2035 lines
54 KiB
JSON
2035 lines
54 KiB
JSON
{
|
|
"best_global_step": null,
|
|
"best_metric": null,
|
|
"best_model_checkpoint": null,
|
|
"epoch": 0.06609166914510425,
|
|
"eval_steps": 500,
|
|
"global_step": 1000,
|
|
"is_hyper_param_search": false,
|
|
"is_local_process_zero": true,
|
|
"is_world_process_zero": true,
|
|
"log_history": [
|
|
{
|
|
"entropy": 10.742584800720214,
|
|
"epoch": 0.0003304583457255213,
|
|
"grad_norm": 5.03125,
|
|
"learning_rate": 2e-06,
|
|
"loss": 10.7734,
|
|
"mean_token_accuracy": 0.0,
|
|
"num_tokens": 10288.0,
|
|
"step": 5
|
|
},
|
|
{
|
|
"entropy": 10.742583179473877,
|
|
"epoch": 0.0006609166914510426,
|
|
"grad_norm": 4.28125,
|
|
"learning_rate": 4.5e-06,
|
|
"loss": 10.7648,
|
|
"mean_token_accuracy": 8.474576170556248e-05,
|
|
"num_tokens": 20647.0,
|
|
"step": 10
|
|
},
|
|
{
|
|
"entropy": 10.74263048171997,
|
|
"epoch": 0.000991375037176564,
|
|
"grad_norm": 4.5,
|
|
"learning_rate": 7e-06,
|
|
"loss": 10.7389,
|
|
"mean_token_accuracy": 0.00012787723680958153,
|
|
"num_tokens": 30934.0,
|
|
"step": 15
|
|
},
|
|
{
|
|
"entropy": 10.742611503601074,
|
|
"epoch": 0.0013218333829020852,
|
|
"grad_norm": 4.8125,
|
|
"learning_rate": 9.5e-06,
|
|
"loss": 10.7077,
|
|
"mean_token_accuracy": 0.0007004500948823988,
|
|
"num_tokens": 41985.0,
|
|
"step": 20
|
|
},
|
|
{
|
|
"entropy": 10.742591762542725,
|
|
"epoch": 0.0016522917286276066,
|
|
"grad_norm": 4.5,
|
|
"learning_rate": 1.2e-05,
|
|
"loss": 10.5981,
|
|
"mean_token_accuracy": 0.005805681901983916,
|
|
"num_tokens": 52249.0,
|
|
"step": 25
|
|
},
|
|
{
|
|
"entropy": 10.742315578460694,
|
|
"epoch": 0.001982750074353128,
|
|
"grad_norm": 3.765625,
|
|
"learning_rate": 1.4500000000000002e-05,
|
|
"loss": 10.4964,
|
|
"mean_token_accuracy": 0.02333841286599636,
|
|
"num_tokens": 63022.0,
|
|
"step": 30
|
|
},
|
|
{
|
|
"entropy": 10.741240406036377,
|
|
"epoch": 0.002313208420078649,
|
|
"grad_norm": 3.34375,
|
|
"learning_rate": 1.7000000000000003e-05,
|
|
"loss": 10.3567,
|
|
"mean_token_accuracy": 0.041300085932016374,
|
|
"num_tokens": 74862.0,
|
|
"step": 35
|
|
},
|
|
{
|
|
"entropy": 10.738348007202148,
|
|
"epoch": 0.0026436667658041703,
|
|
"grad_norm": 2.5,
|
|
"learning_rate": 1.95e-05,
|
|
"loss": 10.2419,
|
|
"mean_token_accuracy": 0.03633426204323768,
|
|
"num_tokens": 86588.0,
|
|
"step": 40
|
|
},
|
|
{
|
|
"entropy": 10.734376811981202,
|
|
"epoch": 0.0029741251115296915,
|
|
"grad_norm": 2.328125,
|
|
"learning_rate": 2.2e-05,
|
|
"loss": 10.1017,
|
|
"mean_token_accuracy": 0.037262484431266785,
|
|
"num_tokens": 96248.0,
|
|
"step": 45
|
|
},
|
|
{
|
|
"entropy": 10.731352519989013,
|
|
"epoch": 0.003304583457255213,
|
|
"grad_norm": 1.953125,
|
|
"learning_rate": 2.4500000000000003e-05,
|
|
"loss": 10.0565,
|
|
"mean_token_accuracy": 0.03895317334681749,
|
|
"num_tokens": 107167.0,
|
|
"step": 50
|
|
},
|
|
{
|
|
"entropy": 10.729418754577637,
|
|
"epoch": 0.0036350418029807343,
|
|
"grad_norm": 1.9609375,
|
|
"learning_rate": 2.7e-05,
|
|
"loss": 9.943,
|
|
"mean_token_accuracy": 0.04272819831967354,
|
|
"num_tokens": 117207.0,
|
|
"step": 55
|
|
},
|
|
{
|
|
"entropy": 10.726603794097901,
|
|
"epoch": 0.003965500148706256,
|
|
"grad_norm": 1.78125,
|
|
"learning_rate": 2.95e-05,
|
|
"loss": 9.9323,
|
|
"mean_token_accuracy": 0.034439234621822835,
|
|
"num_tokens": 127335.0,
|
|
"step": 60
|
|
},
|
|
{
|
|
"entropy": 10.724784851074219,
|
|
"epoch": 0.004295958494431777,
|
|
"grad_norm": 1.890625,
|
|
"learning_rate": 3.2e-05,
|
|
"loss": 9.8653,
|
|
"mean_token_accuracy": 0.04371197037398815,
|
|
"num_tokens": 138329.0,
|
|
"step": 65
|
|
},
|
|
{
|
|
"entropy": 10.722426414489746,
|
|
"epoch": 0.004626416840157298,
|
|
"grad_norm": 1.8046875,
|
|
"learning_rate": 3.4500000000000005e-05,
|
|
"loss": 9.8277,
|
|
"mean_token_accuracy": 0.040007464960217475,
|
|
"num_tokens": 149993.0,
|
|
"step": 70
|
|
},
|
|
{
|
|
"entropy": 10.71851463317871,
|
|
"epoch": 0.00495687518588282,
|
|
"grad_norm": 1.7890625,
|
|
"learning_rate": 3.7e-05,
|
|
"loss": 9.7389,
|
|
"mean_token_accuracy": 0.0400556480512023,
|
|
"num_tokens": 160872.0,
|
|
"step": 75
|
|
},
|
|
{
|
|
"entropy": 10.713956546783447,
|
|
"epoch": 0.005287333531608341,
|
|
"grad_norm": 1.703125,
|
|
"learning_rate": 3.95e-05,
|
|
"loss": 9.7214,
|
|
"mean_token_accuracy": 0.037986579909920694,
|
|
"num_tokens": 172141.0,
|
|
"step": 80
|
|
},
|
|
{
|
|
"entropy": 10.708513736724854,
|
|
"epoch": 0.005617791877333862,
|
|
"grad_norm": 1.71875,
|
|
"learning_rate": 4.2000000000000004e-05,
|
|
"loss": 9.5993,
|
|
"mean_token_accuracy": 0.0436050308868289,
|
|
"num_tokens": 183762.0,
|
|
"step": 85
|
|
},
|
|
{
|
|
"entropy": 10.700162410736084,
|
|
"epoch": 0.005948250223059383,
|
|
"grad_norm": 1.6953125,
|
|
"learning_rate": 4.45e-05,
|
|
"loss": 9.5659,
|
|
"mean_token_accuracy": 0.039148312993347645,
|
|
"num_tokens": 196109.0,
|
|
"step": 90
|
|
},
|
|
{
|
|
"entropy": 10.689516925811768,
|
|
"epoch": 0.006278708568784905,
|
|
"grad_norm": 1.7890625,
|
|
"learning_rate": 4.7000000000000004e-05,
|
|
"loss": 9.4243,
|
|
"mean_token_accuracy": 0.040336356125772,
|
|
"num_tokens": 205671.0,
|
|
"step": 95
|
|
},
|
|
{
|
|
"entropy": 10.678232765197754,
|
|
"epoch": 0.006609166914510426,
|
|
"grad_norm": 1.6328125,
|
|
"learning_rate": 4.9500000000000004e-05,
|
|
"loss": 9.4066,
|
|
"mean_token_accuracy": 0.04090222828090191,
|
|
"num_tokens": 217256.0,
|
|
"step": 100
|
|
},
|
|
{
|
|
"entropy": 10.663106441497803,
|
|
"epoch": 0.006939625260235947,
|
|
"grad_norm": 1.828125,
|
|
"learning_rate": 5.2e-05,
|
|
"loss": 9.2659,
|
|
"mean_token_accuracy": 0.051931383460760115,
|
|
"num_tokens": 228066.0,
|
|
"step": 105
|
|
},
|
|
{
|
|
"entropy": 10.637630844116211,
|
|
"epoch": 0.007270083605961469,
|
|
"grad_norm": 1.7578125,
|
|
"learning_rate": 5.45e-05,
|
|
"loss": 9.2057,
|
|
"mean_token_accuracy": 0.0478770911693573,
|
|
"num_tokens": 238643.0,
|
|
"step": 110
|
|
},
|
|
{
|
|
"entropy": 10.609512996673583,
|
|
"epoch": 0.00760054195168699,
|
|
"grad_norm": 1.765625,
|
|
"learning_rate": 5.7e-05,
|
|
"loss": 9.1192,
|
|
"mean_token_accuracy": 0.04669556953012943,
|
|
"num_tokens": 249812.0,
|
|
"step": 115
|
|
},
|
|
{
|
|
"entropy": 10.571122074127198,
|
|
"epoch": 0.007931000297412512,
|
|
"grad_norm": 1.546875,
|
|
"learning_rate": 5.9499999999999996e-05,
|
|
"loss": 8.9947,
|
|
"mean_token_accuracy": 0.04946679100394249,
|
|
"num_tokens": 261942.0,
|
|
"step": 120
|
|
},
|
|
{
|
|
"entropy": 10.517377948760986,
|
|
"epoch": 0.008261458643138032,
|
|
"grad_norm": 1.671875,
|
|
"learning_rate": 6.2e-05,
|
|
"loss": 8.8178,
|
|
"mean_token_accuracy": 0.053763845935463905,
|
|
"num_tokens": 272537.0,
|
|
"step": 125
|
|
},
|
|
{
|
|
"entropy": 10.431371021270753,
|
|
"epoch": 0.008591916988863553,
|
|
"grad_norm": 1.5625,
|
|
"learning_rate": 6.450000000000001e-05,
|
|
"loss": 8.7253,
|
|
"mean_token_accuracy": 0.051454832404851915,
|
|
"num_tokens": 283610.0,
|
|
"step": 130
|
|
},
|
|
{
|
|
"entropy": 10.352199172973632,
|
|
"epoch": 0.008922375334589075,
|
|
"grad_norm": 1.4296875,
|
|
"learning_rate": 6.7e-05,
|
|
"loss": 8.6516,
|
|
"mean_token_accuracy": 0.048623890802264216,
|
|
"num_tokens": 294257.0,
|
|
"step": 135
|
|
},
|
|
{
|
|
"entropy": 10.283658027648926,
|
|
"epoch": 0.009252833680314597,
|
|
"grad_norm": 1.578125,
|
|
"learning_rate": 6.950000000000001e-05,
|
|
"loss": 8.5669,
|
|
"mean_token_accuracy": 0.04829129688441754,
|
|
"num_tokens": 304405.0,
|
|
"step": 140
|
|
},
|
|
{
|
|
"entropy": 10.165488147735596,
|
|
"epoch": 0.009583292026040118,
|
|
"grad_norm": 1.28125,
|
|
"learning_rate": 7.2e-05,
|
|
"loss": 8.3473,
|
|
"mean_token_accuracy": 0.05426618978381157,
|
|
"num_tokens": 314413.0,
|
|
"step": 145
|
|
},
|
|
{
|
|
"entropy": 10.042571830749512,
|
|
"epoch": 0.00991375037176564,
|
|
"grad_norm": 1.3125,
|
|
"learning_rate": 7.45e-05,
|
|
"loss": 8.3318,
|
|
"mean_token_accuracy": 0.05118043087422848,
|
|
"num_tokens": 325010.0,
|
|
"step": 150
|
|
},
|
|
{
|
|
"entropy": 9.901912784576416,
|
|
"epoch": 0.01024420871749116,
|
|
"grad_norm": 1.2890625,
|
|
"learning_rate": 7.7e-05,
|
|
"loss": 8.2007,
|
|
"mean_token_accuracy": 0.05208889953792095,
|
|
"num_tokens": 336421.0,
|
|
"step": 155
|
|
},
|
|
{
|
|
"entropy": 9.73304557800293,
|
|
"epoch": 0.010574667063216681,
|
|
"grad_norm": 1.28125,
|
|
"learning_rate": 7.950000000000001e-05,
|
|
"loss": 8.0228,
|
|
"mean_token_accuracy": 0.06282360814511775,
|
|
"num_tokens": 346930.0,
|
|
"step": 160
|
|
},
|
|
{
|
|
"entropy": 9.518652629852294,
|
|
"epoch": 0.010905125408942203,
|
|
"grad_norm": 1.25,
|
|
"learning_rate": 8.2e-05,
|
|
"loss": 7.9883,
|
|
"mean_token_accuracy": 0.062138063833117485,
|
|
"num_tokens": 357874.0,
|
|
"step": 165
|
|
},
|
|
{
|
|
"entropy": 9.32524814605713,
|
|
"epoch": 0.011235583754667725,
|
|
"grad_norm": 1.3125,
|
|
"learning_rate": 8.450000000000001e-05,
|
|
"loss": 8.0045,
|
|
"mean_token_accuracy": 0.05923859886825085,
|
|
"num_tokens": 370138.0,
|
|
"step": 170
|
|
},
|
|
{
|
|
"entropy": 9.140588092803956,
|
|
"epoch": 0.011566042100393246,
|
|
"grad_norm": 1.125,
|
|
"learning_rate": 8.7e-05,
|
|
"loss": 7.9149,
|
|
"mean_token_accuracy": 0.061116486042737964,
|
|
"num_tokens": 379999.0,
|
|
"step": 175
|
|
},
|
|
{
|
|
"entropy": 8.904661655426025,
|
|
"epoch": 0.011896500446118766,
|
|
"grad_norm": 0.80078125,
|
|
"learning_rate": 8.95e-05,
|
|
"loss": 7.7459,
|
|
"mean_token_accuracy": 0.0650369856506586,
|
|
"num_tokens": 392754.0,
|
|
"step": 180
|
|
},
|
|
{
|
|
"entropy": 8.754450225830078,
|
|
"epoch": 0.012226958791844288,
|
|
"grad_norm": 0.96875,
|
|
"learning_rate": 9.2e-05,
|
|
"loss": 7.7896,
|
|
"mean_token_accuracy": 0.06155076548457146,
|
|
"num_tokens": 403771.0,
|
|
"step": 185
|
|
},
|
|
{
|
|
"entropy": 8.645609664916993,
|
|
"epoch": 0.01255741713756981,
|
|
"grad_norm": 1.4765625,
|
|
"learning_rate": 9.45e-05,
|
|
"loss": 7.7912,
|
|
"mean_token_accuracy": 0.06344522796571254,
|
|
"num_tokens": 415381.0,
|
|
"step": 190
|
|
},
|
|
{
|
|
"entropy": 8.569410800933838,
|
|
"epoch": 0.01288787548329533,
|
|
"grad_norm": 0.91796875,
|
|
"learning_rate": 9.7e-05,
|
|
"loss": 7.6569,
|
|
"mean_token_accuracy": 0.06627831235527992,
|
|
"num_tokens": 425919.0,
|
|
"step": 195
|
|
},
|
|
{
|
|
"entropy": 8.510480690002442,
|
|
"epoch": 0.013218333829020852,
|
|
"grad_norm": 0.96875,
|
|
"learning_rate": 9.95e-05,
|
|
"loss": 7.6221,
|
|
"mean_token_accuracy": 0.06085339933633804,
|
|
"num_tokens": 437753.0,
|
|
"step": 200
|
|
},
|
|
{
|
|
"entropy": 8.409182167053222,
|
|
"epoch": 0.013548792174746372,
|
|
"grad_norm": 1.0078125,
|
|
"learning_rate": 0.000102,
|
|
"loss": 7.6117,
|
|
"mean_token_accuracy": 0.06501428484916687,
|
|
"num_tokens": 449079.0,
|
|
"step": 205
|
|
},
|
|
{
|
|
"entropy": 8.41590747833252,
|
|
"epoch": 0.013879250520471894,
|
|
"grad_norm": 1.3359375,
|
|
"learning_rate": 0.00010449999999999999,
|
|
"loss": 7.5895,
|
|
"mean_token_accuracy": 0.06874292232096195,
|
|
"num_tokens": 459250.0,
|
|
"step": 210
|
|
},
|
|
{
|
|
"entropy": 8.25771951675415,
|
|
"epoch": 0.014209708866197416,
|
|
"grad_norm": 0.83203125,
|
|
"learning_rate": 0.000107,
|
|
"loss": 7.5524,
|
|
"mean_token_accuracy": 0.06953966170549393,
|
|
"num_tokens": 470898.0,
|
|
"step": 215
|
|
},
|
|
{
|
|
"entropy": 8.192992210388184,
|
|
"epoch": 0.014540167211922937,
|
|
"grad_norm": 0.8828125,
|
|
"learning_rate": 0.0001095,
|
|
"loss": 7.5374,
|
|
"mean_token_accuracy": 0.07173516266047955,
|
|
"num_tokens": 481116.0,
|
|
"step": 220
|
|
},
|
|
{
|
|
"entropy": 8.23768253326416,
|
|
"epoch": 0.014870625557648459,
|
|
"grad_norm": 0.890625,
|
|
"learning_rate": 0.000112,
|
|
"loss": 7.5324,
|
|
"mean_token_accuracy": 0.0693703394383192,
|
|
"num_tokens": 492639.0,
|
|
"step": 225
|
|
},
|
|
{
|
|
"entropy": 8.176224327087402,
|
|
"epoch": 0.01520108390337398,
|
|
"grad_norm": 1.2109375,
|
|
"learning_rate": 0.0001145,
|
|
"loss": 7.5754,
|
|
"mean_token_accuracy": 0.06600871868431568,
|
|
"num_tokens": 504838.0,
|
|
"step": 230
|
|
},
|
|
{
|
|
"entropy": 8.241323566436767,
|
|
"epoch": 0.0155315422490995,
|
|
"grad_norm": 1.09375,
|
|
"learning_rate": 0.00011700000000000001,
|
|
"loss": 7.5075,
|
|
"mean_token_accuracy": 0.05977615527808666,
|
|
"num_tokens": 515982.0,
|
|
"step": 235
|
|
},
|
|
{
|
|
"entropy": 8.07399296760559,
|
|
"epoch": 0.015862000594825024,
|
|
"grad_norm": 0.8359375,
|
|
"learning_rate": 0.00011949999999999999,
|
|
"loss": 7.4775,
|
|
"mean_token_accuracy": 0.06939641572535038,
|
|
"num_tokens": 526830.0,
|
|
"step": 240
|
|
},
|
|
{
|
|
"entropy": 8.070019817352295,
|
|
"epoch": 0.016192458940550544,
|
|
"grad_norm": 0.921875,
|
|
"learning_rate": 0.000122,
|
|
"loss": 7.5155,
|
|
"mean_token_accuracy": 0.06779546402394772,
|
|
"num_tokens": 538111.0,
|
|
"step": 245
|
|
},
|
|
{
|
|
"entropy": 8.138901233673096,
|
|
"epoch": 0.016522917286276063,
|
|
"grad_norm": 0.8828125,
|
|
"learning_rate": 0.0001245,
|
|
"loss": 7.4836,
|
|
"mean_token_accuracy": 0.06842113994061946,
|
|
"num_tokens": 548286.0,
|
|
"step": 250
|
|
},
|
|
{
|
|
"entropy": 8.062221097946168,
|
|
"epoch": 0.016853375632001587,
|
|
"grad_norm": 0.89453125,
|
|
"learning_rate": 0.000127,
|
|
"loss": 7.4483,
|
|
"mean_token_accuracy": 0.07109758108854294,
|
|
"num_tokens": 558638.0,
|
|
"step": 255
|
|
},
|
|
{
|
|
"entropy": 8.011436653137206,
|
|
"epoch": 0.017183833977727107,
|
|
"grad_norm": 0.78125,
|
|
"learning_rate": 0.0001295,
|
|
"loss": 7.418,
|
|
"mean_token_accuracy": 0.07253723293542862,
|
|
"num_tokens": 568909.0,
|
|
"step": 260
|
|
},
|
|
{
|
|
"entropy": 8.036324262619019,
|
|
"epoch": 0.01751429232345263,
|
|
"grad_norm": 0.890625,
|
|
"learning_rate": 0.000132,
|
|
"loss": 7.3883,
|
|
"mean_token_accuracy": 0.07206913344562053,
|
|
"num_tokens": 580392.0,
|
|
"step": 265
|
|
},
|
|
{
|
|
"entropy": 7.999330711364746,
|
|
"epoch": 0.01784475066917815,
|
|
"grad_norm": 0.9453125,
|
|
"learning_rate": 0.00013450000000000002,
|
|
"loss": 7.44,
|
|
"mean_token_accuracy": 0.068961101770401,
|
|
"num_tokens": 590160.0,
|
|
"step": 270
|
|
},
|
|
{
|
|
"entropy": 7.934969806671143,
|
|
"epoch": 0.01817520901490367,
|
|
"grad_norm": 1.1484375,
|
|
"learning_rate": 0.00013700000000000002,
|
|
"loss": 7.3588,
|
|
"mean_token_accuracy": 0.0708784569054842,
|
|
"num_tokens": 601054.0,
|
|
"step": 275
|
|
},
|
|
{
|
|
"entropy": 7.9982236385345455,
|
|
"epoch": 0.018505667360629193,
|
|
"grad_norm": 1.078125,
|
|
"learning_rate": 0.0001395,
|
|
"loss": 7.4727,
|
|
"mean_token_accuracy": 0.06899206414818763,
|
|
"num_tokens": 611908.0,
|
|
"step": 280
|
|
},
|
|
{
|
|
"entropy": 7.999399185180664,
|
|
"epoch": 0.018836125706354713,
|
|
"grad_norm": 0.8828125,
|
|
"learning_rate": 0.00014199999999999998,
|
|
"loss": 7.4195,
|
|
"mean_token_accuracy": 0.07329106517136097,
|
|
"num_tokens": 622259.0,
|
|
"step": 285
|
|
},
|
|
{
|
|
"entropy": 7.868047714233398,
|
|
"epoch": 0.019166584052080236,
|
|
"grad_norm": 0.94140625,
|
|
"learning_rate": 0.0001445,
|
|
"loss": 7.2845,
|
|
"mean_token_accuracy": 0.0750714760273695,
|
|
"num_tokens": 632728.0,
|
|
"step": 290
|
|
},
|
|
{
|
|
"entropy": 7.929320621490478,
|
|
"epoch": 0.019497042397805756,
|
|
"grad_norm": 0.9375,
|
|
"learning_rate": 0.000147,
|
|
"loss": 7.4673,
|
|
"mean_token_accuracy": 0.07401705160737038,
|
|
"num_tokens": 644600.0,
|
|
"step": 295
|
|
},
|
|
{
|
|
"entropy": 7.919670248031617,
|
|
"epoch": 0.01982750074353128,
|
|
"grad_norm": 1.140625,
|
|
"learning_rate": 0.0001495,
|
|
"loss": 7.3135,
|
|
"mean_token_accuracy": 0.07708279564976692,
|
|
"num_tokens": 654004.0,
|
|
"step": 300
|
|
},
|
|
{
|
|
"entropy": 7.94478325843811,
|
|
"epoch": 0.0201579590892568,
|
|
"grad_norm": 0.8359375,
|
|
"learning_rate": 0.000152,
|
|
"loss": 7.3297,
|
|
"mean_token_accuracy": 0.07512539699673652,
|
|
"num_tokens": 665275.0,
|
|
"step": 305
|
|
},
|
|
{
|
|
"entropy": 7.877188873291016,
|
|
"epoch": 0.02048841743498232,
|
|
"grad_norm": 1.1484375,
|
|
"learning_rate": 0.00015450000000000001,
|
|
"loss": 7.3807,
|
|
"mean_token_accuracy": 0.07405197396874427,
|
|
"num_tokens": 677397.0,
|
|
"step": 310
|
|
},
|
|
{
|
|
"entropy": 7.885754251480103,
|
|
"epoch": 0.020818875780707843,
|
|
"grad_norm": 0.9453125,
|
|
"learning_rate": 0.000157,
|
|
"loss": 7.2468,
|
|
"mean_token_accuracy": 0.08046135641634464,
|
|
"num_tokens": 687975.0,
|
|
"step": 315
|
|
},
|
|
{
|
|
"entropy": 7.784027528762818,
|
|
"epoch": 0.021149334126433363,
|
|
"grad_norm": 1.0625,
|
|
"learning_rate": 0.0001595,
|
|
"loss": 7.2053,
|
|
"mean_token_accuracy": 0.08028579428792,
|
|
"num_tokens": 699102.0,
|
|
"step": 320
|
|
},
|
|
{
|
|
"entropy": 7.772454404830933,
|
|
"epoch": 0.021479792472158886,
|
|
"grad_norm": 0.82421875,
|
|
"learning_rate": 0.000162,
|
|
"loss": 7.2547,
|
|
"mean_token_accuracy": 0.07418597042560578,
|
|
"num_tokens": 712007.0,
|
|
"step": 325
|
|
},
|
|
{
|
|
"entropy": 7.787219858169555,
|
|
"epoch": 0.021810250817884406,
|
|
"grad_norm": 0.95703125,
|
|
"learning_rate": 0.00016450000000000001,
|
|
"loss": 7.2527,
|
|
"mean_token_accuracy": 0.08365209773182869,
|
|
"num_tokens": 723757.0,
|
|
"step": 330
|
|
},
|
|
{
|
|
"entropy": 7.8013787269592285,
|
|
"epoch": 0.022140709163609926,
|
|
"grad_norm": 1.078125,
|
|
"learning_rate": 0.00016700000000000002,
|
|
"loss": 7.2636,
|
|
"mean_token_accuracy": 0.08137313425540924,
|
|
"num_tokens": 734353.0,
|
|
"step": 335
|
|
},
|
|
{
|
|
"entropy": 7.695002651214599,
|
|
"epoch": 0.02247116750933545,
|
|
"grad_norm": 0.953125,
|
|
"learning_rate": 0.00016950000000000003,
|
|
"loss": 7.1554,
|
|
"mean_token_accuracy": 0.08258310481905937,
|
|
"num_tokens": 744673.0,
|
|
"step": 340
|
|
},
|
|
{
|
|
"entropy": 7.7564044952392575,
|
|
"epoch": 0.02280162585506097,
|
|
"grad_norm": 0.96875,
|
|
"learning_rate": 0.00017199999999999998,
|
|
"loss": 7.1906,
|
|
"mean_token_accuracy": 0.07295942604541779,
|
|
"num_tokens": 755181.0,
|
|
"step": 345
|
|
},
|
|
{
|
|
"entropy": 7.736034250259399,
|
|
"epoch": 0.023132084200786492,
|
|
"grad_norm": 0.98046875,
|
|
"learning_rate": 0.00017449999999999999,
|
|
"loss": 7.21,
|
|
"mean_token_accuracy": 0.08095265403389931,
|
|
"num_tokens": 765558.0,
|
|
"step": 350
|
|
},
|
|
{
|
|
"entropy": 7.752234363555909,
|
|
"epoch": 0.023462542546512012,
|
|
"grad_norm": 0.921875,
|
|
"learning_rate": 0.000177,
|
|
"loss": 7.2351,
|
|
"mean_token_accuracy": 0.07383172325789929,
|
|
"num_tokens": 778224.0,
|
|
"step": 355
|
|
},
|
|
{
|
|
"entropy": 7.743930006027222,
|
|
"epoch": 0.023793000892237532,
|
|
"grad_norm": 0.92578125,
|
|
"learning_rate": 0.0001795,
|
|
"loss": 7.1656,
|
|
"mean_token_accuracy": 0.07890524119138717,
|
|
"num_tokens": 789586.0,
|
|
"step": 360
|
|
},
|
|
{
|
|
"entropy": 7.688452529907226,
|
|
"epoch": 0.024123459237963055,
|
|
"grad_norm": 0.96875,
|
|
"learning_rate": 0.000182,
|
|
"loss": 7.2174,
|
|
"mean_token_accuracy": 0.07619251199066639,
|
|
"num_tokens": 801595.0,
|
|
"step": 365
|
|
},
|
|
{
|
|
"entropy": 7.692495965957642,
|
|
"epoch": 0.024453917583688575,
|
|
"grad_norm": 0.74609375,
|
|
"learning_rate": 0.0001845,
|
|
"loss": 7.2437,
|
|
"mean_token_accuracy": 0.08221447616815566,
|
|
"num_tokens": 812850.0,
|
|
"step": 370
|
|
},
|
|
{
|
|
"entropy": 7.707998418807984,
|
|
"epoch": 0.0247843759294141,
|
|
"grad_norm": 0.83984375,
|
|
"learning_rate": 0.000187,
|
|
"loss": 7.2205,
|
|
"mean_token_accuracy": 0.0778277974575758,
|
|
"num_tokens": 823125.0,
|
|
"step": 375
|
|
},
|
|
{
|
|
"entropy": 7.7771612167358395,
|
|
"epoch": 0.02511483427513962,
|
|
"grad_norm": 0.890625,
|
|
"learning_rate": 0.0001895,
|
|
"loss": 7.1487,
|
|
"mean_token_accuracy": 0.08065758869051934,
|
|
"num_tokens": 834838.0,
|
|
"step": 380
|
|
},
|
|
{
|
|
"entropy": 7.648675203323364,
|
|
"epoch": 0.02544529262086514,
|
|
"grad_norm": 1.1015625,
|
|
"learning_rate": 0.000192,
|
|
"loss": 7.1992,
|
|
"mean_token_accuracy": 0.08622882142663002,
|
|
"num_tokens": 845976.0,
|
|
"step": 385
|
|
},
|
|
{
|
|
"entropy": 7.634294128417968,
|
|
"epoch": 0.02577575096659066,
|
|
"grad_norm": 1.109375,
|
|
"learning_rate": 0.0001945,
|
|
"loss": 7.1816,
|
|
"mean_token_accuracy": 0.08291547074913978,
|
|
"num_tokens": 857193.0,
|
|
"step": 390
|
|
},
|
|
{
|
|
"entropy": 7.7122198104858395,
|
|
"epoch": 0.02610620931231618,
|
|
"grad_norm": 0.8984375,
|
|
"learning_rate": 0.00019700000000000002,
|
|
"loss": 7.1746,
|
|
"mean_token_accuracy": 0.0808206096291542,
|
|
"num_tokens": 868471.0,
|
|
"step": 395
|
|
},
|
|
{
|
|
"entropy": 7.663157796859741,
|
|
"epoch": 0.026436667658041705,
|
|
"grad_norm": 1.0234375,
|
|
"learning_rate": 0.00019950000000000002,
|
|
"loss": 7.1046,
|
|
"mean_token_accuracy": 0.08876307979226113,
|
|
"num_tokens": 879339.0,
|
|
"step": 400
|
|
},
|
|
{
|
|
"entropy": 7.579256820678711,
|
|
"epoch": 0.026767126003767225,
|
|
"grad_norm": 0.90625,
|
|
"learning_rate": 0.000202,
|
|
"loss": 7.0452,
|
|
"mean_token_accuracy": 0.08175178617238998,
|
|
"num_tokens": 890081.0,
|
|
"step": 405
|
|
},
|
|
{
|
|
"entropy": 7.595317792892456,
|
|
"epoch": 0.027097584349492745,
|
|
"grad_norm": 1.15625,
|
|
"learning_rate": 0.00020449999999999998,
|
|
"loss": 7.0651,
|
|
"mean_token_accuracy": 0.08560339361429214,
|
|
"num_tokens": 899967.0,
|
|
"step": 410
|
|
},
|
|
{
|
|
"entropy": 7.630031490325928,
|
|
"epoch": 0.027428042695218268,
|
|
"grad_norm": 1.0390625,
|
|
"learning_rate": 0.000207,
|
|
"loss": 7.1014,
|
|
"mean_token_accuracy": 0.08190736398100854,
|
|
"num_tokens": 910697.0,
|
|
"step": 415
|
|
},
|
|
{
|
|
"entropy": 7.6542994499206545,
|
|
"epoch": 0.027758501040943788,
|
|
"grad_norm": 1.1015625,
|
|
"learning_rate": 0.0002095,
|
|
"loss": 7.1607,
|
|
"mean_token_accuracy": 0.08872515186667443,
|
|
"num_tokens": 920879.0,
|
|
"step": 420
|
|
},
|
|
{
|
|
"entropy": 7.579720592498779,
|
|
"epoch": 0.02808895938666931,
|
|
"grad_norm": 0.9921875,
|
|
"learning_rate": 0.000212,
|
|
"loss": 7.0708,
|
|
"mean_token_accuracy": 0.09260708317160607,
|
|
"num_tokens": 931566.0,
|
|
"step": 425
|
|
},
|
|
{
|
|
"entropy": 7.549066686630249,
|
|
"epoch": 0.02841941773239483,
|
|
"grad_norm": 0.9375,
|
|
"learning_rate": 0.0002145,
|
|
"loss": 7.0116,
|
|
"mean_token_accuracy": 0.08882770761847496,
|
|
"num_tokens": 941736.0,
|
|
"step": 430
|
|
},
|
|
{
|
|
"entropy": 7.600239658355713,
|
|
"epoch": 0.028749876078120355,
|
|
"grad_norm": 1.078125,
|
|
"learning_rate": 0.00021700000000000002,
|
|
"loss": 7.0376,
|
|
"mean_token_accuracy": 0.08477143570780754,
|
|
"num_tokens": 951652.0,
|
|
"step": 435
|
|
},
|
|
{
|
|
"entropy": 7.493011617660523,
|
|
"epoch": 0.029080334423845874,
|
|
"grad_norm": 0.9375,
|
|
"learning_rate": 0.0002195,
|
|
"loss": 6.9774,
|
|
"mean_token_accuracy": 0.08458059877157212,
|
|
"num_tokens": 963242.0,
|
|
"step": 440
|
|
},
|
|
{
|
|
"entropy": 7.5284465789794925,
|
|
"epoch": 0.029410792769571394,
|
|
"grad_norm": 1.0,
|
|
"learning_rate": 0.000222,
|
|
"loss": 7.0074,
|
|
"mean_token_accuracy": 0.08982692956924439,
|
|
"num_tokens": 974078.0,
|
|
"step": 445
|
|
},
|
|
{
|
|
"entropy": 7.516528034210205,
|
|
"epoch": 0.029741251115296918,
|
|
"grad_norm": 0.75,
|
|
"learning_rate": 0.0002245,
|
|
"loss": 6.9731,
|
|
"mean_token_accuracy": 0.0934697650372982,
|
|
"num_tokens": 986324.0,
|
|
"step": 450
|
|
},
|
|
{
|
|
"entropy": 7.518308115005493,
|
|
"epoch": 0.030071709461022438,
|
|
"grad_norm": 0.94140625,
|
|
"learning_rate": 0.00022700000000000002,
|
|
"loss": 6.9854,
|
|
"mean_token_accuracy": 0.0866778276860714,
|
|
"num_tokens": 997625.0,
|
|
"step": 455
|
|
},
|
|
{
|
|
"entropy": 7.468284225463867,
|
|
"epoch": 0.03040216780674796,
|
|
"grad_norm": 0.9765625,
|
|
"learning_rate": 0.00022950000000000002,
|
|
"loss": 6.9674,
|
|
"mean_token_accuracy": 0.08939552903175355,
|
|
"num_tokens": 1007914.0,
|
|
"step": 460
|
|
},
|
|
{
|
|
"entropy": 7.482946729660034,
|
|
"epoch": 0.03073262615247348,
|
|
"grad_norm": 0.91015625,
|
|
"learning_rate": 0.00023200000000000003,
|
|
"loss": 6.9344,
|
|
"mean_token_accuracy": 0.08928446583449841,
|
|
"num_tokens": 1018510.0,
|
|
"step": 465
|
|
},
|
|
{
|
|
"entropy": 7.432837677001953,
|
|
"epoch": 0.031063084498199,
|
|
"grad_norm": 1.2109375,
|
|
"learning_rate": 0.00023449999999999998,
|
|
"loss": 6.9009,
|
|
"mean_token_accuracy": 0.09617745354771615,
|
|
"num_tokens": 1028532.0,
|
|
"step": 470
|
|
},
|
|
{
|
|
"entropy": 7.4032214164733885,
|
|
"epoch": 0.031393542843924524,
|
|
"grad_norm": 0.93359375,
|
|
"learning_rate": 0.000237,
|
|
"loss": 7.0566,
|
|
"mean_token_accuracy": 0.0882367193698883,
|
|
"num_tokens": 1040252.0,
|
|
"step": 475
|
|
},
|
|
{
|
|
"entropy": 7.425141048431397,
|
|
"epoch": 0.03172400118965005,
|
|
"grad_norm": 1.03125,
|
|
"learning_rate": 0.0002395,
|
|
"loss": 6.856,
|
|
"mean_token_accuracy": 0.09330087304115295,
|
|
"num_tokens": 1050486.0,
|
|
"step": 480
|
|
},
|
|
{
|
|
"entropy": 7.379268646240234,
|
|
"epoch": 0.032054459535375564,
|
|
"grad_norm": 1.0703125,
|
|
"learning_rate": 0.000242,
|
|
"loss": 6.9016,
|
|
"mean_token_accuracy": 0.09040206372737884,
|
|
"num_tokens": 1061218.0,
|
|
"step": 485
|
|
},
|
|
{
|
|
"entropy": 7.510239171981811,
|
|
"epoch": 0.03238491788110109,
|
|
"grad_norm": 1.265625,
|
|
"learning_rate": 0.0002445,
|
|
"loss": 7.0114,
|
|
"mean_token_accuracy": 0.08968377225100994,
|
|
"num_tokens": 1071077.0,
|
|
"step": 490
|
|
},
|
|
{
|
|
"entropy": 7.438390588760376,
|
|
"epoch": 0.03271537622682661,
|
|
"grad_norm": 1.078125,
|
|
"learning_rate": 0.000247,
|
|
"loss": 6.9756,
|
|
"mean_token_accuracy": 0.08917649313807488,
|
|
"num_tokens": 1081434.0,
|
|
"step": 495
|
|
},
|
|
{
|
|
"entropy": 7.337517833709716,
|
|
"epoch": 0.03304583457255213,
|
|
"grad_norm": 1.0390625,
|
|
"learning_rate": 0.0002495,
|
|
"loss": 6.9296,
|
|
"mean_token_accuracy": 0.09428177699446678,
|
|
"num_tokens": 1092237.0,
|
|
"step": 500
|
|
},
|
|
{
|
|
"entropy": 7.469271516799926,
|
|
"epoch": 0.03337629291827765,
|
|
"grad_norm": 0.828125,
|
|
"learning_rate": 0.000252,
|
|
"loss": 7.0125,
|
|
"mean_token_accuracy": 0.08716159909963608,
|
|
"num_tokens": 1102368.0,
|
|
"step": 505
|
|
},
|
|
{
|
|
"entropy": 7.31656436920166,
|
|
"epoch": 0.033706751264003174,
|
|
"grad_norm": 1.09375,
|
|
"learning_rate": 0.0002545,
|
|
"loss": 6.8987,
|
|
"mean_token_accuracy": 0.09418084174394607,
|
|
"num_tokens": 1113174.0,
|
|
"step": 510
|
|
},
|
|
{
|
|
"entropy": 7.376170969009399,
|
|
"epoch": 0.0340372096097287,
|
|
"grad_norm": 1.0546875,
|
|
"learning_rate": 0.000257,
|
|
"loss": 6.9226,
|
|
"mean_token_accuracy": 0.08961384892463684,
|
|
"num_tokens": 1124468.0,
|
|
"step": 515
|
|
},
|
|
{
|
|
"entropy": 7.417706298828125,
|
|
"epoch": 0.03436766795545421,
|
|
"grad_norm": 1.0625,
|
|
"learning_rate": 0.0002595,
|
|
"loss": 6.8741,
|
|
"mean_token_accuracy": 0.08852398991584778,
|
|
"num_tokens": 1134691.0,
|
|
"step": 520
|
|
},
|
|
{
|
|
"entropy": 7.428259468078613,
|
|
"epoch": 0.03469812630117974,
|
|
"grad_norm": 0.84375,
|
|
"learning_rate": 0.000262,
|
|
"loss": 7.0223,
|
|
"mean_token_accuracy": 0.08580638542771339,
|
|
"num_tokens": 1146487.0,
|
|
"step": 525
|
|
},
|
|
{
|
|
"entropy": 7.3816627025604244,
|
|
"epoch": 0.03502858464690526,
|
|
"grad_norm": 1.015625,
|
|
"learning_rate": 0.00026450000000000003,
|
|
"loss": 6.8918,
|
|
"mean_token_accuracy": 0.09239922240376472,
|
|
"num_tokens": 1156897.0,
|
|
"step": 530
|
|
},
|
|
{
|
|
"entropy": 7.373755741119385,
|
|
"epoch": 0.035359042992630776,
|
|
"grad_norm": 1.0546875,
|
|
"learning_rate": 0.00026700000000000004,
|
|
"loss": 6.8115,
|
|
"mean_token_accuracy": 0.09454052895307541,
|
|
"num_tokens": 1167884.0,
|
|
"step": 535
|
|
},
|
|
{
|
|
"entropy": 7.220916080474853,
|
|
"epoch": 0.0356895013383563,
|
|
"grad_norm": 0.9375,
|
|
"learning_rate": 0.00026950000000000005,
|
|
"loss": 6.8725,
|
|
"mean_token_accuracy": 0.09437957406044006,
|
|
"num_tokens": 1178214.0,
|
|
"step": 540
|
|
},
|
|
{
|
|
"entropy": 7.502536869049072,
|
|
"epoch": 0.03601995968408182,
|
|
"grad_norm": 1.0390625,
|
|
"learning_rate": 0.00027200000000000005,
|
|
"loss": 6.9663,
|
|
"mean_token_accuracy": 0.08928380161523819,
|
|
"num_tokens": 1188139.0,
|
|
"step": 545
|
|
},
|
|
{
|
|
"entropy": 7.327961492538452,
|
|
"epoch": 0.03635041802980734,
|
|
"grad_norm": 1.0,
|
|
"learning_rate": 0.0002745,
|
|
"loss": 6.8942,
|
|
"mean_token_accuracy": 0.09445121213793754,
|
|
"num_tokens": 1198349.0,
|
|
"step": 550
|
|
},
|
|
{
|
|
"entropy": 7.297813129425049,
|
|
"epoch": 0.03668087637553286,
|
|
"grad_norm": 0.87109375,
|
|
"learning_rate": 0.000277,
|
|
"loss": 6.868,
|
|
"mean_token_accuracy": 0.09068818464875221,
|
|
"num_tokens": 1212208.0,
|
|
"step": 555
|
|
},
|
|
{
|
|
"entropy": 7.395832061767578,
|
|
"epoch": 0.037011334721258386,
|
|
"grad_norm": 1.0078125,
|
|
"learning_rate": 0.0002795,
|
|
"loss": 6.8705,
|
|
"mean_token_accuracy": 0.09580742493271828,
|
|
"num_tokens": 1222600.0,
|
|
"step": 560
|
|
},
|
|
{
|
|
"entropy": 7.209005451202392,
|
|
"epoch": 0.03734179306698391,
|
|
"grad_norm": 1.046875,
|
|
"learning_rate": 0.00028199999999999997,
|
|
"loss": 6.7736,
|
|
"mean_token_accuracy": 0.09626576229929924,
|
|
"num_tokens": 1233383.0,
|
|
"step": 565
|
|
},
|
|
{
|
|
"entropy": 7.295228719711304,
|
|
"epoch": 0.037672251412709426,
|
|
"grad_norm": 0.99609375,
|
|
"learning_rate": 0.0002845,
|
|
"loss": 6.7866,
|
|
"mean_token_accuracy": 0.09773707985877991,
|
|
"num_tokens": 1243571.0,
|
|
"step": 570
|
|
},
|
|
{
|
|
"entropy": 7.239785385131836,
|
|
"epoch": 0.03800270975843495,
|
|
"grad_norm": 1.2109375,
|
|
"learning_rate": 0.000287,
|
|
"loss": 6.7997,
|
|
"mean_token_accuracy": 0.0885373555123806,
|
|
"num_tokens": 1255150.0,
|
|
"step": 575
|
|
},
|
|
{
|
|
"entropy": 7.133972263336181,
|
|
"epoch": 0.03833316810416047,
|
|
"grad_norm": 0.98046875,
|
|
"learning_rate": 0.0002895,
|
|
"loss": 6.8149,
|
|
"mean_token_accuracy": 0.09696940034627914,
|
|
"num_tokens": 1265994.0,
|
|
"step": 580
|
|
},
|
|
{
|
|
"entropy": 7.3289577007293705,
|
|
"epoch": 0.03866362644988599,
|
|
"grad_norm": 1.1015625,
|
|
"learning_rate": 0.000292,
|
|
"loss": 6.8097,
|
|
"mean_token_accuracy": 0.09445883333683014,
|
|
"num_tokens": 1276244.0,
|
|
"step": 585
|
|
},
|
|
{
|
|
"entropy": 7.231806230545044,
|
|
"epoch": 0.03899408479561151,
|
|
"grad_norm": 1.2421875,
|
|
"learning_rate": 0.0002945,
|
|
"loss": 6.7065,
|
|
"mean_token_accuracy": 0.09788127020001411,
|
|
"num_tokens": 1285479.0,
|
|
"step": 590
|
|
},
|
|
{
|
|
"entropy": 7.2062479019165036,
|
|
"epoch": 0.039324543141337036,
|
|
"grad_norm": 1.015625,
|
|
"learning_rate": 0.000297,
|
|
"loss": 6.8178,
|
|
"mean_token_accuracy": 0.09362647384405136,
|
|
"num_tokens": 1295591.0,
|
|
"step": 595
|
|
},
|
|
{
|
|
"entropy": 7.217821264266968,
|
|
"epoch": 0.03965500148706256,
|
|
"grad_norm": 0.90625,
|
|
"learning_rate": 0.0002995,
|
|
"loss": 6.7382,
|
|
"mean_token_accuracy": 0.10009469687938691,
|
|
"num_tokens": 1305516.0,
|
|
"step": 600
|
|
},
|
|
{
|
|
"entropy": 7.20532751083374,
|
|
"epoch": 0.039985459832788076,
|
|
"grad_norm": 1.0078125,
|
|
"learning_rate": 0.000302,
|
|
"loss": 6.7464,
|
|
"mean_token_accuracy": 0.09776026606559754,
|
|
"num_tokens": 1316345.0,
|
|
"step": 605
|
|
},
|
|
{
|
|
"entropy": 7.265520668029785,
|
|
"epoch": 0.0403159181785136,
|
|
"grad_norm": 0.99609375,
|
|
"learning_rate": 0.0003045,
|
|
"loss": 6.896,
|
|
"mean_token_accuracy": 0.09152787327766418,
|
|
"num_tokens": 1327332.0,
|
|
"step": 610
|
|
},
|
|
{
|
|
"entropy": 7.19921088218689,
|
|
"epoch": 0.04064637652423912,
|
|
"grad_norm": 1.0546875,
|
|
"learning_rate": 0.000307,
|
|
"loss": 6.7434,
|
|
"mean_token_accuracy": 0.09418585896492004,
|
|
"num_tokens": 1337572.0,
|
|
"step": 615
|
|
},
|
|
{
|
|
"entropy": 7.084140110015869,
|
|
"epoch": 0.04097683486996464,
|
|
"grad_norm": 1.046875,
|
|
"learning_rate": 0.0003095,
|
|
"loss": 6.7097,
|
|
"mean_token_accuracy": 0.09913697764277458,
|
|
"num_tokens": 1347712.0,
|
|
"step": 620
|
|
},
|
|
{
|
|
"entropy": 7.243695592880249,
|
|
"epoch": 0.04130729321569016,
|
|
"grad_norm": 0.859375,
|
|
"learning_rate": 0.000312,
|
|
"loss": 6.7867,
|
|
"mean_token_accuracy": 0.0924969807267189,
|
|
"num_tokens": 1359318.0,
|
|
"step": 625
|
|
},
|
|
{
|
|
"entropy": 7.179306936264038,
|
|
"epoch": 0.041637751561415685,
|
|
"grad_norm": 0.92578125,
|
|
"learning_rate": 0.0003145,
|
|
"loss": 6.7094,
|
|
"mean_token_accuracy": 0.09823217391967773,
|
|
"num_tokens": 1370322.0,
|
|
"step": 630
|
|
},
|
|
{
|
|
"entropy": 7.07304048538208,
|
|
"epoch": 0.0419682099071412,
|
|
"grad_norm": 0.92578125,
|
|
"learning_rate": 0.000317,
|
|
"loss": 6.7183,
|
|
"mean_token_accuracy": 0.09707441851496697,
|
|
"num_tokens": 1382660.0,
|
|
"step": 635
|
|
},
|
|
{
|
|
"entropy": 7.1794665336608885,
|
|
"epoch": 0.042298668252866725,
|
|
"grad_norm": 0.88671875,
|
|
"learning_rate": 0.0003195,
|
|
"loss": 6.7788,
|
|
"mean_token_accuracy": 0.09463305547833442,
|
|
"num_tokens": 1394377.0,
|
|
"step": 640
|
|
},
|
|
{
|
|
"entropy": 7.054516887664795,
|
|
"epoch": 0.04262912659859225,
|
|
"grad_norm": 0.90625,
|
|
"learning_rate": 0.000322,
|
|
"loss": 6.7162,
|
|
"mean_token_accuracy": 0.09280202016234398,
|
|
"num_tokens": 1404680.0,
|
|
"step": 645
|
|
},
|
|
{
|
|
"entropy": 7.1342448711395265,
|
|
"epoch": 0.04295958494431777,
|
|
"grad_norm": 1.2109375,
|
|
"learning_rate": 0.00032450000000000003,
|
|
"loss": 6.7346,
|
|
"mean_token_accuracy": 0.09771808311343193,
|
|
"num_tokens": 1416715.0,
|
|
"step": 650
|
|
},
|
|
{
|
|
"entropy": 7.097233819961548,
|
|
"epoch": 0.04329004329004329,
|
|
"grad_norm": 0.921875,
|
|
"learning_rate": 0.00032700000000000003,
|
|
"loss": 6.6857,
|
|
"mean_token_accuracy": 0.10034304857254028,
|
|
"num_tokens": 1427354.0,
|
|
"step": 655
|
|
},
|
|
{
|
|
"entropy": 7.1087141036987305,
|
|
"epoch": 0.04362050163576881,
|
|
"grad_norm": 1.0859375,
|
|
"learning_rate": 0.00032950000000000004,
|
|
"loss": 6.6542,
|
|
"mean_token_accuracy": 0.09519000574946404,
|
|
"num_tokens": 1438554.0,
|
|
"step": 660
|
|
},
|
|
{
|
|
"entropy": 7.193477153778076,
|
|
"epoch": 0.043950959981494335,
|
|
"grad_norm": 1.03125,
|
|
"learning_rate": 0.00033200000000000005,
|
|
"loss": 6.8155,
|
|
"mean_token_accuracy": 0.09368375465273857,
|
|
"num_tokens": 1449213.0,
|
|
"step": 665
|
|
},
|
|
{
|
|
"entropy": 7.187811279296875,
|
|
"epoch": 0.04428141832721985,
|
|
"grad_norm": 0.98046875,
|
|
"learning_rate": 0.00033450000000000005,
|
|
"loss": 6.7209,
|
|
"mean_token_accuracy": 0.09319828450679779,
|
|
"num_tokens": 1460571.0,
|
|
"step": 670
|
|
},
|
|
{
|
|
"entropy": 7.06358699798584,
|
|
"epoch": 0.044611876672945375,
|
|
"grad_norm": 1.046875,
|
|
"learning_rate": 0.000337,
|
|
"loss": 6.7633,
|
|
"mean_token_accuracy": 0.09224831759929657,
|
|
"num_tokens": 1470638.0,
|
|
"step": 675
|
|
},
|
|
{
|
|
"entropy": 7.195232009887695,
|
|
"epoch": 0.0449423350186709,
|
|
"grad_norm": 1.0,
|
|
"learning_rate": 0.0003395,
|
|
"loss": 6.7537,
|
|
"mean_token_accuracy": 0.09635311663150788,
|
|
"num_tokens": 1481773.0,
|
|
"step": 680
|
|
},
|
|
{
|
|
"entropy": 7.032809543609619,
|
|
"epoch": 0.045272793364396414,
|
|
"grad_norm": 0.95703125,
|
|
"learning_rate": 0.000342,
|
|
"loss": 6.6993,
|
|
"mean_token_accuracy": 0.09740457832813262,
|
|
"num_tokens": 1491982.0,
|
|
"step": 685
|
|
},
|
|
{
|
|
"entropy": 7.1097876071929935,
|
|
"epoch": 0.04560325171012194,
|
|
"grad_norm": 0.94921875,
|
|
"learning_rate": 0.00034449999999999997,
|
|
"loss": 6.6839,
|
|
"mean_token_accuracy": 0.10262897387146949,
|
|
"num_tokens": 1503027.0,
|
|
"step": 690
|
|
},
|
|
{
|
|
"entropy": 7.084706974029541,
|
|
"epoch": 0.04593371005584746,
|
|
"grad_norm": 0.98046875,
|
|
"learning_rate": 0.000347,
|
|
"loss": 6.7336,
|
|
"mean_token_accuracy": 0.0956760212779045,
|
|
"num_tokens": 1515977.0,
|
|
"step": 695
|
|
},
|
|
{
|
|
"entropy": 6.980908727645874,
|
|
"epoch": 0.046264168401572985,
|
|
"grad_norm": 1.0078125,
|
|
"learning_rate": 0.0003495,
|
|
"loss": 6.6262,
|
|
"mean_token_accuracy": 0.09501065239310265,
|
|
"num_tokens": 1527309.0,
|
|
"step": 700
|
|
},
|
|
{
|
|
"entropy": 6.991337585449219,
|
|
"epoch": 0.0465946267472985,
|
|
"grad_norm": 1.109375,
|
|
"learning_rate": 0.000352,
|
|
"loss": 6.6428,
|
|
"mean_token_accuracy": 0.10282542034983636,
|
|
"num_tokens": 1538233.0,
|
|
"step": 705
|
|
},
|
|
{
|
|
"entropy": 7.046284151077271,
|
|
"epoch": 0.046925085093024024,
|
|
"grad_norm": 0.87890625,
|
|
"learning_rate": 0.0003545,
|
|
"loss": 6.6239,
|
|
"mean_token_accuracy": 0.09860012754797935,
|
|
"num_tokens": 1549893.0,
|
|
"step": 710
|
|
},
|
|
{
|
|
"entropy": 7.061589002609253,
|
|
"epoch": 0.04725554343874955,
|
|
"grad_norm": 1.0234375,
|
|
"learning_rate": 0.000357,
|
|
"loss": 6.6848,
|
|
"mean_token_accuracy": 0.09669168144464493,
|
|
"num_tokens": 1560926.0,
|
|
"step": 715
|
|
},
|
|
{
|
|
"entropy": 6.923026990890503,
|
|
"epoch": 0.047586001784475064,
|
|
"grad_norm": 0.9453125,
|
|
"learning_rate": 0.0003595,
|
|
"loss": 6.5353,
|
|
"mean_token_accuracy": 0.10450014173984527,
|
|
"num_tokens": 1571826.0,
|
|
"step": 720
|
|
},
|
|
{
|
|
"entropy": 7.062051105499267,
|
|
"epoch": 0.04791646013020059,
|
|
"grad_norm": 0.875,
|
|
"learning_rate": 0.000362,
|
|
"loss": 6.7041,
|
|
"mean_token_accuracy": 0.10204312726855277,
|
|
"num_tokens": 1582710.0,
|
|
"step": 725
|
|
},
|
|
{
|
|
"entropy": 7.0487127780914305,
|
|
"epoch": 0.04824691847592611,
|
|
"grad_norm": 0.93359375,
|
|
"learning_rate": 0.0003645,
|
|
"loss": 6.5516,
|
|
"mean_token_accuracy": 0.09832823127508164,
|
|
"num_tokens": 1593148.0,
|
|
"step": 730
|
|
},
|
|
{
|
|
"entropy": 6.99473762512207,
|
|
"epoch": 0.048577376821651634,
|
|
"grad_norm": 1.140625,
|
|
"learning_rate": 0.000367,
|
|
"loss": 6.6138,
|
|
"mean_token_accuracy": 0.10036826729774476,
|
|
"num_tokens": 1603286.0,
|
|
"step": 735
|
|
},
|
|
{
|
|
"entropy": 6.8419677734375,
|
|
"epoch": 0.04890783516737715,
|
|
"grad_norm": 1.03125,
|
|
"learning_rate": 0.0003695,
|
|
"loss": 6.5385,
|
|
"mean_token_accuracy": 0.10537053346633911,
|
|
"num_tokens": 1613341.0,
|
|
"step": 740
|
|
},
|
|
{
|
|
"entropy": 7.0657447338104244,
|
|
"epoch": 0.049238293513102674,
|
|
"grad_norm": 0.9375,
|
|
"learning_rate": 0.000372,
|
|
"loss": 6.6533,
|
|
"mean_token_accuracy": 0.09263939708471298,
|
|
"num_tokens": 1624733.0,
|
|
"step": 745
|
|
},
|
|
{
|
|
"entropy": 6.944655704498291,
|
|
"epoch": 0.0495687518588282,
|
|
"grad_norm": 0.85546875,
|
|
"learning_rate": 0.0003745,
|
|
"loss": 6.5633,
|
|
"mean_token_accuracy": 0.0979912631213665,
|
|
"num_tokens": 1635369.0,
|
|
"step": 750
|
|
},
|
|
{
|
|
"entropy": 6.999823999404907,
|
|
"epoch": 0.049899210204553714,
|
|
"grad_norm": 1.0546875,
|
|
"learning_rate": 0.000377,
|
|
"loss": 6.6318,
|
|
"mean_token_accuracy": 0.1003281831741333,
|
|
"num_tokens": 1646391.0,
|
|
"step": 755
|
|
},
|
|
{
|
|
"entropy": 6.9726362228393555,
|
|
"epoch": 0.05022966855027924,
|
|
"grad_norm": 0.94140625,
|
|
"learning_rate": 0.0003795,
|
|
"loss": 6.6721,
|
|
"mean_token_accuracy": 0.09612520188093185,
|
|
"num_tokens": 1657923.0,
|
|
"step": 760
|
|
},
|
|
{
|
|
"entropy": 6.905797386169434,
|
|
"epoch": 0.05056012689600476,
|
|
"grad_norm": 0.86328125,
|
|
"learning_rate": 0.000382,
|
|
"loss": 6.5601,
|
|
"mean_token_accuracy": 0.10316932722926139,
|
|
"num_tokens": 1669732.0,
|
|
"step": 765
|
|
},
|
|
{
|
|
"entropy": 6.8569001197814945,
|
|
"epoch": 0.05089058524173028,
|
|
"grad_norm": 0.96484375,
|
|
"learning_rate": 0.0003845,
|
|
"loss": 6.5063,
|
|
"mean_token_accuracy": 0.09963946118950844,
|
|
"num_tokens": 1680646.0,
|
|
"step": 770
|
|
},
|
|
{
|
|
"entropy": 6.952554941177368,
|
|
"epoch": 0.0512210435874558,
|
|
"grad_norm": 0.95703125,
|
|
"learning_rate": 0.00038700000000000003,
|
|
"loss": 6.62,
|
|
"mean_token_accuracy": 0.10107743218541146,
|
|
"num_tokens": 1690855.0,
|
|
"step": 775
|
|
},
|
|
{
|
|
"entropy": 6.992598533630371,
|
|
"epoch": 0.05155150193318132,
|
|
"grad_norm": 0.9140625,
|
|
"learning_rate": 0.00038950000000000003,
|
|
"loss": 6.5594,
|
|
"mean_token_accuracy": 0.10028711706399918,
|
|
"num_tokens": 1701798.0,
|
|
"step": 780
|
|
},
|
|
{
|
|
"entropy": 6.819104719161987,
|
|
"epoch": 0.05188196027890685,
|
|
"grad_norm": 0.875,
|
|
"learning_rate": 0.00039200000000000004,
|
|
"loss": 6.5012,
|
|
"mean_token_accuracy": 0.1022326961159706,
|
|
"num_tokens": 1713307.0,
|
|
"step": 785
|
|
},
|
|
{
|
|
"entropy": 6.940773916244507,
|
|
"epoch": 0.05221241862463236,
|
|
"grad_norm": 1.0625,
|
|
"learning_rate": 0.00039450000000000005,
|
|
"loss": 6.6239,
|
|
"mean_token_accuracy": 0.09705120101571083,
|
|
"num_tokens": 1724158.0,
|
|
"step": 790
|
|
},
|
|
{
|
|
"entropy": 6.82653489112854,
|
|
"epoch": 0.05254287697035789,
|
|
"grad_norm": 0.9609375,
|
|
"learning_rate": 0.00039700000000000005,
|
|
"loss": 6.5774,
|
|
"mean_token_accuracy": 0.10342817828059196,
|
|
"num_tokens": 1734978.0,
|
|
"step": 795
|
|
},
|
|
{
|
|
"entropy": 7.033143377304077,
|
|
"epoch": 0.05287333531608341,
|
|
"grad_norm": 0.8984375,
|
|
"learning_rate": 0.0003995,
|
|
"loss": 6.5952,
|
|
"mean_token_accuracy": 0.09391836002469063,
|
|
"num_tokens": 1745757.0,
|
|
"step": 800
|
|
},
|
|
{
|
|
"entropy": 6.914671754837036,
|
|
"epoch": 0.053203793661808926,
|
|
"grad_norm": 1.03125,
|
|
"learning_rate": 0.000402,
|
|
"loss": 6.5992,
|
|
"mean_token_accuracy": 0.10751016363501549,
|
|
"num_tokens": 1755370.0,
|
|
"step": 805
|
|
},
|
|
{
|
|
"entropy": 6.898441886901855,
|
|
"epoch": 0.05353425200753445,
|
|
"grad_norm": 1.140625,
|
|
"learning_rate": 0.0004045,
|
|
"loss": 6.539,
|
|
"mean_token_accuracy": 0.10239029303193092,
|
|
"num_tokens": 1766031.0,
|
|
"step": 810
|
|
},
|
|
{
|
|
"entropy": 6.88755989074707,
|
|
"epoch": 0.05386471035325997,
|
|
"grad_norm": 1.03125,
|
|
"learning_rate": 0.00040699999999999997,
|
|
"loss": 6.5829,
|
|
"mean_token_accuracy": 0.10488124564290047,
|
|
"num_tokens": 1778672.0,
|
|
"step": 815
|
|
},
|
|
{
|
|
"entropy": 6.787546920776367,
|
|
"epoch": 0.05419516869898549,
|
|
"grad_norm": 1.03125,
|
|
"learning_rate": 0.0004095,
|
|
"loss": 6.5286,
|
|
"mean_token_accuracy": 0.10559183657169342,
|
|
"num_tokens": 1789912.0,
|
|
"step": 820
|
|
},
|
|
{
|
|
"entropy": 6.817356538772583,
|
|
"epoch": 0.05452562704471101,
|
|
"grad_norm": 0.96484375,
|
|
"learning_rate": 0.000412,
|
|
"loss": 6.4381,
|
|
"mean_token_accuracy": 0.10110691860318184,
|
|
"num_tokens": 1800350.0,
|
|
"step": 825
|
|
},
|
|
{
|
|
"entropy": 6.801516056060791,
|
|
"epoch": 0.054856085390436536,
|
|
"grad_norm": 1.0859375,
|
|
"learning_rate": 0.0004145,
|
|
"loss": 6.4627,
|
|
"mean_token_accuracy": 0.10351857915520668,
|
|
"num_tokens": 1811353.0,
|
|
"step": 830
|
|
},
|
|
{
|
|
"entropy": 6.85851640701294,
|
|
"epoch": 0.05518654373616206,
|
|
"grad_norm": 1.015625,
|
|
"learning_rate": 0.000417,
|
|
"loss": 6.597,
|
|
"mean_token_accuracy": 0.1018621951341629,
|
|
"num_tokens": 1822702.0,
|
|
"step": 835
|
|
},
|
|
{
|
|
"entropy": 6.856171178817749,
|
|
"epoch": 0.055517002081887576,
|
|
"grad_norm": 1.09375,
|
|
"learning_rate": 0.0004195,
|
|
"loss": 6.6124,
|
|
"mean_token_accuracy": 0.10211483016610146,
|
|
"num_tokens": 1833520.0,
|
|
"step": 840
|
|
},
|
|
{
|
|
"entropy": 6.887248086929321,
|
|
"epoch": 0.0558474604276131,
|
|
"grad_norm": 1.0234375,
|
|
"learning_rate": 0.000422,
|
|
"loss": 6.5151,
|
|
"mean_token_accuracy": 0.10373834893107414,
|
|
"num_tokens": 1843905.0,
|
|
"step": 845
|
|
},
|
|
{
|
|
"entropy": 6.8532782077789305,
|
|
"epoch": 0.05617791877333862,
|
|
"grad_norm": 1.296875,
|
|
"learning_rate": 0.0004245,
|
|
"loss": 6.6166,
|
|
"mean_token_accuracy": 0.10005839541554451,
|
|
"num_tokens": 1854350.0,
|
|
"step": 850
|
|
},
|
|
{
|
|
"entropy": 6.9412201881408695,
|
|
"epoch": 0.05650837711906414,
|
|
"grad_norm": 1.125,
|
|
"learning_rate": 0.000427,
|
|
"loss": 6.5499,
|
|
"mean_token_accuracy": 0.10348896309733391,
|
|
"num_tokens": 1864279.0,
|
|
"step": 855
|
|
},
|
|
{
|
|
"entropy": 6.724023771286011,
|
|
"epoch": 0.05683883546478966,
|
|
"grad_norm": 0.87890625,
|
|
"learning_rate": 0.0004295,
|
|
"loss": 6.5691,
|
|
"mean_token_accuracy": 0.10336827114224434,
|
|
"num_tokens": 1876618.0,
|
|
"step": 860
|
|
},
|
|
{
|
|
"entropy": 6.99163589477539,
|
|
"epoch": 0.057169293810515186,
|
|
"grad_norm": 0.984375,
|
|
"learning_rate": 0.000432,
|
|
"loss": 6.5638,
|
|
"mean_token_accuracy": 0.10379870086908341,
|
|
"num_tokens": 1888689.0,
|
|
"step": 865
|
|
},
|
|
{
|
|
"entropy": 6.671361923217773,
|
|
"epoch": 0.05749975215624071,
|
|
"grad_norm": 1.0078125,
|
|
"learning_rate": 0.0004345,
|
|
"loss": 6.4902,
|
|
"mean_token_accuracy": 0.09828232303261757,
|
|
"num_tokens": 1899527.0,
|
|
"step": 870
|
|
},
|
|
{
|
|
"entropy": 6.922105932235718,
|
|
"epoch": 0.057830210501966225,
|
|
"grad_norm": 1.140625,
|
|
"learning_rate": 0.000437,
|
|
"loss": 6.5955,
|
|
"mean_token_accuracy": 0.09187263101339341,
|
|
"num_tokens": 1910212.0,
|
|
"step": 875
|
|
},
|
|
{
|
|
"entropy": 6.837419080734253,
|
|
"epoch": 0.05816066884769175,
|
|
"grad_norm": 0.8984375,
|
|
"learning_rate": 0.0004395,
|
|
"loss": 6.4729,
|
|
"mean_token_accuracy": 0.10792759358882904,
|
|
"num_tokens": 1919938.0,
|
|
"step": 880
|
|
},
|
|
{
|
|
"entropy": 6.734318637847901,
|
|
"epoch": 0.05849112719341727,
|
|
"grad_norm": 1.0390625,
|
|
"learning_rate": 0.000442,
|
|
"loss": 6.4306,
|
|
"mean_token_accuracy": 0.10667424649000168,
|
|
"num_tokens": 1929984.0,
|
|
"step": 885
|
|
},
|
|
{
|
|
"entropy": 6.738415002822876,
|
|
"epoch": 0.05882158553914279,
|
|
"grad_norm": 0.97265625,
|
|
"learning_rate": 0.0004445,
|
|
"loss": 6.569,
|
|
"mean_token_accuracy": 0.09858254492282867,
|
|
"num_tokens": 1941683.0,
|
|
"step": 890
|
|
},
|
|
{
|
|
"entropy": 6.901988840103149,
|
|
"epoch": 0.05915204388486831,
|
|
"grad_norm": 0.9296875,
|
|
"learning_rate": 0.000447,
|
|
"loss": 6.5382,
|
|
"mean_token_accuracy": 0.09865912571549415,
|
|
"num_tokens": 1952847.0,
|
|
"step": 895
|
|
},
|
|
{
|
|
"entropy": 6.761583042144776,
|
|
"epoch": 0.059482502230593835,
|
|
"grad_norm": 1.171875,
|
|
"learning_rate": 0.00044950000000000003,
|
|
"loss": 6.4625,
|
|
"mean_token_accuracy": 0.10366626009345055,
|
|
"num_tokens": 1963449.0,
|
|
"step": 900
|
|
},
|
|
{
|
|
"entropy": 6.7592977523803714,
|
|
"epoch": 0.05981296057631935,
|
|
"grad_norm": 0.9453125,
|
|
"learning_rate": 0.00045200000000000004,
|
|
"loss": 6.4768,
|
|
"mean_token_accuracy": 0.10366302728652954,
|
|
"num_tokens": 1973855.0,
|
|
"step": 905
|
|
},
|
|
{
|
|
"entropy": 6.7528948307037355,
|
|
"epoch": 0.060143418922044875,
|
|
"grad_norm": 1.0625,
|
|
"learning_rate": 0.00045450000000000004,
|
|
"loss": 6.472,
|
|
"mean_token_accuracy": 0.10176084563136101,
|
|
"num_tokens": 1985015.0,
|
|
"step": 910
|
|
},
|
|
{
|
|
"entropy": 6.7731701850891115,
|
|
"epoch": 0.0604738772677704,
|
|
"grad_norm": 0.97265625,
|
|
"learning_rate": 0.00045700000000000005,
|
|
"loss": 6.5439,
|
|
"mean_token_accuracy": 0.10046328157186508,
|
|
"num_tokens": 1996042.0,
|
|
"step": 915
|
|
},
|
|
{
|
|
"entropy": 6.770084190368652,
|
|
"epoch": 0.06080433561349592,
|
|
"grad_norm": 0.91015625,
|
|
"learning_rate": 0.00045950000000000006,
|
|
"loss": 6.4981,
|
|
"mean_token_accuracy": 0.10596563145518303,
|
|
"num_tokens": 2008155.0,
|
|
"step": 920
|
|
},
|
|
{
|
|
"entropy": 6.788337850570679,
|
|
"epoch": 0.06113479395922144,
|
|
"grad_norm": 0.953125,
|
|
"learning_rate": 0.000462,
|
|
"loss": 6.4521,
|
|
"mean_token_accuracy": 0.1128300666809082,
|
|
"num_tokens": 2017975.0,
|
|
"step": 925
|
|
},
|
|
{
|
|
"entropy": 6.695605039596558,
|
|
"epoch": 0.06146525230494696,
|
|
"grad_norm": 0.92578125,
|
|
"learning_rate": 0.0004645,
|
|
"loss": 6.363,
|
|
"mean_token_accuracy": 0.10490168184041977,
|
|
"num_tokens": 2029210.0,
|
|
"step": 930
|
|
},
|
|
{
|
|
"entropy": 6.72561674118042,
|
|
"epoch": 0.061795710650672485,
|
|
"grad_norm": 1.0546875,
|
|
"learning_rate": 0.000467,
|
|
"loss": 6.4598,
|
|
"mean_token_accuracy": 0.10339713767170906,
|
|
"num_tokens": 2039131.0,
|
|
"step": 935
|
|
},
|
|
{
|
|
"entropy": 6.664773797988891,
|
|
"epoch": 0.062126168996398,
|
|
"grad_norm": 1.0703125,
|
|
"learning_rate": 0.0004695,
|
|
"loss": 6.501,
|
|
"mean_token_accuracy": 0.10852292701601982,
|
|
"num_tokens": 2049865.0,
|
|
"step": 940
|
|
},
|
|
{
|
|
"entropy": 6.772005081176758,
|
|
"epoch": 0.062456627342123525,
|
|
"grad_norm": 1.1640625,
|
|
"learning_rate": 0.000472,
|
|
"loss": 6.5005,
|
|
"mean_token_accuracy": 0.1033253699541092,
|
|
"num_tokens": 2061151.0,
|
|
"step": 945
|
|
},
|
|
{
|
|
"entropy": 6.687468147277832,
|
|
"epoch": 0.06278708568784905,
|
|
"grad_norm": 1.046875,
|
|
"learning_rate": 0.0004745,
|
|
"loss": 6.3654,
|
|
"mean_token_accuracy": 0.10364857614040375,
|
|
"num_tokens": 2071492.0,
|
|
"step": 950
|
|
},
|
|
{
|
|
"entropy": 6.708350610733032,
|
|
"epoch": 0.06311754403357457,
|
|
"grad_norm": 1.015625,
|
|
"learning_rate": 0.000477,
|
|
"loss": 6.4088,
|
|
"mean_token_accuracy": 0.10490176752209664,
|
|
"num_tokens": 2083291.0,
|
|
"step": 955
|
|
},
|
|
{
|
|
"entropy": 6.738237047195435,
|
|
"epoch": 0.0634480023793001,
|
|
"grad_norm": 0.96484375,
|
|
"learning_rate": 0.0004795,
|
|
"loss": 6.5716,
|
|
"mean_token_accuracy": 0.10328001081943512,
|
|
"num_tokens": 2094515.0,
|
|
"step": 960
|
|
},
|
|
{
|
|
"entropy": 6.7747293472290036,
|
|
"epoch": 0.0637784607250256,
|
|
"grad_norm": 1.109375,
|
|
"learning_rate": 0.000482,
|
|
"loss": 6.4469,
|
|
"mean_token_accuracy": 0.103257467597723,
|
|
"num_tokens": 2104774.0,
|
|
"step": 965
|
|
},
|
|
{
|
|
"entropy": 6.619922113418579,
|
|
"epoch": 0.06410891907075113,
|
|
"grad_norm": 1.0,
|
|
"learning_rate": 0.0004845,
|
|
"loss": 6.3375,
|
|
"mean_token_accuracy": 0.11283469498157501,
|
|
"num_tokens": 2114751.0,
|
|
"step": 970
|
|
},
|
|
{
|
|
"entropy": 6.717543745040894,
|
|
"epoch": 0.06443937741647665,
|
|
"grad_norm": 0.81640625,
|
|
"learning_rate": 0.000487,
|
|
"loss": 6.3979,
|
|
"mean_token_accuracy": 0.10386243537068367,
|
|
"num_tokens": 2126815.0,
|
|
"step": 975
|
|
},
|
|
{
|
|
"entropy": 6.669224214553833,
|
|
"epoch": 0.06476983576220217,
|
|
"grad_norm": 0.90234375,
|
|
"learning_rate": 0.0004895,
|
|
"loss": 6.4599,
|
|
"mean_token_accuracy": 0.10235354974865914,
|
|
"num_tokens": 2137951.0,
|
|
"step": 980
|
|
},
|
|
{
|
|
"entropy": 6.731771755218506,
|
|
"epoch": 0.0651002941079277,
|
|
"grad_norm": 1.0,
|
|
"learning_rate": 0.000492,
|
|
"loss": 6.5617,
|
|
"mean_token_accuracy": 0.10181663483381272,
|
|
"num_tokens": 2149291.0,
|
|
"step": 985
|
|
},
|
|
{
|
|
"entropy": 6.82238655090332,
|
|
"epoch": 0.06543075245365322,
|
|
"grad_norm": 0.921875,
|
|
"learning_rate": 0.0004945,
|
|
"loss": 6.5691,
|
|
"mean_token_accuracy": 0.10031550377607346,
|
|
"num_tokens": 2161351.0,
|
|
"step": 990
|
|
},
|
|
{
|
|
"entropy": 6.644531011581421,
|
|
"epoch": 0.06576121079937874,
|
|
"grad_norm": 0.9140625,
|
|
"learning_rate": 0.000497,
|
|
"loss": 6.3416,
|
|
"mean_token_accuracy": 0.1120948150753975,
|
|
"num_tokens": 2172832.0,
|
|
"step": 995
|
|
},
|
|
{
|
|
"entropy": 6.686471271514892,
|
|
"epoch": 0.06609166914510425,
|
|
"grad_norm": 1.0546875,
|
|
"learning_rate": 0.0004995,
|
|
"loss": 6.4627,
|
|
"mean_token_accuracy": 0.09900413081049919,
|
|
"num_tokens": 2183151.0,
|
|
"step": 1000
|
|
}
|
|
],
|
|
"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": 535588844175360.0,
|
|
"train_batch_size": 16,
|
|
"trial_name": null,
|
|
"trial_params": null
|
|
}
|