1035 lines
27 KiB
JSON
1035 lines
27 KiB
JSON
{
|
|
"best_global_step": null,
|
|
"best_metric": null,
|
|
"best_model_checkpoint": null,
|
|
"epoch": 0.0292243848266994,
|
|
"eval_steps": 500,
|
|
"global_step": 500,
|
|
"is_hyper_param_search": false,
|
|
"is_local_process_zero": true,
|
|
"is_world_process_zero": true,
|
|
"log_history": [
|
|
{
|
|
"entropy": 10.742551231384278,
|
|
"epoch": 0.000292243848266994,
|
|
"grad_norm": 4.9375,
|
|
"learning_rate": 2e-06,
|
|
"loss": 10.7723,
|
|
"mean_token_accuracy": 9.259259095415474e-05,
|
|
"num_tokens": 11114.0,
|
|
"step": 5
|
|
},
|
|
{
|
|
"entropy": 10.742591094970702,
|
|
"epoch": 0.000584487696533988,
|
|
"grad_norm": 5.4375,
|
|
"learning_rate": 4.5e-06,
|
|
"loss": 10.7466,
|
|
"mean_token_accuracy": 0.0,
|
|
"num_tokens": 20935.0,
|
|
"step": 10
|
|
},
|
|
{
|
|
"entropy": 10.742585849761962,
|
|
"epoch": 0.000876731544800982,
|
|
"grad_norm": 5.09375,
|
|
"learning_rate": 7e-06,
|
|
"loss": 10.7483,
|
|
"mean_token_accuracy": 0.0,
|
|
"num_tokens": 31930.0,
|
|
"step": 15
|
|
},
|
|
{
|
|
"entropy": 10.742614936828613,
|
|
"epoch": 0.001168975393067976,
|
|
"grad_norm": 5.125,
|
|
"learning_rate": 9.5e-06,
|
|
"loss": 10.6676,
|
|
"mean_token_accuracy": 0.0005568421445786953,
|
|
"num_tokens": 41672.0,
|
|
"step": 20
|
|
},
|
|
{
|
|
"entropy": 10.742519950866699,
|
|
"epoch": 0.00146121924133497,
|
|
"grad_norm": 4.9375,
|
|
"learning_rate": 1.2e-05,
|
|
"loss": 10.5834,
|
|
"mean_token_accuracy": 0.013186294632032514,
|
|
"num_tokens": 52510.0,
|
|
"step": 25
|
|
},
|
|
{
|
|
"entropy": 10.742103958129883,
|
|
"epoch": 0.001753463089601964,
|
|
"grad_norm": 4.3125,
|
|
"learning_rate": 1.4500000000000002e-05,
|
|
"loss": 10.4654,
|
|
"mean_token_accuracy": 0.04129958841949701,
|
|
"num_tokens": 63214.0,
|
|
"step": 30
|
|
},
|
|
{
|
|
"entropy": 10.740110874176025,
|
|
"epoch": 0.002045706937868958,
|
|
"grad_norm": 3.203125,
|
|
"learning_rate": 1.7000000000000003e-05,
|
|
"loss": 10.305,
|
|
"mean_token_accuracy": 0.05048500820994377,
|
|
"num_tokens": 73942.0,
|
|
"step": 35
|
|
},
|
|
{
|
|
"entropy": 10.733905601501466,
|
|
"epoch": 0.002337950786135952,
|
|
"grad_norm": 2.4375,
|
|
"learning_rate": 1.95e-05,
|
|
"loss": 10.1895,
|
|
"mean_token_accuracy": 0.051587145403027536,
|
|
"num_tokens": 84526.0,
|
|
"step": 40
|
|
},
|
|
{
|
|
"entropy": 10.72324333190918,
|
|
"epoch": 0.0026301946344029457,
|
|
"grad_norm": 2.125,
|
|
"learning_rate": 2.2e-05,
|
|
"loss": 10.1139,
|
|
"mean_token_accuracy": 0.04853160977363587,
|
|
"num_tokens": 95178.0,
|
|
"step": 45
|
|
},
|
|
{
|
|
"entropy": 10.714754772186279,
|
|
"epoch": 0.00292243848266994,
|
|
"grad_norm": 2.140625,
|
|
"learning_rate": 2.4500000000000003e-05,
|
|
"loss": 10.0146,
|
|
"mean_token_accuracy": 0.05312369242310524,
|
|
"num_tokens": 106653.0,
|
|
"step": 50
|
|
},
|
|
{
|
|
"entropy": 10.708310794830322,
|
|
"epoch": 0.0032146823309369336,
|
|
"grad_norm": 1.96875,
|
|
"learning_rate": 2.7e-05,
|
|
"loss": 9.9113,
|
|
"mean_token_accuracy": 0.050825309753417966,
|
|
"num_tokens": 115969.0,
|
|
"step": 55
|
|
},
|
|
{
|
|
"entropy": 10.704185581207275,
|
|
"epoch": 0.003506926179203928,
|
|
"grad_norm": 1.7734375,
|
|
"learning_rate": 2.95e-05,
|
|
"loss": 9.8939,
|
|
"mean_token_accuracy": 0.048680877685546874,
|
|
"num_tokens": 125826.0,
|
|
"step": 60
|
|
},
|
|
{
|
|
"entropy": 10.701387882232666,
|
|
"epoch": 0.0037991700274709215,
|
|
"grad_norm": 1.890625,
|
|
"learning_rate": 3.2e-05,
|
|
"loss": 9.872,
|
|
"mean_token_accuracy": 0.04808175079524517,
|
|
"num_tokens": 136677.0,
|
|
"step": 65
|
|
},
|
|
{
|
|
"entropy": 10.697122478485108,
|
|
"epoch": 0.004091413875737916,
|
|
"grad_norm": 1.8125,
|
|
"learning_rate": 3.4500000000000005e-05,
|
|
"loss": 9.8236,
|
|
"mean_token_accuracy": 0.04725121222436428,
|
|
"num_tokens": 147535.0,
|
|
"step": 70
|
|
},
|
|
{
|
|
"entropy": 10.691372871398926,
|
|
"epoch": 0.00438365772400491,
|
|
"grad_norm": 1.8046875,
|
|
"learning_rate": 3.7e-05,
|
|
"loss": 9.722,
|
|
"mean_token_accuracy": 0.04985377714037895,
|
|
"num_tokens": 157524.0,
|
|
"step": 75
|
|
},
|
|
{
|
|
"entropy": 10.681602478027344,
|
|
"epoch": 0.004675901572271904,
|
|
"grad_norm": 1.6875,
|
|
"learning_rate": 3.95e-05,
|
|
"loss": 9.6896,
|
|
"mean_token_accuracy": 0.05114552080631256,
|
|
"num_tokens": 168344.0,
|
|
"step": 80
|
|
},
|
|
{
|
|
"entropy": 10.67192096710205,
|
|
"epoch": 0.004968145420538897,
|
|
"grad_norm": 1.7109375,
|
|
"learning_rate": 4.2000000000000004e-05,
|
|
"loss": 9.6618,
|
|
"mean_token_accuracy": 0.04409446604549885,
|
|
"num_tokens": 178976.0,
|
|
"step": 85
|
|
},
|
|
{
|
|
"entropy": 10.664519691467286,
|
|
"epoch": 0.0052603892688058915,
|
|
"grad_norm": 1.8671875,
|
|
"learning_rate": 4.45e-05,
|
|
"loss": 9.5639,
|
|
"mean_token_accuracy": 0.05372482612729072,
|
|
"num_tokens": 190567.0,
|
|
"step": 90
|
|
},
|
|
{
|
|
"entropy": 10.647130489349365,
|
|
"epoch": 0.005552633117072886,
|
|
"grad_norm": 1.75,
|
|
"learning_rate": 4.7000000000000004e-05,
|
|
"loss": 9.4984,
|
|
"mean_token_accuracy": 0.05118483640253544,
|
|
"num_tokens": 201587.0,
|
|
"step": 95
|
|
},
|
|
{
|
|
"entropy": 10.62646837234497,
|
|
"epoch": 0.00584487696533988,
|
|
"grad_norm": 1.671875,
|
|
"learning_rate": 4.9500000000000004e-05,
|
|
"loss": 9.4193,
|
|
"mean_token_accuracy": 0.04868996068835259,
|
|
"num_tokens": 211913.0,
|
|
"step": 100
|
|
},
|
|
{
|
|
"entropy": 10.610049438476562,
|
|
"epoch": 0.006137120813606874,
|
|
"grad_norm": 1.8125,
|
|
"learning_rate": 5.2e-05,
|
|
"loss": 9.3409,
|
|
"mean_token_accuracy": 0.05402343980967998,
|
|
"num_tokens": 222247.0,
|
|
"step": 105
|
|
},
|
|
{
|
|
"entropy": 10.584276962280274,
|
|
"epoch": 0.006429364661873867,
|
|
"grad_norm": 1.625,
|
|
"learning_rate": 5.45e-05,
|
|
"loss": 9.2142,
|
|
"mean_token_accuracy": 0.06452971324324608,
|
|
"num_tokens": 232361.0,
|
|
"step": 110
|
|
},
|
|
{
|
|
"entropy": 10.533616924285889,
|
|
"epoch": 0.006721608510140861,
|
|
"grad_norm": 1.671875,
|
|
"learning_rate": 5.7e-05,
|
|
"loss": 9.1281,
|
|
"mean_token_accuracy": 0.0630628488957882,
|
|
"num_tokens": 242885.0,
|
|
"step": 115
|
|
},
|
|
{
|
|
"entropy": 10.462300109863282,
|
|
"epoch": 0.007013852358407856,
|
|
"grad_norm": 1.6015625,
|
|
"learning_rate": 5.9499999999999996e-05,
|
|
"loss": 9.0436,
|
|
"mean_token_accuracy": 0.060920076444745064,
|
|
"num_tokens": 254482.0,
|
|
"step": 120
|
|
},
|
|
{
|
|
"entropy": 10.408651351928711,
|
|
"epoch": 0.00730609620667485,
|
|
"grad_norm": 1.6171875,
|
|
"learning_rate": 6.2e-05,
|
|
"loss": 8.93,
|
|
"mean_token_accuracy": 0.059300025925040245,
|
|
"num_tokens": 265744.0,
|
|
"step": 125
|
|
},
|
|
{
|
|
"entropy": 10.367331790924073,
|
|
"epoch": 0.007598340054941843,
|
|
"grad_norm": 1.421875,
|
|
"learning_rate": 6.450000000000001e-05,
|
|
"loss": 8.8327,
|
|
"mean_token_accuracy": 0.058924198523163794,
|
|
"num_tokens": 276924.0,
|
|
"step": 130
|
|
},
|
|
{
|
|
"entropy": 10.272025775909423,
|
|
"epoch": 0.007890583903208837,
|
|
"grad_norm": 1.375,
|
|
"learning_rate": 6.7e-05,
|
|
"loss": 8.7304,
|
|
"mean_token_accuracy": 0.0622296292334795,
|
|
"num_tokens": 287779.0,
|
|
"step": 135
|
|
},
|
|
{
|
|
"entropy": 10.183625984191895,
|
|
"epoch": 0.008182827751475831,
|
|
"grad_norm": 1.3984375,
|
|
"learning_rate": 6.950000000000001e-05,
|
|
"loss": 8.6335,
|
|
"mean_token_accuracy": 0.06356315091252326,
|
|
"num_tokens": 298807.0,
|
|
"step": 140
|
|
},
|
|
{
|
|
"entropy": 10.097103500366211,
|
|
"epoch": 0.008475071599742826,
|
|
"grad_norm": 1.2109375,
|
|
"learning_rate": 7.2e-05,
|
|
"loss": 8.5907,
|
|
"mean_token_accuracy": 0.0559815414249897,
|
|
"num_tokens": 309760.0,
|
|
"step": 145
|
|
},
|
|
{
|
|
"entropy": 10.004411506652833,
|
|
"epoch": 0.00876731544800982,
|
|
"grad_norm": 1.1796875,
|
|
"learning_rate": 7.45e-05,
|
|
"loss": 8.397,
|
|
"mean_token_accuracy": 0.06126284077763557,
|
|
"num_tokens": 319750.0,
|
|
"step": 150
|
|
},
|
|
{
|
|
"entropy": 9.832055950164795,
|
|
"epoch": 0.009059559296276814,
|
|
"grad_norm": 1.1484375,
|
|
"learning_rate": 7.7e-05,
|
|
"loss": 8.2924,
|
|
"mean_token_accuracy": 0.0603407833725214,
|
|
"num_tokens": 329931.0,
|
|
"step": 155
|
|
},
|
|
{
|
|
"entropy": 9.66302547454834,
|
|
"epoch": 0.009351803144543808,
|
|
"grad_norm": 1.0078125,
|
|
"learning_rate": 7.950000000000001e-05,
|
|
"loss": 8.2478,
|
|
"mean_token_accuracy": 0.06180734224617481,
|
|
"num_tokens": 340457.0,
|
|
"step": 160
|
|
},
|
|
{
|
|
"entropy": 9.49033498764038,
|
|
"epoch": 0.009644046992810802,
|
|
"grad_norm": 0.9296875,
|
|
"learning_rate": 8.2e-05,
|
|
"loss": 8.1199,
|
|
"mean_token_accuracy": 0.06424234174191952,
|
|
"num_tokens": 352450.0,
|
|
"step": 165
|
|
},
|
|
{
|
|
"entropy": 9.311860942840577,
|
|
"epoch": 0.009936290841077795,
|
|
"grad_norm": 0.85546875,
|
|
"learning_rate": 8.450000000000001e-05,
|
|
"loss": 8.1159,
|
|
"mean_token_accuracy": 0.06317566409707069,
|
|
"num_tokens": 363204.0,
|
|
"step": 170
|
|
},
|
|
{
|
|
"entropy": 9.153958892822265,
|
|
"epoch": 0.010228534689344789,
|
|
"grad_norm": 0.80078125,
|
|
"learning_rate": 8.7e-05,
|
|
"loss": 8.0882,
|
|
"mean_token_accuracy": 0.062278729677200315,
|
|
"num_tokens": 373577.0,
|
|
"step": 175
|
|
},
|
|
{
|
|
"entropy": 8.929795932769775,
|
|
"epoch": 0.010520778537611783,
|
|
"grad_norm": 0.76171875,
|
|
"learning_rate": 8.95e-05,
|
|
"loss": 7.9663,
|
|
"mean_token_accuracy": 0.07208616435527801,
|
|
"num_tokens": 384267.0,
|
|
"step": 180
|
|
},
|
|
{
|
|
"entropy": 8.768066787719727,
|
|
"epoch": 0.010813022385878777,
|
|
"grad_norm": 0.75390625,
|
|
"learning_rate": 9.2e-05,
|
|
"loss": 7.9324,
|
|
"mean_token_accuracy": 0.06511378176510334,
|
|
"num_tokens": 394767.0,
|
|
"step": 185
|
|
},
|
|
{
|
|
"entropy": 8.666393280029297,
|
|
"epoch": 0.011105266234145771,
|
|
"grad_norm": 0.94140625,
|
|
"learning_rate": 9.45e-05,
|
|
"loss": 7.9172,
|
|
"mean_token_accuracy": 0.07024285569787025,
|
|
"num_tokens": 405807.0,
|
|
"step": 190
|
|
},
|
|
{
|
|
"entropy": 8.585619068145752,
|
|
"epoch": 0.011397510082412765,
|
|
"grad_norm": 0.68359375,
|
|
"learning_rate": 9.7e-05,
|
|
"loss": 7.869,
|
|
"mean_token_accuracy": 0.0643521461635828,
|
|
"num_tokens": 416381.0,
|
|
"step": 195
|
|
},
|
|
{
|
|
"entropy": 8.565132427215577,
|
|
"epoch": 0.01168975393067976,
|
|
"grad_norm": 0.75390625,
|
|
"learning_rate": 9.95e-05,
|
|
"loss": 7.9632,
|
|
"mean_token_accuracy": 0.06446416340768338,
|
|
"num_tokens": 427262.0,
|
|
"step": 200
|
|
},
|
|
{
|
|
"entropy": 8.418638515472413,
|
|
"epoch": 0.011981997778946754,
|
|
"grad_norm": 0.78515625,
|
|
"learning_rate": 0.000102,
|
|
"loss": 7.8962,
|
|
"mean_token_accuracy": 0.06822493113577366,
|
|
"num_tokens": 438504.0,
|
|
"step": 205
|
|
},
|
|
{
|
|
"entropy": 8.47780933380127,
|
|
"epoch": 0.012274241627213748,
|
|
"grad_norm": 0.765625,
|
|
"learning_rate": 0.00010449999999999999,
|
|
"loss": 7.8873,
|
|
"mean_token_accuracy": 0.06689661256968975,
|
|
"num_tokens": 449923.0,
|
|
"step": 210
|
|
},
|
|
{
|
|
"entropy": 8.430081558227538,
|
|
"epoch": 0.01256648547548074,
|
|
"grad_norm": 0.80078125,
|
|
"learning_rate": 0.000107,
|
|
"loss": 7.8463,
|
|
"mean_token_accuracy": 0.0716497365385294,
|
|
"num_tokens": 460736.0,
|
|
"step": 215
|
|
},
|
|
{
|
|
"entropy": 8.367777156829835,
|
|
"epoch": 0.012858729323747735,
|
|
"grad_norm": 0.80078125,
|
|
"learning_rate": 0.0001095,
|
|
"loss": 7.9365,
|
|
"mean_token_accuracy": 0.0605232123285532,
|
|
"num_tokens": 471513.0,
|
|
"step": 220
|
|
},
|
|
{
|
|
"entropy": 8.417449378967286,
|
|
"epoch": 0.013150973172014729,
|
|
"grad_norm": 0.73828125,
|
|
"learning_rate": 0.000112,
|
|
"loss": 7.7879,
|
|
"mean_token_accuracy": 0.07277968265116215,
|
|
"num_tokens": 482115.0,
|
|
"step": 225
|
|
},
|
|
{
|
|
"entropy": 8.359326553344726,
|
|
"epoch": 0.013443217020281723,
|
|
"grad_norm": 0.7890625,
|
|
"learning_rate": 0.0001145,
|
|
"loss": 7.8079,
|
|
"mean_token_accuracy": 0.07002997919917106,
|
|
"num_tokens": 493273.0,
|
|
"step": 230
|
|
},
|
|
{
|
|
"entropy": 8.321810531616212,
|
|
"epoch": 0.013735460868548717,
|
|
"grad_norm": 0.7578125,
|
|
"learning_rate": 0.00011700000000000001,
|
|
"loss": 7.8188,
|
|
"mean_token_accuracy": 0.07201721332967281,
|
|
"num_tokens": 503939.0,
|
|
"step": 235
|
|
},
|
|
{
|
|
"entropy": 8.311158466339112,
|
|
"epoch": 0.014027704716815711,
|
|
"grad_norm": 0.81640625,
|
|
"learning_rate": 0.00011949999999999999,
|
|
"loss": 7.7369,
|
|
"mean_token_accuracy": 0.06772977598011494,
|
|
"num_tokens": 515215.0,
|
|
"step": 240
|
|
},
|
|
{
|
|
"entropy": 8.281570434570312,
|
|
"epoch": 0.014319948565082705,
|
|
"grad_norm": 0.77734375,
|
|
"learning_rate": 0.000122,
|
|
"loss": 7.8223,
|
|
"mean_token_accuracy": 0.07254013754427432,
|
|
"num_tokens": 526485.0,
|
|
"step": 245
|
|
},
|
|
{
|
|
"entropy": 8.277181148529053,
|
|
"epoch": 0.0146121924133497,
|
|
"grad_norm": 0.77734375,
|
|
"learning_rate": 0.0001245,
|
|
"loss": 7.7672,
|
|
"mean_token_accuracy": 0.06958542391657829,
|
|
"num_tokens": 536735.0,
|
|
"step": 250
|
|
},
|
|
{
|
|
"entropy": 8.277847576141358,
|
|
"epoch": 0.014904436261616694,
|
|
"grad_norm": 0.82421875,
|
|
"learning_rate": 0.000127,
|
|
"loss": 7.8541,
|
|
"mean_token_accuracy": 0.06499786637723445,
|
|
"num_tokens": 547326.0,
|
|
"step": 255
|
|
},
|
|
{
|
|
"entropy": 8.180327033996582,
|
|
"epoch": 0.015196680109883686,
|
|
"grad_norm": 0.9375,
|
|
"learning_rate": 0.0001295,
|
|
"loss": 7.7502,
|
|
"mean_token_accuracy": 0.06869623251259327,
|
|
"num_tokens": 558275.0,
|
|
"step": 260
|
|
},
|
|
{
|
|
"entropy": 8.257558250427246,
|
|
"epoch": 0.01548892395815068,
|
|
"grad_norm": 0.74609375,
|
|
"learning_rate": 0.000132,
|
|
"loss": 7.7096,
|
|
"mean_token_accuracy": 0.07516702339053154,
|
|
"num_tokens": 569711.0,
|
|
"step": 265
|
|
},
|
|
{
|
|
"entropy": 8.150344276428223,
|
|
"epoch": 0.015781167806417674,
|
|
"grad_norm": 0.7421875,
|
|
"learning_rate": 0.00013450000000000002,
|
|
"loss": 7.6881,
|
|
"mean_token_accuracy": 0.07785917520523071,
|
|
"num_tokens": 579602.0,
|
|
"step": 270
|
|
},
|
|
{
|
|
"entropy": 8.24105463027954,
|
|
"epoch": 0.01607341165468467,
|
|
"grad_norm": 0.75,
|
|
"learning_rate": 0.00013700000000000002,
|
|
"loss": 7.7286,
|
|
"mean_token_accuracy": 0.07401030212640762,
|
|
"num_tokens": 590899.0,
|
|
"step": 275
|
|
},
|
|
{
|
|
"entropy": 8.173699569702148,
|
|
"epoch": 0.016365655502951663,
|
|
"grad_norm": 0.87890625,
|
|
"learning_rate": 0.0001395,
|
|
"loss": 7.726,
|
|
"mean_token_accuracy": 0.06962668001651764,
|
|
"num_tokens": 602080.0,
|
|
"step": 280
|
|
},
|
|
{
|
|
"entropy": 8.16687994003296,
|
|
"epoch": 0.016657899351218655,
|
|
"grad_norm": 0.765625,
|
|
"learning_rate": 0.00014199999999999998,
|
|
"loss": 7.7605,
|
|
"mean_token_accuracy": 0.07070777006447315,
|
|
"num_tokens": 613692.0,
|
|
"step": 285
|
|
},
|
|
{
|
|
"entropy": 8.145956993103027,
|
|
"epoch": 0.01695014319948565,
|
|
"grad_norm": 0.88671875,
|
|
"learning_rate": 0.0001445,
|
|
"loss": 7.7305,
|
|
"mean_token_accuracy": 0.07069781571626663,
|
|
"num_tokens": 625019.0,
|
|
"step": 290
|
|
},
|
|
{
|
|
"entropy": 8.123618793487548,
|
|
"epoch": 0.017242387047752643,
|
|
"grad_norm": 0.80078125,
|
|
"learning_rate": 0.000147,
|
|
"loss": 7.6456,
|
|
"mean_token_accuracy": 0.07288045659661294,
|
|
"num_tokens": 636453.0,
|
|
"step": 295
|
|
},
|
|
{
|
|
"entropy": 8.108728456497193,
|
|
"epoch": 0.01753463089601964,
|
|
"grad_norm": 0.89453125,
|
|
"learning_rate": 0.0001495,
|
|
"loss": 7.6918,
|
|
"mean_token_accuracy": 0.07240097671747207,
|
|
"num_tokens": 647272.0,
|
|
"step": 300
|
|
},
|
|
{
|
|
"entropy": 8.040448427200317,
|
|
"epoch": 0.017826874744286632,
|
|
"grad_norm": 1.0859375,
|
|
"learning_rate": 0.000152,
|
|
"loss": 7.701,
|
|
"mean_token_accuracy": 0.07080473527312278,
|
|
"num_tokens": 657724.0,
|
|
"step": 305
|
|
},
|
|
{
|
|
"entropy": 8.166200351715087,
|
|
"epoch": 0.018119118592553628,
|
|
"grad_norm": 0.78125,
|
|
"learning_rate": 0.00015450000000000001,
|
|
"loss": 7.7053,
|
|
"mean_token_accuracy": 0.07137923240661621,
|
|
"num_tokens": 669183.0,
|
|
"step": 310
|
|
},
|
|
{
|
|
"entropy": 8.084780406951904,
|
|
"epoch": 0.01841136244082062,
|
|
"grad_norm": 1.0703125,
|
|
"learning_rate": 0.000157,
|
|
"loss": 7.6679,
|
|
"mean_token_accuracy": 0.06901021562516689,
|
|
"num_tokens": 680439.0,
|
|
"step": 315
|
|
},
|
|
{
|
|
"entropy": 8.157064056396484,
|
|
"epoch": 0.018703606289087616,
|
|
"grad_norm": 0.9765625,
|
|
"learning_rate": 0.0001595,
|
|
"loss": 7.712,
|
|
"mean_token_accuracy": 0.0706634670495987,
|
|
"num_tokens": 690789.0,
|
|
"step": 320
|
|
},
|
|
{
|
|
"entropy": 8.046548128128052,
|
|
"epoch": 0.01899585013735461,
|
|
"grad_norm": 0.8671875,
|
|
"learning_rate": 0.000162,
|
|
"loss": 7.6831,
|
|
"mean_token_accuracy": 0.07187991701066494,
|
|
"num_tokens": 701818.0,
|
|
"step": 325
|
|
},
|
|
{
|
|
"entropy": 8.062455224990845,
|
|
"epoch": 0.019288093985621604,
|
|
"grad_norm": 0.71484375,
|
|
"learning_rate": 0.00016450000000000001,
|
|
"loss": 7.6316,
|
|
"mean_token_accuracy": 0.07437365353107453,
|
|
"num_tokens": 713344.0,
|
|
"step": 330
|
|
},
|
|
{
|
|
"entropy": 8.084509325027465,
|
|
"epoch": 0.019580337833888597,
|
|
"grad_norm": 1.0390625,
|
|
"learning_rate": 0.00016700000000000002,
|
|
"loss": 7.6384,
|
|
"mean_token_accuracy": 0.07300028279423713,
|
|
"num_tokens": 723279.0,
|
|
"step": 335
|
|
},
|
|
{
|
|
"entropy": 8.018547964096069,
|
|
"epoch": 0.01987258168215559,
|
|
"grad_norm": 0.73046875,
|
|
"learning_rate": 0.00016950000000000003,
|
|
"loss": 7.625,
|
|
"mean_token_accuracy": 0.07260205298662185,
|
|
"num_tokens": 733875.0,
|
|
"step": 340
|
|
},
|
|
{
|
|
"entropy": 8.072830390930175,
|
|
"epoch": 0.020164825530422585,
|
|
"grad_norm": 0.94140625,
|
|
"learning_rate": 0.00017199999999999998,
|
|
"loss": 7.6669,
|
|
"mean_token_accuracy": 0.0713542215526104,
|
|
"num_tokens": 745449.0,
|
|
"step": 345
|
|
},
|
|
{
|
|
"entropy": 8.05644793510437,
|
|
"epoch": 0.020457069378689578,
|
|
"grad_norm": 0.7734375,
|
|
"learning_rate": 0.00017449999999999999,
|
|
"loss": 7.6741,
|
|
"mean_token_accuracy": 0.0740118071436882,
|
|
"num_tokens": 757570.0,
|
|
"step": 350
|
|
},
|
|
{
|
|
"entropy": 8.017089939117431,
|
|
"epoch": 0.020749313226956573,
|
|
"grad_norm": 0.8671875,
|
|
"learning_rate": 0.000177,
|
|
"loss": 7.5737,
|
|
"mean_token_accuracy": 0.08025477081537247,
|
|
"num_tokens": 767601.0,
|
|
"step": 355
|
|
},
|
|
{
|
|
"entropy": 8.00361680984497,
|
|
"epoch": 0.021041557075223566,
|
|
"grad_norm": 0.95703125,
|
|
"learning_rate": 0.0001795,
|
|
"loss": 7.6323,
|
|
"mean_token_accuracy": 0.0752041395753622,
|
|
"num_tokens": 777852.0,
|
|
"step": 360
|
|
},
|
|
{
|
|
"entropy": 8.018228721618652,
|
|
"epoch": 0.021333800923490562,
|
|
"grad_norm": 0.890625,
|
|
"learning_rate": 0.000182,
|
|
"loss": 7.5859,
|
|
"mean_token_accuracy": 0.0751498755067587,
|
|
"num_tokens": 788339.0,
|
|
"step": 365
|
|
},
|
|
{
|
|
"entropy": 7.970228672027588,
|
|
"epoch": 0.021626044771757554,
|
|
"grad_norm": 0.984375,
|
|
"learning_rate": 0.0001845,
|
|
"loss": 7.6588,
|
|
"mean_token_accuracy": 0.07353257052600384,
|
|
"num_tokens": 799791.0,
|
|
"step": 370
|
|
},
|
|
{
|
|
"entropy": 8.015653133392334,
|
|
"epoch": 0.02191828862002455,
|
|
"grad_norm": 0.96875,
|
|
"learning_rate": 0.000187,
|
|
"loss": 7.5746,
|
|
"mean_token_accuracy": 0.07374845892190933,
|
|
"num_tokens": 810141.0,
|
|
"step": 375
|
|
},
|
|
{
|
|
"entropy": 8.040093088150025,
|
|
"epoch": 0.022210532468291543,
|
|
"grad_norm": 0.8046875,
|
|
"learning_rate": 0.0001895,
|
|
"loss": 7.6182,
|
|
"mean_token_accuracy": 0.07429262697696686,
|
|
"num_tokens": 820929.0,
|
|
"step": 380
|
|
},
|
|
{
|
|
"entropy": 7.9526266098022464,
|
|
"epoch": 0.022502776316558535,
|
|
"grad_norm": 1.265625,
|
|
"learning_rate": 0.000192,
|
|
"loss": 7.5519,
|
|
"mean_token_accuracy": 0.07891798727214336,
|
|
"num_tokens": 831340.0,
|
|
"step": 385
|
|
},
|
|
{
|
|
"entropy": 7.935062026977539,
|
|
"epoch": 0.02279502016482553,
|
|
"grad_norm": 0.8984375,
|
|
"learning_rate": 0.0001945,
|
|
"loss": 7.5636,
|
|
"mean_token_accuracy": 0.0735638827085495,
|
|
"num_tokens": 841846.0,
|
|
"step": 390
|
|
},
|
|
{
|
|
"entropy": 8.029356002807617,
|
|
"epoch": 0.023087264013092523,
|
|
"grad_norm": 0.90234375,
|
|
"learning_rate": 0.00019700000000000002,
|
|
"loss": 7.5434,
|
|
"mean_token_accuracy": 0.07343484573066235,
|
|
"num_tokens": 852089.0,
|
|
"step": 395
|
|
},
|
|
{
|
|
"entropy": 7.838059997558593,
|
|
"epoch": 0.02337950786135952,
|
|
"grad_norm": 0.97265625,
|
|
"learning_rate": 0.00019950000000000002,
|
|
"loss": 7.5241,
|
|
"mean_token_accuracy": 0.07688843682408333,
|
|
"num_tokens": 862568.0,
|
|
"step": 400
|
|
},
|
|
{
|
|
"entropy": 8.034818124771117,
|
|
"epoch": 0.02367175170962651,
|
|
"grad_norm": 0.8046875,
|
|
"learning_rate": 0.000202,
|
|
"loss": 7.5905,
|
|
"mean_token_accuracy": 0.07528307065367698,
|
|
"num_tokens": 874676.0,
|
|
"step": 405
|
|
},
|
|
{
|
|
"entropy": 7.867450904846192,
|
|
"epoch": 0.023963995557893508,
|
|
"grad_norm": 0.8515625,
|
|
"learning_rate": 0.00020449999999999998,
|
|
"loss": 7.5115,
|
|
"mean_token_accuracy": 0.08100188933312893,
|
|
"num_tokens": 886338.0,
|
|
"step": 410
|
|
},
|
|
{
|
|
"entropy": 7.946144676208496,
|
|
"epoch": 0.0242562394061605,
|
|
"grad_norm": 0.8671875,
|
|
"learning_rate": 0.000207,
|
|
"loss": 7.5287,
|
|
"mean_token_accuracy": 0.07384060397744178,
|
|
"num_tokens": 896671.0,
|
|
"step": 415
|
|
},
|
|
{
|
|
"entropy": 7.923208522796631,
|
|
"epoch": 0.024548483254427496,
|
|
"grad_norm": 0.90625,
|
|
"learning_rate": 0.0002095,
|
|
"loss": 7.5025,
|
|
"mean_token_accuracy": 0.07231347896158695,
|
|
"num_tokens": 908265.0,
|
|
"step": 420
|
|
},
|
|
{
|
|
"entropy": 7.851961994171143,
|
|
"epoch": 0.02484072710269449,
|
|
"grad_norm": 0.8828125,
|
|
"learning_rate": 0.000212,
|
|
"loss": 7.4692,
|
|
"mean_token_accuracy": 0.07885380312800408,
|
|
"num_tokens": 919024.0,
|
|
"step": 425
|
|
},
|
|
{
|
|
"entropy": 7.894201850891113,
|
|
"epoch": 0.02513297095096148,
|
|
"grad_norm": 0.9296875,
|
|
"learning_rate": 0.0002145,
|
|
"loss": 7.4639,
|
|
"mean_token_accuracy": 0.08095669150352477,
|
|
"num_tokens": 930837.0,
|
|
"step": 430
|
|
},
|
|
{
|
|
"entropy": 7.845162343978882,
|
|
"epoch": 0.025425214799228477,
|
|
"grad_norm": 1.0390625,
|
|
"learning_rate": 0.00021700000000000002,
|
|
"loss": 7.4312,
|
|
"mean_token_accuracy": 0.076503673940897,
|
|
"num_tokens": 941374.0,
|
|
"step": 435
|
|
},
|
|
{
|
|
"entropy": 7.838292598724365,
|
|
"epoch": 0.02571745864749547,
|
|
"grad_norm": 0.859375,
|
|
"learning_rate": 0.0002195,
|
|
"loss": 7.3796,
|
|
"mean_token_accuracy": 0.0836145281791687,
|
|
"num_tokens": 953273.0,
|
|
"step": 440
|
|
},
|
|
{
|
|
"entropy": 7.836203670501709,
|
|
"epoch": 0.026009702495762465,
|
|
"grad_norm": 1.6015625,
|
|
"learning_rate": 0.000222,
|
|
"loss": 7.3659,
|
|
"mean_token_accuracy": 0.09230264946818352,
|
|
"num_tokens": 964882.0,
|
|
"step": 445
|
|
},
|
|
{
|
|
"entropy": 7.860473203659057,
|
|
"epoch": 0.026301946344029457,
|
|
"grad_norm": 1.203125,
|
|
"learning_rate": 0.0002245,
|
|
"loss": 7.5208,
|
|
"mean_token_accuracy": 0.08153782263398171,
|
|
"num_tokens": 974732.0,
|
|
"step": 450
|
|
},
|
|
{
|
|
"entropy": 7.774071216583252,
|
|
"epoch": 0.026594190192296453,
|
|
"grad_norm": 0.94140625,
|
|
"learning_rate": 0.00022700000000000002,
|
|
"loss": 7.4513,
|
|
"mean_token_accuracy": 0.07568450495600701,
|
|
"num_tokens": 985709.0,
|
|
"step": 455
|
|
},
|
|
{
|
|
"entropy": 7.999471426010132,
|
|
"epoch": 0.026886434040563446,
|
|
"grad_norm": 0.8046875,
|
|
"learning_rate": 0.00022950000000000002,
|
|
"loss": 7.4197,
|
|
"mean_token_accuracy": 0.07974797300994396,
|
|
"num_tokens": 998177.0,
|
|
"step": 460
|
|
},
|
|
{
|
|
"entropy": 7.711365842819214,
|
|
"epoch": 0.02717867788883044,
|
|
"grad_norm": 0.8671875,
|
|
"learning_rate": 0.00023200000000000003,
|
|
"loss": 7.3536,
|
|
"mean_token_accuracy": 0.08135458827018738,
|
|
"num_tokens": 1008676.0,
|
|
"step": 465
|
|
},
|
|
{
|
|
"entropy": 7.800386238098144,
|
|
"epoch": 0.027470921737097434,
|
|
"grad_norm": 0.875,
|
|
"learning_rate": 0.00023449999999999998,
|
|
"loss": 7.4651,
|
|
"mean_token_accuracy": 0.07976322658360005,
|
|
"num_tokens": 1019951.0,
|
|
"step": 470
|
|
},
|
|
{
|
|
"entropy": 7.785122346878052,
|
|
"epoch": 0.027763165585364426,
|
|
"grad_norm": 0.9453125,
|
|
"learning_rate": 0.000237,
|
|
"loss": 7.3981,
|
|
"mean_token_accuracy": 0.0761278249323368,
|
|
"num_tokens": 1030219.0,
|
|
"step": 475
|
|
},
|
|
{
|
|
"entropy": 7.781033182144165,
|
|
"epoch": 0.028055409433631422,
|
|
"grad_norm": 0.9296875,
|
|
"learning_rate": 0.0002395,
|
|
"loss": 7.4141,
|
|
"mean_token_accuracy": 0.07795999571681023,
|
|
"num_tokens": 1041911.0,
|
|
"step": 480
|
|
},
|
|
{
|
|
"entropy": 7.891617631912231,
|
|
"epoch": 0.028347653281898415,
|
|
"grad_norm": 0.94140625,
|
|
"learning_rate": 0.000242,
|
|
"loss": 7.4312,
|
|
"mean_token_accuracy": 0.08173990175127983,
|
|
"num_tokens": 1052638.0,
|
|
"step": 485
|
|
},
|
|
{
|
|
"entropy": 7.774001789093018,
|
|
"epoch": 0.02863989713016541,
|
|
"grad_norm": 0.94140625,
|
|
"learning_rate": 0.0002445,
|
|
"loss": 7.3755,
|
|
"mean_token_accuracy": 0.08225451484322548,
|
|
"num_tokens": 1063465.0,
|
|
"step": 490
|
|
},
|
|
{
|
|
"entropy": 7.861321496963501,
|
|
"epoch": 0.028932140978432403,
|
|
"grad_norm": 1.1796875,
|
|
"learning_rate": 0.000247,
|
|
"loss": 7.3888,
|
|
"mean_token_accuracy": 0.08350497037172318,
|
|
"num_tokens": 1073583.0,
|
|
"step": 495
|
|
},
|
|
{
|
|
"entropy": 7.868850564956665,
|
|
"epoch": 0.0292243848266994,
|
|
"grad_norm": 1.3515625,
|
|
"learning_rate": 0.0002495,
|
|
"loss": 7.4377,
|
|
"mean_token_accuracy": 0.08040728494524955,
|
|
"num_tokens": 1083961.0,
|
|
"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": 237556316897280.0,
|
|
"train_batch_size": 16,
|
|
"trial_name": null,
|
|
"trial_params": null
|
|
}
|