1035 lines
27 KiB
JSON
1035 lines
27 KiB
JSON
{
|
|
"best_global_step": null,
|
|
"best_metric": null,
|
|
"best_model_checkpoint": null,
|
|
"epoch": 0.03304583457255213,
|
|
"eval_steps": 500,
|
|
"global_step": 500,
|
|
"is_hyper_param_search": false,
|
|
"is_local_process_zero": true,
|
|
"is_world_process_zero": true,
|
|
"log_history": [
|
|
{
|
|
"entropy": 10.742584800720214,
|
|
"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
|
|
}
|
|
],
|
|
"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": 265734562775040.0,
|
|
"train_batch_size": 16,
|
|
"trial_name": null,
|
|
"trial_params": null
|
|
}
|