Files
oh-dcft-v3.1-llama-3.1-nemo…/trainer_state.json

4743 lines
104 KiB
JSON
Raw Permalink Normal View History

{
"best_metric": null,
"best_model_checkpoint": null,
"epoch": 2.9993269015032533,
"eval_steps": 500,
"global_step": 6684,
"is_hyper_param_search": false,
"is_local_process_zero": true,
"is_world_process_zero": true,
"log_history": [
{
"epoch": 0.004487323311644604,
"grad_norm": 1.3059244169345843,
"learning_rate": 5e-06,
"loss": 0.7774,
"step": 10
},
{
"epoch": 0.008974646623289208,
"grad_norm": 1.0232642607902347,
"learning_rate": 5e-06,
"loss": 0.6958,
"step": 20
},
{
"epoch": 0.013461969934933812,
"grad_norm": 0.8401058983048771,
"learning_rate": 5e-06,
"loss": 0.6992,
"step": 30
},
{
"epoch": 0.017949293246578416,
"grad_norm": 0.8139785554230714,
"learning_rate": 5e-06,
"loss": 0.6749,
"step": 40
},
{
"epoch": 0.02243661655822302,
"grad_norm": 0.8129256440286593,
"learning_rate": 5e-06,
"loss": 0.7083,
"step": 50
},
{
"epoch": 0.026923939869867623,
"grad_norm": 0.8418551736226986,
"learning_rate": 5e-06,
"loss": 0.6846,
"step": 60
},
{
"epoch": 0.03141126318151223,
"grad_norm": 0.7167670779403538,
"learning_rate": 5e-06,
"loss": 0.6574,
"step": 70
},
{
"epoch": 0.03589858649315683,
"grad_norm": 0.5953427667880735,
"learning_rate": 5e-06,
"loss": 0.653,
"step": 80
},
{
"epoch": 0.04038590980480144,
"grad_norm": 0.5652705295757201,
"learning_rate": 5e-06,
"loss": 0.6452,
"step": 90
},
{
"epoch": 0.04487323311644604,
"grad_norm": 0.5863707257105095,
"learning_rate": 5e-06,
"loss": 0.6424,
"step": 100
},
{
"epoch": 0.049360556428090646,
"grad_norm": 0.5827496122357744,
"learning_rate": 5e-06,
"loss": 0.6325,
"step": 110
},
{
"epoch": 0.05384787973973525,
"grad_norm": 0.5692209350328149,
"learning_rate": 5e-06,
"loss": 0.6389,
"step": 120
},
{
"epoch": 0.058335203051379854,
"grad_norm": 0.5801005846647308,
"learning_rate": 5e-06,
"loss": 0.6394,
"step": 130
},
{
"epoch": 0.06282252636302446,
"grad_norm": 0.6162942143257907,
"learning_rate": 5e-06,
"loss": 0.6487,
"step": 140
},
{
"epoch": 0.06730984967466906,
"grad_norm": 0.5910482239279874,
"learning_rate": 5e-06,
"loss": 0.6535,
"step": 150
},
{
"epoch": 0.07179717298631366,
"grad_norm": 0.5628620440134962,
"learning_rate": 5e-06,
"loss": 0.6613,
"step": 160
},
{
"epoch": 0.07628449629795826,
"grad_norm": 0.5853996899496572,
"learning_rate": 5e-06,
"loss": 0.6357,
"step": 170
},
{
"epoch": 0.08077181960960288,
"grad_norm": 0.5878540945448613,
"learning_rate": 5e-06,
"loss": 0.636,
"step": 180
},
{
"epoch": 0.08525914292124748,
"grad_norm": 0.5862646504872097,
"learning_rate": 5e-06,
"loss": 0.6312,
"step": 190
},
{
"epoch": 0.08974646623289208,
"grad_norm": 0.5673857464664466,
"learning_rate": 5e-06,
"loss": 0.6332,
"step": 200
},
{
"epoch": 0.09423378954453668,
"grad_norm": 0.6082706115579384,
"learning_rate": 5e-06,
"loss": 0.6092,
"step": 210
},
{
"epoch": 0.09872111285618129,
"grad_norm": 0.6196844470280329,
"learning_rate": 5e-06,
"loss": 0.6421,
"step": 220
},
{
"epoch": 0.10320843616782589,
"grad_norm": 0.5579066730260925,
"learning_rate": 5e-06,
"loss": 0.6281,
"step": 230
},
{
"epoch": 0.1076957594794705,
"grad_norm": 0.565529818208485,
"learning_rate": 5e-06,
"loss": 0.6172,
"step": 240
},
{
"epoch": 0.1121830827911151,
"grad_norm": 0.6071432928320446,
"learning_rate": 5e-06,
"loss": 0.6016,
"step": 250
},
{
"epoch": 0.11667040610275971,
"grad_norm": 0.6073563540142406,
"learning_rate": 5e-06,
"loss": 0.6126,
"step": 260
},
{
"epoch": 0.12115772941440431,
"grad_norm": 0.569801618469301,
"learning_rate": 5e-06,
"loss": 0.6251,
"step": 270
},
{
"epoch": 0.12564505272604892,
"grad_norm": 0.591367661084153,
"learning_rate": 5e-06,
"loss": 0.625,
"step": 280
},
{
"epoch": 0.1301323760376935,
"grad_norm": 0.5757700659124422,
"learning_rate": 5e-06,
"loss": 0.619,
"step": 290
},
{
"epoch": 0.13461969934933812,
"grad_norm": 0.564239118303286,
"learning_rate": 5e-06,
"loss": 0.6073,
"step": 300
},
{
"epoch": 0.13910702266098274,
"grad_norm": 0.5885189461499749,
"learning_rate": 5e-06,
"loss": 0.6269,
"step": 310
},
{
"epoch": 0.14359434597262732,
"grad_norm": 0.5563218966218223,
"learning_rate": 5e-06,
"loss": 0.6294,
"step": 320
},
{
"epoch": 0.14808166928427194,
"grad_norm": 0.6172338154521191,
"learning_rate": 5e-06,
"loss": 0.6175,
"step": 330
},
{
"epoch": 0.15256899259591652,
"grad_norm": 0.5746423605096699,
"learning_rate": 5e-06,
"loss": 0.6,
"step": 340
},
{
"epoch": 0.15705631590756114,
"grad_norm": 0.5780025404911214,
"learning_rate": 5e-06,
"loss": 0.606,
"step": 350
},
{
"epoch": 0.16154363921920575,
"grad_norm": 0.5700015689377997,
"learning_rate": 5e-06,
"loss": 0.5955,
"step": 360
},
{
"epoch": 0.16603096253085034,
"grad_norm": 0.5807610554503659,
"learning_rate": 5e-06,
"loss": 0.5993,
"step": 370
},
{
"epoch": 0.17051828584249495,
"grad_norm": 0.5588205610118117,
"learning_rate": 5e-06,
"loss": 0.5941,
"step": 380
},
{
"epoch": 0.17500560915413957,
"grad_norm": 0.571147485187563,
"learning_rate": 5e-06,
"loss": 0.6086,
"step": 390
},
{
"epoch": 0.17949293246578416,
"grad_norm": 0.5874218028496501,
"learning_rate": 5e-06,
"loss": 0.6298,
"step": 400
},
{
"epoch": 0.18398025577742877,
"grad_norm": 0.553081232586883,
"learning_rate": 5e-06,
"loss": 0.6128,
"step": 410
},
{
"epoch": 0.18846757908907336,
"grad_norm": 0.5922159817168337,
"learning_rate": 5e-06,
"loss": 0.6199,
"step": 420
},
{
"epoch": 0.19295490240071797,
"grad_norm": 0.5941689160082125,
"learning_rate": 5e-06,
"loss": 0.6123,
"step": 430
},
{
"epoch": 0.19744222571236258,
"grad_norm": 0.5837547863994649,
"learning_rate": 5e-06,
"loss": 0.632,
"step": 440
},
{
"epoch": 0.20192954902400717,
"grad_norm": 0.5493502840784451,
"learning_rate": 5e-06,
"loss": 0.5918,
"step": 450
},
{
"epoch": 0.20641687233565179,
"grad_norm": 0.5843347451961302,
"learning_rate": 5e-06,
"loss": 0.6151,
"step": 460
},
{
"epoch": 0.2109041956472964,
"grad_norm": 0.5417935808080204,
"learning_rate": 5e-06,
"loss": 0.6059,
"step": 470
},
{
"epoch": 0.215391518958941,
"grad_norm": 0.5698753919399157,
"learning_rate": 5e-06,
"loss": 0.6135,
"step": 480
},
{
"epoch": 0.2198788422705856,
"grad_norm": 0.5244740288157084,
"learning_rate": 5e-06,
"loss": 0.5821,
"step": 490
},
{
"epoch": 0.2243661655822302,
"grad_norm": 0.6029129072997347,
"learning_rate": 5e-06,
"loss": 0.6116,
"step": 500
},
{
"epoch": 0.2288534888938748,
"grad_norm": 0.543722899676561,
"learning_rate": 5e-06,
"loss": 0.5907,
"step": 510
},
{
"epoch": 0.23334081220551942,
"grad_norm": 0.5619984349965198,
"learning_rate": 5e-06,
"loss": 0.6038,
"step": 520
},
{
"epoch": 0.237828135517164,
"grad_norm": 0.5859927720447103,
"learning_rate": 5e-06,
"loss": 0.6244,
"step": 530
},
{
"epoch": 0.24231545882880862,
"grad_norm": 0.5583414821346395,
"learning_rate": 5e-06,
"loss": 0.5981,
"step": 540
},
{
"epoch": 0.24680278214045323,
"grad_norm": 0.5499638333486776,
"learning_rate": 5e-06,
"loss": 0.5908,
"step": 550
},
{
"epoch": 0.25129010545209784,
"grad_norm": 0.5826333088813789,
"learning_rate": 5e-06,
"loss": 0.5683,
"step": 560
},
{
"epoch": 0.25577742876374243,
"grad_norm": 0.5492837199828549,
"learning_rate": 5e-06,
"loss": 0.622,
"step": 570
},
{
"epoch": 0.260264752075387,
"grad_norm": 0.5829754943916923,
"learning_rate": 5e-06,
"loss": 0.5964,
"step": 580
},
{
"epoch": 0.26475207538703166,
"grad_norm": 0.5639030175066925,
"learning_rate": 5e-06,
"loss": 0.5894,
"step": 590
},
{
"epoch": 0.26923939869867625,
"grad_norm": 0.5840101964805389,
"learning_rate": 5e-06,
"loss": 0.5879,
"step": 600
},
{
"epoch": 0.27372672201032083,
"grad_norm": 0.5654294930486207,
"learning_rate": 5e-06,
"loss": 0.5821,
"step": 610
},
{
"epoch": 0.2782140453219655,
"grad_norm": 0.6015547946403642,
"learning_rate": 5e-06,
"loss": 0.5988,
"step": 620
},
{
"epoch": 0.28270136863361006,
"grad_norm": 0.5475239684469887,
"learning_rate": 5e-06,
"loss": 0.5984,
"step": 630
},
{
"epoch": 0.28718869194525465,
"grad_norm": 0.5984753640582714,
"learning_rate": 5e-06,
"loss": 0.6093,
"step": 640
},
{
"epoch": 0.29167601525689923,
"grad_norm": 0.5743058756802668,
"learning_rate": 5e-06,
"loss": 0.6061,
"step": 650
},
{
"epoch": 0.2961633385685439,
"grad_norm": 0.5643830386334935,
"learning_rate": 5e-06,
"loss": 0.6119,
"step": 660
},
{
"epoch": 0.30065066188018846,
"grad_norm": 0.6133686427622306,
"learning_rate": 5e-06,
"loss": 0.5866,
"step": 670
},
{
"epoch": 0.30513798519183305,
"grad_norm": 0.5891262468962493,
"learning_rate": 5e-06,
"loss": 0.5856,
"step": 680
},
{
"epoch": 0.3096253085034777,
"grad_norm": 0.5667738482065338,
"learning_rate": 5e-06,
"loss": 0.6159,
"step": 690
},
{
"epoch": 0.3141126318151223,
"grad_norm": 0.5659614870185288,
"learning_rate": 5e-06,
"loss": 0.6026,
"step": 700
},
{
"epoch": 0.31859995512676687,
"grad_norm": 0.5708631949188694,
"learning_rate": 5e-06,
"loss": 0.5728,
"step": 710
},
{
"epoch": 0.3230872784384115,
"grad_norm": 0.582979417517679,
"learning_rate": 5e-06,
"loss": 0.6193,
"step": 720
},
{
"epoch": 0.3275746017500561,
"grad_norm": 0.5945610104189938,
"learning_rate": 5e-06,
"loss": 0.5901,
"step": 730
},
{
"epoch": 0.3320619250617007,
"grad_norm": 0.5716342616547154,
"learning_rate": 5e-06,
"loss": 0.62,
"step": 740
},
{
"epoch": 0.3365492483733453,
"grad_norm": 0.5733402898671368,
"learning_rate": 5e-06,
"loss": 0.5918,
"step": 750
},
{
"epoch": 0.3410365716849899,
"grad_norm": 0.572434524506764,
"learning_rate": 5e-06,
"loss": 0.584,
"step": 760
},
{
"epoch": 0.3455238949966345,
"grad_norm": 0.5821520713178042,
"learning_rate": 5e-06,
"loss": 0.5826,
"step": 770
},
{
"epoch": 0.35001121830827914,
"grad_norm": 0.5502868561175508,
"learning_rate": 5e-06,
"loss": 0.6246,
"step": 780
},
{
"epoch": 0.3544985416199237,
"grad_norm": 0.5978195056323264,
"learning_rate": 5e-06,
"loss": 0.6169,
"step": 790
},
{
"epoch": 0.3589858649315683,
"grad_norm": 0.5838319332934967,
"learning_rate": 5e-06,
"loss": 0.5911,
"step": 800
},
{
"epoch": 0.3634731882432129,
"grad_norm": 0.572267142733801,
"learning_rate": 5e-06,
"loss": 0.599,
"step": 810
},
{
"epoch": 0.36796051155485754,
"grad_norm": 0.582203599292086,
"learning_rate": 5e-06,
"loss": 0.6276,
"step": 820
},
{
"epoch": 0.3724478348665021,
"grad_norm": 0.5828348677910858,
"learning_rate": 5e-06,
"loss": 0.5758,
"step": 830
},
{
"epoch": 0.3769351581781467,
"grad_norm": 0.6384385145884847,
"learning_rate": 5e-06,
"loss": 0.5818,
"step": 840
},
{
"epoch": 0.38142248148979135,
"grad_norm": 0.6063697413207821,
"learning_rate": 5e-06,
"loss": 0.5833,
"step": 850
},
{
"epoch": 0.38590980480143594,
"grad_norm": 0.5829350365562836,
"learning_rate": 5e-06,
"loss": 0.5896,
"step": 860
},
{
"epoch": 0.3903971281130805,
"grad_norm": 0.5533097693747884,
"learning_rate": 5e-06,
"loss": 0.6351,
"step": 870
},
{
"epoch": 0.39488445142472517,
"grad_norm": 0.560685123123169,
"learning_rate": 5e-06,
"loss": 0.6013,
"step": 880
},
{
"epoch": 0.39937177473636976,
"grad_norm": 0.5878919825433736,
"learning_rate": 5e-06,
"loss": 0.5834,
"step": 890
},
{
"epoch": 0.40385909804801434,
"grad_norm": 0.6071601416861148,
"learning_rate": 5e-06,
"loss": 0.5952,
"step": 900
},
{
"epoch": 0.408346421359659,
"grad_norm": 0.5743329576838876,
"learning_rate": 5e-06,
"loss": 0.5707,
"step": 910
},
{
"epoch": 0.41283374467130357,
"grad_norm": 0.5838548938969085,
"learning_rate": 5e-06,
"loss": 0.5963,
"step": 920
},
{
"epoch": 0.41732106798294816,
"grad_norm": 0.6266644917633146,
"learning_rate": 5e-06,
"loss": 0.6116,
"step": 930
},
{
"epoch": 0.4218083912945928,
"grad_norm": 0.5997476072987922,
"learning_rate": 5e-06,
"loss": 0.5949,
"step": 940
},
{
"epoch": 0.4262957146062374,
"grad_norm": 0.5325526105961608,
"learning_rate": 5e-06,
"loss": 0.578,
"step": 950
},
{
"epoch": 0.430783037917882,
"grad_norm": 0.5789065042536049,
"learning_rate": 5e-06,
"loss": 0.5756,
"step": 960
},
{
"epoch": 0.4352703612295266,
"grad_norm": 0.5810941623037826,
"learning_rate": 5e-06,
"loss": 0.5773,
"step": 970
},
{
"epoch": 0.4397576845411712,
"grad_norm": 0.5704375735854125,
"learning_rate": 5e-06,
"loss": 0.6106,
"step": 980
},
{
"epoch": 0.4442450078528158,
"grad_norm": 0.5807401734392655,
"learning_rate": 5e-06,
"loss": 0.5867,
"step": 990
},
{
"epoch": 0.4487323311644604,
"grad_norm": 0.5328866205208039,
"learning_rate": 5e-06,
"loss": 0.5752,
"step": 1000
},
{
"epoch": 0.453219654476105,
"grad_norm": 0.5872115159587232,
"learning_rate": 5e-06,
"loss": 0.6077,
"step": 1010
},
{
"epoch": 0.4577069777877496,
"grad_norm": 0.5467361495539828,
"learning_rate": 5e-06,
"loss": 0.6021,
"step": 1020
},
{
"epoch": 0.4621943010993942,
"grad_norm": 0.5824366492306351,
"learning_rate": 5e-06,
"loss": 0.5895,
"step": 1030
},
{
"epoch": 0.46668162441103883,
"grad_norm": 0.5730302930825869,
"learning_rate": 5e-06,
"loss": 0.5904,
"step": 1040
},
{
"epoch": 0.4711689477226834,
"grad_norm": 0.5741405557180129,
"learning_rate": 5e-06,
"loss": 0.5774,
"step": 1050
},
{
"epoch": 0.475656271034328,
"grad_norm": 0.550592306843404,
"learning_rate": 5e-06,
"loss": 0.5936,
"step": 1060
},
{
"epoch": 0.48014359434597265,
"grad_norm": 0.5833572077757988,
"learning_rate": 5e-06,
"loss": 0.5762,
"step": 1070
},
{
"epoch": 0.48463091765761723,
"grad_norm": 0.5737636703367411,
"learning_rate": 5e-06,
"loss": 0.5923,
"step": 1080
},
{
"epoch": 0.4891182409692618,
"grad_norm": 0.5871383202715521,
"learning_rate": 5e-06,
"loss": 0.5893,
"step": 1090
},
{
"epoch": 0.49360556428090646,
"grad_norm": 0.5749395258974863,
"learning_rate": 5e-06,
"loss": 0.5675,
"step": 1100
},
{
"epoch": 0.49809288759255105,
"grad_norm": 0.5977797308447645,
"learning_rate": 5e-06,
"loss": 0.5702,
"step": 1110
},
{
"epoch": 0.5025802109041957,
"grad_norm": 0.5693088698665462,
"learning_rate": 5e-06,
"loss": 0.5783,
"step": 1120
},
{
"epoch": 0.5070675342158403,
"grad_norm": 0.5666383038783265,
"learning_rate": 5e-06,
"loss": 0.579,
"step": 1130
},
{
"epoch": 0.5115548575274849,
"grad_norm": 0.5685405367124683,
"learning_rate": 5e-06,
"loss": 0.5824,
"step": 1140
},
{
"epoch": 0.5160421808391294,
"grad_norm": 0.6132416315716593,
"learning_rate": 5e-06,
"loss": 0.5656,
"step": 1150
},
{
"epoch": 0.520529504150774,
"grad_norm": 0.5517718231004817,
"learning_rate": 5e-06,
"loss": 0.5874,
"step": 1160
},
{
"epoch": 0.5250168274624186,
"grad_norm": 0.5387150051806513,
"learning_rate": 5e-06,
"loss": 0.599,
"step": 1170
},
{
"epoch": 0.5295041507740633,
"grad_norm": 0.5807000848606523,
"learning_rate": 5e-06,
"loss": 0.5657,
"step": 1180
},
{
"epoch": 0.5339914740857079,
"grad_norm": 0.5963571574776558,
"learning_rate": 5e-06,
"loss": 0.5823,
"step": 1190
},
{
"epoch": 0.5384787973973525,
"grad_norm": 0.5454964206038695,
"learning_rate": 5e-06,
"loss": 0.609,
"step": 1200
},
{
"epoch": 0.5429661207089971,
"grad_norm": 0.5727578636243832,
"learning_rate": 5e-06,
"loss": 0.5757,
"step": 1210
},
{
"epoch": 0.5474534440206417,
"grad_norm": 0.5942434570927121,
"learning_rate": 5e-06,
"loss": 0.5939,
"step": 1220
},
{
"epoch": 0.5519407673322863,
"grad_norm": 0.601970394484814,
"learning_rate": 5e-06,
"loss": 0.5702,
"step": 1230
},
{
"epoch": 0.556428090643931,
"grad_norm": 0.5683956652736455,
"learning_rate": 5e-06,
"loss": 0.5761,
"step": 1240
},
{
"epoch": 0.5609154139555755,
"grad_norm": 0.5484215866941293,
"learning_rate": 5e-06,
"loss": 0.57,
"step": 1250
},
{
"epoch": 0.5654027372672201,
"grad_norm": 0.5566860752701241,
"learning_rate": 5e-06,
"loss": 0.5957,
"step": 1260
},
{
"epoch": 0.5698900605788647,
"grad_norm": 0.6230757081721424,
"learning_rate": 5e-06,
"loss": 0.5913,
"step": 1270
},
{
"epoch": 0.5743773838905093,
"grad_norm": 0.5831220627839732,
"learning_rate": 5e-06,
"loss": 0.5694,
"step": 1280
},
{
"epoch": 0.5788647072021539,
"grad_norm": 0.5836944514712498,
"learning_rate": 5e-06,
"loss": 0.5679,
"step": 1290
},
{
"epoch": 0.5833520305137985,
"grad_norm": 0.5606507135283174,
"learning_rate": 5e-06,
"loss": 0.5687,
"step": 1300
},
{
"epoch": 0.5878393538254432,
"grad_norm": 0.6095163090362549,
"learning_rate": 5e-06,
"loss": 0.6019,
"step": 1310
},
{
"epoch": 0.5923266771370878,
"grad_norm": 0.59010448543007,
"learning_rate": 5e-06,
"loss": 0.5885,
"step": 1320
},
{
"epoch": 0.5968140004487323,
"grad_norm": 0.5880393988574679,
"learning_rate": 5e-06,
"loss": 0.6015,
"step": 1330
},
{
"epoch": 0.6013013237603769,
"grad_norm": 0.5951700057903788,
"learning_rate": 5e-06,
"loss": 0.5778,
"step": 1340
},
{
"epoch": 0.6057886470720215,
"grad_norm": 0.5520813189968636,
"learning_rate": 5e-06,
"loss": 0.5577,
"step": 1350
},
{
"epoch": 0.6102759703836661,
"grad_norm": 0.5706267699210373,
"learning_rate": 5e-06,
"loss": 0.5666,
"step": 1360
},
{
"epoch": 0.6147632936953108,
"grad_norm": 0.5873241252786081,
"learning_rate": 5e-06,
"loss": 0.579,
"step": 1370
},
{
"epoch": 0.6192506170069554,
"grad_norm": 0.5919098224825123,
"learning_rate": 5e-06,
"loss": 0.597,
"step": 1380
},
{
"epoch": 0.6237379403186,
"grad_norm": 0.5593146398046412,
"learning_rate": 5e-06,
"loss": 0.5764,
"step": 1390
},
{
"epoch": 0.6282252636302446,
"grad_norm": 0.5211444666215148,
"learning_rate": 5e-06,
"loss": 0.5834,
"step": 1400
},
{
"epoch": 0.6327125869418891,
"grad_norm": 0.5464287006558793,
"learning_rate": 5e-06,
"loss": 0.565,
"step": 1410
},
{
"epoch": 0.6371999102535337,
"grad_norm": 0.5955258518895813,
"learning_rate": 5e-06,
"loss": 0.5686,
"step": 1420
},
{
"epoch": 0.6416872335651783,
"grad_norm": 0.5950629609101662,
"learning_rate": 5e-06,
"loss": 0.5591,
"step": 1430
},
{
"epoch": 0.646174556876823,
"grad_norm": 0.5764649145742474,
"learning_rate": 5e-06,
"loss": 0.5646,
"step": 1440
},
{
"epoch": 0.6506618801884676,
"grad_norm": 0.5623492091421548,
"learning_rate": 5e-06,
"loss": 0.5753,
"step": 1450
},
{
"epoch": 0.6551492035001122,
"grad_norm": 0.5691792772789492,
"learning_rate": 5e-06,
"loss": 0.6129,
"step": 1460
},
{
"epoch": 0.6596365268117568,
"grad_norm": 0.5729790062068799,
"learning_rate": 5e-06,
"loss": 0.5825,
"step": 1470
},
{
"epoch": 0.6641238501234014,
"grad_norm": 0.5516413746220878,
"learning_rate": 5e-06,
"loss": 0.5832,
"step": 1480
},
{
"epoch": 0.668611173435046,
"grad_norm": 0.5952217543049041,
"learning_rate": 5e-06,
"loss": 0.6061,
"step": 1490
},
{
"epoch": 0.6730984967466906,
"grad_norm": 0.5430982582264119,
"learning_rate": 5e-06,
"loss": 0.579,
"step": 1500
},
{
"epoch": 0.6775858200583352,
"grad_norm": 0.5863720016018602,
"learning_rate": 5e-06,
"loss": 0.5872,
"step": 1510
},
{
"epoch": 0.6820731433699798,
"grad_norm": 0.5613897501209862,
"learning_rate": 5e-06,
"loss": 0.5694,
"step": 1520
},
{
"epoch": 0.6865604666816244,
"grad_norm": 0.5650891221037946,
"learning_rate": 5e-06,
"loss": 0.5591,
"step": 1530
},
{
"epoch": 0.691047789993269,
"grad_norm": 0.5648114046496064,
"learning_rate": 5e-06,
"loss": 0.5543,
"step": 1540
},
{
"epoch": 0.6955351133049136,
"grad_norm": 0.5403902955132622,
"learning_rate": 5e-06,
"loss": 0.5584,
"step": 1550
},
{
"epoch": 0.7000224366165583,
"grad_norm": 0.569377834200537,
"learning_rate": 5e-06,
"loss": 0.5697,
"step": 1560
},
{
"epoch": 0.7045097599282029,
"grad_norm": 0.5567609550962811,
"learning_rate": 5e-06,
"loss": 0.5913,
"step": 1570
},
{
"epoch": 0.7089970832398474,
"grad_norm": 0.5460331170553634,
"learning_rate": 5e-06,
"loss": 0.5852,
"step": 1580
},
{
"epoch": 0.713484406551492,
"grad_norm": 0.5804324166647036,
"learning_rate": 5e-06,
"loss": 0.559,
"step": 1590
},
{
"epoch": 0.7179717298631366,
"grad_norm": 0.5960848198273799,
"learning_rate": 5e-06,
"loss": 0.5591,
"step": 1600
},
{
"epoch": 0.7224590531747812,
"grad_norm": 0.5934788516849224,
"learning_rate": 5e-06,
"loss": 0.5757,
"step": 1610
},
{
"epoch": 0.7269463764864258,
"grad_norm": 0.5634154921972939,
"learning_rate": 5e-06,
"loss": 0.5764,
"step": 1620
},
{
"epoch": 0.7314336997980705,
"grad_norm": 0.5608877054747304,
"learning_rate": 5e-06,
"loss": 0.5726,
"step": 1630
},
{
"epoch": 0.7359210231097151,
"grad_norm": 0.5819040189686953,
"learning_rate": 5e-06,
"loss": 0.5761,
"step": 1640
},
{
"epoch": 0.7404083464213597,
"grad_norm": 0.5497077389456424,
"learning_rate": 5e-06,
"loss": 0.5892,
"step": 1650
},
{
"epoch": 0.7448956697330043,
"grad_norm": 0.5540304283341861,
"learning_rate": 5e-06,
"loss": 0.5717,
"step": 1660
},
{
"epoch": 0.7493829930446488,
"grad_norm": 0.555029973847418,
"learning_rate": 5e-06,
"loss": 0.5817,
"step": 1670
},
{
"epoch": 0.7538703163562934,
"grad_norm": 0.5656773354976847,
"learning_rate": 5e-06,
"loss": 0.5592,
"step": 1680
},
{
"epoch": 0.7583576396679381,
"grad_norm": 0.5614465252120793,
"learning_rate": 5e-06,
"loss": 0.5862,
"step": 1690
},
{
"epoch": 0.7628449629795827,
"grad_norm": 0.5971084841293993,
"learning_rate": 5e-06,
"loss": 0.5477,
"step": 1700
},
{
"epoch": 0.7673322862912273,
"grad_norm": 0.5602547594616473,
"learning_rate": 5e-06,
"loss": 0.5515,
"step": 1710
},
{
"epoch": 0.7718196096028719,
"grad_norm": 0.5761549476698519,
"learning_rate": 5e-06,
"loss": 0.5545,
"step": 1720
},
{
"epoch": 0.7763069329145165,
"grad_norm": 0.5623153462700364,
"learning_rate": 5e-06,
"loss": 0.5638,
"step": 1730
},
{
"epoch": 0.780794256226161,
"grad_norm": 0.5595040683918967,
"learning_rate": 5e-06,
"loss": 0.5706,
"step": 1740
},
{
"epoch": 0.7852815795378058,
"grad_norm": 0.6214567614298656,
"learning_rate": 5e-06,
"loss": 0.5759,
"step": 1750
},
{
"epoch": 0.7897689028494503,
"grad_norm": 0.5790695246671757,
"learning_rate": 5e-06,
"loss": 0.5618,
"step": 1760
},
{
"epoch": 0.7942562261610949,
"grad_norm": 0.5868799878975245,
"learning_rate": 5e-06,
"loss": 0.5794,
"step": 1770
},
{
"epoch": 0.7987435494727395,
"grad_norm": 0.5552402081587721,
"learning_rate": 5e-06,
"loss": 0.5725,
"step": 1780
},
{
"epoch": 0.8032308727843841,
"grad_norm": 0.5465116389934637,
"learning_rate": 5e-06,
"loss": 0.5651,
"step": 1790
},
{
"epoch": 0.8077181960960287,
"grad_norm": 0.5830251282771208,
"learning_rate": 5e-06,
"loss": 0.5575,
"step": 1800
},
{
"epoch": 0.8122055194076733,
"grad_norm": 0.5712510816652253,
"learning_rate": 5e-06,
"loss": 0.5598,
"step": 1810
},
{
"epoch": 0.816692842719318,
"grad_norm": 0.562885410999007,
"learning_rate": 5e-06,
"loss": 0.5598,
"step": 1820
},
{
"epoch": 0.8211801660309626,
"grad_norm": 0.5779523907871871,
"learning_rate": 5e-06,
"loss": 0.5723,
"step": 1830
},
{
"epoch": 0.8256674893426071,
"grad_norm": 0.5189856325650857,
"learning_rate": 5e-06,
"loss": 0.5452,
"step": 1840
},
{
"epoch": 0.8301548126542517,
"grad_norm": 0.5720061007902056,
"learning_rate": 5e-06,
"loss": 0.6093,
"step": 1850
},
{
"epoch": 0.8346421359658963,
"grad_norm": 0.5695943485875125,
"learning_rate": 5e-06,
"loss": 0.5652,
"step": 1860
},
{
"epoch": 0.8391294592775409,
"grad_norm": 0.5655340064627559,
"learning_rate": 5e-06,
"loss": 0.5578,
"step": 1870
},
{
"epoch": 0.8436167825891856,
"grad_norm": 0.6040653302056215,
"learning_rate": 5e-06,
"loss": 0.5913,
"step": 1880
},
{
"epoch": 0.8481041059008302,
"grad_norm": 0.5358932702911336,
"learning_rate": 5e-06,
"loss": 0.5506,
"step": 1890
},
{
"epoch": 0.8525914292124748,
"grad_norm": 0.5623759018768127,
"learning_rate": 5e-06,
"loss": 0.5755,
"step": 1900
},
{
"epoch": 0.8570787525241194,
"grad_norm": 0.5478233896665743,
"learning_rate": 5e-06,
"loss": 0.5705,
"step": 1910
},
{
"epoch": 0.861566075835764,
"grad_norm": 0.5581741180688066,
"learning_rate": 5e-06,
"loss": 0.5744,
"step": 1920
},
{
"epoch": 0.8660533991474085,
"grad_norm": 0.581165938497761,
"learning_rate": 5e-06,
"loss": 0.5551,
"step": 1930
},
{
"epoch": 0.8705407224590532,
"grad_norm": 0.5923735410956669,
"learning_rate": 5e-06,
"loss": 0.5724,
"step": 1940
},
{
"epoch": 0.8750280457706978,
"grad_norm": 0.5450424051736013,
"learning_rate": 5e-06,
"loss": 0.5509,
"step": 1950
},
{
"epoch": 0.8795153690823424,
"grad_norm": 0.6040204710835786,
"learning_rate": 5e-06,
"loss": 0.5712,
"step": 1960
},
{
"epoch": 0.884002692393987,
"grad_norm": 0.5655431066508525,
"learning_rate": 5e-06,
"loss": 0.5841,
"step": 1970
},
{
"epoch": 0.8884900157056316,
"grad_norm": 0.5648606449816002,
"learning_rate": 5e-06,
"loss": 0.5559,
"step": 1980
},
{
"epoch": 0.8929773390172762,
"grad_norm": 0.565947961900647,
"learning_rate": 5e-06,
"loss": 0.5424,
"step": 1990
},
{
"epoch": 0.8974646623289207,
"grad_norm": 0.5574359758809583,
"learning_rate": 5e-06,
"loss": 0.5614,
"step": 2000
},
{
"epoch": 0.9019519856405654,
"grad_norm": 0.5511165034666912,
"learning_rate": 5e-06,
"loss": 0.5756,
"step": 2010
},
{
"epoch": 0.90643930895221,
"grad_norm": 0.5700569987335933,
"learning_rate": 5e-06,
"loss": 0.5938,
"step": 2020
},
{
"epoch": 0.9109266322638546,
"grad_norm": 0.5586037159206011,
"learning_rate": 5e-06,
"loss": 0.5793,
"step": 2030
},
{
"epoch": 0.9154139555754992,
"grad_norm": 0.5621567704840573,
"learning_rate": 5e-06,
"loss": 0.5702,
"step": 2040
},
{
"epoch": 0.9199012788871438,
"grad_norm": 0.5706551688766498,
"learning_rate": 5e-06,
"loss": 0.5443,
"step": 2050
},
{
"epoch": 0.9243886021987884,
"grad_norm": 0.5532451823002996,
"learning_rate": 5e-06,
"loss": 0.5726,
"step": 2060
},
{
"epoch": 0.9288759255104331,
"grad_norm": 0.5485701610143561,
"learning_rate": 5e-06,
"loss": 0.5605,
"step": 2070
},
{
"epoch": 0.9333632488220777,
"grad_norm": 0.6047400972527601,
"learning_rate": 5e-06,
"loss": 0.5799,
"step": 2080
},
{
"epoch": 0.9378505721337222,
"grad_norm": 0.5428858709451123,
"learning_rate": 5e-06,
"loss": 0.5745,
"step": 2090
},
{
"epoch": 0.9423378954453668,
"grad_norm": 0.5524203935747322,
"learning_rate": 5e-06,
"loss": 0.5465,
"step": 2100
},
{
"epoch": 0.9468252187570114,
"grad_norm": 0.5515427513485951,
"learning_rate": 5e-06,
"loss": 0.5526,
"step": 2110
},
{
"epoch": 0.951312542068656,
"grad_norm": 0.5853063415860351,
"learning_rate": 5e-06,
"loss": 0.5737,
"step": 2120
},
{
"epoch": 0.9557998653803007,
"grad_norm": 0.5369290552095258,
"learning_rate": 5e-06,
"loss": 0.5493,
"step": 2130
},
{
"epoch": 0.9602871886919453,
"grad_norm": 0.5520734247548831,
"learning_rate": 5e-06,
"loss": 0.5461,
"step": 2140
},
{
"epoch": 0.9647745120035899,
"grad_norm": 0.5653346233403189,
"learning_rate": 5e-06,
"loss": 0.6057,
"step": 2150
},
{
"epoch": 0.9692618353152345,
"grad_norm": 0.582218445061444,
"learning_rate": 5e-06,
"loss": 0.5459,
"step": 2160
},
{
"epoch": 0.973749158626879,
"grad_norm": 0.5787212589606142,
"learning_rate": 5e-06,
"loss": 0.5471,
"step": 2170
},
{
"epoch": 0.9782364819385236,
"grad_norm": 0.5594191735315223,
"learning_rate": 5e-06,
"loss": 0.5689,
"step": 2180
},
{
"epoch": 0.9827238052501682,
"grad_norm": 0.5396068252067977,
"learning_rate": 5e-06,
"loss": 0.5622,
"step": 2190
},
{
"epoch": 0.9872111285618129,
"grad_norm": 0.5376425274574365,
"learning_rate": 5e-06,
"loss": 0.5519,
"step": 2200
},
{
"epoch": 0.9916984518734575,
"grad_norm": 0.5920263370787465,
"learning_rate": 5e-06,
"loss": 0.5561,
"step": 2210
},
{
"epoch": 0.9961857751851021,
"grad_norm": 0.580623876573116,
"learning_rate": 5e-06,
"loss": 0.5712,
"step": 2220
},
{
"epoch": 0.9997756338344178,
"eval_loss": 0.567927360534668,
"eval_runtime": 559.1888,
"eval_samples_per_second": 26.842,
"eval_steps_per_second": 0.42,
"step": 2228
},
{
"epoch": 1.0006730984967467,
"grad_norm": 0.5734872435921652,
"learning_rate": 5e-06,
"loss": 0.566,
"step": 2230
},
{
"epoch": 1.0051604218083914,
"grad_norm": 0.5741533955437085,
"learning_rate": 5e-06,
"loss": 0.5081,
"step": 2240
},
{
"epoch": 1.0096477451200359,
"grad_norm": 0.5286982651391207,
"learning_rate": 5e-06,
"loss": 0.4975,
"step": 2250
},
{
"epoch": 1.0141350684316806,
"grad_norm": 0.5565402789518736,
"learning_rate": 5e-06,
"loss": 0.4866,
"step": 2260
},
{
"epoch": 1.018622391743325,
"grad_norm": 0.5418393585734336,
"learning_rate": 5e-06,
"loss": 0.4855,
"step": 2270
},
{
"epoch": 1.0231097150549697,
"grad_norm": 0.5272758697526944,
"learning_rate": 5e-06,
"loss": 0.5484,
"step": 2280
},
{
"epoch": 1.0275970383666144,
"grad_norm": 0.5666635742640517,
"learning_rate": 5e-06,
"loss": 0.5021,
"step": 2290
},
{
"epoch": 1.032084361678259,
"grad_norm": 0.5482469107158694,
"learning_rate": 5e-06,
"loss": 0.541,
"step": 2300
},
{
"epoch": 1.0365716849899036,
"grad_norm": 0.5785689508918422,
"learning_rate": 5e-06,
"loss": 0.4939,
"step": 2310
},
{
"epoch": 1.041059008301548,
"grad_norm": 0.5479491492317479,
"learning_rate": 5e-06,
"loss": 0.4923,
"step": 2320
},
{
"epoch": 1.0455463316131928,
"grad_norm": 0.596193057255721,
"learning_rate": 5e-06,
"loss": 0.5309,
"step": 2330
},
{
"epoch": 1.0500336549248372,
"grad_norm": 0.5684218587581331,
"learning_rate": 5e-06,
"loss": 0.4871,
"step": 2340
},
{
"epoch": 1.054520978236482,
"grad_norm": 0.5501353350393796,
"learning_rate": 5e-06,
"loss": 0.5042,
"step": 2350
},
{
"epoch": 1.0590083015481266,
"grad_norm": 0.5651420414453633,
"learning_rate": 5e-06,
"loss": 0.5268,
"step": 2360
},
{
"epoch": 1.0634956248597711,
"grad_norm": 0.5556768039050937,
"learning_rate": 5e-06,
"loss": 0.4849,
"step": 2370
},
{
"epoch": 1.0679829481714158,
"grad_norm": 0.5177547386554735,
"learning_rate": 5e-06,
"loss": 0.4897,
"step": 2380
},
{
"epoch": 1.0724702714830603,
"grad_norm": 0.5651839563997499,
"learning_rate": 5e-06,
"loss": 0.524,
"step": 2390
},
{
"epoch": 1.076957594794705,
"grad_norm": 0.5326902266396754,
"learning_rate": 5e-06,
"loss": 0.4881,
"step": 2400
},
{
"epoch": 1.0814449181063495,
"grad_norm": 0.5628872446586874,
"learning_rate": 5e-06,
"loss": 0.5297,
"step": 2410
},
{
"epoch": 1.0859322414179942,
"grad_norm": 0.5477565748095592,
"learning_rate": 5e-06,
"loss": 0.5155,
"step": 2420
},
{
"epoch": 1.0904195647296389,
"grad_norm": 0.5156657552293519,
"learning_rate": 5e-06,
"loss": 0.4981,
"step": 2430
},
{
"epoch": 1.0949068880412833,
"grad_norm": 0.5499438723209844,
"learning_rate": 5e-06,
"loss": 0.5053,
"step": 2440
},
{
"epoch": 1.099394211352928,
"grad_norm": 0.5600071230838901,
"learning_rate": 5e-06,
"loss": 0.5056,
"step": 2450
},
{
"epoch": 1.1038815346645725,
"grad_norm": 0.5379185956294618,
"learning_rate": 5e-06,
"loss": 0.5204,
"step": 2460
},
{
"epoch": 1.1083688579762172,
"grad_norm": 0.5640271259467887,
"learning_rate": 5e-06,
"loss": 0.5004,
"step": 2470
},
{
"epoch": 1.112856181287862,
"grad_norm": 0.5502816116129311,
"learning_rate": 5e-06,
"loss": 0.499,
"step": 2480
},
{
"epoch": 1.1173435045995064,
"grad_norm": 0.5682876811264557,
"learning_rate": 5e-06,
"loss": 0.4921,
"step": 2490
},
{
"epoch": 1.121830827911151,
"grad_norm": 0.5421816523423539,
"learning_rate": 5e-06,
"loss": 0.5108,
"step": 2500
},
{
"epoch": 1.1263181512227955,
"grad_norm": 0.5105369153949931,
"learning_rate": 5e-06,
"loss": 0.5003,
"step": 2510
},
{
"epoch": 1.1308054745344402,
"grad_norm": 0.5588492929357224,
"learning_rate": 5e-06,
"loss": 0.475,
"step": 2520
},
{
"epoch": 1.1352927978460847,
"grad_norm": 0.5716988346827195,
"learning_rate": 5e-06,
"loss": 0.5209,
"step": 2530
},
{
"epoch": 1.1397801211577294,
"grad_norm": 0.5169436013613863,
"learning_rate": 5e-06,
"loss": 0.4916,
"step": 2540
},
{
"epoch": 1.144267444469374,
"grad_norm": 0.6145618845838399,
"learning_rate": 5e-06,
"loss": 0.5013,
"step": 2550
},
{
"epoch": 1.1487547677810186,
"grad_norm": 0.5444973960505051,
"learning_rate": 5e-06,
"loss": 0.5332,
"step": 2560
},
{
"epoch": 1.1532420910926633,
"grad_norm": 0.5520753016233167,
"learning_rate": 5e-06,
"loss": 0.5048,
"step": 2570
},
{
"epoch": 1.1577294144043078,
"grad_norm": 0.5750697187874539,
"learning_rate": 5e-06,
"loss": 0.5141,
"step": 2580
},
{
"epoch": 1.1622167377159525,
"grad_norm": 0.5747532663976821,
"learning_rate": 5e-06,
"loss": 0.5144,
"step": 2590
},
{
"epoch": 1.166704061027597,
"grad_norm": 0.5345651363842143,
"learning_rate": 5e-06,
"loss": 0.5208,
"step": 2600
},
{
"epoch": 1.1711913843392416,
"grad_norm": 0.5280475672990748,
"learning_rate": 5e-06,
"loss": 0.4987,
"step": 2610
},
{
"epoch": 1.1756787076508863,
"grad_norm": 0.6039148518875526,
"learning_rate": 5e-06,
"loss": 0.5132,
"step": 2620
},
{
"epoch": 1.1801660309625308,
"grad_norm": 0.5422913796982765,
"learning_rate": 5e-06,
"loss": 0.496,
"step": 2630
},
{
"epoch": 1.1846533542741755,
"grad_norm": 0.564401368588004,
"learning_rate": 5e-06,
"loss": 0.4827,
"step": 2640
},
{
"epoch": 1.18914067758582,
"grad_norm": 0.5373400269555145,
"learning_rate": 5e-06,
"loss": 0.5073,
"step": 2650
},
{
"epoch": 1.1936280008974647,
"grad_norm": 0.5314197538619471,
"learning_rate": 5e-06,
"loss": 0.5422,
"step": 2660
},
{
"epoch": 1.1981153242091094,
"grad_norm": 0.5166900359174719,
"learning_rate": 5e-06,
"loss": 0.519,
"step": 2670
},
{
"epoch": 1.2026026475207539,
"grad_norm": 0.5180554826133104,
"learning_rate": 5e-06,
"loss": 0.497,
"step": 2680
},
{
"epoch": 1.2070899708323986,
"grad_norm": 0.5323969904323123,
"learning_rate": 5e-06,
"loss": 0.5058,
"step": 2690
},
{
"epoch": 1.211577294144043,
"grad_norm": 0.5424465309236546,
"learning_rate": 5e-06,
"loss": 0.4958,
"step": 2700
},
{
"epoch": 1.2160646174556877,
"grad_norm": 0.5412841285613176,
"learning_rate": 5e-06,
"loss": 0.4948,
"step": 2710
},
{
"epoch": 1.2205519407673322,
"grad_norm": 0.5411400879988145,
"learning_rate": 5e-06,
"loss": 0.4965,
"step": 2720
},
{
"epoch": 1.225039264078977,
"grad_norm": 0.5659357562265832,
"learning_rate": 5e-06,
"loss": 0.4772,
"step": 2730
},
{
"epoch": 1.2295265873906214,
"grad_norm": 0.5512745847285553,
"learning_rate": 5e-06,
"loss": 0.5273,
"step": 2740
},
{
"epoch": 1.234013910702266,
"grad_norm": 0.5704497342942455,
"learning_rate": 5e-06,
"loss": 0.5089,
"step": 2750
},
{
"epoch": 1.2385012340139108,
"grad_norm": 0.5398045427001416,
"learning_rate": 5e-06,
"loss": 0.4945,
"step": 2760
},
{
"epoch": 1.2429885573255552,
"grad_norm": 0.5334476606303412,
"learning_rate": 5e-06,
"loss": 0.4884,
"step": 2770
},
{
"epoch": 1.2474758806372,
"grad_norm": 0.589924660616353,
"learning_rate": 5e-06,
"loss": 0.505,
"step": 2780
},
{
"epoch": 1.2519632039488444,
"grad_norm": 0.5585345752197195,
"learning_rate": 5e-06,
"loss": 0.5335,
"step": 2790
},
{
"epoch": 1.2564505272604891,
"grad_norm": 0.5654189187890049,
"learning_rate": 5e-06,
"loss": 0.477,
"step": 2800
},
{
"epoch": 1.2609378505721338,
"grad_norm": 0.5631419512470206,
"learning_rate": 5e-06,
"loss": 0.4984,
"step": 2810
},
{
"epoch": 1.2654251738837783,
"grad_norm": 0.5686873835040517,
"learning_rate": 5e-06,
"loss": 0.5073,
"step": 2820
},
{
"epoch": 1.269912497195423,
"grad_norm": 0.5503855071141434,
"learning_rate": 5e-06,
"loss": 0.4989,
"step": 2830
},
{
"epoch": 1.2743998205070675,
"grad_norm": 0.5620043676621576,
"learning_rate": 5e-06,
"loss": 0.5013,
"step": 2840
},
{
"epoch": 1.2788871438187122,
"grad_norm": 0.5502017343987001,
"learning_rate": 5e-06,
"loss": 0.4843,
"step": 2850
},
{
"epoch": 1.2833744671303569,
"grad_norm": 0.5377919239436089,
"learning_rate": 5e-06,
"loss": 0.5102,
"step": 2860
},
{
"epoch": 1.2878617904420013,
"grad_norm": 0.5423206907759274,
"learning_rate": 5e-06,
"loss": 0.5308,
"step": 2870
},
{
"epoch": 1.292349113753646,
"grad_norm": 0.553618965490239,
"learning_rate": 5e-06,
"loss": 0.5239,
"step": 2880
},
{
"epoch": 1.2968364370652905,
"grad_norm": 0.5712510750822322,
"learning_rate": 5e-06,
"loss": 0.5021,
"step": 2890
},
{
"epoch": 1.3013237603769352,
"grad_norm": 0.5634666949947841,
"learning_rate": 5e-06,
"loss": 0.4941,
"step": 2900
},
{
"epoch": 1.30581108368858,
"grad_norm": 0.568485750592611,
"learning_rate": 5e-06,
"loss": 0.505,
"step": 2910
},
{
"epoch": 1.3102984070002244,
"grad_norm": 0.5671935619893487,
"learning_rate": 5e-06,
"loss": 0.4912,
"step": 2920
},
{
"epoch": 1.3147857303118688,
"grad_norm": 0.5757679724498568,
"learning_rate": 5e-06,
"loss": 0.4913,
"step": 2930
},
{
"epoch": 1.3192730536235135,
"grad_norm": 0.5590352144821122,
"learning_rate": 5e-06,
"loss": 0.486,
"step": 2940
},
{
"epoch": 1.3237603769351582,
"grad_norm": 0.5445238317180622,
"learning_rate": 5e-06,
"loss": 0.4908,
"step": 2950
},
{
"epoch": 1.3282477002468027,
"grad_norm": 0.5380943498565294,
"learning_rate": 5e-06,
"loss": 0.5293,
"step": 2960
},
{
"epoch": 1.3327350235584474,
"grad_norm": 0.5573459369134599,
"learning_rate": 5e-06,
"loss": 0.4739,
"step": 2970
},
{
"epoch": 1.337222346870092,
"grad_norm": 0.5563120439249387,
"learning_rate": 5e-06,
"loss": 0.482,
"step": 2980
},
{
"epoch": 1.3417096701817366,
"grad_norm": 0.5545396609770007,
"learning_rate": 5e-06,
"loss": 0.4936,
"step": 2990
},
{
"epoch": 1.3461969934933813,
"grad_norm": 0.536222681258455,
"learning_rate": 5e-06,
"loss": 0.4977,
"step": 3000
},
{
"epoch": 1.3506843168050258,
"grad_norm": 0.5622333924200364,
"learning_rate": 5e-06,
"loss": 0.5232,
"step": 3010
},
{
"epoch": 1.3551716401166705,
"grad_norm": 0.5477326991210051,
"learning_rate": 5e-06,
"loss": 0.5085,
"step": 3020
},
{
"epoch": 1.359658963428315,
"grad_norm": 0.543317448941311,
"learning_rate": 5e-06,
"loss": 0.4971,
"step": 3030
},
{
"epoch": 1.3641462867399596,
"grad_norm": 0.5400423612006745,
"learning_rate": 5e-06,
"loss": 0.5122,
"step": 3040
},
{
"epoch": 1.3686336100516043,
"grad_norm": 0.5578167396365121,
"learning_rate": 5e-06,
"loss": 0.5167,
"step": 3050
},
{
"epoch": 1.3731209333632488,
"grad_norm": 0.5740701717143266,
"learning_rate": 5e-06,
"loss": 0.509,
"step": 3060
},
{
"epoch": 1.3776082566748935,
"grad_norm": 0.5570929052762126,
"learning_rate": 5e-06,
"loss": 0.5103,
"step": 3070
},
{
"epoch": 1.382095579986538,
"grad_norm": 0.5433927111213747,
"learning_rate": 5e-06,
"loss": 0.5198,
"step": 3080
},
{
"epoch": 1.3865829032981827,
"grad_norm": 0.5710035732297696,
"learning_rate": 5e-06,
"loss": 0.5064,
"step": 3090
},
{
"epoch": 1.3910702266098274,
"grad_norm": 0.5932400385935832,
"learning_rate": 5e-06,
"loss": 0.5012,
"step": 3100
},
{
"epoch": 1.3955575499214719,
"grad_norm": 0.545506710464315,
"learning_rate": 5e-06,
"loss": 0.4892,
"step": 3110
},
{
"epoch": 1.4000448732331163,
"grad_norm": 0.5752471090405673,
"learning_rate": 5e-06,
"loss": 0.505,
"step": 3120
},
{
"epoch": 1.404532196544761,
"grad_norm": 0.5628082472998481,
"learning_rate": 5e-06,
"loss": 0.4866,
"step": 3130
},
{
"epoch": 1.4090195198564057,
"grad_norm": 0.5755016704798377,
"learning_rate": 5e-06,
"loss": 0.5056,
"step": 3140
},
{
"epoch": 1.4135068431680502,
"grad_norm": 0.5280132793104151,
"learning_rate": 5e-06,
"loss": 0.5281,
"step": 3150
},
{
"epoch": 1.417994166479695,
"grad_norm": 0.5466619698880913,
"learning_rate": 5e-06,
"loss": 0.5207,
"step": 3160
},
{
"epoch": 1.4224814897913394,
"grad_norm": 0.5282843782818277,
"learning_rate": 5e-06,
"loss": 0.5298,
"step": 3170
},
{
"epoch": 1.426968813102984,
"grad_norm": 0.5734506299177996,
"learning_rate": 5e-06,
"loss": 0.5124,
"step": 3180
},
{
"epoch": 1.4314561364146288,
"grad_norm": 0.5344798626824697,
"learning_rate": 5e-06,
"loss": 0.4963,
"step": 3190
},
{
"epoch": 1.4359434597262732,
"grad_norm": 0.5473981763834849,
"learning_rate": 5e-06,
"loss": 0.5054,
"step": 3200
},
{
"epoch": 1.440430783037918,
"grad_norm": 0.5424308579398186,
"learning_rate": 5e-06,
"loss": 0.5019,
"step": 3210
},
{
"epoch": 1.4449181063495624,
"grad_norm": 0.5719832828803095,
"learning_rate": 5e-06,
"loss": 0.4982,
"step": 3220
},
{
"epoch": 1.4494054296612071,
"grad_norm": 0.5572404240421417,
"learning_rate": 5e-06,
"loss": 0.5147,
"step": 3230
},
{
"epoch": 1.4538927529728518,
"grad_norm": 0.5458124274523334,
"learning_rate": 5e-06,
"loss": 0.5168,
"step": 3240
},
{
"epoch": 1.4583800762844963,
"grad_norm": 0.5193755152140953,
"learning_rate": 5e-06,
"loss": 0.5223,
"step": 3250
},
{
"epoch": 1.462867399596141,
"grad_norm": 0.5233432783340276,
"learning_rate": 5e-06,
"loss": 0.5167,
"step": 3260
},
{
"epoch": 1.4673547229077855,
"grad_norm": 0.544740366629981,
"learning_rate": 5e-06,
"loss": 0.4906,
"step": 3270
},
{
"epoch": 1.4718420462194302,
"grad_norm": 0.5995125947333418,
"learning_rate": 5e-06,
"loss": 0.5063,
"step": 3280
},
{
"epoch": 1.4763293695310746,
"grad_norm": 0.5580064274640354,
"learning_rate": 5e-06,
"loss": 0.4932,
"step": 3290
},
{
"epoch": 1.4808166928427193,
"grad_norm": 0.5515028035867199,
"learning_rate": 5e-06,
"loss": 0.4976,
"step": 3300
},
{
"epoch": 1.4853040161543638,
"grad_norm": 0.5326953694404295,
"learning_rate": 5e-06,
"loss": 0.5001,
"step": 3310
},
{
"epoch": 1.4897913394660085,
"grad_norm": 0.5531717436070133,
"learning_rate": 5e-06,
"loss": 0.5157,
"step": 3320
},
{
"epoch": 1.4942786627776532,
"grad_norm": 0.5536934590854812,
"learning_rate": 5e-06,
"loss": 0.5011,
"step": 3330
},
{
"epoch": 1.4987659860892977,
"grad_norm": 0.537594784304719,
"learning_rate": 5e-06,
"loss": 0.4988,
"step": 3340
},
{
"epoch": 1.5032533094009424,
"grad_norm": 0.5325933270053125,
"learning_rate": 5e-06,
"loss": 0.5263,
"step": 3350
},
{
"epoch": 1.5077406327125868,
"grad_norm": 0.517841662464493,
"learning_rate": 5e-06,
"loss": 0.4995,
"step": 3360
},
{
"epoch": 1.5122279560242315,
"grad_norm": 0.5154921484724686,
"learning_rate": 5e-06,
"loss": 0.4933,
"step": 3370
},
{
"epoch": 1.5167152793358762,
"grad_norm": 0.5775973989195,
"learning_rate": 5e-06,
"loss": 0.5068,
"step": 3380
},
{
"epoch": 1.5212026026475207,
"grad_norm": 0.5854301357165765,
"learning_rate": 5e-06,
"loss": 0.5028,
"step": 3390
},
{
"epoch": 1.5256899259591652,
"grad_norm": 0.5438296714544919,
"learning_rate": 5e-06,
"loss": 0.4992,
"step": 3400
},
{
"epoch": 1.53017724927081,
"grad_norm": 0.5581750773098291,
"learning_rate": 5e-06,
"loss": 0.4774,
"step": 3410
},
{
"epoch": 1.5346645725824546,
"grad_norm": 0.5832173054693874,
"learning_rate": 5e-06,
"loss": 0.5061,
"step": 3420
},
{
"epoch": 1.5391518958940993,
"grad_norm": 0.5753758202144109,
"learning_rate": 5e-06,
"loss": 0.4999,
"step": 3430
},
{
"epoch": 1.5436392192057438,
"grad_norm": 0.544533713206346,
"learning_rate": 5e-06,
"loss": 0.5088,
"step": 3440
},
{
"epoch": 1.5481265425173882,
"grad_norm": 0.571084602500549,
"learning_rate": 5e-06,
"loss": 0.4818,
"step": 3450
},
{
"epoch": 1.552613865829033,
"grad_norm": 0.584923156106211,
"learning_rate": 5e-06,
"loss": 0.4813,
"step": 3460
},
{
"epoch": 1.5571011891406776,
"grad_norm": 0.5305219598557945,
"learning_rate": 5e-06,
"loss": 0.5034,
"step": 3470
},
{
"epoch": 1.5615885124523223,
"grad_norm": 0.5488575678648898,
"learning_rate": 5e-06,
"loss": 0.4968,
"step": 3480
},
{
"epoch": 1.5660758357639668,
"grad_norm": 0.5700887842489093,
"learning_rate": 5e-06,
"loss": 0.4957,
"step": 3490
},
{
"epoch": 1.5705631590756113,
"grad_norm": 0.5493679649855058,
"learning_rate": 5e-06,
"loss": 0.4945,
"step": 3500
},
{
"epoch": 1.575050482387256,
"grad_norm": 0.5791638763067023,
"learning_rate": 5e-06,
"loss": 0.5118,
"step": 3510
},
{
"epoch": 1.5795378056989007,
"grad_norm": 0.5916566338039917,
"learning_rate": 5e-06,
"loss": 0.5194,
"step": 3520
},
{
"epoch": 1.5840251290105454,
"grad_norm": 0.5570616792034053,
"learning_rate": 5e-06,
"loss": 0.4857,
"step": 3530
},
{
"epoch": 1.5885124523221898,
"grad_norm": 0.5461551153883559,
"learning_rate": 5e-06,
"loss": 0.4713,
"step": 3540
},
{
"epoch": 1.5929997756338343,
"grad_norm": 0.5515261498525429,
"learning_rate": 5e-06,
"loss": 0.5297,
"step": 3550
},
{
"epoch": 1.597487098945479,
"grad_norm": 0.57188102614297,
"learning_rate": 5e-06,
"loss": 0.4888,
"step": 3560
},
{
"epoch": 1.6019744222571237,
"grad_norm": 0.5171680954916367,
"learning_rate": 5e-06,
"loss": 0.4842,
"step": 3570
},
{
"epoch": 1.6064617455687682,
"grad_norm": 0.5756688858390624,
"learning_rate": 5e-06,
"loss": 0.4943,
"step": 3580
},
{
"epoch": 1.6109490688804127,
"grad_norm": 0.5749499103685126,
"learning_rate": 5e-06,
"loss": 0.5379,
"step": 3590
},
{
"epoch": 1.6154363921920574,
"grad_norm": 0.5351176936292328,
"learning_rate": 5e-06,
"loss": 0.4859,
"step": 3600
},
{
"epoch": 1.619923715503702,
"grad_norm": 0.5465963706973009,
"learning_rate": 5e-06,
"loss": 0.5253,
"step": 3610
},
{
"epoch": 1.6244110388153468,
"grad_norm": 0.5952713410678487,
"learning_rate": 5e-06,
"loss": 0.5248,
"step": 3620
},
{
"epoch": 1.6288983621269912,
"grad_norm": 0.5693133264089241,
"learning_rate": 5e-06,
"loss": 0.4947,
"step": 3630
},
{
"epoch": 1.6333856854386357,
"grad_norm": 0.532801262620042,
"learning_rate": 5e-06,
"loss": 0.4846,
"step": 3640
},
{
"epoch": 1.6378730087502804,
"grad_norm": 0.5731108270619412,
"learning_rate": 5e-06,
"loss": 0.5178,
"step": 3650
},
{
"epoch": 1.642360332061925,
"grad_norm": 0.5566162421839872,
"learning_rate": 5e-06,
"loss": 0.5142,
"step": 3660
},
{
"epoch": 1.6468476553735698,
"grad_norm": 0.5176692150882419,
"learning_rate": 5e-06,
"loss": 0.5146,
"step": 3670
},
{
"epoch": 1.6513349786852143,
"grad_norm": 0.5712166747441446,
"learning_rate": 5e-06,
"loss": 0.4942,
"step": 3680
},
{
"epoch": 1.6558223019968588,
"grad_norm": 0.5411076738264401,
"learning_rate": 5e-06,
"loss": 0.5109,
"step": 3690
},
{
"epoch": 1.6603096253085035,
"grad_norm": 0.5527323928497764,
"learning_rate": 5e-06,
"loss": 0.5189,
"step": 3700
},
{
"epoch": 1.6647969486201482,
"grad_norm": 0.5475126879488802,
"learning_rate": 5e-06,
"loss": 0.498,
"step": 3710
},
{
"epoch": 1.6692842719317929,
"grad_norm": 0.5226831637595657,
"learning_rate": 5e-06,
"loss": 0.4834,
"step": 3720
},
{
"epoch": 1.6737715952434373,
"grad_norm": 0.5378242696447812,
"learning_rate": 5e-06,
"loss": 0.5418,
"step": 3730
},
{
"epoch": 1.6782589185550818,
"grad_norm": 0.5949971573967483,
"learning_rate": 5e-06,
"loss": 0.508,
"step": 3740
},
{
"epoch": 1.6827462418667265,
"grad_norm": 0.5358908747440865,
"learning_rate": 5e-06,
"loss": 0.4967,
"step": 3750
},
{
"epoch": 1.6872335651783712,
"grad_norm": 0.5474090787609206,
"learning_rate": 5e-06,
"loss": 0.4933,
"step": 3760
},
{
"epoch": 1.6917208884900157,
"grad_norm": 0.5914274676963112,
"learning_rate": 5e-06,
"loss": 0.4949,
"step": 3770
},
{
"epoch": 1.6962082118016601,
"grad_norm": 0.5537449538419119,
"learning_rate": 5e-06,
"loss": 0.4966,
"step": 3780
},
{
"epoch": 1.7006955351133048,
"grad_norm": 0.5448028303916069,
"learning_rate": 5e-06,
"loss": 0.5074,
"step": 3790
},
{
"epoch": 1.7051828584249495,
"grad_norm": 0.5708643249016583,
"learning_rate": 5e-06,
"loss": 0.4896,
"step": 3800
},
{
"epoch": 1.7096701817365942,
"grad_norm": 0.5745684091394735,
"learning_rate": 5e-06,
"loss": 0.5045,
"step": 3810
},
{
"epoch": 1.7141575050482387,
"grad_norm": 0.5512474274982683,
"learning_rate": 5e-06,
"loss": 0.5089,
"step": 3820
},
{
"epoch": 1.7186448283598832,
"grad_norm": 0.5430706302627716,
"learning_rate": 5e-06,
"loss": 0.4994,
"step": 3830
},
{
"epoch": 1.723132151671528,
"grad_norm": 0.5556308718806792,
"learning_rate": 5e-06,
"loss": 0.5025,
"step": 3840
},
{
"epoch": 1.7276194749831726,
"grad_norm": 0.5451097170074302,
"learning_rate": 5e-06,
"loss": 0.5041,
"step": 3850
},
{
"epoch": 1.7321067982948173,
"grad_norm": 0.5482562609039291,
"learning_rate": 5e-06,
"loss": 0.505,
"step": 3860
},
{
"epoch": 1.7365941216064618,
"grad_norm": 0.5708328715347044,
"learning_rate": 5e-06,
"loss": 0.4998,
"step": 3870
},
{
"epoch": 1.7410814449181062,
"grad_norm": 0.5491007607865179,
"learning_rate": 5e-06,
"loss": 0.5203,
"step": 3880
},
{
"epoch": 1.745568768229751,
"grad_norm": 0.539262203178338,
"learning_rate": 5e-06,
"loss": 0.4938,
"step": 3890
},
{
"epoch": 1.7500560915413956,
"grad_norm": 0.587964940856532,
"learning_rate": 5e-06,
"loss": 0.5138,
"step": 3900
},
{
"epoch": 1.75454341485304,
"grad_norm": 0.603989953035275,
"learning_rate": 5e-06,
"loss": 0.5019,
"step": 3910
},
{
"epoch": 1.7590307381646848,
"grad_norm": 0.5936506149594802,
"learning_rate": 5e-06,
"loss": 0.4941,
"step": 3920
},
{
"epoch": 1.7635180614763293,
"grad_norm": 0.5461819280550173,
"learning_rate": 5e-06,
"loss": 0.5206,
"step": 3930
},
{
"epoch": 1.768005384787974,
"grad_norm": 0.5639017847680231,
"learning_rate": 5e-06,
"loss": 0.4974,
"step": 3940
},
{
"epoch": 1.7724927080996187,
"grad_norm": 0.5290104593192931,
"learning_rate": 5e-06,
"loss": 0.4939,
"step": 3950
},
{
"epoch": 1.7769800314112632,
"grad_norm": 0.5831983935779752,
"learning_rate": 5e-06,
"loss": 0.5019,
"step": 3960
},
{
"epoch": 1.7814673547229076,
"grad_norm": 0.5538657179797891,
"learning_rate": 5e-06,
"loss": 0.4852,
"step": 3970
},
{
"epoch": 1.7859546780345523,
"grad_norm": 0.568557365614838,
"learning_rate": 5e-06,
"loss": 0.5183,
"step": 3980
},
{
"epoch": 1.790442001346197,
"grad_norm": 0.5362764060596701,
"learning_rate": 5e-06,
"loss": 0.51,
"step": 3990
},
{
"epoch": 1.7949293246578417,
"grad_norm": 0.5592584139795364,
"learning_rate": 5e-06,
"loss": 0.504,
"step": 4000
},
{
"epoch": 1.7994166479694862,
"grad_norm": 0.555910637560083,
"learning_rate": 5e-06,
"loss": 0.494,
"step": 4010
},
{
"epoch": 1.8039039712811307,
"grad_norm": 0.551977757056201,
"learning_rate": 5e-06,
"loss": 0.5222,
"step": 4020
},
{
"epoch": 1.8083912945927754,
"grad_norm": 0.537652808354882,
"learning_rate": 5e-06,
"loss": 0.5047,
"step": 4030
},
{
"epoch": 1.81287861790442,
"grad_norm": 0.5606846471432076,
"learning_rate": 5e-06,
"loss": 0.5148,
"step": 4040
},
{
"epoch": 1.8173659412160648,
"grad_norm": 0.5316042317862866,
"learning_rate": 5e-06,
"loss": 0.4958,
"step": 4050
},
{
"epoch": 1.8218532645277092,
"grad_norm": 0.6009251472832223,
"learning_rate": 5e-06,
"loss": 0.4887,
"step": 4060
},
{
"epoch": 1.8263405878393537,
"grad_norm": 0.5833028652502357,
"learning_rate": 5e-06,
"loss": 0.5026,
"step": 4070
},
{
"epoch": 1.8308279111509984,
"grad_norm": 0.5303089821472199,
"learning_rate": 5e-06,
"loss": 0.4802,
"step": 4080
},
{
"epoch": 1.835315234462643,
"grad_norm": 0.5356894166936327,
"learning_rate": 5e-06,
"loss": 0.4948,
"step": 4090
},
{
"epoch": 1.8398025577742876,
"grad_norm": 0.5399931872531933,
"learning_rate": 5e-06,
"loss": 0.4954,
"step": 4100
},
{
"epoch": 1.8442898810859323,
"grad_norm": 0.562838831677771,
"learning_rate": 5e-06,
"loss": 0.5215,
"step": 4110
},
{
"epoch": 1.8487772043975768,
"grad_norm": 0.5767666769206414,
"learning_rate": 5e-06,
"loss": 0.4895,
"step": 4120
},
{
"epoch": 1.8532645277092215,
"grad_norm": 0.5290702500324674,
"learning_rate": 5e-06,
"loss": 0.5029,
"step": 4130
},
{
"epoch": 1.8577518510208662,
"grad_norm": 0.5308632664860179,
"learning_rate": 5e-06,
"loss": 0.4919,
"step": 4140
},
{
"epoch": 1.8622391743325106,
"grad_norm": 0.5413420247585403,
"learning_rate": 5e-06,
"loss": 0.5034,
"step": 4150
},
{
"epoch": 1.866726497644155,
"grad_norm": 0.5538840065119353,
"learning_rate": 5e-06,
"loss": 0.5002,
"step": 4160
},
{
"epoch": 1.8712138209557998,
"grad_norm": 0.5211702522584797,
"learning_rate": 5e-06,
"loss": 0.505,
"step": 4170
},
{
"epoch": 1.8757011442674445,
"grad_norm": 0.5446822204678821,
"learning_rate": 5e-06,
"loss": 0.5287,
"step": 4180
},
{
"epoch": 1.8801884675790892,
"grad_norm": 0.5561626601823235,
"learning_rate": 5e-06,
"loss": 0.5205,
"step": 4190
},
{
"epoch": 1.8846757908907337,
"grad_norm": 0.5537666305289907,
"learning_rate": 5e-06,
"loss": 0.4989,
"step": 4200
},
{
"epoch": 1.8891631142023781,
"grad_norm": 0.5685163585083653,
"learning_rate": 5e-06,
"loss": 0.5154,
"step": 4210
},
{
"epoch": 1.8936504375140228,
"grad_norm": 0.5451373353451423,
"learning_rate": 5e-06,
"loss": 0.5065,
"step": 4220
},
{
"epoch": 1.8981377608256675,
"grad_norm": 0.5518210301636858,
"learning_rate": 5e-06,
"loss": 0.5001,
"step": 4230
},
{
"epoch": 1.9026250841373122,
"grad_norm": 0.5470491973339323,
"learning_rate": 5e-06,
"loss": 0.4953,
"step": 4240
},
{
"epoch": 1.9071124074489567,
"grad_norm": 0.5402094383394249,
"learning_rate": 5e-06,
"loss": 0.4881,
"step": 4250
},
{
"epoch": 1.9115997307606012,
"grad_norm": 0.5356916841246723,
"learning_rate": 5e-06,
"loss": 0.498,
"step": 4260
},
{
"epoch": 1.9160870540722459,
"grad_norm": 0.6011677528029307,
"learning_rate": 5e-06,
"loss": 0.5128,
"step": 4270
},
{
"epoch": 1.9205743773838906,
"grad_norm": 0.533641599988238,
"learning_rate": 5e-06,
"loss": 0.4942,
"step": 4280
},
{
"epoch": 1.925061700695535,
"grad_norm": 0.561171129511058,
"learning_rate": 5e-06,
"loss": 0.5069,
"step": 4290
},
{
"epoch": 1.9295490240071798,
"grad_norm": 0.5648499086549563,
"learning_rate": 5e-06,
"loss": 0.5264,
"step": 4300
},
{
"epoch": 1.9340363473188242,
"grad_norm": 0.568255235637858,
"learning_rate": 5e-06,
"loss": 0.4944,
"step": 4310
},
{
"epoch": 1.938523670630469,
"grad_norm": 0.558570542074393,
"learning_rate": 5e-06,
"loss": 0.5116,
"step": 4320
},
{
"epoch": 1.9430109939421136,
"grad_norm": 0.5335131103567927,
"learning_rate": 5e-06,
"loss": 0.5308,
"step": 4330
},
{
"epoch": 1.947498317253758,
"grad_norm": 0.528601279188925,
"learning_rate": 5e-06,
"loss": 0.5038,
"step": 4340
},
{
"epoch": 1.9519856405654026,
"grad_norm": 0.5651768549322472,
"learning_rate": 5e-06,
"loss": 0.5391,
"step": 4350
},
{
"epoch": 1.9564729638770473,
"grad_norm": 0.5209641936278054,
"learning_rate": 5e-06,
"loss": 0.4916,
"step": 4360
},
{
"epoch": 1.960960287188692,
"grad_norm": 0.6121127336148168,
"learning_rate": 5e-06,
"loss": 0.4959,
"step": 4370
},
{
"epoch": 1.9654476105003367,
"grad_norm": 0.5295573811299278,
"learning_rate": 5e-06,
"loss": 0.4851,
"step": 4380
},
{
"epoch": 1.9699349338119811,
"grad_norm": 0.5768584348850149,
"learning_rate": 5e-06,
"loss": 0.5266,
"step": 4390
},
{
"epoch": 1.9744222571236256,
"grad_norm": 0.5174975104621315,
"learning_rate": 5e-06,
"loss": 0.4978,
"step": 4400
},
{
"epoch": 1.9789095804352703,
"grad_norm": 0.5546606880895456,
"learning_rate": 5e-06,
"loss": 0.4928,
"step": 4410
},
{
"epoch": 1.983396903746915,
"grad_norm": 0.5683833011043679,
"learning_rate": 5e-06,
"loss": 0.4989,
"step": 4420
},
{
"epoch": 1.9878842270585597,
"grad_norm": 0.537165046986452,
"learning_rate": 5e-06,
"loss": 0.516,
"step": 4430
},
{
"epoch": 1.9923715503702042,
"grad_norm": 0.5319744139724734,
"learning_rate": 5e-06,
"loss": 0.5309,
"step": 4440
},
{
"epoch": 1.9968588736818487,
"grad_norm": 0.5451938729099154,
"learning_rate": 5e-06,
"loss": 0.5079,
"step": 4450
},
{
"epoch": 2.0,
"eval_loss": 0.5559915900230408,
"eval_runtime": 559.3369,
"eval_samples_per_second": 26.835,
"eval_steps_per_second": 0.42,
"step": 4457
},
{
"epoch": 2.0013461969934934,
"grad_norm": 0.5652296466620075,
"learning_rate": 5e-06,
"loss": 0.489,
"step": 4460
},
{
"epoch": 2.005833520305138,
"grad_norm": 0.5447379459508764,
"learning_rate": 5e-06,
"loss": 0.4355,
"step": 4470
},
{
"epoch": 2.0103208436167828,
"grad_norm": 0.5178404377199393,
"learning_rate": 5e-06,
"loss": 0.4286,
"step": 4480
},
{
"epoch": 2.014808166928427,
"grad_norm": 0.5422732055965347,
"learning_rate": 5e-06,
"loss": 0.4383,
"step": 4490
},
{
"epoch": 2.0192954902400717,
"grad_norm": 0.535310702889713,
"learning_rate": 5e-06,
"loss": 0.4392,
"step": 4500
},
{
"epoch": 2.0237828135517164,
"grad_norm": 0.5465033381736284,
"learning_rate": 5e-06,
"loss": 0.4322,
"step": 4510
},
{
"epoch": 2.028270136863361,
"grad_norm": 0.5407880814912572,
"learning_rate": 5e-06,
"loss": 0.422,
"step": 4520
},
{
"epoch": 2.032757460175006,
"grad_norm": 0.5475175525712156,
"learning_rate": 5e-06,
"loss": 0.4228,
"step": 4530
},
{
"epoch": 2.03724478348665,
"grad_norm": 0.5305519155722267,
"learning_rate": 5e-06,
"loss": 0.4414,
"step": 4540
},
{
"epoch": 2.0417321067982948,
"grad_norm": 0.5525760243070599,
"learning_rate": 5e-06,
"loss": 0.4319,
"step": 4550
},
{
"epoch": 2.0462194301099395,
"grad_norm": 0.5380138271735849,
"learning_rate": 5e-06,
"loss": 0.4184,
"step": 4560
},
{
"epoch": 2.050706753421584,
"grad_norm": 0.5393541372377034,
"learning_rate": 5e-06,
"loss": 0.4267,
"step": 4570
},
{
"epoch": 2.055194076733229,
"grad_norm": 0.5280423864839874,
"learning_rate": 5e-06,
"loss": 0.4492,
"step": 4580
},
{
"epoch": 2.059681400044873,
"grad_norm": 0.5598943021301682,
"learning_rate": 5e-06,
"loss": 0.4445,
"step": 4590
},
{
"epoch": 2.064168723356518,
"grad_norm": 0.5110154136642925,
"learning_rate": 5e-06,
"loss": 0.4279,
"step": 4600
},
{
"epoch": 2.0686560466681625,
"grad_norm": 0.521113203676518,
"learning_rate": 5e-06,
"loss": 0.476,
"step": 4610
},
{
"epoch": 2.073143369979807,
"grad_norm": 0.509177043617814,
"learning_rate": 5e-06,
"loss": 0.4511,
"step": 4620
},
{
"epoch": 2.0776306932914514,
"grad_norm": 0.5204617332251407,
"learning_rate": 5e-06,
"loss": 0.4385,
"step": 4630
},
{
"epoch": 2.082118016603096,
"grad_norm": 0.5600686745711032,
"learning_rate": 5e-06,
"loss": 0.4409,
"step": 4640
},
{
"epoch": 2.086605339914741,
"grad_norm": 0.551649701345727,
"learning_rate": 5e-06,
"loss": 0.4235,
"step": 4650
},
{
"epoch": 2.0910926632263855,
"grad_norm": 0.555346212302441,
"learning_rate": 5e-06,
"loss": 0.4367,
"step": 4660
},
{
"epoch": 2.0955799865380302,
"grad_norm": 0.5333054935894155,
"learning_rate": 5e-06,
"loss": 0.4458,
"step": 4670
},
{
"epoch": 2.1000673098496745,
"grad_norm": 0.5663799977134729,
"learning_rate": 5e-06,
"loss": 0.4415,
"step": 4680
},
{
"epoch": 2.104554633161319,
"grad_norm": 0.5768162610420671,
"learning_rate": 5e-06,
"loss": 0.4646,
"step": 4690
},
{
"epoch": 2.109041956472964,
"grad_norm": 0.5204844232448562,
"learning_rate": 5e-06,
"loss": 0.4725,
"step": 4700
},
{
"epoch": 2.1135292797846086,
"grad_norm": 0.5462747740338038,
"learning_rate": 5e-06,
"loss": 0.4342,
"step": 4710
},
{
"epoch": 2.1180166030962533,
"grad_norm": 0.5308200939238422,
"learning_rate": 5e-06,
"loss": 0.4226,
"step": 4720
},
{
"epoch": 2.1225039264078975,
"grad_norm": 0.5689657779220455,
"learning_rate": 5e-06,
"loss": 0.4387,
"step": 4730
},
{
"epoch": 2.1269912497195422,
"grad_norm": 0.5261178859362788,
"learning_rate": 5e-06,
"loss": 0.4343,
"step": 4740
},
{
"epoch": 2.131478573031187,
"grad_norm": 0.5802820337415188,
"learning_rate": 5e-06,
"loss": 0.452,
"step": 4750
},
{
"epoch": 2.1359658963428316,
"grad_norm": 0.5660982173345542,
"learning_rate": 5e-06,
"loss": 0.4405,
"step": 4760
},
{
"epoch": 2.140453219654476,
"grad_norm": 0.5611165514102701,
"learning_rate": 5e-06,
"loss": 0.4426,
"step": 4770
},
{
"epoch": 2.1449405429661206,
"grad_norm": 0.5362977002914265,
"learning_rate": 5e-06,
"loss": 0.4277,
"step": 4780
},
{
"epoch": 2.1494278662777653,
"grad_norm": 0.5775247895144668,
"learning_rate": 5e-06,
"loss": 0.417,
"step": 4790
},
{
"epoch": 2.15391518958941,
"grad_norm": 0.5049281569355238,
"learning_rate": 5e-06,
"loss": 0.4432,
"step": 4800
},
{
"epoch": 2.1584025129010547,
"grad_norm": 0.5101798893420769,
"learning_rate": 5e-06,
"loss": 0.4211,
"step": 4810
},
{
"epoch": 2.162889836212699,
"grad_norm": 0.5297993400461247,
"learning_rate": 5e-06,
"loss": 0.4359,
"step": 4820
},
{
"epoch": 2.1673771595243436,
"grad_norm": 0.543138800578885,
"learning_rate": 5e-06,
"loss": 0.438,
"step": 4830
},
{
"epoch": 2.1718644828359883,
"grad_norm": 0.5405162833857642,
"learning_rate": 5e-06,
"loss": 0.4372,
"step": 4840
},
{
"epoch": 2.176351806147633,
"grad_norm": 0.5772194473884518,
"learning_rate": 5e-06,
"loss": 0.4247,
"step": 4850
},
{
"epoch": 2.1808391294592777,
"grad_norm": 0.5366732857273809,
"learning_rate": 5e-06,
"loss": 0.4439,
"step": 4860
},
{
"epoch": 2.185326452770922,
"grad_norm": 0.5728184560332118,
"learning_rate": 5e-06,
"loss": 0.4537,
"step": 4870
},
{
"epoch": 2.1898137760825667,
"grad_norm": 0.5397099116720598,
"learning_rate": 5e-06,
"loss": 0.4333,
"step": 4880
},
{
"epoch": 2.1943010993942114,
"grad_norm": 0.5303476384455712,
"learning_rate": 5e-06,
"loss": 0.457,
"step": 4890
},
{
"epoch": 2.198788422705856,
"grad_norm": 0.5325455890564232,
"learning_rate": 5e-06,
"loss": 0.4537,
"step": 4900
},
{
"epoch": 2.2032757460175008,
"grad_norm": 0.530437891891213,
"learning_rate": 5e-06,
"loss": 0.4417,
"step": 4910
},
{
"epoch": 2.207763069329145,
"grad_norm": 0.5237148221222313,
"learning_rate": 5e-06,
"loss": 0.4518,
"step": 4920
},
{
"epoch": 2.2122503926407897,
"grad_norm": 0.7599811908495089,
"learning_rate": 5e-06,
"loss": 0.4292,
"step": 4930
},
{
"epoch": 2.2167377159524344,
"grad_norm": 0.5679168932907183,
"learning_rate": 5e-06,
"loss": 0.4337,
"step": 4940
},
{
"epoch": 2.221225039264079,
"grad_norm": 0.5598357464655034,
"learning_rate": 5e-06,
"loss": 0.4453,
"step": 4950
},
{
"epoch": 2.225712362575724,
"grad_norm": 0.5974896598464358,
"learning_rate": 5e-06,
"loss": 0.46,
"step": 4960
},
{
"epoch": 2.230199685887368,
"grad_norm": 0.5483222823612722,
"learning_rate": 5e-06,
"loss": 0.4452,
"step": 4970
},
{
"epoch": 2.2346870091990128,
"grad_norm": 0.5624401258029339,
"learning_rate": 5e-06,
"loss": 0.4324,
"step": 4980
},
{
"epoch": 2.2391743325106575,
"grad_norm": 0.5454688396112486,
"learning_rate": 5e-06,
"loss": 0.4322,
"step": 4990
},
{
"epoch": 2.243661655822302,
"grad_norm": 0.5630702076751016,
"learning_rate": 5e-06,
"loss": 0.4229,
"step": 5000
},
{
"epoch": 2.2481489791339464,
"grad_norm": 0.5401237147561903,
"learning_rate": 5e-06,
"loss": 0.4177,
"step": 5010
},
{
"epoch": 2.252636302445591,
"grad_norm": 0.5424589473983259,
"learning_rate": 5e-06,
"loss": 0.4303,
"step": 5020
},
{
"epoch": 2.257123625757236,
"grad_norm": 0.5577033752080041,
"learning_rate": 5e-06,
"loss": 0.4272,
"step": 5030
},
{
"epoch": 2.2616109490688805,
"grad_norm": 0.5085433579295487,
"learning_rate": 5e-06,
"loss": 0.456,
"step": 5040
},
{
"epoch": 2.266098272380525,
"grad_norm": 0.5262218825649722,
"learning_rate": 5e-06,
"loss": 0.4398,
"step": 5050
},
{
"epoch": 2.2705855956921694,
"grad_norm": 0.5644024837435924,
"learning_rate": 5e-06,
"loss": 0.4144,
"step": 5060
},
{
"epoch": 2.275072919003814,
"grad_norm": 0.5830728870184229,
"learning_rate": 5e-06,
"loss": 0.4433,
"step": 5070
},
{
"epoch": 2.279560242315459,
"grad_norm": 0.5493777480947901,
"learning_rate": 5e-06,
"loss": 0.4606,
"step": 5080
},
{
"epoch": 2.2840475656271035,
"grad_norm": 0.5488762074369681,
"learning_rate": 5e-06,
"loss": 0.4237,
"step": 5090
},
{
"epoch": 2.288534888938748,
"grad_norm": 0.5265349880885405,
"learning_rate": 5e-06,
"loss": 0.426,
"step": 5100
},
{
"epoch": 2.2930222122503925,
"grad_norm": 0.5478130062869077,
"learning_rate": 5e-06,
"loss": 0.4451,
"step": 5110
},
{
"epoch": 2.297509535562037,
"grad_norm": 0.5167579957148292,
"learning_rate": 5e-06,
"loss": 0.4161,
"step": 5120
},
{
"epoch": 2.301996858873682,
"grad_norm": 0.537587588073114,
"learning_rate": 5e-06,
"loss": 0.4269,
"step": 5130
},
{
"epoch": 2.3064841821853266,
"grad_norm": 0.5351966570168155,
"learning_rate": 5e-06,
"loss": 0.4606,
"step": 5140
},
{
"epoch": 2.310971505496971,
"grad_norm": 0.5542030553459122,
"learning_rate": 5e-06,
"loss": 0.4325,
"step": 5150
},
{
"epoch": 2.3154588288086155,
"grad_norm": 0.5657224761819103,
"learning_rate": 5e-06,
"loss": 0.436,
"step": 5160
},
{
"epoch": 2.3199461521202602,
"grad_norm": 0.5189966523850335,
"learning_rate": 5e-06,
"loss": 0.4271,
"step": 5170
},
{
"epoch": 2.324433475431905,
"grad_norm": 0.5490822174982533,
"learning_rate": 5e-06,
"loss": 0.4495,
"step": 5180
},
{
"epoch": 2.3289207987435496,
"grad_norm": 0.5495336668696191,
"learning_rate": 5e-06,
"loss": 0.4184,
"step": 5190
},
{
"epoch": 2.333408122055194,
"grad_norm": 0.5404149701496175,
"learning_rate": 5e-06,
"loss": 0.4536,
"step": 5200
},
{
"epoch": 2.3378954453668386,
"grad_norm": 0.5495815706921072,
"learning_rate": 5e-06,
"loss": 0.4247,
"step": 5210
},
{
"epoch": 2.3423827686784833,
"grad_norm": 0.5223528149857176,
"learning_rate": 5e-06,
"loss": 0.446,
"step": 5220
},
{
"epoch": 2.346870091990128,
"grad_norm": 0.5320050485503626,
"learning_rate": 5e-06,
"loss": 0.4332,
"step": 5230
},
{
"epoch": 2.3513574153017727,
"grad_norm": 0.5299871177712144,
"learning_rate": 5e-06,
"loss": 0.4639,
"step": 5240
},
{
"epoch": 2.355844738613417,
"grad_norm": 0.5205906167014355,
"learning_rate": 5e-06,
"loss": 0.4363,
"step": 5250
},
{
"epoch": 2.3603320619250616,
"grad_norm": 0.5176082572897545,
"learning_rate": 5e-06,
"loss": 0.4216,
"step": 5260
},
{
"epoch": 2.3648193852367063,
"grad_norm": 0.5342565079745604,
"learning_rate": 5e-06,
"loss": 0.4338,
"step": 5270
},
{
"epoch": 2.369306708548351,
"grad_norm": 0.5803420836992969,
"learning_rate": 5e-06,
"loss": 0.4239,
"step": 5280
},
{
"epoch": 2.3737940318599957,
"grad_norm": 0.5287302044926194,
"learning_rate": 5e-06,
"loss": 0.4475,
"step": 5290
},
{
"epoch": 2.37828135517164,
"grad_norm": 0.5645792638215165,
"learning_rate": 5e-06,
"loss": 0.4392,
"step": 5300
},
{
"epoch": 2.3827686784832847,
"grad_norm": 0.5282020656735624,
"learning_rate": 5e-06,
"loss": 0.4245,
"step": 5310
},
{
"epoch": 2.3872560017949294,
"grad_norm": 0.5571286168651558,
"learning_rate": 5e-06,
"loss": 0.4358,
"step": 5320
},
{
"epoch": 2.391743325106574,
"grad_norm": 0.5551698031627329,
"learning_rate": 5e-06,
"loss": 0.4369,
"step": 5330
},
{
"epoch": 2.3962306484182188,
"grad_norm": 0.5387097738562745,
"learning_rate": 5e-06,
"loss": 0.4524,
"step": 5340
},
{
"epoch": 2.400717971729863,
"grad_norm": 0.546173575982491,
"learning_rate": 5e-06,
"loss": 0.4625,
"step": 5350
},
{
"epoch": 2.4052052950415077,
"grad_norm": 0.5495376681388755,
"learning_rate": 5e-06,
"loss": 0.4434,
"step": 5360
},
{
"epoch": 2.4096926183531524,
"grad_norm": 0.5377972436367183,
"learning_rate": 5e-06,
"loss": 0.4217,
"step": 5370
},
{
"epoch": 2.414179941664797,
"grad_norm": 0.5350882677408955,
"learning_rate": 5e-06,
"loss": 0.4487,
"step": 5380
},
{
"epoch": 2.418667264976442,
"grad_norm": 0.5517707610352112,
"learning_rate": 5e-06,
"loss": 0.4352,
"step": 5390
},
{
"epoch": 2.423154588288086,
"grad_norm": 0.567154096040849,
"learning_rate": 5e-06,
"loss": 0.4288,
"step": 5400
},
{
"epoch": 2.4276419115997308,
"grad_norm": 0.5373918223412871,
"learning_rate": 5e-06,
"loss": 0.4741,
"step": 5410
},
{
"epoch": 2.4321292349113754,
"grad_norm": 0.5327337280014666,
"learning_rate": 5e-06,
"loss": 0.4692,
"step": 5420
},
{
"epoch": 2.43661655822302,
"grad_norm": 0.5629600789156937,
"learning_rate": 5e-06,
"loss": 0.4508,
"step": 5430
},
{
"epoch": 2.4411038815346644,
"grad_norm": 0.5536985413385566,
"learning_rate": 5e-06,
"loss": 0.4275,
"step": 5440
},
{
"epoch": 2.445591204846309,
"grad_norm": 0.5312387011564881,
"learning_rate": 5e-06,
"loss": 0.448,
"step": 5450
},
{
"epoch": 2.450078528157954,
"grad_norm": 0.5538065322301513,
"learning_rate": 5e-06,
"loss": 0.4346,
"step": 5460
},
{
"epoch": 2.4545658514695985,
"grad_norm": 0.541885849026688,
"learning_rate": 5e-06,
"loss": 0.4399,
"step": 5470
},
{
"epoch": 2.4590531747812427,
"grad_norm": 0.5284552961380801,
"learning_rate": 5e-06,
"loss": 0.4181,
"step": 5480
},
{
"epoch": 2.4635404980928874,
"grad_norm": 0.5805304303114005,
"learning_rate": 5e-06,
"loss": 0.4585,
"step": 5490
},
{
"epoch": 2.468027821404532,
"grad_norm": 0.5451696084898805,
"learning_rate": 5e-06,
"loss": 0.457,
"step": 5500
},
{
"epoch": 2.472515144716177,
"grad_norm": 0.5556157590446383,
"learning_rate": 5e-06,
"loss": 0.4385,
"step": 5510
},
{
"epoch": 2.4770024680278215,
"grad_norm": 0.5158180051239353,
"learning_rate": 5e-06,
"loss": 0.4827,
"step": 5520
},
{
"epoch": 2.481489791339466,
"grad_norm": 0.5383037166066315,
"learning_rate": 5e-06,
"loss": 0.4524,
"step": 5530
},
{
"epoch": 2.4859771146511105,
"grad_norm": 0.5459360505156834,
"learning_rate": 5e-06,
"loss": 0.4366,
"step": 5540
},
{
"epoch": 2.490464437962755,
"grad_norm": 0.56460798618322,
"learning_rate": 5e-06,
"loss": 0.4487,
"step": 5550
},
{
"epoch": 2.4949517612744,
"grad_norm": 0.5242645705325473,
"learning_rate": 5e-06,
"loss": 0.4318,
"step": 5560
},
{
"epoch": 2.4994390845860446,
"grad_norm": 0.547149907690897,
"learning_rate": 5e-06,
"loss": 0.4362,
"step": 5570
},
{
"epoch": 2.503926407897689,
"grad_norm": 0.5402748092495164,
"learning_rate": 5e-06,
"loss": 0.4623,
"step": 5580
},
{
"epoch": 2.5084137312093335,
"grad_norm": 0.5222621859452697,
"learning_rate": 5e-06,
"loss": 0.4223,
"step": 5590
},
{
"epoch": 2.5129010545209782,
"grad_norm": 0.5310521022656224,
"learning_rate": 5e-06,
"loss": 0.4371,
"step": 5600
},
{
"epoch": 2.517388377832623,
"grad_norm": 0.5335126989409179,
"learning_rate": 5e-06,
"loss": 0.4398,
"step": 5610
},
{
"epoch": 2.5218757011442676,
"grad_norm": 0.5359705116070118,
"learning_rate": 5e-06,
"loss": 0.4397,
"step": 5620
},
{
"epoch": 2.526363024455912,
"grad_norm": 0.5489052783275447,
"learning_rate": 5e-06,
"loss": 0.4542,
"step": 5630
},
{
"epoch": 2.5308503477675566,
"grad_norm": 0.5470901455508153,
"learning_rate": 5e-06,
"loss": 0.4277,
"step": 5640
},
{
"epoch": 2.5353376710792013,
"grad_norm": 0.5346733284005247,
"learning_rate": 5e-06,
"loss": 0.4438,
"step": 5650
},
{
"epoch": 2.539824994390846,
"grad_norm": 0.5164803047551324,
"learning_rate": 5e-06,
"loss": 0.4477,
"step": 5660
},
{
"epoch": 2.5443123177024907,
"grad_norm": 0.5566837513995293,
"learning_rate": 5e-06,
"loss": 0.4513,
"step": 5670
},
{
"epoch": 2.548799641014135,
"grad_norm": 0.5699497948415865,
"learning_rate": 5e-06,
"loss": 0.4162,
"step": 5680
},
{
"epoch": 2.5532869643257796,
"grad_norm": 0.52545278326248,
"learning_rate": 5e-06,
"loss": 0.4422,
"step": 5690
},
{
"epoch": 2.5577742876374243,
"grad_norm": 0.5158599480336954,
"learning_rate": 5e-06,
"loss": 0.4508,
"step": 5700
},
{
"epoch": 2.562261610949069,
"grad_norm": 0.5517542572393412,
"learning_rate": 5e-06,
"loss": 0.4393,
"step": 5710
},
{
"epoch": 2.5667489342607137,
"grad_norm": 0.5405823805282446,
"learning_rate": 5e-06,
"loss": 0.4336,
"step": 5720
},
{
"epoch": 2.571236257572358,
"grad_norm": 0.5173834422219281,
"learning_rate": 5e-06,
"loss": 0.4273,
"step": 5730
},
{
"epoch": 2.5757235808840027,
"grad_norm": 0.5400412377561089,
"learning_rate": 5e-06,
"loss": 0.4468,
"step": 5740
},
{
"epoch": 2.5802109041956474,
"grad_norm": 0.5337516297281755,
"learning_rate": 5e-06,
"loss": 0.4474,
"step": 5750
},
{
"epoch": 2.584698227507292,
"grad_norm": 0.5527299364440325,
"learning_rate": 5e-06,
"loss": 0.4345,
"step": 5760
},
{
"epoch": 2.5891855508189368,
"grad_norm": 0.5315767138582382,
"learning_rate": 5e-06,
"loss": 0.4311,
"step": 5770
},
{
"epoch": 2.593672874130581,
"grad_norm": 0.5515892148242644,
"learning_rate": 5e-06,
"loss": 0.4534,
"step": 5780
},
{
"epoch": 2.5981601974422257,
"grad_norm": 0.5710350598486404,
"learning_rate": 5e-06,
"loss": 0.451,
"step": 5790
},
{
"epoch": 2.6026475207538704,
"grad_norm": 0.5592453853735161,
"learning_rate": 5e-06,
"loss": 0.4274,
"step": 5800
},
{
"epoch": 2.6071348440655147,
"grad_norm": 0.5791874292264974,
"learning_rate": 5e-06,
"loss": 0.4361,
"step": 5810
},
{
"epoch": 2.61162216737716,
"grad_norm": 0.5459606205708067,
"learning_rate": 5e-06,
"loss": 0.4344,
"step": 5820
},
{
"epoch": 2.616109490688804,
"grad_norm": 0.5263598745373247,
"learning_rate": 5e-06,
"loss": 0.443,
"step": 5830
},
{
"epoch": 2.6205968140004487,
"grad_norm": 0.5318408035767697,
"learning_rate": 5e-06,
"loss": 0.455,
"step": 5840
},
{
"epoch": 2.6250841373120934,
"grad_norm": 0.5621927236582933,
"learning_rate": 5e-06,
"loss": 0.4338,
"step": 5850
},
{
"epoch": 2.6295714606237377,
"grad_norm": 0.5443265572382032,
"learning_rate": 5e-06,
"loss": 0.458,
"step": 5860
},
{
"epoch": 2.6340587839353824,
"grad_norm": 0.5291943252034395,
"learning_rate": 5e-06,
"loss": 0.4524,
"step": 5870
},
{
"epoch": 2.638546107247027,
"grad_norm": 0.5344516122229699,
"learning_rate": 5e-06,
"loss": 0.4399,
"step": 5880
},
{
"epoch": 2.643033430558672,
"grad_norm": 0.5148344986062239,
"learning_rate": 5e-06,
"loss": 0.4385,
"step": 5890
},
{
"epoch": 2.6475207538703165,
"grad_norm": 0.517635318320743,
"learning_rate": 5e-06,
"loss": 0.4601,
"step": 5900
},
{
"epoch": 2.6520080771819607,
"grad_norm": 0.5517916805298932,
"learning_rate": 5e-06,
"loss": 0.4323,
"step": 5910
},
{
"epoch": 2.6564954004936054,
"grad_norm": 0.5611554225539297,
"learning_rate": 5e-06,
"loss": 0.4479,
"step": 5920
},
{
"epoch": 2.66098272380525,
"grad_norm": 0.5312102185907968,
"learning_rate": 5e-06,
"loss": 0.4518,
"step": 5930
},
{
"epoch": 2.665470047116895,
"grad_norm": 0.5492020012649198,
"learning_rate": 5e-06,
"loss": 0.4424,
"step": 5940
},
{
"epoch": 2.6699573704285395,
"grad_norm": 0.5253538068318323,
"learning_rate": 5e-06,
"loss": 0.4481,
"step": 5950
},
{
"epoch": 2.674444693740184,
"grad_norm": 0.5433459391246607,
"learning_rate": 5e-06,
"loss": 0.4438,
"step": 5960
},
{
"epoch": 2.6789320170518285,
"grad_norm": 0.5250296745067685,
"learning_rate": 5e-06,
"loss": 0.4417,
"step": 5970
},
{
"epoch": 2.683419340363473,
"grad_norm": 0.57845411310223,
"learning_rate": 5e-06,
"loss": 0.462,
"step": 5980
},
{
"epoch": 2.687906663675118,
"grad_norm": 0.5316069733603933,
"learning_rate": 5e-06,
"loss": 0.4385,
"step": 5990
},
{
"epoch": 2.6923939869867626,
"grad_norm": 0.5309938089529066,
"learning_rate": 5e-06,
"loss": 0.4419,
"step": 6000
},
{
"epoch": 2.696881310298407,
"grad_norm": 0.5562625161020116,
"learning_rate": 5e-06,
"loss": 0.435,
"step": 6010
},
{
"epoch": 2.7013686336100515,
"grad_norm": 0.5599859836426428,
"learning_rate": 5e-06,
"loss": 0.4324,
"step": 6020
},
{
"epoch": 2.7058559569216962,
"grad_norm": 0.5688105639070913,
"learning_rate": 5e-06,
"loss": 0.4344,
"step": 6030
},
{
"epoch": 2.710343280233341,
"grad_norm": 0.5626537832912196,
"learning_rate": 5e-06,
"loss": 0.4373,
"step": 6040
},
{
"epoch": 2.7148306035449856,
"grad_norm": 0.55324964941932,
"learning_rate": 5e-06,
"loss": 0.4317,
"step": 6050
},
{
"epoch": 2.71931792685663,
"grad_norm": 0.5401719719418708,
"learning_rate": 5e-06,
"loss": 0.4494,
"step": 6060
},
{
"epoch": 2.7238052501682746,
"grad_norm": 0.5150225135697268,
"learning_rate": 5e-06,
"loss": 0.4539,
"step": 6070
},
{
"epoch": 2.7282925734799193,
"grad_norm": 0.5740983627117513,
"learning_rate": 5e-06,
"loss": 0.4391,
"step": 6080
},
{
"epoch": 2.732779896791564,
"grad_norm": 0.5398182644675611,
"learning_rate": 5e-06,
"loss": 0.449,
"step": 6090
},
{
"epoch": 2.7372672201032087,
"grad_norm": 0.5443362706259044,
"learning_rate": 5e-06,
"loss": 0.4301,
"step": 6100
},
{
"epoch": 2.741754543414853,
"grad_norm": 0.5455257679601512,
"learning_rate": 5e-06,
"loss": 0.4226,
"step": 6110
},
{
"epoch": 2.7462418667264976,
"grad_norm": 0.5260931529295294,
"learning_rate": 5e-06,
"loss": 0.4512,
"step": 6120
},
{
"epoch": 2.7507291900381423,
"grad_norm": 0.5472643003685588,
"learning_rate": 5e-06,
"loss": 0.4434,
"step": 6130
},
{
"epoch": 2.755216513349787,
"grad_norm": 0.5579861053588995,
"learning_rate": 5e-06,
"loss": 0.4405,
"step": 6140
},
{
"epoch": 2.7597038366614317,
"grad_norm": 0.545878751971252,
"learning_rate": 5e-06,
"loss": 0.4366,
"step": 6150
},
{
"epoch": 2.764191159973076,
"grad_norm": 0.532094976762934,
"learning_rate": 5e-06,
"loss": 0.4414,
"step": 6160
},
{
"epoch": 2.7686784832847207,
"grad_norm": 0.5836755224967016,
"learning_rate": 5e-06,
"loss": 0.4534,
"step": 6170
},
{
"epoch": 2.7731658065963654,
"grad_norm": 0.5342141820947959,
"learning_rate": 5e-06,
"loss": 0.4488,
"step": 6180
},
{
"epoch": 2.7776531299080096,
"grad_norm": 0.565521800045049,
"learning_rate": 5e-06,
"loss": 0.4331,
"step": 6190
},
{
"epoch": 2.7821404532196548,
"grad_norm": 0.5230804833062074,
"learning_rate": 5e-06,
"loss": 0.442,
"step": 6200
},
{
"epoch": 2.786627776531299,
"grad_norm": 0.5470772975018182,
"learning_rate": 5e-06,
"loss": 0.4439,
"step": 6210
},
{
"epoch": 2.7911150998429437,
"grad_norm": 0.5450425972085206,
"learning_rate": 5e-06,
"loss": 0.4471,
"step": 6220
},
{
"epoch": 2.7956024231545884,
"grad_norm": 0.5083879580994983,
"learning_rate": 5e-06,
"loss": 0.45,
"step": 6230
},
{
"epoch": 2.8000897464662327,
"grad_norm": 0.538016313129477,
"learning_rate": 5e-06,
"loss": 0.4369,
"step": 6240
},
{
"epoch": 2.8045770697778774,
"grad_norm": 0.5582050739389205,
"learning_rate": 5e-06,
"loss": 0.4763,
"step": 6250
},
{
"epoch": 2.809064393089522,
"grad_norm": 0.5349488919272378,
"learning_rate": 5e-06,
"loss": 0.4363,
"step": 6260
},
{
"epoch": 2.8135517164011667,
"grad_norm": 0.5264389993349703,
"learning_rate": 5e-06,
"loss": 0.438,
"step": 6270
},
{
"epoch": 2.8180390397128114,
"grad_norm": 0.5498499847639682,
"learning_rate": 5e-06,
"loss": 0.4295,
"step": 6280
},
{
"epoch": 2.8225263630244557,
"grad_norm": 0.5449602025844419,
"learning_rate": 5e-06,
"loss": 0.448,
"step": 6290
},
{
"epoch": 2.8270136863361004,
"grad_norm": 0.5396372393247543,
"learning_rate": 5e-06,
"loss": 0.4601,
"step": 6300
},
{
"epoch": 2.831501009647745,
"grad_norm": 0.5285660625062669,
"learning_rate": 5e-06,
"loss": 0.4256,
"step": 6310
},
{
"epoch": 2.83598833295939,
"grad_norm": 0.5338101430481593,
"learning_rate": 5e-06,
"loss": 0.4461,
"step": 6320
},
{
"epoch": 2.8404756562710345,
"grad_norm": 0.5464049325628102,
"learning_rate": 5e-06,
"loss": 0.4312,
"step": 6330
},
{
"epoch": 2.8449629795826787,
"grad_norm": 0.5462792499593093,
"learning_rate": 5e-06,
"loss": 0.4418,
"step": 6340
},
{
"epoch": 2.8494503028943234,
"grad_norm": 0.5516791871027837,
"learning_rate": 5e-06,
"loss": 0.4449,
"step": 6350
},
{
"epoch": 2.853937626205968,
"grad_norm": 0.5576513237004156,
"learning_rate": 5e-06,
"loss": 0.4453,
"step": 6360
},
{
"epoch": 2.858424949517613,
"grad_norm": 0.5368302048584942,
"learning_rate": 5e-06,
"loss": 0.4418,
"step": 6370
},
{
"epoch": 2.8629122728292575,
"grad_norm": 0.5166792247647458,
"learning_rate": 5e-06,
"loss": 0.4486,
"step": 6380
},
{
"epoch": 2.867399596140902,
"grad_norm": 0.5283542555206898,
"learning_rate": 5e-06,
"loss": 0.4305,
"step": 6390
},
{
"epoch": 2.8718869194525465,
"grad_norm": 0.5745921221883099,
"learning_rate": 5e-06,
"loss": 0.4421,
"step": 6400
},
{
"epoch": 2.876374242764191,
"grad_norm": 0.5610874860309188,
"learning_rate": 5e-06,
"loss": 0.4482,
"step": 6410
},
{
"epoch": 2.880861566075836,
"grad_norm": 0.562508114298865,
"learning_rate": 5e-06,
"loss": 0.4357,
"step": 6420
},
{
"epoch": 2.8853488893874806,
"grad_norm": 0.5404474497571172,
"learning_rate": 5e-06,
"loss": 0.4901,
"step": 6430
},
{
"epoch": 2.889836212699125,
"grad_norm": 0.5731369450030108,
"learning_rate": 5e-06,
"loss": 0.4523,
"step": 6440
},
{
"epoch": 2.8943235360107695,
"grad_norm": 0.5283239685292113,
"learning_rate": 5e-06,
"loss": 0.445,
"step": 6450
},
{
"epoch": 2.8988108593224142,
"grad_norm": 0.555347180047328,
"learning_rate": 5e-06,
"loss": 0.4396,
"step": 6460
},
{
"epoch": 2.903298182634059,
"grad_norm": 0.5337754968953309,
"learning_rate": 5e-06,
"loss": 0.4659,
"step": 6470
},
{
"epoch": 2.9077855059457036,
"grad_norm": 0.5482307100875392,
"learning_rate": 5e-06,
"loss": 0.4386,
"step": 6480
},
{
"epoch": 2.912272829257348,
"grad_norm": 0.5137780447940196,
"learning_rate": 5e-06,
"loss": 0.4498,
"step": 6490
},
{
"epoch": 2.9167601525689926,
"grad_norm": 0.5410402939710804,
"learning_rate": 5e-06,
"loss": 0.4447,
"step": 6500
},
{
"epoch": 2.9212474758806373,
"grad_norm": 0.5512710516362269,
"learning_rate": 5e-06,
"loss": 0.4415,
"step": 6510
},
{
"epoch": 2.925734799192282,
"grad_norm": 0.5378334272814395,
"learning_rate": 5e-06,
"loss": 0.4278,
"step": 6520
},
{
"epoch": 2.9302221225039267,
"grad_norm": 0.5525646027445259,
"learning_rate": 5e-06,
"loss": 0.439,
"step": 6530
},
{
"epoch": 2.934709445815571,
"grad_norm": 0.573171926611343,
"learning_rate": 5e-06,
"loss": 0.4361,
"step": 6540
},
{
"epoch": 2.9391967691272156,
"grad_norm": 0.5217333672218816,
"learning_rate": 5e-06,
"loss": 0.4621,
"step": 6550
},
{
"epoch": 2.9436840924388603,
"grad_norm": 0.5342154460861862,
"learning_rate": 5e-06,
"loss": 0.4646,
"step": 6560
},
{
"epoch": 2.9481714157505046,
"grad_norm": 0.5545259487380719,
"learning_rate": 5e-06,
"loss": 0.4658,
"step": 6570
},
{
"epoch": 2.9526587390621493,
"grad_norm": 0.5469560259016665,
"learning_rate": 5e-06,
"loss": 0.455,
"step": 6580
},
{
"epoch": 2.957146062373794,
"grad_norm": 0.5483835668636673,
"learning_rate": 5e-06,
"loss": 0.4502,
"step": 6590
},
{
"epoch": 2.9616333856854387,
"grad_norm": 0.614582634766974,
"learning_rate": 5e-06,
"loss": 0.4846,
"step": 6600
},
{
"epoch": 2.9661207089970834,
"grad_norm": 0.5262842688899035,
"learning_rate": 5e-06,
"loss": 0.4374,
"step": 6610
},
{
"epoch": 2.9706080323087276,
"grad_norm": 0.5571152950087439,
"learning_rate": 5e-06,
"loss": 0.4568,
"step": 6620
},
{
"epoch": 2.9750953556203723,
"grad_norm": 0.5654247741253923,
"learning_rate": 5e-06,
"loss": 0.4543,
"step": 6630
},
{
"epoch": 2.979582678932017,
"grad_norm": 0.5557134515495112,
"learning_rate": 5e-06,
"loss": 0.4413,
"step": 6640
},
{
"epoch": 2.9840700022436617,
"grad_norm": 0.5100342280894994,
"learning_rate": 5e-06,
"loss": 0.4297,
"step": 6650
},
{
"epoch": 2.9885573255553064,
"grad_norm": 0.5717765932430845,
"learning_rate": 5e-06,
"loss": 0.4274,
"step": 6660
},
{
"epoch": 2.9930446488669507,
"grad_norm": 0.5179292681178803,
"learning_rate": 5e-06,
"loss": 0.4425,
"step": 6670
},
{
"epoch": 2.9975319721785953,
"grad_norm": 0.5427917144920164,
"learning_rate": 5e-06,
"loss": 0.4538,
"step": 6680
},
{
"epoch": 2.9993269015032533,
"eval_loss": 0.5616346001625061,
"eval_runtime": 559.8562,
"eval_samples_per_second": 26.81,
"eval_steps_per_second": 0.42,
"step": 6684
},
{
"epoch": 2.9993269015032533,
"step": 6684,
"total_flos": 3503854586953728.0,
"train_loss": 0.5125751896649332,
"train_runtime": 89473.4361,
"train_samples_per_second": 9.562,
"train_steps_per_second": 0.075
}
],
"logging_steps": 10,
"max_steps": 6684,
"num_input_tokens_seen": 0,
"num_train_epochs": 3,
"save_steps": 500,
"stateful_callbacks": {
"TrainerControl": {
"args": {
"should_epoch_stop": false,
"should_evaluate": false,
"should_log": false,
"should_save": true,
"should_training_stop": true
},
"attributes": {}
}
},
"total_flos": 3503854586953728.0,
"train_batch_size": 8,
"trial_name": null,
"trial_params": null
}