1035 lines
27 KiB
JSON
1035 lines
27 KiB
JSON
{
|
|
"best_global_step": null,
|
|
"best_metric": null,
|
|
"best_model_checkpoint": null,
|
|
"epoch": 0.032988058322887114,
|
|
"eval_steps": 500,
|
|
"global_step": 500,
|
|
"is_hyper_param_search": false,
|
|
"is_local_process_zero": true,
|
|
"is_world_process_zero": true,
|
|
"log_history": [
|
|
{
|
|
"entropy": 10.691983032226563,
|
|
"epoch": 0.00032988058322887113,
|
|
"grad_norm": 12.625,
|
|
"learning_rate": 2e-06,
|
|
"loss": 10.8364,
|
|
"mean_token_accuracy": 7.923930534161627e-05,
|
|
"num_tokens": 11020.0,
|
|
"step": 5
|
|
},
|
|
{
|
|
"entropy": 10.691853046417236,
|
|
"epoch": 0.0006597611664577423,
|
|
"grad_norm": 13.0625,
|
|
"learning_rate": 4.5e-06,
|
|
"loss": 10.7946,
|
|
"mean_token_accuracy": 0.0,
|
|
"num_tokens": 23411.0,
|
|
"step": 10
|
|
},
|
|
{
|
|
"entropy": 10.691090774536132,
|
|
"epoch": 0.0009896417496866134,
|
|
"grad_norm": 8.75,
|
|
"learning_rate": 7e-06,
|
|
"loss": 10.5709,
|
|
"mean_token_accuracy": 0.019937679241411387,
|
|
"num_tokens": 35592.0,
|
|
"step": 15
|
|
},
|
|
{
|
|
"entropy": 10.680480861663819,
|
|
"epoch": 0.0013195223329154845,
|
|
"grad_norm": 5.90625,
|
|
"learning_rate": 9.5e-06,
|
|
"loss": 10.2415,
|
|
"mean_token_accuracy": 0.038927487842738626,
|
|
"num_tokens": 46417.0,
|
|
"step": 20
|
|
},
|
|
{
|
|
"entropy": 10.63813009262085,
|
|
"epoch": 0.0016494029161443558,
|
|
"grad_norm": 3.96875,
|
|
"learning_rate": 1.2e-05,
|
|
"loss": 9.9858,
|
|
"mean_token_accuracy": 0.035798029974102974,
|
|
"num_tokens": 57261.0,
|
|
"step": 25
|
|
},
|
|
{
|
|
"entropy": 10.610331535339355,
|
|
"epoch": 0.001979283499373227,
|
|
"grad_norm": 3.15625,
|
|
"learning_rate": 1.4500000000000002e-05,
|
|
"loss": 9.8202,
|
|
"mean_token_accuracy": 0.04165456015616655,
|
|
"num_tokens": 68120.0,
|
|
"step": 30
|
|
},
|
|
{
|
|
"entropy": 10.573247146606445,
|
|
"epoch": 0.002309164082602098,
|
|
"grad_norm": 2.875,
|
|
"learning_rate": 1.7000000000000003e-05,
|
|
"loss": 9.7586,
|
|
"mean_token_accuracy": 0.03941070232540369,
|
|
"num_tokens": 80384.0,
|
|
"step": 35
|
|
},
|
|
{
|
|
"entropy": 10.594020080566406,
|
|
"epoch": 0.002639044665830969,
|
|
"grad_norm": 2.390625,
|
|
"learning_rate": 1.95e-05,
|
|
"loss": 9.7087,
|
|
"mean_token_accuracy": 0.04071928896009922,
|
|
"num_tokens": 90416.0,
|
|
"step": 40
|
|
},
|
|
{
|
|
"entropy": 10.597117710113526,
|
|
"epoch": 0.0029689252490598403,
|
|
"grad_norm": 2.796875,
|
|
"learning_rate": 2.2e-05,
|
|
"loss": 9.6704,
|
|
"mean_token_accuracy": 0.037894781678915024,
|
|
"num_tokens": 101944.0,
|
|
"step": 45
|
|
},
|
|
{
|
|
"entropy": 10.580680561065673,
|
|
"epoch": 0.0032988058322887116,
|
|
"grad_norm": 2.234375,
|
|
"learning_rate": 2.4500000000000003e-05,
|
|
"loss": 9.6271,
|
|
"mean_token_accuracy": 0.042053837701678275,
|
|
"num_tokens": 113439.0,
|
|
"step": 50
|
|
},
|
|
{
|
|
"entropy": 10.554423809051514,
|
|
"epoch": 0.0036286864155175825,
|
|
"grad_norm": 2.15625,
|
|
"learning_rate": 2.7e-05,
|
|
"loss": 9.5826,
|
|
"mean_token_accuracy": 0.039308933168649675,
|
|
"num_tokens": 124817.0,
|
|
"step": 55
|
|
},
|
|
{
|
|
"entropy": 10.562382888793945,
|
|
"epoch": 0.003958566998746454,
|
|
"grad_norm": 2.171875,
|
|
"learning_rate": 2.95e-05,
|
|
"loss": 9.5126,
|
|
"mean_token_accuracy": 0.03932496458292008,
|
|
"num_tokens": 136784.0,
|
|
"step": 60
|
|
},
|
|
{
|
|
"entropy": 10.557148647308349,
|
|
"epoch": 0.004288447581975325,
|
|
"grad_norm": 2.328125,
|
|
"learning_rate": 3.2e-05,
|
|
"loss": 9.4286,
|
|
"mean_token_accuracy": 0.04576659500598908,
|
|
"num_tokens": 148412.0,
|
|
"step": 65
|
|
},
|
|
{
|
|
"entropy": 10.451301574707031,
|
|
"epoch": 0.004618328165204196,
|
|
"grad_norm": 2.15625,
|
|
"learning_rate": 3.4500000000000005e-05,
|
|
"loss": 9.3304,
|
|
"mean_token_accuracy": 0.04561230204999447,
|
|
"num_tokens": 159001.0,
|
|
"step": 70
|
|
},
|
|
{
|
|
"entropy": 10.413534927368165,
|
|
"epoch": 0.004948208748433067,
|
|
"grad_norm": 2.125,
|
|
"learning_rate": 3.7e-05,
|
|
"loss": 9.2421,
|
|
"mean_token_accuracy": 0.046519069373607634,
|
|
"num_tokens": 170301.0,
|
|
"step": 75
|
|
},
|
|
{
|
|
"entropy": 10.452965450286865,
|
|
"epoch": 0.005278089331661938,
|
|
"grad_norm": 1.921875,
|
|
"learning_rate": 3.95e-05,
|
|
"loss": 9.1493,
|
|
"mean_token_accuracy": 0.04921787865459919,
|
|
"num_tokens": 180966.0,
|
|
"step": 80
|
|
},
|
|
{
|
|
"entropy": 10.412489986419677,
|
|
"epoch": 0.00560796991489081,
|
|
"grad_norm": 2.09375,
|
|
"learning_rate": 4.2000000000000004e-05,
|
|
"loss": 9.095,
|
|
"mean_token_accuracy": 0.04874967411160469,
|
|
"num_tokens": 192843.0,
|
|
"step": 85
|
|
},
|
|
{
|
|
"entropy": 10.429522895812989,
|
|
"epoch": 0.005937850498119681,
|
|
"grad_norm": 2.28125,
|
|
"learning_rate": 4.45e-05,
|
|
"loss": 8.9693,
|
|
"mean_token_accuracy": 0.054649117216467855,
|
|
"num_tokens": 203167.0,
|
|
"step": 90
|
|
},
|
|
{
|
|
"entropy": 10.29054889678955,
|
|
"epoch": 0.0062677310813485515,
|
|
"grad_norm": 2.03125,
|
|
"learning_rate": 4.7000000000000004e-05,
|
|
"loss": 8.856,
|
|
"mean_token_accuracy": 0.05385527983307838,
|
|
"num_tokens": 214704.0,
|
|
"step": 95
|
|
},
|
|
{
|
|
"entropy": 10.288533592224121,
|
|
"epoch": 0.006597611664577423,
|
|
"grad_norm": 2.03125,
|
|
"learning_rate": 4.9500000000000004e-05,
|
|
"loss": 8.7532,
|
|
"mean_token_accuracy": 0.05811888948082924,
|
|
"num_tokens": 224848.0,
|
|
"step": 100
|
|
},
|
|
{
|
|
"entropy": 10.19245138168335,
|
|
"epoch": 0.006927492247806294,
|
|
"grad_norm": 1.7734375,
|
|
"learning_rate": 5.2e-05,
|
|
"loss": 8.6653,
|
|
"mean_token_accuracy": 0.058622144907712934,
|
|
"num_tokens": 236213.0,
|
|
"step": 105
|
|
},
|
|
{
|
|
"entropy": 10.099221515655518,
|
|
"epoch": 0.007257372831035165,
|
|
"grad_norm": 2.21875,
|
|
"learning_rate": 5.45e-05,
|
|
"loss": 8.4791,
|
|
"mean_token_accuracy": 0.06761969737708569,
|
|
"num_tokens": 246918.0,
|
|
"step": 110
|
|
},
|
|
{
|
|
"entropy": 10.06296901702881,
|
|
"epoch": 0.007587253414264037,
|
|
"grad_norm": 2.140625,
|
|
"learning_rate": 5.7e-05,
|
|
"loss": 8.3357,
|
|
"mean_token_accuracy": 0.06780013889074325,
|
|
"num_tokens": 256940.0,
|
|
"step": 115
|
|
},
|
|
{
|
|
"entropy": 9.910451412200928,
|
|
"epoch": 0.007917133997492908,
|
|
"grad_norm": 2.46875,
|
|
"learning_rate": 5.9499999999999996e-05,
|
|
"loss": 8.289,
|
|
"mean_token_accuracy": 0.06589791402220727,
|
|
"num_tokens": 267317.0,
|
|
"step": 120
|
|
},
|
|
{
|
|
"entropy": 9.842613124847412,
|
|
"epoch": 0.008247014580721778,
|
|
"grad_norm": 1.6015625,
|
|
"learning_rate": 6.2e-05,
|
|
"loss": 8.2062,
|
|
"mean_token_accuracy": 0.057442688196897504,
|
|
"num_tokens": 277636.0,
|
|
"step": 125
|
|
},
|
|
{
|
|
"entropy": 9.712352275848389,
|
|
"epoch": 0.00857689516395065,
|
|
"grad_norm": 1.3359375,
|
|
"learning_rate": 6.450000000000001e-05,
|
|
"loss": 8.1248,
|
|
"mean_token_accuracy": 0.06353966109454631,
|
|
"num_tokens": 288733.0,
|
|
"step": 130
|
|
},
|
|
{
|
|
"entropy": 9.581976509094238,
|
|
"epoch": 0.008906775747179522,
|
|
"grad_norm": 1.4609375,
|
|
"learning_rate": 6.7e-05,
|
|
"loss": 8.0115,
|
|
"mean_token_accuracy": 0.06089223399758339,
|
|
"num_tokens": 300288.0,
|
|
"step": 135
|
|
},
|
|
{
|
|
"entropy": 9.340094566345215,
|
|
"epoch": 0.009236656330408393,
|
|
"grad_norm": 1.515625,
|
|
"learning_rate": 6.950000000000001e-05,
|
|
"loss": 7.8688,
|
|
"mean_token_accuracy": 0.0640185870230198,
|
|
"num_tokens": 311399.0,
|
|
"step": 140
|
|
},
|
|
{
|
|
"entropy": 9.135732650756836,
|
|
"epoch": 0.009566536913637264,
|
|
"grad_norm": 1.390625,
|
|
"learning_rate": 7.2e-05,
|
|
"loss": 7.7402,
|
|
"mean_token_accuracy": 0.06681753881275654,
|
|
"num_tokens": 323432.0,
|
|
"step": 145
|
|
},
|
|
{
|
|
"entropy": 8.876615238189697,
|
|
"epoch": 0.009896417496866134,
|
|
"grad_norm": 1.671875,
|
|
"learning_rate": 7.45e-05,
|
|
"loss": 7.6722,
|
|
"mean_token_accuracy": 0.06561163328588009,
|
|
"num_tokens": 334825.0,
|
|
"step": 150
|
|
},
|
|
{
|
|
"entropy": 8.727448081970214,
|
|
"epoch": 0.010226298080095005,
|
|
"grad_norm": 1.265625,
|
|
"learning_rate": 7.7e-05,
|
|
"loss": 7.5926,
|
|
"mean_token_accuracy": 0.06817376501858234,
|
|
"num_tokens": 347005.0,
|
|
"step": 155
|
|
},
|
|
{
|
|
"entropy": 8.559758567810059,
|
|
"epoch": 0.010556178663323876,
|
|
"grad_norm": 1.2734375,
|
|
"learning_rate": 7.950000000000001e-05,
|
|
"loss": 7.62,
|
|
"mean_token_accuracy": 0.06228872612118721,
|
|
"num_tokens": 358362.0,
|
|
"step": 160
|
|
},
|
|
{
|
|
"entropy": 8.359399795532227,
|
|
"epoch": 0.010886059246552749,
|
|
"grad_norm": 1.296875,
|
|
"learning_rate": 8.2e-05,
|
|
"loss": 7.6046,
|
|
"mean_token_accuracy": 0.06165908053517342,
|
|
"num_tokens": 370076.0,
|
|
"step": 165
|
|
},
|
|
{
|
|
"entropy": 8.354181671142578,
|
|
"epoch": 0.01121593982978162,
|
|
"grad_norm": 1.140625,
|
|
"learning_rate": 8.450000000000001e-05,
|
|
"loss": 7.5281,
|
|
"mean_token_accuracy": 0.07050900757312775,
|
|
"num_tokens": 381231.0,
|
|
"step": 170
|
|
},
|
|
{
|
|
"entropy": 8.184219360351562,
|
|
"epoch": 0.01154582041301049,
|
|
"grad_norm": 1.046875,
|
|
"learning_rate": 8.7e-05,
|
|
"loss": 7.4706,
|
|
"mean_token_accuracy": 0.06394450217485428,
|
|
"num_tokens": 392818.0,
|
|
"step": 175
|
|
},
|
|
{
|
|
"entropy": 8.137578392028809,
|
|
"epoch": 0.011875700996239361,
|
|
"grad_norm": 1.1875,
|
|
"learning_rate": 8.95e-05,
|
|
"loss": 7.4081,
|
|
"mean_token_accuracy": 0.07258594520390034,
|
|
"num_tokens": 403013.0,
|
|
"step": 180
|
|
},
|
|
{
|
|
"entropy": 8.080502128601074,
|
|
"epoch": 0.012205581579468232,
|
|
"grad_norm": 1.5,
|
|
"learning_rate": 9.2e-05,
|
|
"loss": 7.3973,
|
|
"mean_token_accuracy": 0.06417961642146111,
|
|
"num_tokens": 415665.0,
|
|
"step": 185
|
|
},
|
|
{
|
|
"entropy": 8.013210487365722,
|
|
"epoch": 0.012535462162697103,
|
|
"grad_norm": 1.03125,
|
|
"learning_rate": 9.45e-05,
|
|
"loss": 7.4584,
|
|
"mean_token_accuracy": 0.06514312215149402,
|
|
"num_tokens": 427732.0,
|
|
"step": 190
|
|
},
|
|
{
|
|
"entropy": 8.03695797920227,
|
|
"epoch": 0.012865342745925976,
|
|
"grad_norm": 1.28125,
|
|
"learning_rate": 9.7e-05,
|
|
"loss": 7.3249,
|
|
"mean_token_accuracy": 0.0695117510855198,
|
|
"num_tokens": 438081.0,
|
|
"step": 195
|
|
},
|
|
{
|
|
"entropy": 7.947143793106079,
|
|
"epoch": 0.013195223329154846,
|
|
"grad_norm": 1.078125,
|
|
"learning_rate": 9.95e-05,
|
|
"loss": 7.3207,
|
|
"mean_token_accuracy": 0.07360398545861244,
|
|
"num_tokens": 449592.0,
|
|
"step": 200
|
|
},
|
|
{
|
|
"entropy": 7.896442890167236,
|
|
"epoch": 0.013525103912383717,
|
|
"grad_norm": 1.265625,
|
|
"learning_rate": 0.000102,
|
|
"loss": 7.3115,
|
|
"mean_token_accuracy": 0.07550386749207974,
|
|
"num_tokens": 459192.0,
|
|
"step": 205
|
|
},
|
|
{
|
|
"entropy": 7.946354150772095,
|
|
"epoch": 0.013854984495612588,
|
|
"grad_norm": 1.109375,
|
|
"learning_rate": 0.00010449999999999999,
|
|
"loss": 7.341,
|
|
"mean_token_accuracy": 0.06514765433967114,
|
|
"num_tokens": 469973.0,
|
|
"step": 210
|
|
},
|
|
{
|
|
"entropy": 7.886367321014404,
|
|
"epoch": 0.014184865078841459,
|
|
"grad_norm": 1.296875,
|
|
"learning_rate": 0.000107,
|
|
"loss": 7.2943,
|
|
"mean_token_accuracy": 0.06789265871047974,
|
|
"num_tokens": 480690.0,
|
|
"step": 215
|
|
},
|
|
{
|
|
"entropy": 7.843042516708374,
|
|
"epoch": 0.01451474566207033,
|
|
"grad_norm": 1.25,
|
|
"learning_rate": 0.0001095,
|
|
"loss": 7.2639,
|
|
"mean_token_accuracy": 0.07212954573333263,
|
|
"num_tokens": 493081.0,
|
|
"step": 220
|
|
},
|
|
{
|
|
"entropy": 7.77789306640625,
|
|
"epoch": 0.014844626245299202,
|
|
"grad_norm": 1.046875,
|
|
"learning_rate": 0.000112,
|
|
"loss": 7.265,
|
|
"mean_token_accuracy": 0.07772576659917832,
|
|
"num_tokens": 504880.0,
|
|
"step": 225
|
|
},
|
|
{
|
|
"entropy": 7.728792381286621,
|
|
"epoch": 0.015174506828528073,
|
|
"grad_norm": 1.328125,
|
|
"learning_rate": 0.0001145,
|
|
"loss": 7.1671,
|
|
"mean_token_accuracy": 0.07903244793415069,
|
|
"num_tokens": 515151.0,
|
|
"step": 230
|
|
},
|
|
{
|
|
"entropy": 7.766470575332642,
|
|
"epoch": 0.015504387411756944,
|
|
"grad_norm": 1.171875,
|
|
"learning_rate": 0.00011700000000000001,
|
|
"loss": 7.2088,
|
|
"mean_token_accuracy": 0.07608162797987461,
|
|
"num_tokens": 526547.0,
|
|
"step": 235
|
|
},
|
|
{
|
|
"entropy": 7.735764694213867,
|
|
"epoch": 0.015834267994985815,
|
|
"grad_norm": 1.203125,
|
|
"learning_rate": 0.00011949999999999999,
|
|
"loss": 7.2032,
|
|
"mean_token_accuracy": 0.0676932182163,
|
|
"num_tokens": 539456.0,
|
|
"step": 240
|
|
},
|
|
{
|
|
"entropy": 7.74909987449646,
|
|
"epoch": 0.016164148578214688,
|
|
"grad_norm": 1.7421875,
|
|
"learning_rate": 0.000122,
|
|
"loss": 7.3264,
|
|
"mean_token_accuracy": 0.0744432620704174,
|
|
"num_tokens": 553097.0,
|
|
"step": 245
|
|
},
|
|
{
|
|
"entropy": 7.753055667877197,
|
|
"epoch": 0.016494029161443557,
|
|
"grad_norm": 1.1796875,
|
|
"learning_rate": 0.0001245,
|
|
"loss": 7.1525,
|
|
"mean_token_accuracy": 0.0776367574930191,
|
|
"num_tokens": 563352.0,
|
|
"step": 250
|
|
},
|
|
{
|
|
"entropy": 7.710342359542847,
|
|
"epoch": 0.01682390974467243,
|
|
"grad_norm": 1.046875,
|
|
"learning_rate": 0.000127,
|
|
"loss": 7.2119,
|
|
"mean_token_accuracy": 0.07701391950249672,
|
|
"num_tokens": 576709.0,
|
|
"step": 255
|
|
},
|
|
{
|
|
"entropy": 7.599351787567139,
|
|
"epoch": 0.0171537903279013,
|
|
"grad_norm": 1.140625,
|
|
"learning_rate": 0.0001295,
|
|
"loss": 7.158,
|
|
"mean_token_accuracy": 0.07675702273845672,
|
|
"num_tokens": 587943.0,
|
|
"step": 260
|
|
},
|
|
{
|
|
"entropy": 7.681706237792969,
|
|
"epoch": 0.01748367091113017,
|
|
"grad_norm": 1.28125,
|
|
"learning_rate": 0.000132,
|
|
"loss": 7.1274,
|
|
"mean_token_accuracy": 0.0756020799279213,
|
|
"num_tokens": 598751.0,
|
|
"step": 265
|
|
},
|
|
{
|
|
"entropy": 7.636404752731323,
|
|
"epoch": 0.017813551494359044,
|
|
"grad_norm": 1.3125,
|
|
"learning_rate": 0.00013450000000000002,
|
|
"loss": 7.1761,
|
|
"mean_token_accuracy": 0.0747161127626896,
|
|
"num_tokens": 610403.0,
|
|
"step": 270
|
|
},
|
|
{
|
|
"entropy": 7.661175394058228,
|
|
"epoch": 0.018143432077587913,
|
|
"grad_norm": 1.3671875,
|
|
"learning_rate": 0.00013700000000000002,
|
|
"loss": 7.1387,
|
|
"mean_token_accuracy": 0.06750445775687694,
|
|
"num_tokens": 620524.0,
|
|
"step": 275
|
|
},
|
|
{
|
|
"entropy": 7.52873387336731,
|
|
"epoch": 0.018473312660816785,
|
|
"grad_norm": 1.2578125,
|
|
"learning_rate": 0.0001395,
|
|
"loss": 7.0716,
|
|
"mean_token_accuracy": 0.07579033151268959,
|
|
"num_tokens": 631821.0,
|
|
"step": 280
|
|
},
|
|
{
|
|
"entropy": 7.555831003189087,
|
|
"epoch": 0.018803193244045655,
|
|
"grad_norm": 1.2265625,
|
|
"learning_rate": 0.00014199999999999998,
|
|
"loss": 7.1448,
|
|
"mean_token_accuracy": 0.08173828572034836,
|
|
"num_tokens": 643721.0,
|
|
"step": 285
|
|
},
|
|
{
|
|
"entropy": 7.593539047241211,
|
|
"epoch": 0.019133073827274527,
|
|
"grad_norm": 1.28125,
|
|
"learning_rate": 0.0001445,
|
|
"loss": 7.1666,
|
|
"mean_token_accuracy": 0.0737812027335167,
|
|
"num_tokens": 654474.0,
|
|
"step": 290
|
|
},
|
|
{
|
|
"entropy": 7.555977010726929,
|
|
"epoch": 0.019462954410503396,
|
|
"grad_norm": 1.1875,
|
|
"learning_rate": 0.000147,
|
|
"loss": 7.0585,
|
|
"mean_token_accuracy": 0.07724994048476219,
|
|
"num_tokens": 664918.0,
|
|
"step": 295
|
|
},
|
|
{
|
|
"entropy": 7.560142946243286,
|
|
"epoch": 0.01979283499373227,
|
|
"grad_norm": 1.171875,
|
|
"learning_rate": 0.0001495,
|
|
"loss": 7.0644,
|
|
"mean_token_accuracy": 0.07510337978601456,
|
|
"num_tokens": 674533.0,
|
|
"step": 300
|
|
},
|
|
{
|
|
"entropy": 7.545318174362182,
|
|
"epoch": 0.02012271557696114,
|
|
"grad_norm": 1.1953125,
|
|
"learning_rate": 0.000152,
|
|
"loss": 7.0351,
|
|
"mean_token_accuracy": 0.08078778870403766,
|
|
"num_tokens": 685394.0,
|
|
"step": 305
|
|
},
|
|
{
|
|
"entropy": 7.4799717426300045,
|
|
"epoch": 0.02045259616019001,
|
|
"grad_norm": 1.21875,
|
|
"learning_rate": 0.00015450000000000001,
|
|
"loss": 7.1152,
|
|
"mean_token_accuracy": 0.07179579921066762,
|
|
"num_tokens": 698698.0,
|
|
"step": 310
|
|
},
|
|
{
|
|
"entropy": 7.544073390960693,
|
|
"epoch": 0.020782476743418883,
|
|
"grad_norm": 0.9921875,
|
|
"learning_rate": 0.000157,
|
|
"loss": 7.0412,
|
|
"mean_token_accuracy": 0.07378829419612884,
|
|
"num_tokens": 709874.0,
|
|
"step": 315
|
|
},
|
|
{
|
|
"entropy": 7.517040634155274,
|
|
"epoch": 0.021112357326647752,
|
|
"grad_norm": 1.203125,
|
|
"learning_rate": 0.0001595,
|
|
"loss": 7.0463,
|
|
"mean_token_accuracy": 0.0847871646285057,
|
|
"num_tokens": 720121.0,
|
|
"step": 320
|
|
},
|
|
{
|
|
"entropy": 7.501154994964599,
|
|
"epoch": 0.021442237909876625,
|
|
"grad_norm": 1.109375,
|
|
"learning_rate": 0.000162,
|
|
"loss": 6.9986,
|
|
"mean_token_accuracy": 0.08447686731815338,
|
|
"num_tokens": 731027.0,
|
|
"step": 325
|
|
},
|
|
{
|
|
"entropy": 7.431021118164063,
|
|
"epoch": 0.021772118493105497,
|
|
"grad_norm": 1.1796875,
|
|
"learning_rate": 0.00016450000000000001,
|
|
"loss": 6.9182,
|
|
"mean_token_accuracy": 0.07528524361550808,
|
|
"num_tokens": 741589.0,
|
|
"step": 330
|
|
},
|
|
{
|
|
"entropy": 7.3853919506073,
|
|
"epoch": 0.022101999076334367,
|
|
"grad_norm": 1.0078125,
|
|
"learning_rate": 0.00016700000000000002,
|
|
"loss": 6.989,
|
|
"mean_token_accuracy": 0.07859820201992988,
|
|
"num_tokens": 752199.0,
|
|
"step": 335
|
|
},
|
|
{
|
|
"entropy": 7.432969665527343,
|
|
"epoch": 0.02243187965956324,
|
|
"grad_norm": 1.2734375,
|
|
"learning_rate": 0.00016950000000000003,
|
|
"loss": 7.0073,
|
|
"mean_token_accuracy": 0.07819744795560837,
|
|
"num_tokens": 762272.0,
|
|
"step": 340
|
|
},
|
|
{
|
|
"entropy": 7.445111465454102,
|
|
"epoch": 0.02276176024279211,
|
|
"grad_norm": 0.98828125,
|
|
"learning_rate": 0.00017199999999999998,
|
|
"loss": 6.9716,
|
|
"mean_token_accuracy": 0.08092061579227447,
|
|
"num_tokens": 773985.0,
|
|
"step": 345
|
|
},
|
|
{
|
|
"entropy": 7.312721014022827,
|
|
"epoch": 0.02309164082602098,
|
|
"grad_norm": 1.1953125,
|
|
"learning_rate": 0.00017449999999999999,
|
|
"loss": 6.9515,
|
|
"mean_token_accuracy": 0.07682415321469308,
|
|
"num_tokens": 785112.0,
|
|
"step": 350
|
|
},
|
|
{
|
|
"entropy": 7.37159481048584,
|
|
"epoch": 0.02342152140924985,
|
|
"grad_norm": 1.125,
|
|
"learning_rate": 0.000177,
|
|
"loss": 6.9117,
|
|
"mean_token_accuracy": 0.07757236957550048,
|
|
"num_tokens": 795541.0,
|
|
"step": 355
|
|
},
|
|
{
|
|
"entropy": 7.423793363571167,
|
|
"epoch": 0.023751401992478723,
|
|
"grad_norm": 1.2109375,
|
|
"learning_rate": 0.0001795,
|
|
"loss": 6.9885,
|
|
"mean_token_accuracy": 0.08327835351228714,
|
|
"num_tokens": 807143.0,
|
|
"step": 360
|
|
},
|
|
{
|
|
"entropy": 7.415313482284546,
|
|
"epoch": 0.024081282575707595,
|
|
"grad_norm": 1.1953125,
|
|
"learning_rate": 0.000182,
|
|
"loss": 7.0019,
|
|
"mean_token_accuracy": 0.08312778137624263,
|
|
"num_tokens": 819095.0,
|
|
"step": 365
|
|
},
|
|
{
|
|
"entropy": 7.265562868118286,
|
|
"epoch": 0.024411163158936464,
|
|
"grad_norm": 1.5703125,
|
|
"learning_rate": 0.0001845,
|
|
"loss": 6.8904,
|
|
"mean_token_accuracy": 0.09070328995585442,
|
|
"num_tokens": 830380.0,
|
|
"step": 370
|
|
},
|
|
{
|
|
"entropy": 7.2948534965515135,
|
|
"epoch": 0.024741043742165337,
|
|
"grad_norm": 1.0546875,
|
|
"learning_rate": 0.000187,
|
|
"loss": 6.941,
|
|
"mean_token_accuracy": 0.07862535268068313,
|
|
"num_tokens": 841405.0,
|
|
"step": 375
|
|
},
|
|
{
|
|
"entropy": 7.351482820510864,
|
|
"epoch": 0.025070924325394206,
|
|
"grad_norm": 1.0703125,
|
|
"learning_rate": 0.0001895,
|
|
"loss": 6.8778,
|
|
"mean_token_accuracy": 0.07986355572938919,
|
|
"num_tokens": 852529.0,
|
|
"step": 380
|
|
},
|
|
{
|
|
"entropy": 7.329665899276733,
|
|
"epoch": 0.02540080490862308,
|
|
"grad_norm": 1.140625,
|
|
"learning_rate": 0.000192,
|
|
"loss": 6.8671,
|
|
"mean_token_accuracy": 0.08470154702663421,
|
|
"num_tokens": 865100.0,
|
|
"step": 385
|
|
},
|
|
{
|
|
"entropy": 7.301132297515869,
|
|
"epoch": 0.02573068549185195,
|
|
"grad_norm": 1.234375,
|
|
"learning_rate": 0.0001945,
|
|
"loss": 6.9132,
|
|
"mean_token_accuracy": 0.0864314541220665,
|
|
"num_tokens": 876385.0,
|
|
"step": 390
|
|
},
|
|
{
|
|
"entropy": 7.182307958602905,
|
|
"epoch": 0.02606056607508082,
|
|
"grad_norm": 1.09375,
|
|
"learning_rate": 0.00019700000000000002,
|
|
"loss": 6.8564,
|
|
"mean_token_accuracy": 0.08138777352869511,
|
|
"num_tokens": 887207.0,
|
|
"step": 395
|
|
},
|
|
{
|
|
"entropy": 7.2932390689849855,
|
|
"epoch": 0.026390446658309693,
|
|
"grad_norm": 1.0234375,
|
|
"learning_rate": 0.00019950000000000002,
|
|
"loss": 6.8291,
|
|
"mean_token_accuracy": 0.08961722925305367,
|
|
"num_tokens": 900124.0,
|
|
"step": 400
|
|
},
|
|
{
|
|
"entropy": 7.234239864349365,
|
|
"epoch": 0.026720327241538562,
|
|
"grad_norm": 1.25,
|
|
"learning_rate": 0.000202,
|
|
"loss": 6.8417,
|
|
"mean_token_accuracy": 0.08171609789133072,
|
|
"num_tokens": 911177.0,
|
|
"step": 405
|
|
},
|
|
{
|
|
"entropy": 7.174425363540649,
|
|
"epoch": 0.027050207824767435,
|
|
"grad_norm": 1.1640625,
|
|
"learning_rate": 0.00020449999999999998,
|
|
"loss": 6.8362,
|
|
"mean_token_accuracy": 0.08683784157037736,
|
|
"num_tokens": 922524.0,
|
|
"step": 410
|
|
},
|
|
{
|
|
"entropy": 7.304110097885132,
|
|
"epoch": 0.027380088407996304,
|
|
"grad_norm": 1.09375,
|
|
"learning_rate": 0.000207,
|
|
"loss": 6.8652,
|
|
"mean_token_accuracy": 0.08327626436948776,
|
|
"num_tokens": 933484.0,
|
|
"step": 415
|
|
},
|
|
{
|
|
"entropy": 7.25415415763855,
|
|
"epoch": 0.027709968991225176,
|
|
"grad_norm": 1.15625,
|
|
"learning_rate": 0.0002095,
|
|
"loss": 6.8579,
|
|
"mean_token_accuracy": 0.08281879425048828,
|
|
"num_tokens": 944996.0,
|
|
"step": 420
|
|
},
|
|
{
|
|
"entropy": 7.105071210861206,
|
|
"epoch": 0.02803984957445405,
|
|
"grad_norm": 1.15625,
|
|
"learning_rate": 0.000212,
|
|
"loss": 6.7473,
|
|
"mean_token_accuracy": 0.08429811634123326,
|
|
"num_tokens": 955943.0,
|
|
"step": 425
|
|
},
|
|
{
|
|
"entropy": 7.203190279006958,
|
|
"epoch": 0.028369730157682918,
|
|
"grad_norm": 1.1953125,
|
|
"learning_rate": 0.0002145,
|
|
"loss": 6.8392,
|
|
"mean_token_accuracy": 0.08637975230813026,
|
|
"num_tokens": 967721.0,
|
|
"step": 430
|
|
},
|
|
{
|
|
"entropy": 7.126667737960815,
|
|
"epoch": 0.02869961074091179,
|
|
"grad_norm": 1.359375,
|
|
"learning_rate": 0.00021700000000000002,
|
|
"loss": 6.7418,
|
|
"mean_token_accuracy": 0.08021172881126404,
|
|
"num_tokens": 978976.0,
|
|
"step": 435
|
|
},
|
|
{
|
|
"entropy": 7.1474371433258055,
|
|
"epoch": 0.02902949132414066,
|
|
"grad_norm": 1.2265625,
|
|
"learning_rate": 0.0002195,
|
|
"loss": 6.8113,
|
|
"mean_token_accuracy": 0.08653630092740058,
|
|
"num_tokens": 990703.0,
|
|
"step": 440
|
|
},
|
|
{
|
|
"entropy": 7.275717878341675,
|
|
"epoch": 0.029359371907369532,
|
|
"grad_norm": 1.296875,
|
|
"learning_rate": 0.000222,
|
|
"loss": 6.912,
|
|
"mean_token_accuracy": 0.0858251266181469,
|
|
"num_tokens": 1002011.0,
|
|
"step": 445
|
|
},
|
|
{
|
|
"entropy": 7.111966323852539,
|
|
"epoch": 0.029689252490598405,
|
|
"grad_norm": 1.140625,
|
|
"learning_rate": 0.0002245,
|
|
"loss": 6.6889,
|
|
"mean_token_accuracy": 0.09279728755354881,
|
|
"num_tokens": 1011843.0,
|
|
"step": 450
|
|
},
|
|
{
|
|
"entropy": 7.017643737792969,
|
|
"epoch": 0.030019133073827274,
|
|
"grad_norm": 1.0703125,
|
|
"learning_rate": 0.00022700000000000002,
|
|
"loss": 6.6816,
|
|
"mean_token_accuracy": 0.09300963133573532,
|
|
"num_tokens": 1023472.0,
|
|
"step": 455
|
|
},
|
|
{
|
|
"entropy": 7.143270444869995,
|
|
"epoch": 0.030349013657056147,
|
|
"grad_norm": 1.1796875,
|
|
"learning_rate": 0.00022950000000000002,
|
|
"loss": 6.7775,
|
|
"mean_token_accuracy": 0.08557102233171462,
|
|
"num_tokens": 1034106.0,
|
|
"step": 460
|
|
},
|
|
{
|
|
"entropy": 7.0807695388793945,
|
|
"epoch": 0.030678894240285016,
|
|
"grad_norm": 1.0859375,
|
|
"learning_rate": 0.00023200000000000003,
|
|
"loss": 6.7295,
|
|
"mean_token_accuracy": 0.09505119472742081,
|
|
"num_tokens": 1045877.0,
|
|
"step": 465
|
|
},
|
|
{
|
|
"entropy": 7.1490395069122314,
|
|
"epoch": 0.03100877482351389,
|
|
"grad_norm": 1.0,
|
|
"learning_rate": 0.00023449999999999998,
|
|
"loss": 6.8494,
|
|
"mean_token_accuracy": 0.0859566606581211,
|
|
"num_tokens": 1058738.0,
|
|
"step": 470
|
|
},
|
|
{
|
|
"entropy": 7.045120143890381,
|
|
"epoch": 0.03133865540674276,
|
|
"grad_norm": 1.296875,
|
|
"learning_rate": 0.000237,
|
|
"loss": 6.6646,
|
|
"mean_token_accuracy": 0.09076537564396858,
|
|
"num_tokens": 1069415.0,
|
|
"step": 475
|
|
},
|
|
{
|
|
"entropy": 7.029132843017578,
|
|
"epoch": 0.03166853598997163,
|
|
"grad_norm": 1.1953125,
|
|
"learning_rate": 0.0002395,
|
|
"loss": 6.7098,
|
|
"mean_token_accuracy": 0.09159902483224869,
|
|
"num_tokens": 1079979.0,
|
|
"step": 480
|
|
},
|
|
{
|
|
"entropy": 7.103505086898804,
|
|
"epoch": 0.0319984165732005,
|
|
"grad_norm": 1.28125,
|
|
"learning_rate": 0.000242,
|
|
"loss": 6.8259,
|
|
"mean_token_accuracy": 0.08354151919484139,
|
|
"num_tokens": 1090946.0,
|
|
"step": 485
|
|
},
|
|
{
|
|
"entropy": 7.008528804779052,
|
|
"epoch": 0.032328297156429375,
|
|
"grad_norm": 1.0546875,
|
|
"learning_rate": 0.0002445,
|
|
"loss": 6.6646,
|
|
"mean_token_accuracy": 0.09265932142734527,
|
|
"num_tokens": 1100926.0,
|
|
"step": 490
|
|
},
|
|
{
|
|
"entropy": 7.005897378921508,
|
|
"epoch": 0.03265817773965824,
|
|
"grad_norm": 0.9921875,
|
|
"learning_rate": 0.000247,
|
|
"loss": 6.6844,
|
|
"mean_token_accuracy": 0.09150492250919343,
|
|
"num_tokens": 1112169.0,
|
|
"step": 495
|
|
},
|
|
{
|
|
"entropy": 7.125348281860352,
|
|
"epoch": 0.032988058322887114,
|
|
"grad_norm": 1.3046875,
|
|
"learning_rate": 0.0002495,
|
|
"loss": 6.793,
|
|
"mean_token_accuracy": 0.08383904099464416,
|
|
"num_tokens": 1123425.0,
|
|
"step": 500
|
|
}
|
|
],
|
|
"logging_steps": 5,
|
|
"max_steps": 4000,
|
|
"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": true,
|
|
"should_training_stop": false
|
|
},
|
|
"attributes": {}
|
|
}
|
|
},
|
|
"total_flos": 1910648991744000.0,
|
|
"train_batch_size": 16,
|
|
"trial_name": null,
|
|
"trial_params": null
|
|
}
|