082282,
"loss": 0.7126,
"step": 1910
},
{
"epoch": 0.6128,
"grad_norm": 0.03873128265195665,
"learning_rate": 0.00011741590639729918,
"loss": 0.7133,
"step": 1915
},
{
"epoch": 0.6144,
"grad_norm": 0.042933631355136996,
"learning_rate": 0.00011659852036250113,
"loss": 0.7118,
"step": 1920
},
{
"epoch": 0.616,
"grad_norm": 0.04491368497490653,
"learning_rate": 0.00011578217658200874,
"loss": 0.7094,
"step": 1925
},
{
"epoch": 0.6176,
"grad_norm": 0.03787882428314268,
"learning_rate": 0.00011496690052887973,
"loss": 0.7213,
"step": 1930
},
{
"epoch": 0.6192,
"grad_norm": 0.037724023271429444,
"learning_rate": 0.00011415271764285474,
"loss": 0.7118,
"step": 1935
},
{
"epoch": 0.6208,
"grad_norm": 0.04085791852018764,
"learning_rate": 0.00011333965332956332,
"loss": 0.7241,
"step": 1940
},
{
"epoch": 0.6224,
"grad_norm": 0.03841379840100555,
"learning_rate": 0.00011252773295973121,
"loss": 0.6915,
"step": 1945
},
{
"epoch": 0.624,
"grad_norm": 0.039180179570884835,
"learning_rate": 0.00011171698186838887,
"loss": 0.6933,
"step": 1950
},
{
"epoch": 0.6256,
"grad_norm": 0.04020685900513363,
"learning_rate": 0.00011090742535408063,
"loss": 0.7099,
"step": 1955
},
{
"epoch": 0.6272,
"grad_norm": 0.0415898031303519,
"learning_rate": 0.00011009908867807556,
"loss": 0.7092,
"step": 1960
},
{
"epoch": 0.6288,
"grad_norm": 0.04650092478952174,
"learning_rate": 0.00010929199706357905,
"loss": 0.6972,
"step": 1965
},
{
"epoch": 0.6304,
"grad_norm": 0.03819491431751356,
"learning_rate": 0.0001084861756949457,
"loss": 0.6933,
"step": 1970
},
{
"epoch": 0.632,
"grad_norm": 0.03674646898808985,
"learning_rate": 0.00010768164971689374,
"loss": 0.6946,
"step": 1975
},
{
"epoch": 0.6336,
"grad_norm": 0.04079648142913826,
"learning_rate": 0.00010687844423372018,
"loss": 0.7057,
"step": 1980
},
{
"epoch": 0.6352,
"grad_norm": 0.04442964097385929,
"learning_rate": 0.00010607658430851744,
"loss": 0.7079,
"step": 1985
},
{
"epoch": 0.6368,
"grad_norm": 0.041439999468769535,
"learning_rate": 0.00010527609496239142,
"loss": 0.7094,
"step": 1990
},
{
"epoch": 0.6384,
"grad_norm": 0.03612721628167792,
"learning_rate": 0.00010447700117368078,
"loss": 0.7088,
"step": 1995
},
{
"epoch": 0.64,
"grad_norm": 0.04369717146674075,
"learning_rate": 0.00010367932787717727,
"loss": 0.7434,
"step": 2000
},
{
"epoch": 0.6416,
"grad_norm": 0.041592772074316386,
"learning_rate": 0.00010288309996334801,
"loss": 0.7133,
"step": 2005
},
{
"epoch": 0.6432,
"grad_norm": 0.041785169745661155,
"learning_rate": 0.00010208834227755847,
"loss": 0.7039,
"step": 2010
},
{
"epoch": 0.6448,
"grad_norm": 0.0382736079703658,
"learning_rate": 0.00010129507961929748,
"loss": 0.7041,
"step": 2015
},
{
"epoch": 0.6464,
"grad_norm": 0.03803427582508488,
"learning_rate": 0.00010050333674140326,
"loss": 0.7349,
"step": 2020
},
{
"epoch": 0.648,
"grad_norm": 0.04429830134367503,
"learning_rate": 9.971313834929099e-05,
"loss": 0.7127,
"step": 2025
},
{
"epoch": 0.6496,
"grad_norm": 0.04092289365426379,
"learning_rate": 9.892450910018203e-05,
"loss": 0.709,
"step": 2030
},
{
"epoch": 0.6512,
"grad_norm": 0.038535908469184546,
"learning_rate": 9.813747360233443e-05,
"loss": 0.6825,
"step": 2035
},
{
"epoch": 0.6528,
"grad_norm": 0.03773550734191166,
"learning_rate": 9.73520564142751e-05,
"loss": 0.6937,
"step": 2040
},
{
"epoch": 0.6544,
"grad_norm": 0.03897370588757223,
"learning_rate": 9.656828204403352e-05,
"loss": 0.7035,
"step": 2045
},
{
"epoch": 0.656,
"grad_norm": 0.03993122832767393,
"learning_rate": 9.578617494837682e-05,
"loss": 0.7098,
"step": 2050
},
{
"epoch": 0.6576,
"grad_norm": 0.04113301544799768,
"learning_rate": 9.500575953204684e-05,
"loss": 0.7067,
"step": 2055
},
{
"epoch": 0.6592,
"grad_norm": 0.04267204755150456,
"learning_rate": 9.422706014699863e-05,
"loss": 0.7192,
"step": 2060
},
{
"epoch": 0.6608,
"grad_norm": 0.038540803442012495,
"learning_rate": 9.345010109164037e-05,
"loss": 0.6948,
"step": 2065
},
{
"epoch": 0.6624,
"grad_norm": 0.03639271405028558,
"learning_rate": 9.267490661007529e-05,
"loss": 0.6982,
"step": 2070
},
{
"epoch": 0.664,
"grad_norm": 0.03764258257256097,
"learning_rate": 9.190150089134512e-05,
"loss": 0.7011,
"step": 2075
},
{
"epoch": 0.6656,
"grad_norm": 0.03741160593359946,
"learning_rate": 9.112990806867543e-05,
"loss": 0.7077,
"step": 2080
},
{
"epoch": 0.6672,
"grad_norm": 0.04138034247014772,
"learning_rate": 9.036015221872238e-05,
"loss": 0.7348,
"step": 2085
},
{
"epoch": 0.6688,
"grad_norm": 0.03629720795501264,
"learning_rate": 8.959225736082158e-05,
"loss": 0.6773,
"step": 2090
},
{
"epoch": 0.6704,
"grad_norm": 0.03804004946419038,
"learning_rate": 8.882624745623856e-05,
"loss": 0.7019,
"step": 2095
},
{
"epoch": 0.672,
"grad_norm": 0.035721267232502185,
"learning_rate": 8.806214640742099e-05,
"loss": 0.7145,
"step": 2100
},
{
"epoch": 0.6736,
"grad_norm": 0.03632799434171817,
"learning_rate": 8.729997805725303e-05,
"loss": 0.6927,
"step": 2105
},
{
"epoch": 0.6752,
"grad_norm": 0.036651796521166076,
"learning_rate": 8.653976618831118e-05,
"loss": 0.7045,
"step": 2110
},
{
"epoch": 0.6768,
"grad_norm": 0.037066902859228194,
"learning_rate": 8.578153452212221e-05,
"loss": 0.7155,
"step": 2115
},
{
"epoch": 0.6784,
"grad_norm": 0.03767995581990963,
"learning_rate": 8.502530671842288e-05,
"loss": 0.7043,
"step": 2120
},
{
"epoch": 0.68,
"grad_norm": 0.03561868304758191,
"learning_rate": 8.427110637442201e-05,
"loss": 0.7129,
"step": 2125
},
{
"epoch": 0.6816,
"grad_norm": 0.038918395104413535,
"learning_rate": 8.351895702406366e-05,
"loss": 0.7143,
"step": 2130
},
{
"epoch": 0.6832,
"grad_norm": 0.035964229362269014,
"learning_rate": 8.276888213729307e-05,
"loss": 0.6861,
"step": 2135
},
{
"epoch": 0.6848,
"grad_norm": 0.03733779735545552,
"learning_rate": 8.202090511932428e-05,
"loss": 0.7153,
"step": 2140
},
{
"epoch": 0.6864,
"grad_norm": 0.03770594702466583,
"learning_rate": 8.127504930990986e-05,
"loss": 0.7121,
"step": 2145
},
{
"epoch": 0.688,
"grad_norm": 0.03768921861691119,
"learning_rate": 8.053133798261233e-05,
"loss": 0.7034,
"step": 2150
},
{
"epoch": 0.6896,
"grad_norm": 0.03512675419142099,
"learning_rate": 7.978979434407842e-05,
"loss": 0.6788,
"step": 2155
},
{
"epoch": 0.6912,
"grad_norm": 0.033678086620005154,
"learning_rate": 7.905044153331441e-05,
"loss": 0.711,
"step": 2160
},
{
"epoch": 0.6928,
"grad_norm": 0.04310639146017272,
"learning_rate": 7.831330262096444e-05,
"loss": 0.7007,
"step": 2165
},
{
"epoch": 0.6944,
"grad_norm": 0.035740468353472145,
"learning_rate": 7.757840060859053e-05,
"loss": 0.6907,
"step": 2170
},
{
"epoch": 0.696,
"grad_norm": 0.0388185330369263,
"learning_rate": 7.684575842795485e-05,
"loss": 0.7066,
"step": 2175
},
{
"epoch": 0.6976,
"grad_norm": 0.035326733296463714,
"learning_rate": 7.611539894030405e-05,
"loss": 0.6856,
"step": 2180
},
{
"epoch": 0.6992,
"grad_norm": 0.03673006246399958,
"learning_rate": 7.53873449356562e-05,
"loss": 0.7035,
"step": 2185
},
{
"epoch": 0.7008,
"grad_norm": 0.04032795219690735,
"learning_rate": 7.466161913208932e-05,
"loss": 0.713,
"step": 2190
},
{
"epoch": 0.7024,
"grad_norm": 0.04007204362489801,
"learning_rate": 7.39382441750326e-05,
"loss": 0.6855,
"step": 2195
},
{
"epoch": 0.704,
"grad_norm": 0.0349153417362628,
"learning_rate": 7.321724263655988e-05,
"loss": 0.6931,
"step": 2200
},
{
"epoch": 0.7056,
"grad_norm": 0.039088044797178166,
"learning_rate": 7.249863701468514e-05,
"loss": 0.7003,
"step": 2205
},
{
"epoch": 0.7072,
"grad_norm": 0.036879283255029253,
"learning_rate": 7.178244973266072e-05,
"loss": 0.7259,
"step": 2210
},
{
"epoch": 0.7088,
"grad_norm": 0.03653281890039639,
"learning_rate": 7.106870313827734e-05,
"loss": 0.7083,
"step": 2215
},
{
"epoch": 0.7104,
"grad_norm": 0.0378300680680633,
"learning_rate": 7.035741950316698e-05,
"loss": 0.688,
"step": 2220
},
{
"epoch": 0.712,
"grad_norm": 0.03938909267628237,
"learning_rate": 6.964862102210773e-05,
"loss": 0.6902,
"step": 2225
},
{
"epoch": 0.7136,
"grad_norm": 0.043292695863837646,
"learning_rate": 6.894232981233154e-05,
"loss": 0.707,
"step": 2230
},
{
"epoch": 0.7152,
"grad_norm": 0.03873590569931821,
"learning_rate": 6.823856791283366e-05,
"loss": 0.7172,
"step": 2235
},
{
"epoch": 0.7168,
"grad_norm": 0.040179163618301025,
"learning_rate": 6.753735728368534e-05,
"loss": 0.7001,
"step": 2240
},
{
"epoch": 0.7184,
"grad_norm": 0.03665089260769722,
"learning_rate": 6.683871980534833e-05,
"loss": 0.7119,
"step": 2245
},
{
"epoch": 0.72,
"grad_norm": 0.04017293034485019,
"learning_rate": 6.614267727799212e-05,
"loss": 0.6988,
"step": 2250
},
{
"epoch": 0.7216,
"grad_norm": 0.03813762526767263,
"learning_rate": 6.544925142081391e-05,
"loss": 0.7121,
"step": 2255
},
{
"epoch": 0.7232,
"grad_norm": 0.03854828556279748,
"learning_rate": 6.475846387136067e-05,
"loss": 0.6962,
"step": 2260
},
{
"epoch": 0.7248,
"grad_norm": 0.03719435244618414,
"learning_rate": 6.4070336184854e-05,
"loss": 0.6927,
"step": 2265
},
{
"epoch": 0.7264,
"grad_norm": 0.04241730525764197,
"learning_rate": 6.338488983351777e-05,
"loss": 0.7053,
"step": 2270
},
{
"epoch": 0.728,
"grad_norm": 0.03572759759362592,
"learning_rate": 6.270214620590773e-05,
"loss": 0.7022,
"step": 2275
},
{
"epoch": 0.7296,
"grad_norm": 0.03946181148151894,
"learning_rate": 6.202212660624432e-05,
"loss": 0.7106,
"step": 2280
},
{
"epoch": 0.7312,
"grad_norm": 0.03882270135457663,
"learning_rate": 6.134485225374787e-05,
"loss": 0.685,
"step": 2285
},
{
"epoch": 0.7328,
"grad_norm": 0.0379435383069947,
"learning_rate": 6.06703442819764e-05,
"loss": 0.6952,
"step": 2290
},
{
"epoch": 0.7344,
"grad_norm": 0.03820113955011566,
"learning_rate": 5.999862373816643e-05,
"loss": 0.6873,
"step": 2295
},
{
"epoch": 0.736,
"grad_norm": 0.04381507714614669,
"learning_rate": 5.932971158257583e-05,
"loss": 0.6982,
"step": 2300
},
{
"epoch": 0.7376,
"grad_norm": 0.045255086563538174,
"learning_rate": 5.866362868783008e-05,
"loss": 0.711,
"step": 2305
},
{
"epoch": 0.7392,
"grad_norm": 0.03831786593553919,
"learning_rate": 5.800039583827083e-05,
"loss": 0.6905,
"step": 2310
},
{
"epoch": 0.7408,
"grad_norm": 0.038176364964366265,
"learning_rate": 5.7340033729307355e-05,
"loss": 0.6952,
"step": 2315
},
{
"epoch": 0.7424,
"grad_norm": 0.03557728343089575,
"learning_rate": 5.668256296677093e-05,
"loss": 0.6942,
"step": 2320
},
{
"epoch": 0.744,
"grad_norm": 0.037453237191514,
"learning_rate": 5.602800406627168e-05,
"loss": 0.704,
"step": 2325
},
{
"epoch": 0.7456,
"grad_norm": 0.04201408146495244,
"learning_rate": 5.5376377452558405e-05,
"loss": 0.6901,
"step": 2330
},
{
"epoch": 0.7472,
"grad_norm": 0.03459644417240455,
"learning_rate": 5.472770345888134e-05,
"loss": 0.6893,
"step": 2335
},
{
"epoch": 0.7488,
"grad_norm": 0.03576418609943801,
"learning_rate": 5.4082002326357655e-05,
"loss": 0.6953,
"step": 2340
},
{
"epoch": 0.7504,
"grad_norm": 0.03654844796403911,
"learning_rate": 5.343929420333984e-05,
"loss": 0.6902,
"step": 2345
},
{
"epoch": 0.752,
"grad_norm": 0.043819614388394754,
"learning_rate": 5.2799599144786966e-05,
"loss": 0.7114,
"step": 2350
},
{
"epoch": 0.7536,
"grad_norm": 0.04040722566989979,
"learning_rate": 5.216293711163908e-05,
"loss": 0.6901,
"step": 2355
},
{
"epoch": 0.7552,
"grad_norm": 0.03649315688342495,
"learning_rate": 5.152932797019403e-05,
"loss": 0.691,
"step": 2360
},
{
"epoch": 0.7568,
"grad_norm": 0.03671257352284008,
"learning_rate": 5.089879149148781e-05,
"loss": 0.6973,
"step": 2365
},
{
"epoch": 0.7584,
"grad_norm": 0.03720399162630074,
"learning_rate": 5.027134735067753e-05,
"loss": 0.669,
"step": 2370
},
{
"epoch": 0.76,
"grad_norm": 0.03563182343583573,
"learning_rate": 4.964701512642746e-05,
"loss": 0.689,
"step": 2375
},
{
"epoch": 0.7616,
"grad_norm": 0.035955548417391445,
"learning_rate": 4.9025814300298265e-05,
"loss": 0.6944,
"step": 2380
},
{
"epoch": 0.7632,
"grad_norm": 0.036492436660829214,
"learning_rate": 4.840776425613886e-05,
"loss": 0.6876,
"step": 2385
},
{
"epoch": 0.7648,
"grad_norm": 0.04191984666054339,
"learning_rate": 4.779288427948168e-05,
"loss": 0.7135,
"step": 2390
},
{
"epoch": 0.7664,
"grad_norm": 0.0389152552227016,
"learning_rate": 4.718119355694091e-05,
"loss": 0.7025,
"step": 2395
},
{
"epoch": 0.768,
"grad_norm": 0.03552598244342018,
"learning_rate": 4.657271117561374e-05,
"loss": 0.6891,
"step": 2400
},
{
"epoch": 0.7696,
"grad_norm": 0.037991199731399375,
"learning_rate": 4.596745612248488e-05,
"loss": 0.7018,
"step": 2405
},
{
"epoch": 0.7712,
"grad_norm": 0.04009265600800646,
"learning_rate": 4.536544728383403e-05,
"loss": 0.6975,
"step": 2410
},
{
"epoch": 0.7728,
"grad_norm": 0.034383530574825634,
"learning_rate": 4.476670344464645e-05,
"loss": 0.6718,
"step": 2415
},
{
"epoch": 0.7744,
"grad_norm": 0.034993269435869126,
"learning_rate": 4.417124328802694e-05,
"loss": 0.684,
"step": 2420
},
{
"epoch": 0.776,
"grad_norm": 0.03425009459589239,
"learning_rate": 4.357908539461678e-05,
"loss": 0.6858,
"step": 2425
},
{
"epoch": 0.7776,
"grad_norm": 0.03919492090647449,
"learning_rate": 4.299024824201402e-05,
"loss": 0.6988,
"step": 2430
},
{
"epoch": 0.7792,
"grad_norm": 0.034476194930932354,
"learning_rate": 4.240475020419675e-05,
"loss": 0.6733,
"step": 2435
},
{
"epoch": 0.7808,
"grad_norm": 0.03498517451850298,
"learning_rate": 4.1822609550950066e-05,
"loss": 0.6847,
"step": 2440
},
{
"epoch": 0.7824,
"grad_norm": 0.03513837580346364,
"learning_rate": 4.124384444729561e-05,
"loss": 0.7061,
"step": 2445
},
{
"epoch": 0.784,
"grad_norm": 0.03568352623910555,
"learning_rate": 4.0668472952924977e-05,
"loss": 0.7133,
"step": 2450
},
{
"epoch": 0.7856,
"grad_norm": 0.03452583660811545,
"learning_rate": 4.009651302163614e-05,
"loss": 0.6906,
"step": 2455
},
{
"epoch": 0.7872,
"grad_norm": 0.037618350024330255,
"learning_rate": 3.952798250077317e-05,
"loss": 0.6919,
"step": 2460
},
{
"epoch": 0.7888,
"grad_norm": 0.03692436380717943,
"learning_rate": 3.896289913066952e-05,
"loss": 0.7076,
"step": 2465
},
{
"epoch": 0.7904,
"grad_norm": 0.03430203099467551,
"learning_rate": 3.840128054409414e-05,
"loss": 0.682,
"step": 2470
},
{
"epoch": 0.792,
"grad_norm": 0.03461351398958901,
"learning_rate": 3.784314426570156e-05,
"loss": 0.6773,
"step": 2475
},
{
"epoch": 0.7936,
"grad_norm": 0.03921640450921252,
"learning_rate": 3.728850771148486e-05,
"loss": 0.6973,
"step": 2480
},
{
"epoch": 0.7952,
"grad_norm": 0.03544120496297954,
"learning_rate": 3.67373881882323e-05,
"loss": 0.71,
"step": 2485
},
{
"epoch": 0.7968,
"grad_norm": 0.036332234631180356,
"learning_rate": 3.618980289298736e-05,
"loss": 0.6869,
"step": 2490
},
{
"epoch": 0.7984,
"grad_norm": 0.0350795962674337,
"learning_rate": 3.5645768912512046e-05,
"loss": 0.6763,
"step": 2495
},
{
"epoch": 0.8,
"grad_norm": 0.03620429105944647,
"learning_rate": 3.510530322275362e-05,
"loss": 0.7201,
"step": 2500
},
{
"epoch": 0.8016,
"grad_norm": 0.034305049981790806,
"learning_rate": 3.4568422688315025e-05,
"loss": 0.6943,
"step": 2505
},
{
"epoch": 0.8032,
"grad_norm": 0.036585621965710045,
"learning_rate": 3.4035144061928635e-05,
"loss": 0.6916,
"step": 2510
},
{
"epoch": 0.8048,
"grad_norm": 0.037124612583122524,
"learning_rate": 3.350548398393345e-05,
"loss": 0.719,
"step": 2515
},
{
"epoch": 0.8064,
"grad_norm": 0.035350388626476145,
"learning_rate": 3.297945898175584e-05,
"loss": 0.6869,
"step": 2520
},
{
"epoch": 0.808,
"grad_norm": 0.0368119047385571,
"learning_rate": 3.245708546939401e-05,
"loss": 0.6981,
"step": 2525
},
{
"epoch": 0.8096,
"grad_norm": 0.038393586777655066,
"learning_rate": 3.19383797469055e-05,
"loss": 0.6906,
"step": 2530
},
{
"epoch": 0.8112,
"grad_norm": 0.034070531981797145,
"learning_rate": 3.142335799989883e-05,
"loss": 0.7026,
"step": 2535
},
{
"epoch": 0.8128,
"grad_norm": 0.03656747081077928,
"learning_rate": 3.091203629902832e-05,
"loss": 0.6936,
"step": 2540
},
{
"epoch": 0.8144,
"grad_norm": 0.03591367118250926,
"learning_rate": 3.0404430599492636e-05,
"loss": 0.6964,
"step": 2545
},
{
"epoch": 0.816,
"grad_norm": 0.03224723089209564,
"learning_rate": 2.9900556740537026e-05,
"loss": 0.6967,
"step": 2550
},
{
"epoch": 0.8176,
"grad_norm": 0.03592379714034871,
"learning_rate": 2.9400430444958932e-05,
"loss": 0.6837,
"step": 2555
},
{
"epoch": 0.8192,
"grad_norm": 0.03564531082517651,
"learning_rate": 2.8904067318617453e-05,
"loss": 0.6863,
"step": 2560
},
{
"epoch": 0.8208,
"grad_norm": 0.036226568505578746,
"learning_rate": 2.841148284994635e-05,
"loss": 0.6803,
"step": 2565
},
{
"epoch": 0.8224,
"grad_norm": 0.036427168710013086,
"learning_rate": 2.792269240947076e-05,
"loss": 0.6884,
"step": 2570
},
{
"epoch": 0.824,
"grad_norm": 0.03873968606981723,
"learning_rate": 2.7437711249327655e-05,
"loss": 0.6861,
"step": 2575
},
{
"epoch": 0.8256,
"grad_norm": 0.040206628177924045,
"learning_rate": 2.6956554502789867e-05,
"loss": 0.678,
"step": 2580
},
{
"epoch": 0.8272,
"grad_norm": 0.035584398042952534,
"learning_rate": 2.6479237183793712e-05,
"loss": 0.693,
"step": 2585
},
{
"epoch": 0.8288,
"grad_norm": 0.03504043384204677,
"learning_rate": 2.6005774186470746e-05,
"loss": 0.6899,
"step": 2590
},
{
"epoch": 0.8304,
"grad_norm": 0.03585999832889161,
"learning_rate": 2.5536180284682857e-05,
"loss": 0.6822,
"step": 2595
},
{
"epoch": 0.832,
"grad_norm": 0.03582872557001751,
"learning_rate": 2.507047013156133e-05,
"loss": 0.7006,
"step": 2600
},
{
"epoch": 0.8336,
"grad_norm": 0.03521207047344509,
"learning_rate": 2.4608658259049542e-05,
"loss": 0.6968,
"step": 2605
},
{
"epoch": 0.8352,
"grad_norm": 0.03638818276349012,
"learning_rate": 2.4150759077449698e-05,
"loss": 0.6858,
"step": 2610
},
{
"epoch": 0.8368,
"grad_norm": 0.03448652695031895,
"learning_rate": 2.369678687497289e-05,
"loss": 0.6923,
"step": 2615
},
{
"epoch": 0.8384,
"grad_norm": 0.03585778645343454,
"learning_rate": 2.3246755817293444e-05,
"loss": 0.6921,
"step": 2620
},
{
"epoch": 0.84,
"grad_norm": 0.035878782315928406,
"learning_rate": 2.2800679947106853e-05,
"loss": 0.6868,
"step": 2625
},
{
"epoch": 0.8416,
"grad_norm": 0.03413705214140856,
"learning_rate": 2.2358573183691565e-05,
"loss": 0.6745,
"step": 2630
},
{
"epoch": 0.8432,
"grad_norm": 0.03741900054647843,
"learning_rate": 2.1920449322474742e-05,
"loss": 0.6944,
"step": 2635
},
{
"epoch": 0.8448,
"grad_norm": 0.033945500957096725,
"learning_rate": 2.1486322034601657e-05,
"loss": 0.6764,
"step": 2640
},
{
"epoch": 0.8464,
"grad_norm": 0.03693693672153618,
"learning_rate": 2.105620486650914e-05,
"loss": 0.6861,
"step": 2645
},
{
"epoch": 0.848,
"grad_norm": 0.035959214488486574,
"learning_rate": 2.063011123950295e-05,
"loss": 0.6907,
"step": 2650
},
{
"epoch": 0.8496,
"grad_norm": 0.03606750368452151,
"learning_rate": 2.0208054449338878e-05,
"loss": 0.6757,
"step": 2655
},
{
"epoch": 0.8512,
"grad_norm": 0.039459520303168856,
"learning_rate": 1.979004766580797e-05,
"loss": 0.6802,
"step": 2660
},
{
"epoch": 0.8528,
"grad_norm": 0.035442956185441674,
"learning_rate": 1.9376103932325547e-05,
"loss": 0.6933,
"step": 2665
},
{
"epoch": 0.8544,
"grad_norm": 0.035197279761445745,
"learning_rate": 1.8966236165524102e-05,
"loss": 0.6948,
"step": 2670
},
{
"epoch": 0.856,
"grad_norm": 0.03580007213986297,
"learning_rate": 1.8560457154850395e-05,
"loss": 0.682,
"step": 2675
},
{
"epoch": 0.8576,
"grad_norm": 0.0339719466539305,
"learning_rate": 1.8158779562166252e-05,
"loss": 0.6871,
"step": 2680
},
{
"epoch": 0.8592,
"grad_norm": 0.0342844166181569,
"learning_rate": 1.7761215921353576e-05,
"loss": 0.6871,
"step": 2685
},
{
"epoch": 0.8608,
"grad_norm": 0.03654992692878969,
"learning_rate": 1.7367778637923135e-05,
"loss": 0.6735,
"step": 2690
},
{
"epoch": 0.8624,
"grad_norm": 0.03205243106560114,
"learning_rate": 1.6978479988627592e-05,
"loss": 0.6778,
"step": 2695
},
{
"epoch": 0.864,
"grad_norm": 0.0356464042393281,
"learning_rate": 1.6593332121078295e-05,
"loss": 0.6893,
"step": 2700
},
{
"epoch": 0.8656,
"grad_norm": 0.03475488641190361,
"learning_rate": 1.6212347053366287e-05,
"loss": 0.6969,
"step": 2705
},
{
"epoch": 0.8672,
"grad_norm": 0.03571595410829047,
"learning_rate": 1.5835536673687278e-05,
"loss": 0.6561,
"step": 2710
},
{
"epoch": 0.8688,
"grad_norm": 0.037110415954442255,
"learning_rate": 1.5462912739970663e-05,
"loss": 0.6739,
"step": 2715
},
{
"epoch": 0.8704,
"grad_norm": 0.03705588213171844,
"learning_rate": 1.5094486879512786e-05,
"loss": 0.6724,
"step": 2720
},
{
"epoch": 0.872,
"grad_norm": 0.0359981143961099,
"learning_rate": 1.4730270588613874e-05,
"loss": 0.6899,
"step": 2725
},
{
"epoch": 0.8736,
"grad_norm": 0.03586953931395446,
"learning_rate": 1.437027523221948e-05,
"loss": 0.68,
"step": 2730
},
{
"epoch": 0.8752,
"grad_norm": 0.03601668331584326,
"learning_rate": 1.4014512043565812e-05,
"loss": 0.6782,
"step": 2735
},
{
"epoch": 0.8768,
"grad_norm": 0.03533498098436524,
"learning_rate": 1.3662992123829164e-05,
"loss": 0.7159,
"step": 2740
},
{
"epoch": 0.8784,
"grad_norm": 0.037640124507066094,
"learning_rate": 1.3315726441779629e-05,
"loss": 0.6846,
"step": 2745
},
{
"epoch": 0.88,
"grad_norm": 0.03482454818070702,
"learning_rate": 1.2972725833438752e-05,
"loss": 0.6819,
"step": 2750
},
{
"epoch": 0.8816,
"grad_norm": 0.0367271221138329,
"learning_rate": 1.2634001001741373e-05,
"loss": 0.6755,
"step": 2755
},
{
"epoch": 0.8832,
"grad_norm": 0.03841584868617722,
"learning_rate": 1.2299562516201739e-05,
"loss": 0.6935,
"step": 2760
},
{
"epoch": 0.8848,
"grad_norm": 0.040854540776410125,
"learning_rate": 1.1969420812583603e-05,
"loss": 0.7012,
"step": 2765
},
{
"epoch": 0.8864,
"grad_norm": 0.0341221051853296,
"learning_rate": 1.1643586192574684e-05,
"loss": 0.6706,
"step": 2770
},
{
"epoch": 0.888,
"grad_norm": 0.03757875689795596,
"learning_rate": 1.1322068823465125e-05,
"loss": 0.6771,
"step": 2775
},
{
"epoch": 0.8896,
"grad_norm": 0.034492555749470376,
"learning_rate": 1.1004878737830363e-05,
"loss": 0.6678,
"step": 2780
},
{
"epoch": 0.8912,
"grad_norm": 0.03331925332418725,
"learning_rate": 1.06920258332179e-05,
"loss": 0.6849,
"step": 2785
},
{
"epoch": 0.8928,
"grad_norm": 0.03367504222803977,
"learning_rate": 1.0383519871838587e-05,
"loss": 0.671,
"step": 2790
},
{
"epoch": 0.8944,
"grad_norm": 0.03373539399249363,
"learning_rate": 1.0079370480261984e-05,
"loss": 0.6742,
"step": 2795
},
{
"epoch": 0.896,
"grad_norm": 0.03247618752201015,
"learning_rate": 9.779587149115891e-06,
"loss": 0.6856,
"step": 2800
},
{
"epoch": 0.8976,
"grad_norm": 0.03584180038198852,
"learning_rate": 9.484179232790389e-06,
"loss": 0.6664,
"step": 2805
},
{
"epoch": 0.8992,
"grad_norm": 0.036302861284426946,
"learning_rate": 9.193155949145736e-06,
"loss": 0.672,
"step": 2810
},
{
"epoch": 0.9008,
"grad_norm": 0.03503030446991528,
"learning_rate": 8.906526379224865e-06,
"loss": 0.7019,
"step": 2815
},
{
"epoch": 0.9024,
"grad_norm": 0.03560018408938535,
"learning_rate": 8.624299466969964e-06,
"loss": 0.6819,
"step": 2820
},
{
"epoch": 0.904,
"grad_norm": 0.03286452854193144,
"learning_rate": 8.346484018943428e-06,
"loss": 0.6785,
"step": 2825
},
{
"epoch": 0.9056,
"grad_norm": 0.03712658893214972,
"learning_rate": 8.073088704053082e-06,
"loss": 0.6868,
"step": 2830
},
{
"epoch": 0.9072,
"grad_norm": 0.035362561023794904,
"learning_rate": 7.80412205328153e-06,
"loss": 0.6847,
"step": 2835
},
{
"epoch": 0.9088,
"grad_norm": 0.03413740311243108,
"learning_rate": 7.539592459420219e-06,
"loss": 0.6694,
"step": 2840
},
{
"epoch": 0.9104,
"grad_norm": 0.03460380103631612,
"learning_rate": 7.279508176807258e-06,
"loss": 0.6814,
"step": 2845
},
{
"epoch": 0.912,
"grad_norm": 0.03345889140853353,
"learning_rate": 7.023877321070071e-06,
"loss": 0.6826,
"step": 2850
},
{
"epoch": 0.9136,
"grad_norm": 0.03658572315985044,
"learning_rate": 6.7727078688720515e-06,
"loss": 0.6942,
"step": 2855
},
{
"epoch": 0.9152,
"grad_norm": 0.033525356039550876,
"learning_rate": 6.526007657663656e-06,
"loss": 0.6743,
"step": 2860
},
{
"epoch": 0.9168,
"grad_norm": 0.03569978757002032,
"learning_rate": 6.283784385437962e-06,
"loss": 0.6824,
"step": 2865
},
{
"epoch": 0.9184,
"grad_norm": 0.034788249129439716,
"learning_rate": 6.046045610490324e-06,
"loss": 0.6843,
"step": 2870
},
{
"epoch": 0.92,
"grad_norm": 0.03856999362984195,
"learning_rate": 5.8127987511825516e-06,
"loss": 0.6919,
"step": 2875
},
{
"epoch": 0.9216,
"grad_norm": 0.03363147811222781,
"learning_rate": 5.584051085711488e-06,
"loss": 0.6676,
"step": 2880
},
{
"epoch": 0.9232,
"grad_norm": 0.03337401697063256,
"learning_rate": 5.359809751881833e-06,
"loss": 0.668,
"step": 2885
},
{
"epoch": 0.9248,
"grad_norm": 0.03391950164002368,
"learning_rate": 5.140081746883546e-06,
"loss": 0.6709,
"step": 2890
},
{
"epoch": 0.9264,
"grad_norm": 0.03628260571840196,
"learning_rate": 4.924873927073292e-06,
"loss": 0.6759,
"step": 2895
},
{
"epoch": 0.928,
"grad_norm": 0.03719727694206389,
"learning_rate": 4.714193007760681e-06,
"loss": 0.6841,
"step": 2900
},
{
"epoch": 0.9296,
"grad_norm": 0.035072599886154625,
"learning_rate": 4.5080455629986465e-06,
"loss": 0.687,
"step": 2905
},
{
"epoch": 0.9312,
"grad_norm": 0.03402712931516279,
"learning_rate": 4.306438025378317e-06,
"loss": 0.6887,
"step": 2910
},
{
"epoch": 0.9328,
"grad_norm": 0.033629393596595857,
"learning_rate": 4.109376685828336e-06,
"loss": 0.6938,
"step": 2915
},
{
"epoch": 0.9344,
"grad_norm": 0.03430870214738047,
"learning_rate": 3.916867693418474e-06,
"loss": 0.6779,
"step": 2920
},
{
"epoch": 0.936,
"grad_norm": 0.03428926816458531,
"learning_rate": 3.7289170551678626e-06,
"loss": 0.683,
"step": 2925
},
{
"epoch": 0.9376,
"grad_norm": 0.03636527753521949,
"learning_rate": 3.545530635857463e-06,
"loss": 0.6849,
"step": 2930
},
{
"epoch": 0.9392,
"grad_norm": 0.037074968354287835,
"learning_rate": 3.366714157847078e-06,
"loss": 0.6787,
"step": 2935
},
{
"epoch": 0.9408,
"grad_norm": 0.03690326258068921,
"learning_rate": 3.192473200896828e-06,
"loss": 0.6912,
"step": 2940
},
{
"epoch": 0.9424,
"grad_norm": 0.03676314823395921,
"learning_rate": 3.0228132019930407e-06,
"loss": 0.6982,
"step": 2945
},
{
"epoch": 0.944,
"grad_norm": 0.03326096113948387,
"learning_rate": 2.8577394551785705e-06,
"loss": 0.6808,
"step": 2950
},
{
"epoch": 0.9456,
"grad_norm": 0.03315391474215544,
"learning_rate": 2.6972571113875798e-06,
"loss": 0.6828,
"step": 2955
},
{
"epoch": 0.9472,
"grad_norm": 0.034162229284439904,
"learning_rate": 2.5413711782848845e-06,
"loss": 0.6769,
"step": 2960
},
{
"epoch": 0.9488,
"grad_norm": 0.03557968800367496,
"learning_rate": 2.39008652010963e-06,
"loss": 0.684,
"step": 2965
},
{
"epoch": 0.9504,
"grad_norm": 0.037053943949424364,
"learning_rate": 2.243407857523544e-06,
"loss": 0.7004,
"step": 2970
},
{
"epoch": 0.952,
"grad_norm": 0.03521116100655522,
"learning_rate": 2.1013397674636057e-06,
"loss": 0.6724,
"step": 2975
},
{
"epoch": 0.9536,
"grad_norm": 0.037461763983045196,
"learning_rate": 1.9638866829992606e-06,
"loss": 0.6813,
"step": 2980
},
{
"epoch": 0.9552,
"grad_norm": 0.0337899195203273,
"learning_rate": 1.8310528931940627e-06,
"loss": 0.6876,
"step": 2985
},
{
"epoch": 0.9568,
"grad_norm": 0.0333450608692831,
"learning_rate": 1.702842542971866e-06,
"loss": 0.679,
"step": 2990
},
{
"epoch": 0.9584,
"grad_norm": 0.03259070901476999,
"learning_rate": 1.5792596329874451e-06,
"loss": 0.6946,
"step": 2995
},
{
"epoch": 0.96,
"grad_norm": 0.0355419164421536,
"learning_rate": 1.4603080195017103e-06,
"loss": 0.687,
"step": 3000
},
{
"epoch": 0.9616,
"grad_norm": 0.03486772045983402,
"learning_rate": 1.3459914142613381e-06,
"loss": 0.6793,
"step": 3005
},
{
"epoch": 0.9632,
"grad_norm": 0.03758596013480105,
"learning_rate": 1.2363133843829976e-06,
"loss": 0.6928,
"step": 3010
},
{
"epoch": 0.9648,
"grad_norm": 0.03419367915456267,
"learning_rate": 1.1312773522419716e-06,
"loss": 0.6823,
"step": 3015
},
{
"epoch": 0.9664,
"grad_norm": 0.035663872743848725,
"learning_rate": 1.0308865953654267e-06,
"loss": 0.6989,
"step": 3020
},
{
"epoch": 0.968,
"grad_norm": 0.03616981749097929,
"learning_rate": 9.351442463301273e-07,
"loss": 0.6731,
"step": 3025
},
{
"epoch": 0.9696,
"grad_norm": 0.03762974647472138,
"learning_rate": 8.440532926646315e-07,
"loss": 0.6924,
"step": 3030
},
{
"epoch": 0.9712,
"grad_norm": 0.03399772759360819,
"learning_rate": 7.576165767561815e-07,
"loss": 0.6958,
"step": 3035
},
{
"epoch": 0.9728,
"grad_norm": 0.03359200040600764,
"learning_rate": 6.758367957619082e-07,
"loss": 0.6755,
"step": 3040
},
{
"epoch": 0.9744,
"grad_norm": 0.03408737827328263,
"learning_rate": 5.98716501524732e-07,
"loss": 0.6727,
"step": 3045
},
{
"epoch": 0.976,
"grad_norm": 0.033986659157017304,
"learning_rate": 5.262581004936927e-07,
"loss": 0.6825,
"step": 3050
},
{
"epoch": 0.9776,
"grad_norm": 0.036097505040925654,
"learning_rate": 4.5846385364886053e-07,
"loss": 0.6807,
"step": 3055
},
{
"epoch": 0.9792,
"grad_norm": 0.03186557437798362,
"learning_rate": 3.953358764308412e-07,
"loss": 0.6746,
"step": 3060
},
{
"epoch": 0.9808,
"grad_norm": 0.033878030478239944,
"learning_rate": 3.3687613867469653e-07,
"loss": 0.6913,
"step": 3065
},
{
"epoch": 0.9824,
"grad_norm": 0.03510026870536183,
"learning_rate": 2.830864645485265e-07,
"loss": 0.6869,
"step": 3070
},
{
"epoch": 0.984,
"grad_norm": 0.03417621913984012,
"learning_rate": 2.3396853249653124e-07,
"loss": 0.688,
"step": 3075
},
{
"epoch": 0.9856,
"grad_norm": 0.03312402040298033,
"learning_rate": 1.895238751866035e-07,
"loss": 0.6843,
"step": 3080
},
{
"epoch": 0.9872,
"grad_norm": 0.03360622380173029,
"learning_rate": 1.4975387946256635e-07,
"loss": 0.6783,
"step": 3085
},
{
"epoch": 0.9888,
"grad_norm": 0.0354700691322541,
"learning_rate": 1.1465978630085803e-07,
"loss": 0.6868,
"step": 3090
},
{
"epoch": 0.9904,
"grad_norm": 0.03295568524608156,
"learning_rate": 8.424269077182965e-08,
"loss": 0.6738,
"step": 3095
},
{
"epoch": 0.992,
"grad_norm": 0.032788816463157545,
"learning_rate": 5.8503542005555734e-08,
"loss": 0.6812,
"step": 3100
},
{
"epoch": 0.9936,
"grad_norm": 0.035024900613057144,
"learning_rate": 3.7443143162207976e-08,
"loss": 0.704,
"step": 3105
},
{
"epoch": 0.9952,
"grad_norm": 0.036688178061528604,
"learning_rate": 2.1062151407025274e-08,
"loss": 0.6963,
"step": 3110
},
{
"epoch": 0.9968,
"grad_norm": 0.03618314272574878,
"learning_rate": 9.36107788981344e-09,
"loss": 0.689,
"step": 3115
},
{
"epoch": 0.9984,
"grad_norm": 0.03425916181867982,
"learning_rate": 2.340287728941348e-09,
"loss": 0.6649,
"step": 3120
},
{
"epoch": 1.0,
"grad_norm": 0.03464178557265085,
"learning_rate": 0.0,
"loss": 0.6967,
"step": 3125
},
{
"epoch": 1.0,
"eval_loss": 0.6901005506515503,
"eval_runtime": 259.4269,
"eval_samples_per_second": 119.849,
"eval_steps_per_second": 3.747,
"step": 3125
},
{
"epoch": 1.0,
"step": 3125,
"total_flos": 514082196226048.0,
"train_loss": 0.7583255160522461,
"train_runtime": 13626.6135,
"train_samples_per_second": 29.354,
"train_steps_per_second": 0.229
}
],
"logging_steps": 5,
"max_steps": 3125,
"num_input_tokens_seen": 0,
"num_train_epochs": 1,
"save_steps": 500,
"stateful_callbacks": {
"TrainerControl": {
"args": {
"should_epoch_stop": false,
"should_evaluate": false,
"should_log": false,
"should_save": false,
"should_training_stop": false
},
"attributes": {}
}
},
"total_flos": 514082196226048.0,
"train_batch_size": 4,
"trial_name": null,
"trial_params": null
}